Promoting Green Urban Development in Africa: Enhancing the relationship between urbanization, environmental assets and ecosystem services RETURN ON INVESTMENT IN GREEN URBAN DEVELOPMENT: AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT, DAR ES SALAAM, TANZANIA Promoting Green Urban Development in Africa: Enhancing the relationship between urbanization, environmental assets and ecosystem services RETURN ON INVESTMENT IN GREEN URBAN DEVELOPMENT: AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT, DAR ES SALAAM, TANZANIA Authors Jane Turpie, Timm Kroeger, Raffaele De Risi, Francesco de Paola, Gwyneth Letley, Katherine Forsythe & Liz Day Prepared for AECOM on behalf of The World Bank by Anchor Environmental with support from The Nature Conservancy Prepared by Anchor Environmental Consultants 8 Steenberg House, Silverwood Close, Tokai 7945 www.anchorenvironmental.co.za 2016 COPYRIGHT © 2016 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. December 2016 Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to the Publishing and Knowledge Division, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org.                                                  Page ii AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT PREFACE AND ACKNOWLEDGMENTS This study forms one of the case studies of a larger study on Green Urban Development commissioned by the World Bank and co-funded by The Nature Conservancy. Anchor Environmental Consultants (Anchor) was subcontracted by AECOM to undertake case studies in three cities: Kampala, Uganda; Dar es Salaam, Tanzania; and Durban, South Africa. Each city was consulted as to the focus of the case study. In the case of Dar es Salaam, the city requested a study to evaluate the potential costs and benefits of rehabilitating the Msimbazi River and catchment to address the flooding problems associated with this river system. This study builds on the preparation of an Environmental Profile for Dar es Salaam by AECOM, as well as on earlier work on flooding in the city led by one of our team members, Raffaele de Risi. The study was led by Drs Jane Turpie of Anchor Environmental Consultants and Timm Kroeger of The Nature Conservancy. Gwyneth Letley and Katherine Forsythe of Anchor and Dr Liz Day of Freshwater Consulting Group undertook the ecological and green urban design aspects of the study and associated costings. Dr Raffaele de Risi of Bristol University and Dr Francesco de Paola of Naples University undertook the hydrological and hydraulic modelling and flood damage estimates. We are grateful to the Dar es Salaam municipal staff for their interest and support of this project, in particular to local World Bank consultant Amy Faust for her valuable inputs and assistance with data collation and to Mary Bitekerezo (social development specialist) for her advice on resettlement. Nancy Lorenzo Garcia of the World Bank kindly provided detailed land cover data. Thanks to Roland White and Chyi-Yun Huang of the World Bank and Diane Dale, Brian Goldberg and John Bachmann of AECOM for inputs and discussions during the project planning phase, as well as to Elizabeth Tellman (Arizona State University) and Daniel Auerbach (U.S. EPA) for their technical guidance on hydrological aspects.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page iii This page intentionally blank.                                                  Page iv AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT EXECUTIVE SUMMARY Introduction The overall approach was to model current flooding and Rapid urbanisation is taking place at an unprecedented expected annual losses (EAL) in the Msimbazi catchment rate throughout the world, with the rate of growth and to determine the potential change in these after often outpacing urban planning and the capacity of city implementation of a range of stormwater management managers. As a result, existing natural areas within cities, scenarios involving different combinations of feasible which provide a range of benefits to urban dwellers measures. The scenarios were then compared in terms are becoming smaller and degraded, and problems of their net present value (NPV), internal rate of return such as flooding, air pollution and water pollution are (IRR) and return on investment (ROI). becoming worse in many places. African cities often lack the resources to deal with these problems. However, a Study area number of studies have suggested that investing in the maintenance or restoration of natural infrastructure Dar es Salaam, the most populous city in Tanzania, has in many cases may not only address given problems at undergone rapid population growth and currently has comparable or lower cost than conventional engineering a population of more than 4.36 million. Much of this projects, but also generate multiple additional benefits growth has taken the form of unplanned residential that ultimately translate into cost savings and increased areas which now account for 75% of the urban area, human wellbeing. and many houses have been built in areas previously considered unsuitable, such as on floodplains and river Meanwhile, great strides have been made in the banks. The infrastructure of the city, which is governed design of sustainable mechanisms to deal with urban by five municipalities (Kinondoni, Ilala, Ubungo, environmental issues, stormwater flows and the Kigamboni and Temeke), has not been able to keep attendant pollution problems, and management up with this growth. Most residents still lack access to and planning of cities is increasingly taking a holistic public services including sanitation and waste collection. approach that includes the use and conservation of During the rainy season, intense rainfall events often semi-natural and natural areas within cities as part cause flooding in certain areas of the city. Of the four of a green urban development strategy. One of the main river systems, the problems are greatest in the challenges of green urban development will be to find Msimbazi river catchment, which floods parts of the city the right balance between ecological infrastructure centre. This study therefore focuses on the Msimbazi (natural systems), “green” (= environmentally friendly) river system. built infrastructure, and conventional (“grey”) built infrastructure. The Msimbazi catchment covers approximately 300 km2 and extends across the Kinondoni and Ilala municipalities Dar es Salaam, located on Africa’s Indian Ocean Coast, and beyond the western boundary of the city. Once faces a multitude of environmental problems. Prominent an important water resource, it is now highly polluted among them is the problem of flooding in and around with both solid waste and effluents. While the lower the city centre, which frequently brings the city to a catchment is highly urbanised, further upstream land standstill, as well as causing infrastructural damage. cover becomes increasingly agricultural, and the source Many factors have contributed to this problem, including of the Msimbazi falls within the Pugu Forest Reserve. unplanned informal settlements in the upper catchment The main river has two tributaries, the Ubungo, which and floodplain areas, a lack of drainage and a lack of flows through a largely cultivated landscape with some solid waste management. The impacts of flooding are woodland/bushland and urban areas, and the Sinza, also exacerbated by high levels of pollution in the rivers, which flows through a mainly urbanised catchment and which increases the risks associated with flooding. joins the Msimbazi closer to its estuary. In consultations for this study, stakeholders in Dar es Salaam identified the Msimbazi River as being among The catchment is densely populated, with densities the most degraded ecosystems in the city and also the increasing towards the city centre. At the source, source of the most serious flooding problems. large areas of the Pugu Forest have been deforested due to charcoal production and agriculture. In the The aim of the study was to explore the potential costs upper catchment, agricultural areas, dump sites and and benefits of undertaking a green urban development quarries border on the river and have degraded the approach, including catchment-to-coast restoration riparian vegetation. Further down the catchment, some measures, to ameliorate flood risk in the Msimbazi River floodplain areas are occupied by dense unplanned catchment. settlements, and others are heavily used for cultivation. The mid to lower reaches enter more densely populated areas which also include an abattoir and other industries                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page v along the river banks. In the lower catchment, where to generate the maps of maximum water height and the Ubungo and Sinza Rivers join the Msimbazi, the velocity for each node of a lattice covering the zone of surrounds are heavily populated and while there is interest for a given return period (the flood hazard map). little residential development in the floodplains, these areas are used for agriculture and appear to have been Historical rainfall data was obtained from the single disconnected from the river channel to some extent meteorological station in the area, located at Dar es by berms. In the Msimbazi estuary, there still exists a Salaam International Airport at 55 m above sea level. large remnant mangrove stand of approximately 0.5 ha. We chose five rainfall durations (1, 3, 6, 12, and 24 The Msimbazi river system is highly contaminated and hours), the typical values adopted. Hydrologic basin pollutant levels exceed many standards for drinking, modelling was then carried out in order to produce irrigation and contact with skin. Pollution levels at the hydrographs for each of the three sub-catchment areas river mouth in some cases are over 1000 times the of the Msimbazi river system at the points where they levels considered safe for human contact. The largest enter the main built environment. The geographic contributors to water pollution are inadequate on-site characteristics of the catchments were used to estimate sanitation systems and industrial areas without sewers. the concentration times of 9.79 hours, 7.80 hours, and 4.16 hours for the main Msimbazi and the Ubungo and Sinza contributory catchments, respectively. Although it Modelling flood risk in the Msimbazi catchment is preferable to use a more comprehensive, distributed or semi-distributed model to estimate the design Flood risk assessment comprises three phases: hazard, hydrograph, the lack of data with which to calibrate such exposure and vulnerability assessment. The hazard is models dictated the choice of a relatively simple tool – generally assessed through physically-based hydraulic the classic Curve Number Method. The Curve Number models, providing the flood depth and the velocity is a function of the major runoff producing watershed for each point within the study area, also accounting characteristics, and is fairly well documented for its for the presence of buildings, infrastructure and soil inputs (soil, land use/treatment, surface condition, and characteristics. Established methods are computationally antecedent soil moisture condition (AMC). Our analysis demanding, and require significant amounts of data. For assumed conservatively (based on the historic record) developing country contexts, simplifications in modeling that the AMC class at the beginning of the modelled hypotheses and assumptions are generally adopted. In extreme rainfall events was AMC III. this study, flood risk is modelled using a physically-based method. Exposure assessments require identification In the next step, the flood discharge estimated using the of the elements at risk, including all the elements of hydrograph was propagated through the zone of interest human, built and natural environments at risk in the in order to delineate the flood prone areas for various flooding area. Finally, given the characterization of the return periods. Flood routing in two dimensions was built environment, vulnerability analysis can be carried accomplished by means of the commercial software FLO- out in order to quantify the adverse effects of flooding. 2D, a flood volume conservation model based on general A vulnerability analysis provides infrastructure fragility constitutive fluid equations of continuity and flood functions, representing the probability of reaching dynamics, i.e. shallow water equations or Saint-Venant or exceeding predefined damage states, for a given equations. The flow is considered variable in space level of flood intensity. The combination of hazard and in time, and the bottom friction is evaluated using and vulnerability returns the mean annual rate of Manning’s formula. The Manning’s coefficients were exceedance of a specific limit state. This rate can then assigned to computational cells based on a literature be used to calculate the probability of exceedance in a review. Conservative estimates of 0.04 and 0.02 were given time window, by adopting a reasonable probability assumed for natural and for urban areas, respectively. distribution describing the event occurrence. This The drainage systems not already incorporated in the probability can then be combined with the exposed 2 m DEM (e.g. the sewage system) were omitted due value in order to quantify the flood risk in the predefined the lack of available data on these systems. In order time window in terms of economic losses or in terms to optimize computational time, the analysis domain of number of casualties. The final result is generally was divided in five sub-domains (ES Figure 1). We then expressed as the expected annual loss (EAL). estimated the mean annual frequency of exceeding a given flood height at any point within each domain In the flood hazard assessment, rainfall intensity- (flood hazard curves). duration-frequency (IDF) curves, geologic and land-use information were used to characterize the hydrograph, leading to the calculation of the discharge (Q) and the total water volume (i.e. the area under the hydrograph) for different return periods. This information, together with the topographic map of the zone of interest was used in a two-dimensional diffusion model in order                                                  Page vi AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT ES Figure 1 Flood domains used in the hydraulic analysis Detailed spatial building information was obtained Evaluation and selection of potential urban stormwater from OpenStreetMap.com, and intersected with GIS management options data on Urban Morphology Type to identify the type Urban drainage management has changed significantly of each building at risk. The buildings at risk in the over the last few decades, from a conventional analysis domain were identified by intersecting the ‘rapid disposal’ approach to a more integrated and map of all buildings with the maximum extent of the sustainable ‘design with nature’ approach. There has baseline flood inundation. A total of 12,744 buildings been a proliferation of related approaches going under fell within this area. Sixty-two potential combinations of terms such as Integrated Urban Water Management available characteristics were recognized. From these, (IUWM), Water Sensitive Urban Design (WSUD), urban three main structural types were identified: informal stormwater Best Management Practices (BMPs), masonry (89.5%), formal masonry (8.8%), and reinforced Sustainable Urban Drainage Systems (SUDS) and Low concrete frame (1.7%). Next, the vulnerability of each Impact Development (LID). These describe a number type of structure was described using published fragility of measures to address flooding and/or water quality functions, which evaluate the probability of reaching problems. These tend to be categorised into passive and or exceeding specific damage states for a given hazard active structural and non-structural measures, and the intensity. These functions were derived in prior studies active measures, which seek to reduce the effects of in Dar es Salaam and elsewhere. urbanisation on the quantity and quality of catchment runoff, can be further categorised into source, local and After estimating the flood hazard curves for each of regional controls, as summarised in ES Figure 2. the buildings at risk, we summarised the flood risk assessment as the mean annual rate of exceedance of a given limit state (critical water height) for a structure beyond which it no longer fulfills a specified functionality. We then estimated the expected annual losses from flooding in the form of damages to buildings, based on the estimated degree of damages and replacement value of the buildings.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page vii ES Figure 2 Different types of measures used in stormwater management Source: This Study While conveyance measures tend to be highly effective both the range of options that could be considered to for reducing flood exposure/risk, they achieve little offset flooding problems and their potential efficacy. water quality improvement, vary in terms of cost- Significant manipulation of flow regime in the upstream effectiveness and generally produce relatively small catchment and/or better conveyance from the flood co-benefits. Indeed, they are more likely to lead to prone areas is required to improve water quality and externalities such as damage to aquatic ecosystems. address flooding in Dar es Salaam. “Green” measures (that seek to ameliorate the impacts of urban development on quantity and quality of flows) Conventional flood conveyance methods are not only also vary in their cost-effectiveness and may have to expensive but would be difficult to establish in this be applied in combination and/or at scale for effective catchment because of the size of the floods that need flood protection, but are important for water quality. to be contained. Very few of the active structural They also present much greater opportunities for options were considered feasible. In some cases this delivering co-benefits, such as water supply (in the case was because of the low location in the catchment of the of rainwater harvesting) and the provision of sports building structures they would be associated with, or and recreational opportunities. The latter is particularly because of the unsuitable soils. Rainwater harvesting the case for the vegetated options which have greater would have limited flood benefits in Dar es Salaam, as aesthetic appeal. Green measures include both the tanks would fill up early in the rainy season. A much engineering solutions and the protection or restoration better option is to have large-volume storage systems of natural systems in riparian and catchment areas. able to absorb high-rainfall events and slowly release Within flood prone areas, conservation of natural green the stored water. Thus the potential for detention infrastructure such as riparian buffers and functional basins was explored. Swales were also considered for floodplain areas can potentially enhance the value of implementation in the lower catchment, high residential, development setbacks and conveyance measures. Non- flood prone areas where they could convey rainfall and structural measures can also be considered “green”. runoff out of these areas as quickly as possible. This study sought to find a suitable set of “green” Among the most feasible options identified were the measures that could be implemented in combination to protection, restoration and/or enhancement of natural address flooding problems in Dar es Salaam, while also systems. There are substantial areas of degraded contributing to a green urban development path for the forest in the catchment that could be restored, and city. Each intervention was assessed on their limitations floodplains lower in the catchment have been artificially and requirements as well as their suitability for disconnected from the river, greatly reducing their application in the Msimbazi catchment in Dar es Salaam. potential for flood mitigation and co-benefits. Restoring The rapid expansion of Dar es Salaam city and the lack the natural hydrological connectivity of the river system of control over settlement patterns, especially in the will provide numerous ecological benefits and the floodplain areas of the lower catchment, has resulted deepening of the floodplain in the lower catchment in increased stormwater runoff, reduced water quality provides an opportunity to develop a wetland park coming from the catchment, and reduced capacity of which would provide inner city recreational green the floodplain to accommodate and convey flows. The open space area. Furthermore, there are a number nature and pattern of development in the catchment of floodplain areas in the mid-lower catchment areas and flood receiving areas therefore severely constrains that could be enhanced to improve their water holding                                                  Page viii AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT capacity at the same time as providing other benefits such 60 m River Reserve areas were estimated to be in the as erosion control and provision of areas for agriculture order of $44 million. Whilst the River Reserve areas are and wetlands. The idea of a mixed use enhanced riparian considered protected areas in which no development is and floodplain area was developed based on the concept allowed, these areas have not been clearly demarcated of a combination of riparian zone rehabilitation and or enforced by government. Therefore, unless floodplain enhancement measures that store and retard government acts to enforce this law, people who have flows but which could easily include opportunities for settled informally in these areas might need to be beneficial uses, including sports fields, agricultural resettled in order to execute certain GUD interventions. lots and parks as well as active riparian buffer zones/ However, it is important to note that compensation conservation corridors. Whilst these beneficial uses of payments are likely to be counterproductive, as they the floodplain could potentially raise initial costs, it is could encourage rent-seeking behaviour in this and expected that they are likely to reduce opportunities for other such reserve areas in the future. The costs unplanned resettlement of the floodplain. In addition, a associated with this scenario are therefore much higher community-based river cleaning programme was included as a result of the River Reserve areas not having been as an essential measure to help deal with the problem of maintained and protected. solid waste in the river system that leads to the clogging of drainage infrastructure such as culverts and channels. This could be considered as an interim measure until Scenario analysis proper municipal waste collection and management Five combinations of stormwater management measures services are implemented. were included in the analysis: The extent and location of each physical intervention 1. Riparian setbacks in the flood prone area; was estimated using Google Earth and GIS land cover maps in combination with the criteria and limitations 2. Green urban development measures (GUD); described for the interventions to identify the most suitable areas within the catchment for implementing 3. GUD measures + riparian setbacks in the flood prone each specific stormwater management measure. The area; costing of the selected interventions was based on a wide range of information sources collated from 4. GUD measures + additional detention basin(s); and literature and various green urban development projects offered in other parts of the world. 5. GUD measures + detention basin(s) + riparian setbacks in the flood prone area. The extent and cost of each proposed GUD intervention is shown in ES Table 1. The total initial investment The scenarios are a combination of interventions that cost of the GUD interventions was estimated to be either reduce exposure to flooding, reduce flood risk, approximately $40 million with annual maintenance or a combination of both (ES Table 2). By removing costs in the order of $1.6 million. Just more than 40% of people from flood prone areas within riparian setback the total investment cost is for the mixed-use enhanced buffers the number of people and structures exposed riparian and floodplain areas, which cover almost 500 to flooding is reduced. By implementing GUD and ha and detain 5 million m3 of runoff. In addition, costs additional storage interventions the flood hydrograph is associated with the resettlement of households from the lowered and flood risk is reduced. ES Table 1 Estimated extent and cost of the proposed GUD interventions Intervention Extent Initial / Annual maintenance cost (ha) construction cost (US$) (US$) Swales to improve drainage in flood prone areas 10 1 800 000 108 000 Catchment reforestation in Pugu Forest Reserve 776 845 000 17 000 Mixed use enhanced riparian and floodplain areas (~1m deep) 488 28 000 000 1 036 000 Rehabilitated floodplain and wetland park (~2m deep) 15 3 130 000 94 000 Enhanced floodplain-recessed gardens (~1m deep) 51 5 360 000 107 000 Community-based river cleaning project - 1 000 000 250 000 Total without resettlement costs 1340 40 135 000 1 612 000 Relocation with compensation 44 000 000 Total with maximum resettlement costs 84 135 000 1 612 000 Source: adapted from TEEB 2010                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page ix ES Table 2 Scenarios 1-5 and their estimated costs Reduce exposure  People and structures No interventions removed from 60m buffer in in flood prone areas flood prone areas Reduce flood risk Scenario 1 ↓ No interventions in catchment $62.6 million Scenario 2 Scenario 3 GUD interventions in catchment1 $84 million $138.5 million2 Scenario 4 Scenario 5 GUD with additional storage $124 million $178.5 million 1 GUD: (a) restoration of forests in upper catchment, (b) rehabilitated and enhanced riparian and floodplain areas in middle catchment, (d) river cleaning in middle catchment, (c) floodplain rehabilitation in lower catchment, (e) swales in flood prone areas. 2 This is less than the sum of 1 and 2 since the number of buildings at risk in the buffer is reduced, and so a reduced number of households need to be resettled. ES Table 3 Impacts of Scenarios 1 to 5 on expected annual losses (EAL), and the percentage change in EAL. Reduce exposure  People and structures No interventions removed from 60m buffer in in flood prone areas flood prone areas Reduce flood risk Baseline Scenario 1 ↓ No interventions in catchment US$47.30 million US$37.24 million (-21%) Scenario 2 Scenario 3 GUD interventions in catchment US$28.87 million (-39%) US$23.16 million (-51%) Scenario 4 Scenario 5 GUD with additional storage US$27.78 million (-41%) US$21.64 million (-54%) Modelling the effect of the riparian setback involved of the lower Msimbazi catchment, resulting in average removal of buildings from within 60 m of the rivers, annual cost savings ranging from $10 million to $26 therefore changing the number of buildings exposed million, or from 21% to 54% of present EAL (ES Table 3). to flooding. The effect of catchment restoration, which improves infiltration capacity of soils, was modelled by Costs generally increased from Scenario 1 to 5 (ES changing the antecedent soil moisture condition from Figure 3). Nevertheless, all the options considered had the baseline, wettest AMC III to the moderately moist positive outcomes, with the time taken for the return on AMC II, resulting in a change in the input hydrograph. For investment to exceed 1 ranging from 7 to 19 years. the floodplain storage area, we used a simple approach of removing from the stream flow the discharge Net present value was highest for Scenarios 2 and 3. accumulated in the floodplain storage. However, return on investment (ROI) was highest for Scenarios 1 and 2. A similar pattern is observed for IRR, The different scenarios were evaluated in terms of their which appears to exceed hurdle rates in most cases. return on investment (ROI), that is, the ratio of benefits and costs. Benefits were taken as the difference in The results suggest that the investment should initially net present value of EAL under each scenario versus be targeted at implementation of GUD measures in the baseline. Costs were calculated as the net present the catchment areas, and that if sufficient funds are value of the life-cycle costs of implementation of the available, these should be used to extend the investment mitigation measures, based on a review of the literature. to include resettlement from a setback zone as well A discount rate of 6% was used. (i.e. scenario 3). Factors such as the availability of financial resources, the desired time for ROI to surpass 1 (breakeven time), the impact on the environment (life Results cycle analysis of the adopted mitigation strategies) and society, should also be taken into account at a definitive All of the scenarios resulted in a significant impact on design stage. EAL associated with flooding in the flood prone areas                                                  Page x AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT ES Figure 3 Graphical representation of scenario results                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page xi Summary and conclusions Project. These include the lining of 8.5 km of the In this study we investigated the potential feasibility main drainage channel of the Sinza River and 5.4 km of investing in green urban development interventions of secondary drainage sections along the Msimbazi to alleviate flooding problems in Dar es Salaam by River. These engineering solutions have been designed analysing a range of stormwater management scenarios for a 1:25 year flood on the Msimbazi River and for a that considered measures that either reduced exposure 1:50 flood on the Sinza River. The unit cost of this is to flooding, reduced flood risk, or a combination estimated to be $1500 - 2500 per m, with a total cost of of these. The three types of measures considered $29 million. This is similar to the cost of the main GUD - implementation of restoration and rehabilitation intervention included in this study; the rehabilitation measures in the catchment, storage basins, moving and enhancement of middle catchment riparian and people away from flood prone areas – all led to floodplain areas which cover 488 ha along the Msimbazi, decreases in the damage costs of flooding. Absolute Sinza and Ubungo Rivers. benefits therefore increase as more measures are The role of catchment riparian and floodplain areas in combined, but so do costs. Taken alone, catchment biodiversity conservation must be emphasised as these rehabilitation measures provided higher net benefit areas are considered critical for maintaining ecological than moving people from the flood prone areas, and connectivity between terrestrial systems, rivers and also yielded the highest rates of return. The addition of estuaries. These areas also include opportunities for a storage basin added least value, but largely because other beneficial uses, such as sports fields and parks, opportunities for the location of such an intervention and are more likely to reduce the chances of informal were too low down in the catchment to be particularly resettlement of the floodplain. Community-based river effective. The results suggest that investment should cleaning programmes also provide important co-benefits be secured for the implementation of a combination including education, social awareness and community of rehabilitation measures in the catchment that are development as evidenced by the effective operation of specifically designed to attenuate flows and improve the Mlalakua River Restoration Project in Dar es Salaam. drainage, including formal solid waste management and However the success of such programmes depends on community-based river cleaning programs, reforestation active support and diversified and resilient funding. in the upper catchment, the rehabilitation of river These green urban development interventions, while buffers in the middle catchment and the reconnection designed to control flooding impacts, also contribute to of floodplains in the lower reaches. This could be part water quality enhancement and present opportunities for of an even broader catchment-to-coast rehabilitation generating amenity value, other ecosystem services, and programme for the Msimbazi River system which also community upliftment. Many of the investments required aims to address water quality problems and the need for in the Msimbazi catchment do involve costly rehabilitation green open space within the rapidly-growing city. (catchment land cover and river-floodplain connection) It is important to note that this analysis did not and relocation of unplanned settlements from river capture all the costs and benefits associated with the margins, demonstrating that better historic protection of implementation of GUD interventions. On the positive both catchment and floodplain areas would have been a side, these include the amenity benefits associated far more efficient development path. This is important for with the creation of green open space areas along the the city to bear in mind as it prepares for rapid expansion, riparian zones as well as improvement in biodiversity. especially toward the south. A green urban development path would offer a variety Due to the limited availability of data, this study by of opportunities for enhancing the livability of the necessity utilized simple models and assumptions. city. On the negative side, it should be acknowledged While the results strongly suggest that catchment that relocation of people away from the setback areas rehabilitation interventions would yield a positive could generate psychological suffering and anxiety in outcome in economic terms, the figures presented the affected individuals that is difficult to quantify or here are preliminary and warrant further investigation compensate in monetary terms. and refinement. The results, do however, provide a Whilst conventional conveyance measures were useful step towards informing policies and contributing not considered during this study it is important to to Dar es Salaam’s green urban development path. acknowledge that solving the flooding and water It is recommended that investment is made in the quality problems in Dar es Salaam will likely require development of better hydrological data, through a combination of conventional and green urban establishment of flow and additional rainfall gauges, as development measures. Within the Msimbazi catchment well as development of detailed spatial datasets on soils, a number of conveyance measures have been designed land cover, the built environment and the city’s drainage as part of the Dar es Salaam Metropolitan Development systems. Moving forward these datasets can then be used to construct a more definitive analysis.                                                  Page xii AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT ACRONYMS AND ABBREVIATIONS AMC Antecedent soil moisture condition LIDAR Light Detection and Ranging BOD Biochemical oxygen demand LULC Land Use Land Cover BMP Best Management Practice MAR Mean Annual Runoff CCIAM Climate Change Impacts Adaptation and NEMC National Environment Management Council Mitigation NMFA Norwegian Ministry of Foreign Affairs CDF Cumulative Distribution Function NTU Nephelometric Turbitity Units CLS Collapse Limit State OSM Open Street Map CLUVA Climate Change and Urban Vulnerability in PAP Project Affected Persons Africa RAP Resettlement Action Plan CVM Curve Number Method RCF Reinforced Concrete Frame DEM Digital Elevation Model REDD Reducing emissions from deforestation and DLS Damage Limit State forest degradation DMDP Dar es Salaam Metropolitan Development ROI Return on investment Project SUDS Sustainable Urban Development Systems EAL Expected Annual Loss TEEB The Economics of Ecosystems and Biodiversity EMA Environmental Management Act TIN Total Inorganic Nitrogen FM Formal Masonry TMA Tanzania Meteorological Agency GDP Gross Domestic Product TN Total Nitrogen GIS Geographic Information System TOC Total Organic Carbon GUD Green Urban Development TP Total Phosphorous IDF Intensity Duration Frequency TSS Total Suspended Solids IM Informal Masonry TZS Tanzanian standard IRR Internal Rate of Return UMT Urban Morphology Type IUWM Integrated Urban Water Management WCST Wildlife Conservation Society of Tanzania LCC Lifecycle cost WHO World Health Organisation LID Low Impact Development WSUD Water Sensitive Urban Design                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page xiii TABLE OF CONTENTS I. INTRODUCTION 1 1.1 Background......................................................................................................................................................................1 1.2 Study aims.......................................................................................................................................................................3 1.2.1 Overall approach ................................................................................................................................................. 3 II. STUDY AREA 5 2.1 Dar es Salaam.................................................................................................................................................................5 2.2 The Msimbazi river system.............................................................................................................................................6 2.2.1 Overview of the river system............................................................................................................................... 6 2.2.2 Condition of the riparian areas........................................................................................................................... 9 2.2.3 Condition of the estuary and mangroves......................................................................................................... 14 2.2.4 Pollution and flooding........................................................................................................................................ 14 III. MODELLING FLOOD RISK IN THE MSIMBAZI CATCHMENT 17 3.1 Overview.........................................................................................................................................................................17 3.2 Data requirements for flood risk assessment.............................................................................................................17 3.2.1 Historical rainfall data........................................................................................................................................ 18 3.2.2 Geomorphologic / biophysical data.................................................................................................................. 18 3.2.3 Land use/land cover data................................................................................................................................. 23 3.3 Flood hazard assessment.............................................................................................................................................25 3.3.1 The rainfall curve................................................................................................................................................ 25 3.3.2 Hydrologic basin modelling............................................................................................................................... 28 3.3.3 The hazard curves.............................................................................................................................................. 36 3.4 The exposure and vulnerability models.......................................................................................................................38 3.4.1 The vulnerability model...................................................................................................................................... 40 3.4.2 The hazard curves for the buildings at risk ..................................................................................................... 42 3.4.3 Risk assessment ............................................................................................................................................... 42 3.4.4 The expected annual losses (EAL) ................................................................................................................... 43 IV. EVALUATION AND SELECTION OF POTENTIAL URBAN STORMWATER MANAGE- MENT OPTIONS 45 4.1 Overview of stormwater management.........................................................................................................................45 4.2 Types of stormwater management measures.............................................................................................................45 4.3 Relative performance of different measures..............................................................................................................46 4.3.1 Average cost effectiveness in terms of peak flow and volume reduction...................................................... 47 4.3.2 Average cost effectiveness in terms of water quality amelioration................................................................ 47 4.3.3 Overall effectiveness, cost-effectiveness and potential co-benefits............................................................... 49                                                  Page xiv AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT 4.4 Selection of suitable measures for Dar es Salaam.....................................................................................................51 4.5 Conceptual design and costing of selected interventions..........................................................................................54 4.5.1 Swales................................................................................................................................................................. 54 4.5.2 Catchment reforestation ................................................................................................................................... 54 4.5.3 Rehabilitation and enhancement of middle catchment riparian and floodplain areas................................ 57 4.5.4 Rehabilitation and enhancement of lower floodplain areas........................................................................... 62 4.5.5 Community-based river cleaning programme.................................................................................................. 64 4.5.6 Summary ........................................................................................................................................................... 64 V. SCENARIOS ANALYSIS 67 5.1 Scenarios.......................................................................................................................................................................67 5.2 Hydrologic modelling assumptions..............................................................................................................................67 5.2.1 Riparian setback in lower floodplain................................................................................................................. 67 5.2.2 Combined GUD interventions............................................................................................................................ 68 5.2.3 Additional storage in Scenarios 4 and 5.......................................................................................................... 69 5.2.4 Caveats............................................................................................................................................................... 69 5.3 Costs of the interventions.............................................................................................................................................70 5.3.1 Riparian setback in lower floodplain................................................................................................................. 70 5.3.2 Combined GUD interventions............................................................................................................................ 70 5.3.3 Additional storage in Scenarios 4 and 5.......................................................................................................... 70 5.4 Scenario evaluation approach .....................................................................................................................................71 5.5 Results...........................................................................................................................................................................71 VI. SUMMARY AND CONCLUSIONS 75 VII. REFERENCES 77                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page xv LIST OF FIGURES Figure 1.1 Schematic diagramme of the range of infrastructure required for Green Urban Development............................................................2 Figure 1.2 Rough sketch of the relationships between environmental assets, ecosystem services and their beneficiaries in Dar es Salaam. Bars provide a qualitative indication of the ecological condition of these ecosystems.......................2 Figure 2.1 Location of the Msimbazi River catchment in relation to the three municipalities of Dar es Salaam, and showing the topographical landscape. ...............................................................................................................................................7 Figure 2.2 Land cover within the Dar es Salaam municipal areas in and around the Msimbazi catchment.............................................................8 Figure 2.3 Msimbazi River at the edge of the Pugu Forest Reserve (on left), flowing through agricultural land and scattered settlements showing Pugu Kinyamwezi dumpsite on the south bank. Location of the image is shown in Figure 2.4 (A).....9 Figure 2.4 Confluence of Msimbazi River and Luhanga tributary showing some riparian vegetation amongst scattered settlements. Location of the image is shown in Figure 2.4 (B)................................................................................................10 Figure 2.5 Msimbazi River flowing through dense residential area and showing industrial areas at the bottom of the image and Vingunguti dumpsite along the banks in the right hand side. Location of the image is shown in Figure 2.4 (C)................11 Figure 2.6 Industrial area surrounding Nelson Mandela Bridge and urban agriculture within the floodplain of the Msimbazi River. Location of the image is shown in Figure 2.4 (D)............................................................................................... 12 Figure 2.7 Confluence of the Ubungo and Sinza Rivers showing large floodplain area and informal settlements. Location of the image is shown in Figure 2.4 (E)..................................................................................................................................... 13 Figure 2.8 Flooding in the Msimbazi Valley causes significant damage and the dumping of solid waste blocks culverts, pipes and channels which further exacerbates flooding. ............................................................................................16 Figure 3.1 DEM NASA SRTM 1 arcsec (i.e. horizontal resolution 30 meters).......................................................................................................... 19 Figure 3.2 DEM acquired from the local city council (vertical and horizontal resolution of 2 meters)................................................................. 19 Figure 3.3 Lidar surveys (vertical resolution 0.5 meters, horizontal resolution 0.5 meters)..................................................................................20 Figure 3.4 DEM for the analyses domain (horizontal resolution 2 meters)..............................................................................................................20 Figure 3.5 Geology: The description of soils..............................................................................................................................................................22 Figure 3.6 Lithology of the study area........................................................................................................................................................................22 Figure 3.7 Land use for the larger Dar es Salaam region.......................................................................................................................................... 23 Figure 3.8 ..............................................................................................................................................................24 Urban Morphology Types (2012). Figure 3.9 Hazard assessment procedure (h = rainfall height, d = duration, Q = discharge, and TR = return period; De Risi et al. 2013a)a....... 25 Figure 3.10 Rainfall Probability Curves for Dar es Salaam based on historical data..................................................................................................27 Figure 3.11 Catchments of the case study river..........................................................................................................................................................28 Figure 3.12 The schematic diagram of a hydrographic basin..................................................................................................................................... 30 Figure 3.13 The hydrographs for the three catchments, for the three AMCs and for all the considered return periods. Note y-axis scales differ among panels ....................................................................................................................................................35 ................................................................................................................................................................................37 Figure 3.14 Analysis sub-domains. Figure 3.15 Schematic representation of the procedure and of the expected output in terms of flood hazard curves. In this example only the red hazard curve will be associated to the analyzed building........................................................................37 Figure 3.16 Geographical localization of the analyses domain.................................................................................................................................. 38 ..................................................................................................39 Figure 3.17 Building footprints in the analyses domain distinguished by typology. Figure 3.18 (a) Buildings at risk, and (b) buildings at risk distinguished by structural typology................................................................................39 Figure 3.19 Fragility curves for the three considered structural typologies and two potential limit states........................................................... 40 Figure 3.20 Hazard curves for the baseline scenario; λ values ≤ 0.2 based on inundation records; values ≥ 0.2 based on extrapolation............42 Figure 4.1 Different types of measures used in stormwater management. These measures are described in Appendix 5................................ 46 Figure 4.2 Comparison of cost per unit volume of runoff reduction for various stormwater management options, based on literature averages......................................................................................................................................................47 Figure 4.3 Comparison of cost per unit mass of pollutant/nutrient reduction for various stormwater management options............................49 Figure 4.4 ...........................................51 Heavy development around the lower floodplain areas of the Msimbazi River system in Dar es Salaam. Figure 4.5 Map showing proposed location of “green” interventions within the Msimbazi River catchment......................................................56 Figure 4.6 ..................... 58 Conceptual plan of mixed use Enhanced Riparian and Floodplain areas. See text for description of zone treatment. Figure 4.7 Rough cross-section sketch of concept................................................................................................................................................... 58 Figure 4.8 Digitised map showing the dwellings and agricultural fields in the setback zones of the middle to upper catchment areas of the Msimbazi, Sinza and Ubungo sub-catchments.........................................................................................................................................61 Figure 5.1 Two different ways to implement the mitigation strategies in the input hydrograph for different return periods............................ 68                                                  Figure 5.2 Graphical representation of the results shown in Table 5.4....................................................................................................................73 Page xvi AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT LIST OF TABLES Table 2.1 Point source and non-point source discharge entering the Msimbazi River ........................................................................................15 Table 2.2 Estimated pollution loads from difference sources into the Msimbazi River........................................................................................16 Table 3.1 Percentage of utilization of the available DEMs......................................................................................................................................21 Table 3.2 The characteristics of the Msimbazi River catchments.......................................................................................................................... 29 Table 4.1 Measured pollutant removal capacities of selected stormwater management options and technologies .......................................48 Table 4.2 Relative merits (indicated by number of “X”) of different measures for stormwater and flood risk management, based on the literature. Measures considered in this study area are marked with an asterisk...........................................................50 Table 4.3 Requirements for different stormwater management measures (apart from financial), and implications/ suitability in Dar es Salaam....................................................................................................................................................................... 53 Table 4.4 Total estimated resettlement costs for the lower catchment and middle to upper catchment areas ............................................... 62 Table 4.5 The estimated extent of enhanced floodplain-recessed gardens, estimated total cost of the intervention and the estimated total amount of runoff retained............................................................................................................................................. 63 Table 4.6 Estimated extent and total cost of the extended shallow wetland and the total stormwater runoff expected to be retained........ 65 Table 4.7 Summary of the extent and costs of the selected GUD interventions in the middle to upper Msimbazi catchment and the resettlement costs involved in relocating households from these areas..............................................................................................65 Table 5.1 Scenarios 1-5 and their estimated costs.................................................................................................................................................. 67 Table 5.2 Cost breakdown of detention basin.........................................................................................................................................................70 Table 5.3 Impacts of Scenarios 1 to 5 on expected annual losses (EAL), and the percentage change in EAL..................................................... 72 Table 5.4 Comparison of scenarios .........................................................................................................................................................................73                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page xvii This page intentionally blank.                                                  Page xviii AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT I. INTRODUCTION Background Urbanisation is taking place at an unprecedented rate The benefits of natural ecosystems, or “ecological throughout the world, with the rate of growth often infrastructure”, and more broadly, of “green outpacing plans and the capacity of city managers, infrastructure” which also includes man-made particularly in developing countries. As a result, ecosystems, are increasingly being recognised in the existing natural areas within cities that provide a range growing area of research and development regarding of benefits are becoming smaller and degraded, and urban stormwater management systems. Great problems such as flooding, air pollution and water strides have been made in the design of sustainable pollution are becoming worse. This has led to negative mechanisms to deal with stormwater flows and the impacts on health, income, productivity and quality of attendant pollution problems, and management life, as well as stretching local and national government and planning of cities is increasingly taking a holistic finances. These problems are likely to escalate with approach that includes the use and conservation of continued movement of the poor into cities. semi-natural and natural areas within cities as part of a green urban development (GUD) strategy. This aligns Environmental problems are particularly acute in well with the concept of “green urban development”, African cities, where the lack of sufficient regulation the essence of which is development that minimizes of urbanization leads to unplanned urban growth impacts on and/or enhances the value of the natural characterized by poor or non-existent construction environment through incorporation of an optimal mix standards (De Risi et al. 2013c), and by structures located of different types of green and grey infrastructure in high-risk areas such as river banks and flood plains (Figure 1.1), in conjunction with supporting non- (Sakijege et al. 2014), which also increases the risks structural interventions (laws, maintenance, etc.). associated with natural events such as floods, leading to “natural disasters”. It also decreases cities’ resilience to The conservation of natural systems and the services climate change, under which the likelihood of extreme they provide is believed to form an important part of this flooding events is expected to increase (Khan & Kelman strategy. However, one of the challenges of green urban 2012, Jalayer et al. 2013a). Developing countries already development will be to find the right balance between tend to be less resilient to natural disasters because of ecological and green or grey engineered infrastructure. their fragile economies, poverty, lack of risk awareness, There is generally a paucity of understanding of both and lack of coping capacities in urban communities (De the ecological functioning and the value of the existing Risi et al. 2013a, 2013b, Jalayer et al. 2015). natural assets in African cities, or of the trade-offs involved in developments that replace or degrade these African cities often lack the resources to deal with these assets (e.g. Daily & Matson 2008). problems. Solutions have traditionally involved costly engineering interventions applied in the developed Dar es Salaam, located on Africa’s Indian Ocean Coast, world. However, several studies have suggested that has a range of ecosystem types within and around investing in the maintenance or restoration of natural the city that could deliver important ecosystem assets may not only offset much larger engineering services (Figure 1.2). Yet many of these have become costs, but bring multiple benefits that ultimately severely degraded. The degradation of the city’s translate into cost savings and increased human forest and river systems is well known, and is believed wellbeing. Various types of natural areas exist within to have contributed to the some of the multitude of urban areas, which yield a range of benefits to different environmental problems that the city now faces. sectors of society. At the top of its list is the problem of flooding in and around the city centre, which frequently brings the city to a standstill, as well as causing infrastructural damages. Many factors have contributed to this problem, including unplanned settlements in both the catchment and floodplain areas, a lack of drainage and a lack of solid waste management. The impacts of flooding are also exacerbated by high levels of pollution in the rivers. In consultations that were held for this study, stakeholders in Dar es Salaam identified the centrally-located Msimbazi River as being among the most degraded ecosystems in the city and also the source of the most serious flooding problems.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 1 Figure 1.1 Schematic diagramme of the range of infrastructure required for Green Urban Development Figure 1.2 Rough sketch of the relationships between environmental assets, ecosystem services and their beneficiaries in Dar es Salaam. Bars provide a qualitative indication of the ecological condition of these ecosystems. Source: Author                                                  Page 2 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Study aims A two-dimensional hydraulic model was set up using The aim of the study was to explore the potential Flo2D software to model flood velocities and depths costs and benefits of undertaking a green urban in the flood risk area and to estimate the expected development approach to addressing flooding problems annual losses in terms of structural damages, based on in the Msimbazi River floodplain area of Dar es Salaam data on building locations, types, values and fragility. through a range of complementary measures, including Design floods were generated from rainfall data using ecological restoration. a hydrological model of the three sub-catchments that discharge into the flood risk area. The models were The study aimed to provide a high level evaluation of built using available data on the study area, which is measures to alleviate flood risk as a first step towards generally poor in quality, necessitating some simplifying the development of more comprehensive plans that assumptions. include the reduction of pollution loads into the system (notably improved sanitation systems and control of A review was carried out on stormwater management point source pollution), in order to realise the potential options, their efficacy in terms of various criteria, capacity of these interventions to also contribute their cost-effectiveness and the necessary or suitable to water quality enhancement and present further conditions for their implementation. Based on this, opportunities for generating amenity value and other and available GIS data on land cover, slope and soils ecosystem services. of the catchment, the long list of possible measures was reduced to a set of measures that had both a high feasibility of implementation in the study area and 1.1.1 Overall approach that would be complementary in terms of their effects on the flood risk area. The potential extent of their The overall approach was to model current flooding and implementation was then estimated and mapped. expected annual losses (EAL) in the Msimbazi catchment Finally, a set of six scenarios was devised which included and to determine the potential change in flooding and the full combination and various subsets of these EAL after implementation of a range of stormwater measures. management scenarios involving different combinations of feasible measures. The scenarios were then The hydrology or hydraulic model parameters were compared in terms of their return on investment (ROI). adjusted, as appropriate, to simulate the effects of the different measures on the flood hydrograph1 and/or The study began with a review of all available information the conveyance of flood water in the receiving area. on the Msimbazi River system and its catchment and The scenarios were compared in terms of their ROI, net floodplain areas. The current state of the river system present value and internal rate of return in order to was assessed on the basis of a brief site visit, published identify the best solution to mitigate/avoid future flood and unpublished information including GIS data layers, catastrophes. and examination of Google Earth imagery. 1 A graph showing the rise and fall in water volume passing a particular point time during a flood event                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 3 This page intentionally blank.                                                  Page 4 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT II. STUDY AREA Dar es Salaam Up to 60% of the solid waste generated in Dar es Salaam Dar es Salaam, situated on the coast of Tanzania, is a remains uncollected with collection rates of 27%, 39% city that has undergone rapid growth and urbanisation and 41% for Temeke, Ilala and Kinondoni respectively in the past few decades. The city has one of the highest (AECOM 2015). The city of Dar es Salaam only has one population growth rates in Tanzania, which has been operational dumpsite, Pugu Kinyamwezi which is located near or above 5% per annum for the past 30 years, to the west of the city in a former sand quarry. Other compared to the national average of 3% (URT 2002, dumpsites have all been shut down through either 2012). Estimates from the 2012 census place the city’s’ community protest or court order, due to poor planning population at around 4.36 million with a density of 3.133 and management (Kihampa 2013). The lack of dumpsites, persons per km2 (URT 2013). along with the lack of access to all areas, inadequate finances and low priorities within municipalities, have The city is governed by three separate municipalities, contributed to the solid waste problem in Dar es Salaam. Kinondoni in the north, Ilala in the centre and Temeke in the south. The rapid growth and expansion of the Lacking proper infrastructure like sewerage systems city has led to the proliferation of unplanned residential or waste removal, unplanned areas are a large source areas, especially on the outskirts of the city and along of pollution within the city. While both raw effluent major roads. In addition there is consolidation in and solid waste can contribute to polluting waterways, already developed areas leading to many houses being solid waste can also lead to clogged drains and canals built in areas previously considered unsuitable, such and exacerbate flooding as it hinders the drainage of as on floodplains and river banks. The result of this water during storm events. Not only do these polluted rapid expansion and consolidation is that unplanned waters pose a health risk to residents that come developments now occupy over three quarters of into contact with the water, but they also pose risks the city (NEMC 2009, Hill & Linder 2010). While plans through contamination of clean water sources as well as are in place to try and formalise the unplanned areas contributing to the spread of diseases like malaria and and provide some infrastructure and services, the lymphatic filarial. municipalities have not been able to keep up with the Dar es Salaam has two rainy seasons, March-April-May growth of the city and, as a result, most residents still and November-December. The mean rainfall intensity have poor sanitation, inadequate infrastructure and a is highest in the March-April-May rainy season (average lack of services (UN Habitat 2010). It should be noted of approximately 50 mm in 24 hrs). The most intense that in Dar es Salaam, unlike many other African cities, single storm events between 1971 and 2009, however, the ‘informal’ structures that are built in unplanned were recorded during the November-December rainy areas are generally constructed using durable building season, reaching 150 mm in 24 hrs (TMA 2011). Up to a materials and are therefore considered to be relatively 6% increase in mean precipitation during the long rainy permanent structures. season is projected to occur within the next 100 years The coverage of storm water drains throughout Dar along Tanzanian coastline (Matari et al. 2008). While es Salaam is extremely limited and the majority of predictions for changes to overall rainfall in Tanzania are the drainage infrastructure, located mainly in the city unclear, it is likely that it will result in great variability centre, is currently in poor condition (Draft Dar es in rainfall which will likely lead to more frequent and Salaam Masterplan 2012-2013). The current sewerage intense droughts and floods (Watkiss et al. 2011). system in Dar es Salaam only services about 10% of the population, leaving 90% of the city relying on on-site sanitation (URT 2014b, TSCP). Even the sewage that does pass through the sewerage system is only partially treated before being discharged into rivers or the ocean. The system consists of 15 pumping stations and 8 independent waste stabilisation ponds that, at best, provide primary treatment of effluent, and a sea sewage outfall (AECOM 2015).                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 5 There are four main rivers draining the city The Msimbazi river system municipalities; the Mpiji, Kizinga and Mzinga and Msimbazi rivers. The Mpiji forms the northern boundary 2.0.1 Overview of the river system of Dar es Salaam, the Kizinga and Mzinga rivers flow into the large harbour south of the city centre, where as the The Msimbazi catchment is approximately 300 km2 and Msimbazi flows through the heart of the city. During the extends across two city municipalities, the Kinondoni rainy season, intense rainfall events often cause flooding and Ilala councils, as well as extending west of Dar es in certain areas of the city. The Msimbazi river valley Salaam city (Figure 2.1). The Msimbazi River stretches tends to flood frequently due to its relatively high-clay approximately 35 km from Kisarawe outside of the Dar soil content compared with the well-drained coastal es Salaam and flows into a mangrove estuary within plain sands which characterise the Kazinga and Mzinga the heart of the city before discharging into the Indian river basins (NEMC 2009). While the river floodplains Ocean. It is joined by the Sinza and Ubungo tributaries would have flooded naturally after large storms, prior to meeting the ocean. the increase in hardened surfaces in the catchment area such as roofs, pavement or compacted dirt in The Msimbazi River was historically important as a conjunction with the inadequate storm water system source of water for Dar es Salaam residents, providing have exacerbated the flooding, effectively extending water for drinking, bathing, industry and agriculture the active floodplain area. However it is not only the (NEMC 2009). The river has, however, become structures that have been built within the original progressively contaminated. Up to 55 tonnes of diffuse floodplain area that are at risk, but also structures that pollution from illegal solid waste and up to 450 tonnes have been built in areas that would have been safe from from other sources enters the Msimbazi River each year flooding before the level of flooding was increased as a (NEMC 2009). While this discourages use as drinking result of increased runoff from the increasingly hardened water especially along the main river, some poorer catchment areas as Dar es Salaam has grown. residents in some tributaries are still using the water for human consumption. The most common usage of water This study focuses on the Msimbazi River system, where is extraction for agriculture. the flooding problems have been greatest. The river system is described in more detail below. Landuse within the Msimbazi catchment is divided amongst cultivated land, woodland/bushland and forest, with urban areas concentrated in the lower catchment area but encroaching up into the whole catchment (ILIR 2007, Figure 2.2). The Ubungo catchment has the highest proportion of cultivated land with some woodland/ bushland and urban areas, whereas the Sinza is highly urbanised (ILIR 2007).                                                  Page 6 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Figure 2.1 Location of the Msimbazi River catchment in relation to the three municipalities of Dar es Salaam, and showing the topographical landscape.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 7 Figure 2.2 Land cover within the Dar es Salaam municipal areas in and around the Msimbazi catchment. The headwaters of the Msimbazi River are situated in the converted. Additionally, from satellite imagery, it is Pugu Forest Reserve at the top of the catchment. This evident that deforestation and thinning of the natural reserve, while somewhat conserved in comparison to vegetation is also taking place over a large portion of the the rest of the catchment, is still considered degraded reserve. Only about 20% of the reserve is considered to (CCIAM 2011). The edge of the reserve has been heavily be in good condition (CCIAM 2011). encroached by the surrounding small-scale agriculture and up to 1 km within the reserve is now completely converted. Additionally, from satellite imagery, it is 2.0.2 Condition of the riparian areas evident that deforestation and thinning of the natural As the Msimbazi flows east out of the Pugu Forest vegetation is also taking place over a large portion of the Reserve it crosses crop fields and agricultural land reserve. Only about 20% of the reserve is considered to (Figure 2.3). South of the river in an old sand quarry, the be in good condition (CCIAM 2011). city’s sole working solid waste dump, Pugu Kinyamwezi dumpsite, can be found. Here the riparian vegetation has The headwaters of the Msimbazi River are situated in the been heavily denuded and little remains. Farming and Pugu Forest Reserve at the top of the catchment. This small settlements frequently extend right up to the river reserve, while somewhat conserved in comparison to banks. As the river approaches the city, the density of the rest of the catchment, is still considered degraded settlements increases. Along this stretch there is some (CCIAM 2011). The edge of the reserve has been heavily remaining riparian vegetation, although it is likely to be encroached by the surrounding small-scale agriculture degraded. and up to 1 km within the reserve is now completely                                                  Page 8 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Figure 2.3 Msimbazi River at the edge of the Pugu Forest Reserve (on left), flowing through agricultural land and scattered settlements showing Pugu Kinyamwezi dumpsite on the south bank. Location of the image is shown in Figure 2.4 (A). Figure 2.4 Locations of the Google Earth images shown in this section                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 9 Figure 2.5 Confluence of Msimbazi River and Luhanga tributary showing some riparian vegetation amongst scattered settlements. Location of the image is shown in Figure 2.4 (B). At the confluence with a Luhanga tributary (Figure 2.5) there is still some remaining riparian vegetation, despite the increases in density of people. This tributary drains a large area of farmland to the north of the main Msimbazi channel. Below this, the river has an extensive floodplain area that is under dense unplanned settlement. Illegal sand mining in the middle catchment has caused severe degradation and destabilisation of the riparian zone.                                                  Page 10 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Figure 2.6 Msimbazi River flowing through dense residential area and showing industrial areas at the bottom of the image and Vingunguti dumpsite along the banks in the right hand side. Location of the image is shown in Figure 2.4 (C). The Msimbazi then passes some industrial areas and the old Vingunguti solid waste disposal site on its south bank (Figure 2.6). Adjacent this site the river has been straightened to allow expansion of the dumpsite. This immediate area also houses the Vingunguti Abattoir and mixed residential area. There is very little riparian vegetation along this stretch of the river and occasional scatted urban agriculture.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 11 Figure 2.7 Industrial area surrounding Nelson Mandela Bridge and urban agriculture within the floodplain of the Msimbazi River. Location of the image is shown in Figure 2.4 (D). Between the dumpsite and the Nelson Mandela Bridge (see bottom left hand corner of Figure 2.7), there are densely populated residential areas on both sides of the river, with occasional small patches of urban agriculture along the banks. At the bridge, there is a large industrial area backing directly onto the river (Figure 2.7). Beyond the industrial area the floodplain opens up into a wide area which is filled with agricultural fields on the south bank, and dense residential areas on the north bank.                                                  Page 12 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Figure 2.8 Confluence of the Ubungo and Sinza Rivers showing large floodplain area and informal settlements. Location of the image is shown in Figure 2.4 (E). Between the confluences with the Ubungo and Sinza The Sinza River originates near the densely vegetated Rivers, the Msimbazi passes through another wide military base and the University of Dar es Salaam. As the floodplain area which is mainly covered with urban Sinza approaches the Msimbazi it passes through formal agriculture and scattered dwellings (Figure 2.8). The residential areas with forested riparian buffers. Closer floodplain areas appear to have been disconnected from to the city centre, the houses become denser and the the river channel in this area by berms. The agricultural riparian vegetation disappears. Before the confluence areas are surrounded by mainly mixed residential area. with the Msimbazi, the river opens into a floodplain that There are also small sections along the river that have is partially cultivated but also contains many informal been raised and converted to industrial areas. dwellings. The Ubungo tributary mainly courses through dense residential areas as well as some industrial areas. There are some tall trees and other riparian vegetation along the banks of this stream. There is limited urban agriculture except in a small floodplain near the confluence with the Msimbazi (Figure 2.8).                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 13 2.0.3 Condition of the estuary and mangroves 2.0.4 Pollution and flooding Natural vegetation along the banks of the Msimbazi Pollution and flooding are both major problems in the River and tributaries is mostly denuded as are natural Msimbazi River system, with pollution also exacerbating floodplains and wetlands. Those that do remain are the impacts of flooding. Pollution comes from a number highly degraded and modified. At the Msimbazi estuary, of different sources including industrial, residential there is still a large mangrove stand of approximately and agricultural effluent, storm water, solid waste and 0.5ha, which is dominated by Avicennia marina (Mremi & wastewater (Table 2.1). As a result, the Msimbazi River Machiwa 2003). is highly contaminated and pollutant levels exceed many standards for drinking, irrigation and contact with skin. The mangrove system in the Msimbazi estuary serves The number of faecal coliform bacteria colony-forming as a filter for pollution before the river water is units varied between 27 000 - 580 000 during the wet discharged into the Indian Ocean. The pollution from season and 37 000 - 117 000 during the dry season residential and industrial effluent as well as that from along the main Msimbazi River (Kassenga & Mbuligwe storm water has the potential to influence not only the 2009). In the dry season the highest values were found health of this mangrove ecosystem, but also the coastal near the Nelson Mandela Road Bridge, whereas in the ecosystems including seagrass beds and coral reefs that wet season the highest values were encountered near are situated along the coast. These ecosystems provide the mouth of the river (Kassenga & Mbuligwe 2009). All important food, shelter and nurseries for a variety of measurements indicated that the water in the Msimbazi different taxa including amongst others, commercially River was not fit for human consumption, and the values important fish species. at the river mouth are over 1000 times the level that is safe for full contact (e.g. swimming). Examination of the pollution levels within the mangroves at the mouth of the Msimbazi has found high levels Lead concentration in the water also exceeds WHO of heavy metals not only in the sediments, but also drinking standards (Mwegoha & Kihampa 2010). The level in mangrove roots and leaves as well as in fauna of lead ranged from 0.083-0.113mg/L, whereas the WHO living amongst the mangroves including crustacean drinking water standard is <0.01. The highest levels of and gastropod species (De Wolf et al. 2001, Mremi contamination were found along the main Msimbazi River & Machiwa 2003, Mrutu et al. 2013). The amount of between the dumpsite and the Nelson Mandela Bridge, these heavy metals that are bioavailable and thus can while the lowest concentrations were found further be transported up the food chain has not yet been downstream between the confluences with the Ubungo determined. While some of the levels of heavy metals and Sinza Rivers (Kassenga & Mbuligwe 2009). in the sediment amongst the mangroves are still lower than those reported in some European and American estuaries (Mrutu et al. 2013), the rate of increase has been quite rapid (De Wolf et al. 2001). The clay content in the soils in the Msimbazi estuary play a large role in filtering these heavy metals out of the polluted river water. The clay content of the top layer ranged from 17-85%, making it potentially effective at retaining heavy metals, however the retention capability of the mangrove is exceeded by the amount of heavy metals being deposited (Mrutu et al. 2013). This indicates that pollution is still passing through the mangrove system and therefore can reach and have the potential to have negative effects on the coastal environments. A few studies of the nearshore environment just off Dar es Salaam have also indeed showed that there is also heavy metal contamination in these environments (Mremi & Machiwa 2003, Mtanga and Machiwa 2007, Muzuka 2007). In addition, Daudi et al. (2012) found that the nutrient enrichment from the waters exiting from the Msimbazi River mouth was having negative impacts of a range of higher trophic levels species that inhabit seagrass beds.                                                  Page 14 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Table 2.1 Point source and non-point source discharge entering the Msimbazi River Category Source Description of effluent/discharge Pollution concern Industrial Contains spent grain, yeast and dilute Tanzania Breweries Ltd High BOD caustic soda Contains detergents, dyes, starch Friendship Textile Factory High pH and coloured effluent. and chemicals Dar Brew Contains spent grain and yeast High BOD, TSS and low pH. Tanzania Dairies Contains fats and waste milk High BOD and high pH. Robbialac Paints Contains waste paints Traces of heavy metals. Ubungo Farm Implements Hot water with traces of metals Hot alkaline effluent. Ubungo Power Station Contains fuel and lubricating oils Diesel and oils. Sewage from industrial premises High BOD, faecal coliforms, TN Diffuse industrial effluent discharge discharged directly to stream or and TP. through groundwater seepage Residential Sewage from residential homes – Diffuse effluent discharge from on- flooded pit latrines, septic systems High BOD, faecal coliforms, TN site sanitation leaking into groundwater or open and TP. drains. Agricultural Run-off contains organics or Diffuse effluent from farm land and fertilizers from farms, gardens and High BOD, TN and TP. animal grazing areas grazing lands. Other Run-off contains oils/fuels due to Petroleum hydrocarbon storage and inappropriate disposal of used oil Hydrocarbons in surface transportation facilities and poor loading and off-loading waters. procedures at depots. Sewage from three waste High BOD, faecal coliforms, TN Wastewater treatment plant stabilisation pond systems. and TP. Wastewater from wards, theatres, High BOD, COD and faecal Institutional effluent e.g. hospitals laboratories, mortuary and laundry. coliforms. Wastewater passes through retention tank, but still contains Vingunguti Abattoir Very high BOD and low DO. blood, manure, urine and animal pieces. Leachates from solid waste disposal Vingunguti solid waste disposal Very high BOD site. Includes roadside run-off and Storm water drainage contains oils and greases and some Oils and heavy metals heavy metal traces. In rains, on-site sanitation may flood Diffuse effluent from areas without High BOD, faecal coliforms, TN and sewage carried with storm water storm water drainage and TP. into the river. Source: NEMC 2009                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 15 Kassenga & Mbuligwe (2009) estimated the loads of Over recent years flooding in the Msimbazi Valley has pollution coming from different sources along the Msimbazi intensified and the damages have increased due to River (Table 2.2). The total pollution load was estimated the continuous influx of people choosing to settle on to be between 93-503 tonnes per year, with the largest unplanned land that is prone to flooding (Figure 2.9). In contributors coming from on-site sanitation systems and fact, areas that were previously not prone to flooding industrial areas without sewers. However, the pollution are now at risk due to increases in amount of hardened yield from these different sources may not necessarily surfaces in the catchment, construction of structures correspond to the severity of the impact that they have on in drainage channels and solid waste blockages in the population. Whilst illegal solid waste disposal has the pipes, culverts and channels (Figure 2.9). Bushesha lowest pollution yield (as shown in Table 2.2) this masks & Mbura (2015) concluded that the main reasons for the high impact it potentially does have on the population persistent floods in Dar es Salaam include inadequate when compared to other pollution sources. enforcement of land policies and legislations, poor institutional capacity in enforcing land use planning, and During flood events, the river water has the potential inadequate infrastructure and services to support rapid to come into contact with a large number of people. In urbanisation. Damages from floods in 2011, 2014 and addition to the costs of rebuilding houses and repairing 2015 were particularly severe. damage to infrastructure, there is an additional human health cost that is not often considered. The potential Table 2.2 Estimated pollution loads from difference sources into health effects associated with contact with this water the Msimbazi River during flood events are cause for concern. In addition to direct contact with river water during flood events, Pollution Source Pollution estimate (t/yr) residents may also come into contact with pollutants through vegetables grown in soils contaminated by the Min Max river water. Studies examining heavy metals in the soil found highest concentrations in the top soil layers which On-site sanitation systems 20.32 101.57 then decreased with depth (Mwegoha & Kihampa 2010). Industrial areas without sewers 17.70 141.56 While lead, chromium and cadmium levels in the soil Informal sector premises 16.12 80.61 were within Tanzanian standards (TZS 2003) permissible Storm water from undrained areas 8.57 42.83 limits for soils, there is still potential for transfer of these heavy metals up the food chain (Mwegoha & Kihampa Farm and animal grazing lands 19.75 80.57 2010). In the Msimbazi Valley there are high incidence Illegal solid waste disposal 11.17 55.86 rates of cholera, diarrhoea, intestinal worms, and Total Pollution Load 93.62 503.01 gastroenteritis. For more detail and a map of the cholera Source: Kassenga & Mbuligwe 2009 prone areas of Dar es Salaam see the Environmental Profile (AECOM 2015). Figure 2.9 Flooding in the Msimbazi Valley causes significant damage and the dumping of solid waste blocks culverts, pipes and channels which further exacerbates flooding. Source: Eric Schaechter UFZ, Resilient Cities: www.100resilientcities.org (top) and Dar Ramani Huria: ramanihuria.org (bottom)                                                  Page 16 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT III. MODELLING FLOOD RISK IN THE MSIMBAZI CATCHMENT Overview Data requirements for flood risk assessment Flood risk assessment encompasses three phases: Riverine flooding phenomena can be triggered by natural hazard, exposure and vulnerability assessment (Leader processes, such as heavy precipitation, or by the failure & Wallingford 2009; Jalayer et al. 2014). The hazard is of man-made structures such as levees, dams, and generally assessed through physically-based hydraulic drainage systems (in urban areas). The impact of flooding models, providing the flood depth and the velocity can be quantified in terms of depth, peak discharge, for each point within the study area, also accounting extent of area inundated, and volume of flows. Floods for the presence of buildings, infrastructure and soil vary in size and scale, ranging from minor waterlogged characteristics (O’Brien et al. 1993; Aronica et al. 1998; fields or briefly-blocked roads to the total inundation Cobby et al. 2003; Fabio et al. 2010; Biscarini et al. and destruction of homes and other structures, involving 2013; Yang et al. 2015). These methods are generally casualties. Catastrophic flooding events take place with computationally demanding, and they require significant a certain regularity over long periods of time (in the amounts of data and parameters in order to describe order of one to hundreds of years). Therefore, one way the morphology and the surface characteristics of the to classify the recurrence characteristics of a flooding flood basin (Bates & De Roo 2000; Bates et al. 2004; event is in terms of its return period (TR). In evaluating Di Baldassare et al. 2009). However, the required data inland river floods, the basic hydrological unit in river and modeling capabilities are not always available in systems is the drainage basin, or catchment (i.e. the area developing countries (Hagen & Lu 2011). Therefore that drains to any defined point along the river network). reasonable simplifications in modeling hypotheses and After a dry spell, rainfall infiltrates into the upper layers assumptions are generally adopted and accepted. For of soil and rock and only a small amount of water runs flood risk assessment, simplified methodologies based off in the catchment; only some soils need to be a little on a basin’s geomorphologic features can be used (De wet to encourage infiltration. However, continuing rain Risi et al. 2014; 2015). Such methodologies are less may lead to saturation of the surface soil layers; and accurate since they rely only on the topography because as a result, the volume of water will eventually exceed of its important role in flood depth and propagation (i.e. the amount that can be absorbed. At this point, surface extension; Gallant & Dowling 2003, Dodov & Foufoula- runoff begins. One way to represent the proportion of Georgiou 2006) and do not allow a straightforward total rainfall that is not infiltrated (i.e. is transformed to probabilistic risk assessment. Therefore, in this study, surface runoff) over a catchment is through the runoff only a physically-based method is adopted. Exposure coefficient. assessments require identification of the elements at risk, including all the elements of human, built and Surface runoff usually starts as sheet flow in which natural environments at risk in the flooding area. water moves as a thin, continuous film over relatively These can be population, buildings, infrastructure, smooth soil or rock surfaces and it is not concentrated economic activities, ecosystems, etc. Finally, given the into channels. As the volume of water increases, it forms characterization of the built environment, vulnerability tiny rills, then gullies, and then flows into small tributary analysis can be carried out in order to quantify the streams. The triggering of a flooding event is determined adverse effects of flooding. A vulnerability analysis by many factors, such as: provides the fragility functions (Porter et al. 2007), representing the probability of reaching or exceeding ƒƒ the amount of prior rainfall, which determines the predefined damage states, for a given level of intensity degree of soil saturation—referred to in this document (i.e. flood depth or inundation velocity). According to the as antecedent moisture condition; integration procedure proposed by De Risi et al. (2013a) ƒƒ topography of the drainage basin; the combination of hazard and vulnerability yields the mean annual rate of exceedance of a specific limit state. ƒƒ type of soil and land use; and This rate can be further used to calculate the probability of exceedance in a given time window, by adopting a ƒƒ intensity of rainfall. reasonable probability distribution describing the event occurrence (e.g. Poisson distribution). This probability can then be combined with the exposed asset value in order to quantify the flood risk in the predefined time window in terms of economic losses or in terms of number of casualties. Generally speaking, considering a time window of one year, the final result is expressed as the Expected Annual Loss (EAL).                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 17 Extended wet periods during any season can lead to 3.0.2 Geomorphologic / biophysical data saturated soil conditions. In such cases, additional Geomorphologic/biophysical spatial data (e.g., rainfall will quickly run off into streams and rivers. At topographic maps, geology maps, land-use, etc.) are some point, the run-off volume is going to exceed the fundamental data requirements in various stages of river channel capacity and lead to flooding. At times, flood risk assessment. These datasets are described in very heavy rainfall is followed almost immediately by more detail below. a large run-off volume, despite highly pervious and dry soil conditions. This condition is usually caused when large volumes of rainfall take place in a short interval of 3.2.2.1 Topography/ digital elevation model (DEM) time. Such bursts of extreme rainfall can produce spatio- Topography plays an important role in flood modeling, temporally localized and devastating flooding events with a demonstrated macro-scale correlation in between known as flash floods. terrain elevation and annual accumulated rainfall It is particularly challenging to create a mathematical (Allamano et al. 2009). Moreover, topography plays a model of flooding due to its spatio-temporal complexity key role in surface runoff and catchment response time and multi-stage triggering process. As a consequence, (i.e. the time between the peak rainfall and the peak quantified flood risk assessment demands a considerable flow discharge). Steeper catchments have higher runoff amount of data and information. Hence, data acquisition coefficients and faster response time, with mountain plays a pivotal role in a quantified flood risk assessment rivers flowing much faster than lowland rivers. The framework. This Section discusses the basic concepts typical instrument used to describe the topography of a and data requirements for a quantified flood risk generic hydrological domain is a digital elevation model assessment procedure. (DEM), a 3D digital representation of a terrain surface. The DEM is used herein for flood diffusion/propagation by employing a classic hydraulic routine. 3.0.1 Historical rainfall data In this study, three topographic datasets were collected: Riverine flooding events are strictly connected to (i) a DEM of 30 m horizontal resolution, obtained rainfall patterns. Therefore, rainfall time-series data from the U.S. National Aeronautics and Space Agency are essential for estimation of total flood discharge. (NASA) SRTM project website (http://gdex.cr.usgs.gov/ These data can be obtained as pluviometer records gdex/) (STRM Project; Figure 3.1); (ii) a contour map from governmental organizations or other sources and with 2 m vertical resolution covering only the central in many cases are available online (e.g., www.tutiempo. part of the city, acquired from the Dar es Salaam city net and www.knmi.nl). Ideally the pluviometric data council (Figure 3.2); (iii) ten LIDAR surveys for strategic should be available as precipitation extremes (maxima) areas in the city of Dar es Salaam, obtained from a recorded over a range of time intervals. The rainfall previous research project (Climate Change and Urban maxima recorded for different intervals are used in Vulnerability in Africa; Figure 3.3). Using GIS techniques, order to construct the rainfall curve, also known as the a final DEM was constructed of the analysis domain, Intensity-Duration-Frequency (IDF) curve. Historical using the best available resolution data1 (Figure 3.4). rainfall data can also be used to evaluate the antecedent soil moisture condition. In hydrological modeling, antecedent moisture condition is usually described as 1 The three DEMs were combined in a Geographic Information System the pre-storm soil moisture deficit. This latter has a (GIS) using a classic mosaic operation. Such a tool, available in many significant effect on the amount of rainfall drained by professional GIS platforms, is based on the pixel of a pre-defined grid, the river network and finally on the flooding potential of stores detailed properties, metadata, and processing information a rainstorm. in order to produce a new dataset containing the combination of the initial data. In this study, this technique was used both to have a In this study, historical rainfall data was obtained global DEM for the catchment identification and to have a final DEM for the geographical domain (hereafter referred to as the analysis from the single existing meteorological station in the domain) in which the inundation analyses are carried out. For the catchment, located in the Dar es Salaam International definition of the analysis domain, a 2 x 2 m grid was constructed. The Airport at 55 m above sea level, 6°86’S and 39°20’E. A resampled data were obtained through a linear interpolation. For detailed description of these data is presented in De each node of the resampled lattice, the topographic elevation then Paola et al. (2014). was obtained selecting the most refined data among the available DEMs (i.e. in sequence 0.5 m, 2 m, and 30 m).                                                  Page 18 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Figure 3.1 DEM NASA SRTM 1 arcsec (i.e. horizontal resolution 30 meters) Figure 3.2 DEM acquired from the local city council (vertical and horizontal resolution of 2 meters)                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 19 Figure 3.3 Lidar surveys (vertical resolution 0.5 meters, horizontal resolution 0.5 meters) Figure 3.4 DEM for the analyses domain (horizontal resolution 2 meters)                                                  Page 20 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Table 3.1 Percentage of utilization of the available DEMs. DEM Res. Area of the analyses domain Percentage of the Area of the analyses Percentage of the analyses covered by DEM analyses domain domain covered by domain covered by the covered by DEM the resampled DEM resampled DEM 1 30 m 1396 ha 100% 258 ha 18% 2 2m 1135 ha 81% 1041 ha 75% 3 0.5 m 97 ha 7% 97 ha 7% Table 3.1 lists the area of the analysis domain covered by Generally the geological data are in form of GIS-based each available DEM and the percentage of area used to shape files or raster files with variable scale --between build the resampled DEM. 1:1000 to 1:200000. In order to gain knowledge about subsoil stratigraphy, appropriate geological sections can be defined inside a geological spatial dataset. In 3.2.2.2 Geological data this study, the source data were acquired from the The underlying geology and dominant soil type of a Geological Survey of Tanzania’s Geological and Mineral catchment area are important factors in determining Information System (http://www.gmis-tanzania.com/). the quantity and proportion of the surface run-off. The The map has a resolution 1:2M and represents the best water infiltration capacity of the soil depends on both data available for the country. This is the official map the soil type (e.g., its grain size distribution and porosity) recognized by the Republic of Tanzania and it is in good and the characteristics of the underlying water-bearing accordance with the ISRIC – World Soil Information rock layers (e.g., porosity and thickness) also known as data (http://www.soilgrids.org/). Figure 3.5 presents “groundwater aquifers.” Large groundwater aquifers the geology and Figure 3.6 shows the lithology of the act as large reservoirs in that they release --over long study area. These show that the entire area of Dar es periods of time-- the water infiltrated after a storm. Salaam is characterized by the presence of Marine and Igneous rocks produced from lava or magma from the Fluvio-Marine sandy-clayey sediments. These soils have earth’s lower crust (e.g. basalt and granite) have very low moderate infiltration rates when thoroughly wetted and infiltration capacity unless they are heavily fractured. a moderate rate of water transmission. On the other hand, chalk and limestone have high infiltration capacity to the extent that they can prevent rivers from forming. Sandstone and shale are the analogs of sand and clay consolidated into rocks over millions of years through a process called lithification. Their infiltration capacity is usually low unless they are heavily fractured by subsequent geologic processes. A common and widely-used way to quantify the effect of soil type and stratification on the quantity of surface run-off is by means of the runoff coefficient that expresses the proportion of total rainfall that is drained by the river network. This coefficient reflects the runoff-relevant effects of the numerous possible combinations of soil types and underlying geology.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 21 Figure 3.5 Geology: The description of soils Figure 3.6 Lithology of the study area                                                  Page 22 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT 3.0.3 Land use/land cover data In this study, two different sources were used The runoff coefficient, catchment discharge and to characterize land use in the study area: (i) catchment response time all depend on land cover. In a CGIAR coarse-resolution map (available at forested areas, tree roots increase the infiltration of http://192.156.137.110/gis/search.asp?id=543) water by channeling it deeply inside the soil layers down showing the land use for the entire Tanzania in 2002 to the ground water. This effect is less pronounced in (Figure 3.7), and (ii) a finer-resolution map of the city of areas covered by shrubs and in pastures where the roots Dar es Salaam (Figure 3.8) obtained from the Climate are much shallower. In urbanized areas, and in particular Change and Urban Vulnerability in Africa (CLUVA) in large cities, the large percentage of paved areas project (www.cluva.eu). The latter is mapped as Urban may increase the runoff coefficient significantly. This Morphology Type (UMT), which is a powerful tool for is because smooth surfaces like asphalt and concrete the representation of the built and natural environment. generally have very low infiltration capacity. This leads UMTs (Pauleit and Duhme 2000; Gill et al. 2008) form to larger flood discharge and shorter response times in the foundation of a classification scheme bringing urban areas. The land-use geo-spatial datasets can be together facets of urban form and function. UMTs have found in GIS-based formats such as shape files and raster characteristic physical features (i.e. land cover) and files. The resolution of these maps may vary between are further differentiated on the basis of the human 1:1000 to 1:100000. activities they accommodate (i.e., land uses). The procedure adopted to obtain UMT for the specific case of Dar es Salaam is described in Cavan et al. (2012). Figure 3.7 Land use for the larger Dar es Salaam region                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 23 Figure 3.8 Urban Morphology Types (2012) The use of two different resources for land use While there is a high level of urbanization in the city characterization was necessitated by the fact that the centre, beyond these areas the study area contains main catchment feeding the Msimbazi River extends areas of bushland, cultivated land, bare soil, forest and beyond the limit of the more refined UMT map which savannas. It is also worth noting that for a large part of is confined to the political borders of Dar es Salaam. the area identified as urbanized, the absence of road Therefore, a mixed resource was created, with the less paving and the lack of proper drainage systems lead to refined data being used for the area outside the city low water run-off towards the main channels. boundary.                                                  Page 24 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Flood hazard assessment The data adopted for constructing the IDF curves In this section, the methodology employed for the are those presented in Box 3.1, and the calibrated hazard assessment is described, with some of the coefficients are those presented in De Paola et al. (2014) technical detail in boxes. A schematic diagram of the for the specific case of Dar es Salaam. The following procedure used for hazard modeling is illustrated in generic procedure for constructing IDF curves was used. Figure 3.9. In summary, intensity-duration-frequency The IDF curve characterizes an area’s rainfall pattern. By (IDF) curves, geologic and land-use information are used analyzing past and projected rainfall events, statistics of to characterize the hydrograph leading to the calculation the recurrence of rainfall extremes can be determined of the discharge (Q) and the total water volume (i.e. the for various return periods (TR) (e.g. 5, 10, 30, 50, 100 and area under the hydrograph) for different return periods. 300 years). The annual extremes are usually obtained This information, together with the topographic map as maximum rainfall over a one-year time period, as of the zone of interest are used in a two-dimensional extreme rainfall height values hr (in mm) calculated diffusion model in order to generate the maps of over time window duration d. The rainfall curve, maximum water height and velocity for each node of a corresponding to a specific return period, is calculated lattice covering the zone of interest for a given return by fitting a suitable probability model to the extreme period (the flood hazard map). rainfall data. Estimation of IDF curves is explained in more detail in Box 3.1. 3.0.1 The rainfall curve The rainfall probability curves for Dar es Salaam based Rainfall curves or intensity-duration-frequency (IDF) on historical data are shown in Figure 3.10. curves are described for various return periods. Rainfall curves are normally used where there is a lack of sufficient flow data for direct probabilistic discharge modeling, in order to evaluate the peak discharge. The IDF curve presents the probability of a given rainfall intensity and duration expected to occur at a particular location. Figure 3.9 Hazard assessment procedure (h = rainfall height, d = duration, Q = discharge, and TR = return period; De Risi et al. 2013a)a                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 25 Box 3.1. Calculation of intensity-duration-frequency (IDF) curves                                                  Page 26 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Box 3.1. Calculation of intensity-duration-frequency (IDF) curves (continued) Figure 3.10 Rainfall Probability Curves for Dar es Salaam based on historical data                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 27 3.0.2 Hydrologic basin modelling The aim of the hydrologic basin modelling is to produce a series of hydrographs (the discharge [volume/time] plotted against time), for different return-period flood events. A systematic description of the procedure is reported below. 3.3.2.1 Identification of catchments Catchment area characterization, which is done based on the topography of the zone, is one of the first steps in hydrographic basin analysis. The catchment refers to the topographical area from which a watercourse, or a water course section, receives surface water from rainfall (and/or melting snow or ice). For the analysis domain within Dar es Salaam, the river representing the greatest threat is the Msimbazi River. Three catchments feed the Msimbazi (Figure 3.11): 1. The northern catchment (Catchment 3 in Figure 3.11) is drained by the Mto Ng’ombe water course that crosses the Kinondoni district, covering 9755 meters between the drainage outlets up to the junction with Figure 3.11 Catchments of the case study river the Msimbazi. It crosses the Ubungo; Manzese; Tandale; Makumbusho; Ndugumbi; Magomeni and Upanga Magharibi sub-wards. In Upanga Magharibi 3.3.2.2 Estimation of the design flood hydrographs sub-ward, the Mto Ng’ombe water course joins the Once the IDF curve has been characterized, a rainfall- Msimbazi River. runoff method must be applied in order to determine 2. The central catchment (Catchment 2 in Figure 3.11) is the hydrograph. The hydrograph refers to the flow drained by the Kibangu water course that crosses the discharge in the drainage outlet of the catchment as Kinondoni and Ilala districts, covering 3363 meters a function of time and constitutes the input for the from the drainage outlet up to the junction with the hydraulic diffusion model (Figure 3.12). The area under Msimbazi. It crosses the Kigongo (Kinondoni); and the hydrograph is equal to the total discharge volume for Mchikichini (Ilala) subwards. In the Mchikichini sub- the basin under study for a given rainfall event. ward, Kibangu water course joins the Msimbazi River. Different approaches can be used to estimate the 3. The southern catchment (Catchment 1 in Figure 3.11) peak flow for a specified return-period event (design is drained by the Msimbazi River that crosses the peak flow), such as the Rational Method, the Curve entire Ilala district covering 12248 meters from the Number Method or more complete approaches such drainage outlet to the ocean. It crosses the Kipawa; as distributed models (e.g. TRibs, available at http:// Vingunguti; Buguruni sub wards (Ilala); and the vivoni.asu.edu/tribs.html) or semi-distributed models Mchikichini; Jangwani; Upanga Magharibi and Upanga (e.g. Hec-HMS, available at http://www.hec.usace.army. Mashariki sub wards (Kivukoni). mil/software/hec-hms/). A detailed discussion about distributed and semi-distributed models is presented The drainage outlets (the black dots in Figure 3.11) in in El‐Nasr et al. (2005). The more sophisticated the the analysis domain are upstream of the main built method, the more data is required for the calibration environment. The morphometric characteristics of of the model’s parameters. In developing countries, the the three catchments of these three points of flow are choice of model is typically constrained by the absence described in Table 3.2. The characteristics presented in of data. Therefore it is generally necessary to employ Table 3.2 were used to calculate the concentration time relatively simple tools. TC for each catchment (see Box 3.2).                                                  Page 28 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Table 3.2 The characteristics of the Msimbazi River catchments Characteristics Msimbazi River Catchment 1 Catchment 2 Catchment 3 Ab Drainage area (km2) 172.6 51.0 11.8 L a Main channel length (km) 25 15 7 Average catchment slope (%) 12.3 8.5 9.0 Average channel slope (%) 7.0 4.0 4.0 zm Maximum height (m.a.s.l.) 347 305 160 Average height (m.a.s.l.) 145 83 103 z0 Drainage outlet height (m.a.s.l.) 12.8 16 50 Box 3.2. Concentration time and its estimation                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 29 Figure 3.12 The schematic diagram of a hydrographic basin For ungauged basins--where water runoff cannot be However, the method has some limitations (Mokus directly measured, such as is the case in Dar es Salaam, 1972; SCS 1972): the classic Curve Number Method (CNM) (Mokus 1972; SCS 1972) is considered suitable for modeling the ƒƒ Rainfall is considered spatially uniform; hydrograph for the following reasons: ƒƒ It does not contain any expression for time and ƒƒ It is the simplest conceptual method for estimation of ignores the impact of rainfall intensity and its the direct runoff amount from a particular storm temporal distribution; rainfall amount, and is well supported by empirical data, ƒƒ There is a lack of clear guidance on how to vary antecedent moisture condition, especially for lower ƒƒ The method relies on only one parameter, the curve curve numbers and rainfall amount; and number (CN), which is a function of the major runoff producing watershed characteristics, ƒƒ The method was originally developed for agricultural areas, and while it performs well also in other ƒƒ It is fairly well documented for its inputs (soil, land contexts, it performs poorly in forest areas. use/treatment, surface condition, and antecedent moisture condition); and Nevertheless, the method is considered acceptable for the examined case in Dar es Salaam. The Curve ƒƒ Its features are readily grasped, well established, and Number Method and assumptions for Dar es Salaam are accepted not only in the US (i.e. the country for which explained in more detail in Box 3.3. the methodology was developed) but also other countries around the world.                                                  Page 30 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Box 3.3. Estimation of design flood hydrographs using the Curve Number Method                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 31 Box 3.3. Estimation of design flood hydrographs using the Curve Number Method (continued)                                                  Page 32 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Box 3.3. Estimation of design flood hydrographs using the Curve Number Method (continued)                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 33 Box 3.3. Estimation of design flood hydrographs using the Curve Number Method (continued)                                                  Page 34 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Figure 3.13 shows the hydrographs obtained for all The characterization of the discharge hydrograph three catchments analyzed, considering all the potential formed the basis for the identification of flood prone conditions in terms of AMCs and for all six return areas. In the next step, the flood discharge estimated periods. The peak discharges increase passing from by the hydrograph was propagated through the zone of AMCI to AMCIII and the durations decrease passing from interest in order to delineate the flood prone areas for Catchment 1 to Catchment 3. various return periods. Figure 3.13 The hydrographs for the three catchments, for the three AMCs and for all the considered return periods. Note y-axis scales differ among panels                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 35 3.3.2.3 The two-dimensional propagation model The simulation duration for each sub-domain is equal to In this work, flood routing in two dimensions was the time necessary to exhaust the related hydrograph. accomplished through the numerical integration of The results of the hydraulic routine are presented in the equations of motion and continuity (dynamic Appendix 1. The most important observations validating wave momentum equation) for the flow. This was the reliability of the simulations are: accomplished by means of the commercial software ƒƒ Inundation depth and flow velocity values increased FLO-2D (O’Brien et al. 1993, FLO-2D 2004) which is a for increasing return period; flood volume conservation model based on general constitutive fluid equations of continuity and flood ƒƒ Maximum depth and flow velocity were generally dynamics, i.e. shallow water equations or Saint-Venant reached in proximity of intersections with equations. The flow is considered variable in space infrastructure (e.g. road or railway embankments), and in time, and the bottom friction is evaluated using especially where reductions of floodable section Manning’s formula. The Manning’s coefficients were occurred; assigned to computational cells based on a literature review. Conservative estimates of 0.04 and 0.02 were ƒƒ All the infrastructure features, well represented in the assumed for natural and for urban areas, respectively. adopted DEM, created overflowing upstream; dam- Drainage features not already incorporated in the DEM effects can be observed first in correspondence of the (e.g. the sewage system) were omitted due the lack of railway embankment of the Nelson Mandela available data on these systems. Expressway and then in correspondence of Morogoro road embankment. A two-dimensional flood simulation is based on a digital elevation model (DEM) overlaid with the surface grid, ƒƒ The final part of the Msimbazi River (i.e. near its aerial photography and orthographic photos, detailed mouth) had its flow well confined in the flood plain topographic maps and digitized mapping. Detailed between the more external stems, even if the flow cartography is needed in order to identify the surface was very high for the contribution of the other two attributes of the grid system; for example, streets, catchments. buildings, bridges, culverts or other flood routing or storage structures. The principal advantages in using a A large part of the built environment in the case study two-dimensional diffusion model is that it can be applied area is affected by the flood events in all the analyzed sub- in special cases such as, unconfined or tributary flow, domains. More detailed descriptions of the exposure and very flat topography and split flow. of the affected assets are given in Section 3.4. In this study, the 2 m resolution DEM presented in Section 3.2.2 was adopted; a computational grid of 10 3.0.3 The hazard curves m resolution was defined within the single domain for Flooding hazard curves were associated with each the simulation code, returning an output having the building in the inundation area. Such curves represent same resolution of the initial DEM. Such hypothesis the mean annual frequency of exceeding (equal to the on the computational resolution allow reasonable inverse of return period for a homogenous Poisson computational time. The buildings in the area were process) a given flood height for a given point within treated in the simulations as obstacles to water the case-study area (e.g. one of the corner of a given accumulation/flow. building), based on the input grid data set described In order to optimize the computational time, the previously. This is done by a spatial interpolation analysis domain was divided in five sub-domains. The between the point (e.g. the corner of the building) and first, second and third sub-domains were related to the the flood height values at the vertex of the lattice grid three main water courses; the last two sub-domains containing the point in question. are defined below the junctions of the previous main For each building, the interpolation with the hazard was water courses. In each drainage outlet the related repeated for all the building’s corners. Then, only the hydrograph was applied; in each junction a hydrograph maximum value of inundation depth was considered was obtained as the sum of the flows from the upstream for each return period. Figure 3.15 below shows the water courses. Figure 3.14 shows the five sub-domains schematic representations of the adopted procedure considered and the junction points. and the final output.                                                  Page 36 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Figure 3.14 Analysis sub-domains Figure 3.15 Schematic representation of the procedure and of the expected output in terms of flood hazard curves. In this example only the red hazard curve will be associated to the analyzed building.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 37 The exposure and vulnerability models The analysis domain and the sub-domains cover portions of all three municipalities in Dar es Salaam, as shown in Figure 3.16, and represent one of the most populated areas in the city. The analysis domain covers the sub- wards of Ilala, Kinondoni, and Temeke. The study area was chosen because it contains the low-lying densely populated areas that have experienced recurrent flood damages in the past, often annually or sub-annually. According to the results presented in Section 3.2.3 and represented in Figure 3.8, it is also possible to observe that the analysis domain covers a big part of the city center in which residential areas and economic activities are concentrated. In this study, the buildings potentially subject to the flood hazard were investigated in detail, and were used as main elements of the exposed asset. Appendix 2 contains a literature survey and presents the estimation of the current building value in Dar es Salaam used in this analysis. Building-by-building identification benefiting from Volunteered Geographic Information (VGI), in particular OpenStreetMap (OSM, www.openstreetmap.org), one form of big data, was used. Such a resource is often very useful in developing countries where there is a lack of Figure 3.16 Geographical localization of the analyses domain. data on exposed assets and it represents an alternative source for detailed spatial building information. The breakdown of the building characteristics based on By intersecting the Urban Morphology Type map available information is reported in Appendix 3. Sixty- with the building footprints it was possible to identify two potential combinations of available characteristics the typology of each building. Figure 3.17 shows the were recognized. Such combinations led to the buildings in the analysis domain; the total number of identification of three main structural types: informal buildings is 209124; and the distribution of building masonry (IM), formal masonry (FM), and reinforced footprints according to their typology. For the buildings concrete frame (RCF). The sixty-two combinations falling in areas of mixed types, we used the additional of characteristics were also analyzed in order to details in the OpenStreetMap shapefiles. estimate an average unit cost using the data reported in Appendix 2. The criteria adopted for identifying the The buildings at risk in the analysis domain were structural typology and the unit cost is documented in identified by intersecting the map of all buildings with Appendix 4. Figure 3.18 (b) shows the distribution of the maximum extent of the baseline flood inundation. A building structural typologies according to the adopted total of 12,744 buildings fell within this area. Figure 3.18 distinction criterion: 89.5% of the buildings analyzed (a) shows the footprints of the buildings at risk. were characterized as IM, 8.8% as FM, and 1.7% as RCF.                                                  Page 38 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Figure 3.17 Building footprints in the analyses domain distinguished by typology Figure 3.18 (a) Buildings at risk, and (b) buildings at risk distinguished by structural typology.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 39 3.0.1 The vulnerability model The vulnerability of the structure was described Once the main structural typologies were identified, using fragility functions. A fragility function evaluates it was necessary to assign a vulnerability model to the probability of reaching or exceeding specific each. According to Durra & Tingsanchali (2003) the damage states for a given hazard intensity (in this case damages caused by floods can be classified in two inundation depth; Porter et al. 2007). Such relationships categories: qualitative and quantitative. The qualitative between hazard and potential damage play a vital role damages include human conditions like anxiety, mental in quantifying damage and losses associated with flood suffering or the inconvenience associated the loss events. In this study, the onset of damage was defined of possessions or with resettlement. On the other as the structure being 50% affected relative to the onset hand, the quantitative damages can be measured in of collapse. economic terms in a more straightforward manner. The fragility curves provide a visual and efficient way of The latter can be subdivided into indirect and direct representing the structural vulnerability. One of their damages. The indirect damages can be quantified --in characteristics is that they correspond to a specific economic terms -- as a function of downtime and its structural limit state. Two limit states were considered consequent disruption of economic and social activities. in this study: a damage limit state (DLS); and the The direct damages are due to the interaction between collapse limit state (CLS). The methods and assumptions the water flow and the physical system (i.e. structure, for estimating the curves are described in Box 3.4. infrastructure, green space, etc.). In this study we focus Figure 3.19 shows the fragility functions adopted in this on direct damages. It is important to note that while this study. represents the most appropriate economic approach to measuring the damages, it may overstate the damages to commercial or industrial businesses, as a result of higher unit costs, and understate the damages to residential unplanned dwellings that are owned with the most vulnerable in society. Figure 3.19 Fragility curves for the three considered structural typologies and two potential limit states                                                  Page 40 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Box 3.4. Estimation of the fragility functions                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 41 Figure 3.20 Hazard curves for the baseline scenario; λ values ≤ 0.2 based on inundation records; values ≥ 0.2 based on extrapolation 3.0.2 The hazard curves for the buildings at risk For all the buildings at risk represented in Figure 3.20 (a) vulnerability assessment taking into account the many the procedure explained in Section 3.3.3 was applied to sources of uncertainties. Nadal et al. (2010) propose generate the respective flood hazard curves. Figure 3.23 a stochastic method for assessing the direct impact shows the hazard curves for the 12 744 buildings at risk. of flooding on buildings. A general methodological approach to flood risk assessment is embedded in the HAZUS procedures for risk assessment (Scawthorn et al. 3.0.3 Risk assessment 2006a, 2006b). Apel et al. (2009) comprehensively assess In recent years, increasing attention has focused the various scales of complexity and precision involved on riverine flood risk assessment. In fact, several in flood risk assessment. publications discuss the consequences of flooding, such as loss of life (Jonkman et al. 2008a), economic Here, following De Risi et al. (2013a), we summarize losses (Pistrika & Tsakiris 2007, Pistrika 2010, Pistrika & flood risk assessment using a single equation (Eq. 3.25), Jonkman 2010) and damage to buildings (Smith 1994, where λ LS denotes the risk expressed as the mean annual Kang et al. 2005, Schwarz & Maiwald 2008, Chang et rate of exceedance of a given limit state (LS). The limit al. 2009). These research efforts have many aspects in state refers to a threshold (e.g. critical water height common, such as a direct link between flood intensity hf,c) for a structure beyond which it no longer fulfills a and duration and the incurred damage, and that they specified functionality. λ(hf) denotes the mean annual are based on real damages observed in the aftermath rate of exceedance of a given flooding height hf at a of flooding events. On the other hand, many research given point in the considered area. P(LS|hf) denotes the efforts are starting to galvanize in the direction of flooding fragility for limit state LS expressed in terms of proposing analytical models for flood hazard and the probability of exceeding the limit state threshold.                                                  Page 42 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Indirect losses are not caused by the disaster itself but rather by its secondary effects. For example, if flooding damages infrastructure (e.g. transportation or utility networks), it often causes business interruptions The risk λ LS is calculated in terms of the mean annual that continue far beyond the duration of the actual frequency of exceeding the limit state LS for each node flooding itself. Likewise, because of economic linkages of the lattice covering the zone of interest by integrating among businesses and economic sectors, flooding may fragility P(LS|hf) and the (absolute value of) hazard cause indirect losses in the form of negative effects increment |dλ(hf)| over all possible values of flooding on economic activity outside of the immediate flood height. The mean annual frequency of exceeding the footprint. limit state λ LS is then transformed into the annual probability of exceeding the limit state assuming a Our analysis only considers direct market losses from homogenous Poisson process as a model for occurrence flooding in the form of damages to buildings. The of limit-state-inducing events, according to Eq. 3.26. expected loss is calculated as the expected repair costs 4 (per building or per unit residential area), E[R] as a function of the limit state probabilities and by defining the damage state i as the structural state between limit states i and i+1: where t is the time in years. It should be noted that Eq. (6.1) manages to divide the flood risk assessment procedure into two main modules, namely, the hazard assessment module which leads to the calculation of the mean annual frequency λ(hf) of exceeding a given flooding height hf, and the vulnerability assessment where NLS is the number of limit states that are used in the module used to calculate the flooding fragility curve in problem in order to discretize the structural damage; Ri is terms of the probability of exceeding a specified limit the repair cost corresponding to damage state i; and state P(LS|hf). . In this study, we set the repair cost associated with the CLS to 100% of the value of the total exposed asset, and the repair costs associated 3.0.4 The expected annual losses (EAL) with the DLS to 50% of that value. Moreover, in the case of collapse, a further cost of 10% of the entire asset is From an economic perspective, the costs of flooding and assumed for the dismantling of the collapsed building other natural disasters can be broadly categorized into and removal of debris. The expected annual loss (EAL) market versus non-market and direct versus indirect then is obtained using probability terms in equation 3.27 losses (Hallegatte & Przyluski 2010). Direct market calculated considering a time window of one year in losses are negative impacts of the disaster on goods and equation 3.26. Computation of EAL is explained in more services commonly bought and sold and whose value detail in Box 3.5. therefore generally can be fairly accurately determined using readily-available, directly-observable price data (e.g. costs of infrastructure repair or medical treatment). Direct non-market losses are costs that are caused by the disaster but whose economic value cannot be readily quantified because they are not themselves traded on markets (e.g. suffering caused by injury or by death of family members or friends; loss of life; loss of amenity). While economic valuation of direct, non-market impacts is possible (e.g. the cost of suffering from specific health effects) and in many cases even fairly common (e.g. the value of a statistical life or of disability-adjusted life years), the resulting values often are contentious because they rely on indirect valuation approaches that utilize people’s stated rather than observed preferences, or because many individuals are uncomfortable with assigning monetary values to these impacts.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 43 Box 3.5. Expected annual loss                                                  Page 44 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT IV. EVALUATION AND SELECTION OF POTENTIAL URBAN STORMWATER MANAGEMENT OPTIONS Overview of stormwater management countries such as Australia, New Zealand, United States, Within urban areas, rain that falls onto impermeable England and Scotland to describe this new approach surfaces such as streets, parking lots, pavements and to urban drainage and are all essentially synonymous. roofs is unable to filtrate into the ground as it would “Green infrastructure” is another term commonly used normally do in undeveloped areas, leading to higher levels in this context, and tends to refer to any environmentally- of surface runoff during storm events than would have friendly stormwater management structures, natural and happened naturally and creating flooding problems in semi-natural systems used in stormwater management. downstream areas (Armitage et al. 2013). This problem Storm water management measures now tend to be is exacerbated by the fact that urban stormwater runoff designed to address both flooding and water quality generally contains litter, debris, and sediments which lead problems, with many measures addressing both of these. to blockages of the systems designed to convey water, and they also contain bacteria, heavy metals and nutrients, Types of stormwater management measures which means that floodwaters can become a pollution hazard. All of this can have negative impacts on property, Stormwater management measures are classified into urban infrastructure and natural habitats, as well as urban structural and non-structural measures, with structural inhabitants. With an increase in urbanisation worldwide measures being further subdivided into passive and active and the associated impact of increasing stormwater measures (Figure 4.1; see Appendix 5 for details). runoff on aquatic ecosystems, the management of urban drainage has become a critically important challenge ƒƒ Passive structural measures aim to convey water and (Fletcher et al. 2015). protect areas from flooding. Examples are levees, increasing the channel capacity by clearing of debris or Urban drainage management has changed significantly increasing its cross-section, and constructing hydraulic over the last few decades, from a conventional ‘rapid bypasses (waterways) to divert high flows. disposal’ approach to a more integrated and sustainable ‘design with nature’ approach (Fletcher et al. 2015). ƒƒ Active structural measures aim to modify the The early traditional attitudes towards urban drainage hydrograph (i.e. reduce flood peak and volume) and management focused on trying to dispose of stormwater address water quality by retarding water movement, in the fastest way possible with not much consideration by increasing infiltration or storage in the catchment for surrounding ecosystems or for downstream water area (Topa et al. 2014). These can be referred to as quality impacts. In the 1980s and 1990s a new focus on “green” (sustainable/environmentally-friendly) urban stormwater runoff and water quality developed engineering measures. around the world, which concentrated on a more catchment-wide management and restoration approach ƒƒ Non-structural measures do not involve physical to urban drainage as opposed to the standard end-of- construction but use knowledge, practice or catchment solution (Fletcher et al. 2015). This embodied agreements to reduce risks and impacts through a more holistic approach which treats stormwater behavioral changes, in particular through policies and runoff problems at source and minimises environmental laws, public awareness raising, training and education degradation, while delivering environmental, economic (Kundzewicz 2002). These include flood warning and social benefits. This rapid development in the field systems, land use regulations such as development of urban drainage saw a number of terms being used setbacks which identify where development can and to define similar concepts (Fletcher et al. 2015). Terms cannot occur, or to what elevation structures should such as “Water Sensitive Urban Design” (WSUD), locate their lowest habitable floor to; flood proofing “Low Impact Development” (LID), “Sustainable Urban and retrofitting of buildings may increase the strength Drainage Systems” (SUDS), “Integrated Urban Water against flood actions; elevation of buildings may avoid Management” (IUWM) and “Best Management completely the inundation. Flood insurance and Practices” (BMPs) were first used by professionals in relocations also belong to this typology of measures.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 45 Figure 4.1 Different types of measures used in stormwater management. These measures are described in Appendix 5. Source: This study The active structures or “green” engineering measures Relative performance of different measures tend to be grouped as either source, local or regional Generally a combination of these measures would controls (Thampapillai & Musgrave 1985, Kundzewicz be applied. There will be trade-offs among the 2002, Armitage et al. 2013; Figure 4.1, see Appendix 5 interventions and finding the correct balance can be for details): a complex task. The active structural measures, such as floodplain storage interventions will reduce the ƒƒ Source controls tend to be used to manage need for extensive passive measures such as levees or stormwater runoff as close to the source as possible, the widening of channels. However, active measures generally within the boundaries of the property such alone often will not be able to eliminate flooding as green roofs, soakaways and permeable paving. problems completely and thus generally will need to ƒƒ Local controls are usually used to manage runoff as a be implemented in conjunction with some conveyance second line of defence typically in public areas, along infrastructure. Generally, the ‘softer’ the intervention roadways and adjacent to parks such as filter strips the less efficient it tends to be in terms of m3 reduction and swales. per unit of space. However, the softer interventions tend to have greater benefits in terms of amenity and ƒƒ Regional controls are used to manage runoff as the social value. Therefore it is necessary to develop a sound last line of defence and are generally large-scale methodology for evaluating the interventions based on interventions constructed on municipal land such as cost-effectiveness, efficiency in removing peak flows, detention ponds and wetlands (Armitage et al. 2013). reducing runoff volume and water quality amelioration, and providing amenity, conservation and social benefits. Measures that retard flows generally also contribute These are discussed in more detail on the following towards improving water quality, and vegetated areas page. further contribute to water quality amelioration where flows are slow enough. The various structural measures are described in more detail in Appendix 5.                                                  Page 46 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT When developing a proposed plan for managing 4.0.1 Average cost effectiveness in terms of peak flow and stormwater it is important to link together the various volume reduction interventions and the benefits that they provide with Cost-effectiveness in terms of runoff reduction ($/ the greatest possible efficiency. That is, identify the m3) is shown in Figure 4.2. These estimates are based combination of interventions for a specific project site on reviews and examples from the stormwater that will achieve the outlined objective in the most cost- management literature (Joksimovic & Alam 2014, Liu effective way, i.e. results are achieved at a lower cost et al. 2015, Jiang et al. 2015, Committee for Climate compared to alternatives. This involves determining the Change 2012, Xiao & McPherson 2002, McPherson et cost effectiveness of different interventions, i.e. a cost al. 1999). From the examples it is clear that green roofs per unit reduction of runoff volume (m3) or cost per unit and permeable pavements are the least cost-effective, reduction in pollutant loads (kg), depending on what even though they are efficient at reducing runoff, due the proposed project is trying to achieve. Outside of to their higher capital and maintenance costs compared cost-effectiveness, interventions need to be assessed in to other options. Soakaways and infiltration trenches terms of any other benefits that they may provide, such are the most cost-effective of the structural engineering as conservation value, amenity value or social benefits methods as they are cheaper to construct and maintain. such as increased water supply. The interventions need Constructed wetlands, sand filters, bioretention areas, to be realistic in terms of what is feasible and practical detention basins, filter strips and swales generally are all within the designated project area. While this study was relatively cost-effective. They are however less efficient focused on the flood amelioration aspect, it is worth at trapping or attenuating peak flows after a large storm. considering the water quality impacts at the same time, Riparian buffers and catchment reforestation represent since these go hand in hand, particularly if the required the most cost-effective option in that they do contribute sanitation and sewage systems are in place. significantly to rainwater infiltration, but they also do not trap or attenuate peak flows (which is not captured in Numerous studies have examined the relative ability the $/m3 assessment). The structural engineering options of different interventions to reduce pollutant loads, are more efficient in this regard. flow volumes and attenuate peak flows during storm events, and their cost-effectiveness. Estimates for cost- effectiveness in terms of cost per unit runoff reduction 4.0.2 Average cost effectiveness in terms of water quality ($/m3) and cost per unit reduction in pollutant loads amelioration ($/kg) were collated from a wide range of stormwater A number of interventions are designed to specifically management literature. These estimates tend to be site control and improve the quality of stormwater runoff. specific based on local costs but nonetheless provide Generally their performance is assessed based on their a clear indication of which interventions are generally pollutant removal capabilities. Some interventions more cost-effective in terms of their ability to reduce may be more efficient at removing suspended runoff and remove pollutants. solids and hydrocarbons, whereas others may be particularly efficient at removing soluble nutrients such as phosphorus. Often this means that a number of interventions are required to achieve a specified outcome. The capacity for pollutant removal of different interventions is summarised in Table 4.1. Figure 4.2 Comparison of cost per unit volume of runoff reduction for various stormwater management options, based on literature averages                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 47 Table 4.1 Measured pollutant removal capacities of selected stormwater management options and technologies Pollutant Removal (%) Option/Technology Hydro- TSS TP TN Faecal Coliforms Heavy Metals carbons Source Controls Green roofs 60-95 - - - - 60-90 Sand filters 80-90 50-80 50-80 25-40 40-50 50-80 Underground sand filters 75-90 - 30-60 30-50 40-70 40-80 Surface sand filters 80-90 - 50-60 30-40 - - Filter drains 50-85 30-70 - - - 50-80 Soakaways 70-80 - 60-80 25-60 60-90 60-90 Oil and grit separators 0-40 40-90 0-5 0-5 - - Modular geocellular structures PS PS PS PS PS PS Stormwater collection and reuse PS PS PS PS PS PS Local controls Bioretention areas 50-80 50-80 50-60 40-50 - 50-90 Filter strips 50-85 70-90 10-20 10-20 - 25-40 Infiltration trenches 70-80 - 60-80 25-60 60-90 60-90 Permeable pavements 60-95 70-90 50-80 65-80 - 60-95 Swales 60-90 70-90 25-80 30-90 - 40-90 Enhanced dry swales 70-90 70-90 30-80 50-90 - 80-90 Wet swales 60-80 70-90 25-35 30-40 - 40-70 Vegetated buffers* 50-85 70-90 10-20 10-20 - 25-40 Regional controls Constructed wetlands 80-90 50-80 30-40 30-60 50-70 50-60 Extended detention shallow wetland 60-70 - 30-40 50-60 - - Pocket wetland* 80-90 50-80 30-40 30-60 50-70 50-60 Submerged gravel wetland 80-90 - 60-70 55-60 - 85-90 Detention ponds* 45-90 30-60 20-70 20-60 50-70 40-90 Extended detention ponds 65-90 30-60 20-50 20-30 50-70 40-90 Infiltration basins 45-75 - 60-70 55-60 - 85-90 Retention ponds 75-90 30-60 30-50 30-50 50-70 50-80 *Estimated values based on similar stormwater technologies TSS – Total Suspended Soilds, TP = Total Phosphorous, TN = Total Nitrogen Source: Armitage et al. 2013                                                  Page 48 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Figure 4.3 Comparison of cost per unit mass of pollutant/nutrient reduction for various stormwater management options Detailed information provided in Armitage et al. 4.0.3 Overall effectiveness, cost-effectiveness and potential (2013) (Table 4.1) was used to assess the water quality co-benefits amelioration performance of the various interventions. The relative effectiveness of different interventions in These data as well as data from Centre for Watershed terms of flood and water quality amelioration, their Protection (2013) were combined with cost data (see cost-effectiveness, and other potential benefits are Appendix 6) to estimate relative cost-effectiveness for a summarised in qualitative terms in Table 4.2. This selected range of interventions (Figure 4.3). suggests that while conveyance measures are highly effective for reducing flood exposure/risk, they make The most cost-effective options in terms of TSS removal little contribution to water quality amelioration, they are filter strips, swales and detention basins. Riparian vary in terms of cost-effectiveness and have relatively buffers and catchment reforestation are also cost- little in the way of co-benefits. Indeed, they are more effective options. In terms of TP and TN removal, likely to lead to negative externalities such as damage riparian buffers and catchment reforestation are very to aquatic ecosystems or acerbation of flooding further cost-effective as are swales, filter strips and sand filters. downstream. Of the conveyance measures, detention Constructed wetlands and bioretention areas are less basins are potentially beneficial in terms of providing cost-effective for all three but they do have higher opportunities for amenity, such as sunken sports fields. amenity values when compared to the other options. The “green” engineering measures are generally less efficient in flood protection, but are important for water quality. Effective flood protection will require these measures to be implemented in combination and/or at scale. Green engineering measures also vary in their cost-effectiveness and may not always compete with conveyance measures. They do however, also present much greater opportunities for delivering co-benefits, such as water supply and the provision of recreational areas. The latter is particularly the case for the vegetated options which have greater aesthetic appeal.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 49 Table 4.2 Relative merits (indicated by number of “X”) of different measures for stormwater and flood risk management, based on the literature. Measures considered in this study area are marked with an asterisk. Option/ technology Conveyance/ Flood Water quality Cost- Water Amenity Conservation Reduction of attenuation/ amelioration effectiveness supply potential value exposure Reduction of flood risk Conveyance measures (lower catchment) Swales/drains* XX XX Channel enlargement/ canalisation/levees XXX X Hydraulic bypass XXX X ‘Green’ engineering measures (mid-upper catchment) Infiltration trenches XXX XXX XXX X Soakaways XXX XX XX X Permeable pavements XXX XXX XX X Rainwater harvesting X X X XXX Bio-retention areas XX XXX XX XX Sand filters XX XXX XX Green roofs XX XXX X XX Filter strips XX XXX XXX X Vegetated swales XX XXX XXX X Constructed wetlands XX XXX XX XXX X Detention basins* XXX XX XXX X XXX Non-structural measures Development X XXX XX X setbacks* Conventional solid XX X XX XX X waste management River cleaning X X XXX XX XX programmes* Protection/restoration of catchment forests XX XXX XXX XX XX XXX and wetlands* Protection/restoration of riparian areas, X XX XXX XXX XXX XXX floodplains*                                                  Page 50 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Figure 4.4 Heavy development around the lower floodplain areas of the Msimbazi River system in Dar es Salaam Source: OpenStreetMap The protection or restoration of natural systems in An important consideration is that in the case of Dar es catchment areas contributes to the reduction and Salaam, not only has the expansion of the city resulted in retardation of flows and to water quality amelioration. increased stormwater runoff and reduced water quality Within the flood prone areas, riparian buffers and coming from the catchment, but encroachment and functional floodplain areas reduce the exposure to reclamation of the floodplains in the lower catchment flooding, and further contribute to water quality has also occurred (Figure 4.4). Thus the flow for any amelioration. In all cases, these areas have the potential given rainfall event has increased, while the capacity to contribute significantly in terms of other co-benefits. of the floodplain to accommodate and convey this flow has been reduced. Both of these effects increase the exposure of people and infrastructure to flooding, so Selection of suitable measures for Dar es Salaam when they occur together they can create an enormous This study sought to find a suitable set of measures problem. In most developed country contexts the that could be implemented in combination to address emphasis has been on amelioration of the effects of flooding problems in Dar es Salaam, while also urbanisation in the catchment, for example by using contributing to a green urban development path measures to try and neutralise the effects of hardening for the city. As well as the relative merits discussed on runoff volumes. But in the majority of developing above, the limitations or requirements for the country cities, not least in Africa, the lack of control over different interventions were taken into account, along settlement patterns has led to much higher levels of with information on the catchment area, in order to encroachment of the original or natural floodplain areas, determine the most suitable types of interventions to let alone the expanded floodplain area required as a include in the analysis. Following this, the potential result of increased runoff. This means that there might extent of each intervention was determined based on still need to be heavy reliance on conveyance measures biophysical criteria. in conjunction with green structural and non-structural measures, or that catchment measures might have to do more than neutralise the effects of catchment urbanisation, reducing storm flows to lower than natural levels. Both of these options may raise costs considerably. In addition, it should be noted that other characteristics of urbanisation in developing countries also affect the relative suitability of different measures compared to developed country cities.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 51 This nature and pattern of development in both the soils. Soakaways, or sub-surface infiltration beds, were catchment and flood receiving areas of Dar es Salaam considered for industrial and commercial areas where severely constrains both the range of options that runoff from extensive roof surfaces could be collected could be considered to offset flooding problems and and slowly infiltrated through soakaway pits. These are their potential efficacy. An effective flood management relatively cost-effective in terms of runoff reduction as scheme would need to offset the effects of floodplain well as in terms of their ability to remove suspended infilling, channelization and canalisation in the lower solids. The amount of roof runoff that could be captured floodplain area, as well as offsetting the impacts of by soakaways during a large stormwater event in Dar urbanisation in the catchment area such as increased es Salaam could be significant. There are five large peak discharges and poor water quality. This means industrial areas in the catchment where soakaways that whatever system is found to meet the combined could possibly be implemented and based on the size of objectives of improving water quality and addressing these buildings it was estimated that collectively they flooding will not be based on mimicking natural could retain more than 30 000 m3 of runoff in a 50mm ecosystem function, but will of necessity require rainfall event. However, the soil in the area where these significant manipulation of flow regime in the upstream industrial sites are located is predominantly clayey, catchment and/or better conveyance from the flood making it unsuitable for implementation due to very prone areas. These limitations must be considered poor infiltration rates associated with clay-type soils. in planning remedial activities for Dar es Salaam, but The industrial areas are also located in the flood prone should also be brought into the planning of any future areas in the lower catchment, thus making these sites urban expansion. impractical and unsuitable for soakaways. There would also be little opportunity to implement measures such as For this study, each intervention was assessed on its green roofs in the catchment areas because of the fact limitations and requirements as well as its suitability that most of the buildings are informal structures. for application in the Msimbazi catchment in Dar es Salaam (Table 4.3). This study was carried out as a Rainwater harvesting is something that could benefit desktop study and as such the selection of suitable households in the catchment area. However, apart stormwater measures was based on available data, from being fairly expensive to implement, it would have expert opinion and local and international literature. limited flood benefits in Dar es Salaam. The rainwater As a result of this, potential administrative issues tanks would fill up quickly during a large storm event surrounding the implementation of these interventions thereafter making them ineffective in trying to dampen in Dar es Salaam have not been accounted for here the effects of increased runoff from the hardened and it is recognised that a more participatory approach surfaces. A much better option is to have large-capacity may be needed to refine the selection or extent of storage systems that release water, so that they are application of GUD measures. ready for the next high rainfall event. Indeed, of the green engineering options, detention basins were Conventional flood conveyance methods are not only considered to have the most promise as an option to expensive but would be difficult to establish in the explore further. study area because of the size of the floods that need to be contained. While dredging of the river channel Among the most feasible options were non-structural may help to mitigate some flooding, it is not seen as measures such as river cleaning programmes, solid a long term solution to the flooding problem in Dar es waste management and the protection, restoration Salaam. Very few of the “green” engineering measures and/or enhancement of natural systems. River cleaning were considered feasible options for the study area. programmes and solid waste management are likely to In some cases this was because of the low location in be very worthwhile considering the immense problem the catchment of the building structures they would created by solid waste and its contribution to flooding. be associated with, or because of the clayey nature of                                                  Page 52 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Table 4.3 Requirements for different stormwater management measures (apart from financial), and implications /suitability in Dar es Salaam Option/ technology Limitations/requirements Potential for Dar es Salaam Included in for implementation this study Conveyance measures (lower catchment) Swales/Drains High – enough space to - Yes implement in flood risk areas Channel modification, Levees Enough space to accommodate the Low – extremely large channel engineering measures and straightening; and/or high levees required to potential damage to ecosystems; cost of contain floods - unrealistic resettlement is a potential limitation Hydraulic bypass Low – requires drainage from Enough space to accommodate the catchment area; altering drainage engineering measures; potential damage path may only shift problem; to ecosystems; cost of resettlement is a negative impacts on lower river potential limitation system ‘Green’ engineering measures (mid-upper catchment) Permeable pavements Not suitable in areas that experience Low-Med – Paving is low priority heavy traffic at present. Infiltration trenches Low gradient, permeable soils. Low – clayey soils Soakaways Would not be combined with other rooftop collection measures such as Low – clayey soils; suitable rainwater harvesting or green roofs. buildings mainly in lower areas Needs to be in catchment area. Green roofs Only on well-constructed, solid buildings, Low – most solid structures are as very heavy low in the Rainwater harvesting Requires adequate space between High – but mainly for water houses. Would not be combined with supply; limited flood benefit other rooftop collection measures Vegetated swales Not suitable for high pedestrian areas; Low – high pedestrian activity hard to retrofit in developed urban areas Filter strips Not suitable for high pedestrian areas Low – high pedestrian activity Sand filters Medium to low gradient. Medium – but little flood benefit Bio-retention areas Low gradient, permeable soils. Low – clayey soils Detention basins Large space requirements High – space exists Yes Constructed wetlands Medium – but lower priority than Large space requirements restoration of existing wetlands/ floodplains Non-structural measures Conventional solid waste Good governance, traffic mobility Low – many obstacles management River cleaning programme High – high rate of Good project management Yes unemployment Protection/restoration of Undeveloped, managed land; cost of High – large areas of degraded Yes catchment forests and wetlands resettlement is a potential limitation forest in upper catchment Protection/restoration of Undeveloped riparian land; restorable High – supported by law; riparian areas, floodplains former/degraded wetland or floodplain significant areas of floodplain Yes areas; cost of resettlement is a potential have been ‘cut off’ or degraded in limitation mid to lower catchment.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 53 There are also substantial areas of degraded forest in the the Sinza River. The channels would probably need to catchment that could be restored, and floodplains lower be relatively small - around 1 x 1 m, as the residential in the catchment have been artificially disconnected areas are densely populated and space is limited. They from the river, greatly reducing their potential benefits. would be graded and sized to convey up to a 1:5 year Furthermore, there are a number of floodplain areas return interval flood, thus decreasing the frequency of in the mid-lower catchment that could be enhanced to flood damage/inconvenience in low-lying areas, which improve their water holding capacity at the same time would remain vulnerable during larger events. The level as providing other benefits such as erosion control and of the base of the channels would need to be such that provision of areas for agriculture and wetland that could they were able to discharge above the 1:5 year flood also enhance water quality. level in the river – this could limit the areas in which this measure is applicable. That is, in low-lying areas, It interesting to note that the most feasible options are drainage may already be below the 1:5 year flood level largely those to do with the protection or restoration of natural capital, rather than engineering measures, even though many of the latter would be “green” 4.5.1.2 Extent and cost (environmentally friendly) options. The measures It is estimated that in these residential flood-prone areas considered feasible and their potential design and extent approximately 100 km of swales could be constructed of implementation in the study area are discussed in adjacent to existing smaller roads and lanes. The cost more detail below. is estimated to be $18 per linear metre of constructed swale. This equates to a total cost of $1.8 million. The primary maintenance objective for swales is to maintain Conceptual design and costing of selected interventions the hydraulic and removal efficiency of the channel The extent and location of each intervention was which involves litter and debris removal, and the estimated using Google Earth and GIS land cover maintenance of vegetation if the swales are vegetated. maps in combination with the criteria and limitations described for the interventions to identify the most suitable areas within the catchment for implementing 4.0.2 Catchment reforestation each specific stormwater management measure. The design of each and the extent is based on current land 4.5.2.1 Effects of reforestation use in the catchment and identifying open space areas most suitable for implementing wetland and floodplain A number of studies have shown that reforestation of measures. The extent, design and location of each of the catchment areas helps to significantly retain and slow selected interventions is described below. down runoff, reducing downstream flooding (Oosterberg 1997, Bahremand 2006, Serrano-Muela et al. 2008, Taylor The costing of the selected interventions was based et al. 2008, Zheng et al. 2008, Olang & Furst 2011, Ouyang on a wide range of information sources collated from et al. 2013, Gageler et al. 2014). Reforestation helps to literature and various green urban development projects reduce peak flows through interception and storage of offered in other parts of the world. The estimated precipitation in the leafy canopy, by slowing down and unit costs for the interventions and the sources of the storing runoff in the thick layer of organic matter on information are outlined in Appendix 6. All costs were the ground (such as leaves and branches), and through expressed in terms of 2015 US Dollars. increased infiltration as a result of improved soil structure (Gageler et al. 2014, Opperman 2014). In contrast, deforested or overgrazed areas within a catchment tend 4.0.1 Swales to be characterized by a limited or absent organic layer and compacted soils which encourage rapid surface 4.5.1.1 Design and function runoff (Opperman 2014). Some of the findings from the literature are summarised below: Swales are useful in built-up, high density areas where flows can be conveyed via small channels. Swales can ƒƒ Ouyang et al. (2013) determined the impacts of be constructed in the lower catchment, high residential, reforestation on runoff attenuation and sediment load flood prone areas with the function of conveying rainfall reduction in the Lower Yazoo River Watershed (LYRW) and runoff out of these areas as quickly as possible. within the Lower Mississippi River Alluvial Valley The residential areas identified are those situated (LMRAV) and found that conversion of agricultural immediately adjacent to the Msimbazi River in the land into forests attenuated runoff and reduced lower inundation area at the confluence of the Sinza, sediment load significantly. A two-fold increase in the residential areas along the lower reaches of the forest land area resulted in approximately a two-fold Ubungo river north of where it joins the Msimbazi and reduction in annual runoff volume and sediment load the densely populated areas along the lower reaches of mass, and on average, over a 10-year simulation, the                                                  Page 54 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT specific runoff attenuation and sediment load control. They found that vegetation structure and reduction were, respectively, 250 m3/ha/y and 4.02 plant life forms were the main factors reducing surface metric ton/ha/y. runoff and the movement of sediments. ƒƒ Gageler et al. (2014) compared remnant riparian Catchment reforestation and riparian revegetation rainforest, pasture and reforestation plantings aged should be seen as a catchment scale tool that can have 2–20 years in an Australian subtropical catchment to a significant beneficial effect on flooding in lowland determine the extent to which reforestation restores areas (Rutherford et al. 2006). At the catchment scale key soil properties. They found that Infiltration rates the effect of land use change, for example reforestation, were significantly lower in pasture than remnant will have a more substantial effect on the depth and riparian rainforest, and that within reforestation duration of flooding (i.e. the amount of water in a flood plantings, infiltration rates increased up to 60-fold event), whereas the effect of riparian vegetation is to with time post reforestation. alter the timing of the delivery of the flood (Rutherford et al. 2006). ƒƒ Olang & Furst (2011) investigated the impacts of historical land cover changes on the hydrologic response in the Nyando River Basin, Kenya using 5.4.2.2 Potential extent in the study area hydrologic models. They found significant and varying The headwaters of the Msimbazi River are located in increases in the runoff peak discharges and volumes the Pugu Forest Reserve at the top of the catchment. within the basin as a result of deforestation. In the The Reserve is heavily degraded. The edges of upstream sub-catchments where there were higher the forest reserve have been encroached by local rates of deforestation, increases between 30 and 47% communities who have cleared vast tracts of forest for were observed in the peak discharge; whereas in the subsistence agriculture. Satellite imagery has shown that entire basin, the flood peak discharges and volumes degradation of the natural vegetation and deforestation increased by at least 16 and 10% respectively during is not only taking place at the edge but occurs in large the study period. areas of the Reserve. The loss of biodiversity and ecosystem services such as soil retention and flow ƒƒ Bahremand (2006) investigated and assessed the regulation could have serious impacts not only for impacts of land use changes (particularly people living in the immediate area but also for those reforestation) on floods by means of distributed living downstream of the forested reserve. In the modelling and GIS in the Hornad River Basin in short term people are receiving immediate benefits Slovakia. Their results showed that 50% reforestation from forests in the form of timber, charcoal making, decreased the peak discharge by 12% and total runoff cultivation, and grazing. However, over the long term the by 4.5%. A 23% reforestation scenario decreased peak impacts are extensive such as increased soil erosion and discharge by 5.2% and a 38% reforestation scenario loss of soil fertility, siltation of rivers, increases in flood decreased peak discharge by 9.1%. The time to peak of events, reduced water availability, and severe fuel wood the simulated hydrograph of the reforestation scenario shortages (IUCN, WWF 2002). A concern in Tanzania was 9 hours longer than for the present landuse. is the overpricing of conventional energy sources via ƒƒ Taylor et al. 2008 investigated the impacts of land use high taxation which makes fuelwood a very attractive change on flood risk, through an assessment of the source of energy, resulting in significant degradation infiltration characteristics of soils in the upper of woodlands and forests (IUCN, WWF 2002). With Waikato, New Zealand under both forest and only 20% (400 ha of a total 2180 ha) of the Pugu Forest agriculture. Infiltration measurements in this study considered to be in reasonably good condition, there is were similar to literature values for a wide range of potential for significant reforestation initiatives within soil textures. Infiltration under grazed pasture was an the Reserve. order of magnitude less than that under forest for all Using Google Earth it was possible to identify the sites, and infiltration rates were significantly greater in degraded areas of the forest reserve that could be the forest sites (671 ± 335 mm.h-1) than in agricultural restored and rehabilitated. In the most northern section sites (47 ± 39 mm.h-1). of the reserve, a total area of 776 ha was identified ƒƒ Zheng et al. 2008 estimated the long-term influences for potential forest restoration (see Figure 4.5). The of regenerating forest cover on soil and water loss areas around the edge of the forest reserve have been from degraded land, the runoff and soil loss in the completely transformed into agricultural fields and context of different forest restoration approaches over present a further but potentially far more costly option. a four-year period (2000–2003) in a hilly red soil region in Southern China. Their results indicated that forest restoration decreased surface runoff by 63.0–88.1% and soil erosion by 75.5–97.1% compared to the                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 55 Figure 4.5 Map showing proposed location of “green” interventions within the Msimbazi River catchment 4.5.2.3 Methods and costs Ideally, the forest restoration process should involve Reforestation projects involve soil conservation and recruitment and training of local people at the district direct seeding and planting of indigenous tree species. and community level. An example of such an initiative The projects also need to go hand in hand with initiatives is from the Upper Tana River Basin in Kenya where local to prevent further deforestation. Soil conservation is women’s groups are engaged and involved in protecting a necessary part of the restoration process because and raising tree seedlings to rehabilitate two degraded deforestation can cause large scale erosion and loss of forest areas (TNC 2015). nutrients. Techniques for improved soil conservation Based on the Upper Tana River watershed project in include terracing in steeper areas to prevent erosion Kenya and projects in Costa Rica and the Philippines and to manipulate the flow of water by slowing it down, (FAO 2011, TNC 2015), it is estimated that basic and to re-establish vegetation cover by using indigenous reforestation would cost in the region of US$1090 per trees and shrubs. Therefore the following actions are hectare. This includes labour and material used for required as part of the restoration initiative: reforesting degraded areas but does not include the ƒƒ Terracing to stabilise steep slopes long term monitoring or maintenance costs, or the costs involved in developing strategies and policies to prevent ƒƒ Direct seeding and planting of indigenous seedlings further encroachment into the forested area. ƒƒ Protection and raising of seedlings ƒƒ Measures to prevent further encroachment and deforestation                                                  Page 56 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT While the primary restoration activities (soil 4.0.3 Rehabilitation and enhancement of middle catchment conservation, planting and raising of seedlings) are riparian and floodplain areas relatively straightforward, the process required to pave the way for these activities, such as relocating 4.5.3.1 Design and function people from the forest, and the implementation of successful measures to prevent further encroachment There are a variety of ways that the riparian and and deforestation present far more of a challenge. It associated floodplain areas of the catchment could is anticipated that for reforestation of the Pugu Forest be treated. A common model is to use a development Reserve to be successful, the resettlement of households setback to create a riparian buffer. Riparian buffer out of the reserve may also be necessary and going areas would have different effects in different parts forward this should be investigated. of the catchment. In the flood prone areas, this would mainly have the effect of reducing the number of In 2011 a “reducing emissions from deforestation and buildings and people at risk. Further up the catchment, forest degradation” (REDD) project entitled “Piloting this could help to reduce flow velocities and improve REDD in the Pugu-Kazimzumbwi Forest Reserves” water quality. Riparian buffer zones along waterways was initiated by the Wildlife Conservation Society of intercept sediments, nutrients, pesticides and litter Tanzania (WCST) and the Norwegian Ministry of Foreign in unchannelled surface runoff, thereby reducing the Affairs (NMFA). Funding amounting to US$3.9 million to amount of pollutants entering rivers and streams. They implement the four year project was provided by the also provide habitat and linear wildlife corridors through NMFA. The main aims of the project were to reduce the landscape – increasingly important functions as CO2 emissions by reducing deforestation and forest adjacent areas are sterilised by urban or agricultural degradation as well as supporting community livelihoods development. Assuming appropriate vegetation types, (Deloitte 2012). However, in 2012 concerns around the riparian buffers can also be important for reducing mismanagement of the project and the misuse of funds surface erosion and providing river bank stabilisation, were raised. Conflicts arose between local communities both by reducing the velocity of overbank runoff from in the study area, government departments and project adjacent areas and by anchoring the soil and reducing staff. As a result, the NMFA stopped all funding and near-bank velocities of water in the channel, through the project ended. It is important that lessons learnt increased channel roughness. during this process are used to facilitate new initiatives so that the complete loss of Pugu Forest Reserve can be The main shortcoming of riparian buffers is that they prevented. do not do much in terms of increasing the storage capacity of the catchment. In the catchment above the However, it is clear from previous attempts at flood-prone area, more is required to reduce run-off reforestation that the situation in Pugu Forest Reserve to natural or lower levels. Therefore, it is suggested is complicated and will require a dedicated and focused that the concept of riparian buffers is extended to the approach to curbing deforestation. Informal land tenure creation of enhanced floodplain areas that are designed within the reserve is a big concern and has damaged in such a way as to store and retard flood flows. This government credibility (Deloitte 2012). A successful entails a combination of riparian zone rehabilitation and reforestation measure will require dialogue between floodplain enhancement measures that are primarily all stakeholders and will require focused government designed to retard flows but which can easily include management and enforcement to address deforestation. opportunities for beneficial uses, including (variously) Forest borders and village boundaries need to be re- sports fields, agricultural lots and parks as well as active established and enforced. The only way to prevent the riparian buffer zones/conservation corridors. These complete loss of the Pugu Forest is to actively control, beneficial uses of the floodplain might raise initial costs, manage and enforce forest boundaries and the use of but are more likely to reduce opportunities for informal natural resources by local communities. Alternative resettlement of the floodplain. livelihood projects (located well away from the forest), and awareness programs could be used to encourage The concept is best applicable along river reaches communities to move away from harvesting forest where the channel gradient is low (1: 1000 or flatter) products. This involves additional costs on top of but because it involves substantial manipulation of monitoring and enforcement. degraded floodplains, it would lend itself to attenuation of runoff into steeper-gradient watercourses, even if attenuation of instream flow through overtopping is less likely. The measures should not be considered where natural habitats of conservation value remain (e.g. in the forested zone in the upper catchment) or where river slopes are steep.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 57 Figure 4.6 Conceptual plan of mixed use Enhanced Riparian and Floodplain areas. See text for description of zone treatment Figure 4.7 Rough cross-section sketch of concept This approach, shown in rough concept in Figure 4.6 The components, to be mirrored on both sides of all and Figure 4.7, would involve the creation of riparian river channels, are described in more detail as follows: buffer areas flanked by a series of bermed lower and upper floodplain areas running alongside the riparian 1. The river banks: these would, where required, be area that have means for slow drainage system back graded (slopes flatter than 1:4 would probably be into the river system after flooding. The lower floodplain appropriate) and stabilised with appropriate (and areas would flood more often, whereas the upper areas preferably locally indigenous) riverine vegetation would capture larger floods. Both the floodplains and including reeds and sedges that will play a role in berms could be designed for a variety of agricultural or preventing bank erosion as a result of root recreational uses. The entire width is envisaged to be penetration and stabilising effects as well as of 60m on either side of the river, with a river buffer of increasing hydraulic roughness in the channel itself. 15 m on each side, and the combined upper and lower floodplains being at least 45 m wide. Ideally this should 2. The riparian buffer: this area would extend a become wider with distance downstream through the minimum of 15 m wide along minor streams and up catchment, but space is at a premium in the study area. to 30 m wide along major streams, and it is assumed The 60m buffer is already written into law, though not that these areas would contain within year floods and yet enforced. possibly up to at least the 1:2 year river floods. It would be designed, so as to include plantings of trees, sedges, shrubs and other types that are locally indigenous/area appropriate, with the following primary functions: soil surface stabilisation,                                                  Page 58 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT prevention of erosion from sheet flows, provision of 5. Longitudinal swales: these would manage the spread ecological habitat and faunal linkages to upstream of flows from areas upslope of the riparian corridor, and downstream areas. Note that provision of a and allow its dissipation as sheet flow into the rehabilitated riparian corridor along at least one side riparian areas, allowing functions such as sediment of any channel is considered important. trapping, prevention of concentrated flows and associated erosion, and nutrient uptake both within 3. Lower floodplain: For rivers that are not steeply the swale and with diffuse passage across the buffer. sloped (see above), this area should accommodate Ideally, the swale should be designed as an infiltration river floods at least up to the 1:5 year flood, and trench that allows water to seep out of multiple ideally up to the 1:10 year flood line. For rivers that porous areas in the trench, potentially created by are steeper, these areas can be narrower, and should stone or gravel packing. The swale would also serve instead be designed to detain surface runoff from the as a final litter collection zone; surrounding catchment - this area and the upper floodplain should together allow attenuation of 6. Lateral swales (towards the river): these should be (ideally) the peak discharge of up to the 1: 50 year designed for water quality improvement, and sized to return interval storm. It can be designed as a mixed convey small storms (< 1-year return intervals) across use area, with seasonal crops included, on the the floodplains and into the infiltration trench (5 in understanding that periodic (less frequent than 1: 2 Figure 4.6). Larger volumes of runoff should overtop years) wet season flooding is likely. Planting of an into the adjacent upper floodplain, and pass as extended riparian buffer along the lower part of this overland flow into the lower floodplain. Channelling system could be envisaged – however it is noted that of runoff into the lateral swales should be via stilling in some situations, densely planted riverine areas are ponds (7 in Figure 4.6), in which litter and sediment viewed with concern by local communities from a can collect, providing zones for concentrated litter security perspective, and this issue should be checked collection activities. locally; Important assumptions of this concept include the 4. Upper floodplain: this area should accommodate less following: frequent floods than those of the lower floodplain. The two could be separated by a berm that would ƒƒ It is assumed that river channels and their riparian allow water levels and detention time in the upper zones, flood plains and abutting terrestrial areas to a floodplain to be controlled; the berm should be distance of 60m to 100m on either side of the channel equipped with pipes or porous areas (e.g. stone (or as wide as the scheme extends) are already so packing) to allow slow downslope drainage of degraded that the significant disturbance and long- floodwaters into the adjacent lower floodplain and term changes in utilisation and management will have thence into the riparian buffer zone. The flood no negative biodiversity or other ecological effects. attenuation capacity of the upper floodplain could be This is an important point and requires thorough increased if it included a shallow basin – significant ground-truthing and verification prior to any earthworks would in any case be associated with the consideration of implementation of this concept; establishment of both floodplains. The passage of runoff from the adjacent catchment into the upper ƒƒ Setting of the width of the intervention at 60m on and lower floodplains could be further controlled and either side of the river is based on legislated setback detention capacity enhanced by using porous berms widths rather than any hydrologically defendable data. to separate “compartments” that are arranged If the concept is considered further, it would be longitudinally down the river/floodplain corridor. necessary for their detailed design to evaluate their Various uses that could be carried out in these effect on flood lines, so that the width of adjacent “compartments” include agriculture, grazing, sports areas, as well as berm and terrace heights can be fields – active, planned establishment of specified designed to accommodate the 1:5, 1:10 and 1: 50 year uses in all such corridors would reduce the likelihood (or higher if required) flood return intervals as of their being resettled on in the future. required.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 59 4.5.3.2 Extent and costs of rehabilitation work 4.5.3.3 Potential resettlement costs This measure has been costed for application for a In addition, there is a potential cost of resettlement. band of 60m either side of all rivers in the study area In Dar es Salaam the areas up to 60 m on either side of above the flood prone area and below Pugu Forest. This all river channels are known as River Reserves and are amounts to 270 ha in the Msimbazi River subcatchment, protected areas in which no development is allowed 110 ha in the Ubungo subcatchment, and 8 ha in the (Environmental Management Act of 2004). However, Sinza subcatchment, a total of 488 ha. The intervention the River Reserve areas have not been demarcated is expected to be very effective as it could detain almost or enforced and as a result have become encroached 5 million m3 of runoff in the middle-upper catchment by unplanned settlement, with densities increasing areas. Aside from the flood mitigating benefits and the downstream. During December 2015 a significant ecological benefits associated with this intervention, number of demolitions (over 700 houses) took place space for multi-use areas such as agriculture, parks and within the River Reserve areas in the lower catchment. sports fields all contribute to improving quality of life in However, the process was stopped pending the use of the urban environment. proper procedures. Therefore, unless government acts to enforce this law, people who have moved into these The plan involves a river rehabilitation component areas might need to be resettled in order to execute the for the area up to about 20 m from the channel, and project. earthworks to create a modified floodplain. In the Msimbazi catchment there are extensive sections The resettlement of affected households requires of river between Pugu Forest and the confluence of a structured, participatory approach following the Sinza that are severely degraded, with eroded international best practices related to displacement banks, and the riparian vegetation either completely and resettlement of people. This may entail significant removed or in poor condition. Costs of rehabilitation compensation costs. World Bank-funded projects in of riparian buffer areas vary greatly depending on Tanzania would estimate compensation costs on the specific site conditions and the level of degradation. basis of: If rehabilitation only includes seeding and planting then the costs involved are relatively low per hectare. 1. Replacement cost of dwellings and other structures; However, these areas would require some landscaping or earth grading as well as seedling protection, which 2. Replacement cost of land (the market value plus can increase the costs significantly. Based on projects transactional and other costs involved in acquiring carried out elsewhere, restoration of the riparian zone, new land); including seeding and planting, is expected to cost 3. Replacement cost of productive assets such as approximately $2376 per ha (Appendix 6), or just under enterprises, water supply facilities, etc.; $400 000 in total. This estimate is based on the fact that the riparian buffer includes riparian zone rehabilitation 4. Projected production losses from land, crops and and floodplain enhancement measures and was based trees; and on the assumption of approximately one third of the riparian buffer zone requiring rehabilitation. The main 5. A disturbance compensation which is a specified cost component is that of excavation and earthworks to percentage of the total of the above costs. create the berms and swales to control water in the rest of the buffer zone. The extent of earthworks required Important note: This study does not provide a formal is dependent on factors such as slope. In some areas estimate of these costs, but provides a preliminary it may be possible to create the stormwater damming estimate in order to obtain a ball-park estimate of the effect simply through creation of berms, whereas in potential additional costs of the project, if all households other areas some degree of excavation may be required. currently within the project area had to be moved Assuming that the latter is minimal, the overall cost of with compensation. It is very important to note that earthworks would be estimated to be in the order of compensation payments under these circumstances are $10.5 million. If all holding areas were to be excavated, likely to be counterproductive, as they might encourage it would be closer to $44.5 million. The overall cost of rent-seeking behaviour in this and other such reserve the works is therefore in the region of $11 – 45 million, areas in the future. This could raise the costs of the with a mid-point estimate of $28 million. Further work is whole exercise. Further work is required to determine required to refine this estimate. what role the Government might play and what compensation would be necessary and appropriate.                                                  Page 60 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Figure 4.8 Digitised map showing the dwellings and agricultural fields in the setback zones of the middle to upper catchment areas of the Msimbazi, Sinza and Ubungo sub-catchments. Replacement cost of dwellings Market value of land The affected houses within the 60m riverine buffer In Mozambique, owners of agricultural land were reserve areas were identified and counted by digitizing compensated for labour invested in land improvements dwellings from the latest Google Earth satellite imagery (clearing, tilling, and grubbing) as a proxy for land value for the middle to upper sections of the three sub- (Mozambique LNG 2015). This follows an approach catchments where the rehabilitation initiatives are used by the World Bank that recognises the farmer’s proposed (Figure 4.8). investment in land, without being a payment for the land itself, which remains vested in the State (Mozambique A total of 4813 residential dwellings were identified for LNG 2015). The compensation amount includes the costs potential resettlement. The majority of these fell within associated with bush clearing, annual clearing, tillage, the Ubungo and Msimbazi sub-catchments. Replacement maintenance and the provision for land investment and costs for living structures was based on costs obtained disturbance. A labour and agricultural disturbance rate from the literature for informal residential dwellings of $1600 per ha was applied (Mozambique LNG 2015). ($120 per m2, Appendix 2), and using an average house Based on Google Earth imagery, the extent of the large size of 70 m2 (based on a sample of actual dwelling areas agricultural fields within the 60 m buffer was estimated in the study area from open street maps). The cost for to be 32.3 ha, with most of these fields being located replacing 4813 dwellings was therefore estimated to be in the Msimbazi sub-catchment (Figure 4.8). Applying $40.4 million. this value to the 32.3 ha of agricultural land, the total replacement cost for lost agricultural land is estimated to be approximately $51 830. It is likely, however, that many or most of the identified fields within the Riparian Reserve could be incorporated into the design described above. Where this is the case, compensation would not be necessary.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 61 Losses of productive assets and projected production Disturbance allowance Many of the affected households are likely to have The disturbance allowance provides support for gardens and at fruit trees at their homesteads, or households during the resettlement process, and is along the river. The main types of trees planted in Dar calculated as a percentage of the total of the above es Salaam include coconut, mango, cashew, papaya costs. The percentage was taken from the Surface Water and avocado. The main crops tend to be leafy green Drainage Systems (SWDS) Project Resettlement Action vegetables such as spinach, sweet potato leaves, Plan, which was specified as 8%. This was applied to the pumpkin leaves and cowpea leaves, as well as other calculated resettlement costs and added to the overall vegetables such as eggplant, okra and tomato. Accurate total cost. The disturbance allowance was estimated to estimation of the numbers and extent of these was be $3.2 million. beyond the scope of this study, but a ball-park estimate of the potential compensation costs is derived from other Resettlement Action Plans (RAPs) in the area. A Total RAP provides an agreed plan for the resettlement and Above the flood prone area in the middle to upper compensation of Project Affected Persons (PAPs) and catchment the resettlement costs are estimated to be aims to ensure that land acquisition is undertaken as per approximately $44 million (Table 4.4). Going forward, specific standards. these preliminary estimates would need to be properly validated. It was estimated that 18% of affected households had productive trees. Based on the findings of Jacobi (1997) that 15-20% of all houses in two unplanned areas of 4.0.4 Rehabilitation and enhancement of lower Dar es Salaam had vegetable gardens and/or fruit/nut floodplain areas trees, we used an estimate of 18%, which translates to 866 households. Average compensation per producer household was obtained from the Kinondoni Municipality Increased stormwater runoff has resulted in increased RAP developed for the Surface Water Drainage System flow into the rivers. These rivers therefore need larger- Subproject under the Dar es Salaam Metropolitan than-predevelopment floodplain areas to enhance Development Project (DMDP) (PMO-RALG 2014a). the storage and capacity of the floodplain to dampen The values taken from the RAP were inflated to 2015 flooding events. Enhancing the floodplain areas involves prices and converted to US Dollars. The RAP included a deepening the floodplain area to increase storage summary of the total number of affected households, capabilities. In certain areas, benefits of increasing the and total compensation costs for the loss of trees/crops. floodplain can be added through the establishment Compensation for crops is determined as the average of agricultural areas, which is already the case in value over the previous year, corrected for inflation and some sections along the river. In other areas along the compensation for trees is based on the type, age and the river system it involves the removal of berms and productive value of affected trees plus 10% premium re-establishing connections with adjacent areas to (PMO-RALG 2014a). The average compensation cost for allow the river to overtop its banks into its floodplain the loss of these was $167 per household. Applying these more frequently. Restoring the natural hydrological values to our sample of affected households, the total connectivity of the system will have numerous ecological compensation cost for loss of trees and crops is estimated benefits. By deepening the floodplain areas in the lower to be around $145073. Table 4.4 Total estimated resettlement costs for the lower catchment and middle to upper catchment areas Items Msimbazi Ubungo Sinza Total Number of households 1939 2580 294 4813 Compensation Cost Loss of dwellings 16 287 600 21 672 000 2 469 600 40 429 200 Loss of productive trees/crops 58 445 77 766 8 862 145 073 Loss of agricultural land* 46 992 2 638 2 200 51 830 Sub-Total 40 626 103 Disturbance allowance (8%) 3 250 088 Total 43 876 191 *As explained in the text only some of this will actually be lost, as it can be incorporated into the floodplain buffer design.                                                  Page 62 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT catchment there is the opportunity to develop a wetland The cost of deepening the floodplain areas for recessed park to provide an important inner city recreational gardens will include excavation costs and possibly some green open space area that could provide numerous top soiling and grassing costs. Market gardens will benefits. Activities such as fish farming could also be provide vegetation cover during non peak-flow periods. considered for such areas, assuming that water quality There are three main areas on the Msimbazi River, two was adequate and that the ponds were designed to on the Ubungo River and two on the Sinza River that include areas of permanent water. have been identified as possible locations for recessed gardens. The enhanced floodplain-recessed gardens in the mid-lower catchment will cover an estimated area 4.5.4.1 Enhanced floodplain-recessed gardens in middle- of 47 ha, costing approximately $5.4 million (Table 4.5). lower catchment Stormwater retention was calculated by assuming a Enhanced floodplains with recessed gardens in the depth of one metre for the floodplain recessed gardens. middle-lower catchment would be a local scale approach to attenuate flood peaks in urban and peri-urban areas It is expected that the recessed gardens will be relatively where households have market gardens situated along cost-effective in that the capital costs associated with the river channel within the floodplain. The main aim of the intervention are relatively low. The interventions can the recessed gardens is to increase the water storage cover a very large area making the overall cost outlay capacity of the floodplain but still maintain functionality rather high. However, the intervention, especially in for crop production during non-flood periods. Currently terms of runoff retention is extremely cost effective. The the market gardens are grown in the floodplain and are intervention is not very cost-effective in terms of soluble constantly flooded during high rainfall events with not nutrient removal. much storage of excess water. By excavating and thus deepening the floodplain area adjacent to the channel the storage capacity of the area is significantly increased, reducing flows downstream. Households would still be able to grow crops within these areas outside of the main flooding season. These areas have the potential to retain a significant amount of runoff during peak flows and will also contribute to the removal of sediments and nutrients through infiltration, although this function is expected to be relatively small. The vegetated floodplain is much the same as a vegetated channel in terms of nutrient and sediment removal. Table 4.5 The estimated extent of enhanced floodplain-recessed gardens, estimated total cost of the intervention and the estimated total amount of runoff retained Potential extent of Total stormwater runoff Identified areas Estimated total cost (US$) intervention (ha) retained (m3) Msimbazi 1 26.63 2 796 150 266 300 Msimbazi 2 1.93 202 650 19 300 Msimbazi 3 5.58 585 900 55 800 Ubungo 1 5.34 560 700 53 400 Ubungo 2 4.43 465 150 44 300 Sinza 1 3.14 329 700 31 400 Sinza 2 4.00 420 000 40 000 Total 47.05 5 360 250 510 500                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 63 4.5.4.2 Rehabilitated floodplain / wetland park areas in lower Community based river cleaning projects have shown catchment to be successful and sustainable. For example the Rehabilitation of floodplain wetland areas can also Mlalakua River Restoration Project initiated in 2012 has be incorporated with extended detention ponds to been successful in raising awareness in communities allow for storage and treatment of a greater volume and in cleaning the Mlalakua River in Kinondoni of stormwater runoff than in a simple shallow Municipal Area in Dar es Salaam (see Appendix 5). The wetland. These are known as “extended detention Sihlanzimvelo Stream Cleaning Project initiated in 2011 shallow wetlands” which are able to store most of the with the aim of maintaining and cleaning approximately stormwater volume above the relatively shallow marshy 490 km of watercourses throughout the eThekwini depths within the macrophyte zone (Armitage et al. Municipality in Durban has been successful too (see 2013). Restoration of wetlands and construction of the Appendix 5). Most of the rivers and streams included extended detention basin would require excavation in the Sihlanzimvelo project are located in the poorer, and earthworks to develop a detention area that is dug more densely populated suburbs of Durban. Both of out and can aid in regulating flow and attenuating flood these projects focus on cleaning rivers of litter and peaks to an extent in combination with the functioning alien vegetation, provide employment opportunities of the restored wetland. for community members and educate communities on the benefits provided by clean and safe environments. In Dar es Salaam there is an area in the lower catchment A project such as this could be initiated in the Msimbazi that has been identified as a possible location for catchment where the rivers and streams are constantly the development of an extended detention wetland. blocked with litter and debris. The significant amounts of The area is frequently flooded during high rainfall rubbish that end up in the rivers and streams exacerbate events. Some informal houses have been constructed the flooding problems in the city, causing widespread on this floodplain area and would therefore need to damage. By generating employment opportunities and be relocated if the wetland were to be constructed providing education and awareness to communities, a here. These informal homes are regularly flooded. It project such as Sihlanzimvelo or Mlalakua, could have is envisioned that the detention wetland will regulate significant positive impacts in the Msimbazi catchment. and attenuate flood peaks, will remove pollutants and Based on the programme in Durban and the Mlalakua sediments from the water and will provide amenity Project in Dar es Salaam, it is expected that the cost of value to the surrounding urban landscape. The shallow setting up and running such as programme would be in detention wetland will cover an area of just under 15 the region of $1 million in year one and thereafter would ha, with an estimated total construction cost of just cost around $250 000 per annum. over $3 million (Table 4.6). This intervention will detain approximately 298 000 m3 of runoff, based on a depth of two metres. 4.0.6 Summary Construction costs and maintenance costs were estimated using examples from the stormwater 4.0.5 Community-based river cleaning programme management literature and from examples of projects One approach to keeping rivers clear of litter and where similar measures have been applied elsewhere debris and maintaining a healthy river system is to in the world. These estimates were used to develop involve communities that live alongside rivers and an average cost for each intervention based on the streams. Community involvement projects can have knowledge of the extent of restoration or the extent multi-sectoral impacts as they generate employment of each intervention required in Dar es Salaam. These opportunities, provide awareness, safeguard estimates and their sources can be found in Appendix communities and provide city-wide services such as 6. Annual maintenance costs were calculated based functioning river systems that are clean and clear of on estimates from the literature described as a litter. Sections of rivers or streams are maintained percentage of overall construction costs. The total initial by cooperatives which are responsible for removing investment cost of these interventions is estimated to alien vegetation, rubble and any solid waste blocking be approximately $40 million (without resettlement) the free flow of water down the stream or river. They and annual maintenance costs were estimated to be in are also responsible for maintaining the grass and the order of $1.6 million (Table 4.7). Around 40% of the other vegetation along the banks of the waterway. total investment cost is for the mixed-use enhanced The cooperatives generally consist of members of the riparian and floodplain areas, which cover almost 500 community that are unemployed and vulnerable and ha and detain 5 million m3 of runoff. Relocation costs the project focuses on raising awareness and generating were estimated to be in the order of $44 million. Total employment. costs associated with the selected GUD interventions and associated resettlement activities was therefore estimated to be $84 million.                                                  Page 64 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Table 4.6 Estimated extent and total cost of the extended shallow wetland and the total stormwater runoff expected to be retained Estimated total cost Total stormwater runoff Intervention Extent (ha) (US$) retained (m3) Floodplain reconnection and 15 3 129 000 298 000 extended detention shallow wetland Table 4.7 Summary of the extent and costs of the selected GUD interventions in the middle to upper Msimbazi catchment and the resettlement costs involved in relocating households from these areas Extent Initial /construction cost Annual maintenance cost Identified areas (ha) (US$) (US$) Swales to improve drainage in flood prone areas 10 1 800 000 108 000 Catchment reforestation in Pugu Forest Reserve 776 845 000 17 000 Mixed use enhanced riparian and floodplain areas (~1m deep) 488 28 000 000 1 036 000 Rehabilitated floodplain and wetland park (~2m deep) 15 3 130 000 94 000 Enhanced floodplain-recessed gardens (~1m deep) 51 5 360 000 107 000 Community-based river cleaning project - 1 000 000 250 000 Total without resettlement costs 1340 40 135 000 1 612 000 Relocation with compensation 44 000 000 Total with maximum resettlement costs 84 135 000 1 612 000                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 65 This page intentionally blank.                                                  Page 66 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT V. SCENARIOS ANALYSIS Scenarios In the engineering design process, different stages Five combinations of stormwater management measures at increasing level of detail can be defined: research, were included in the analysis: conceptualization, feasibility assessment, establishing design requirements, preliminary design, detailed 1. Riparian setbacks in the flood prone area; design, construction/production planning and construction/production (Ertas & Jones 1992). The 2. Green urban development measures (GUD); feasibility study stage narrows the scope of the project in order to identify the best scenarios. The scenarios 3. GUD measures + riparian setbacks in the flood prone conceptualized herein are then implemented in the area; preliminary design, returning the general framework to build the project on (Dym et al. 2009). 4. GUD measures + additional detention basin(s); and 5. GUD measures + detention basin(s) + riparian Hydrologic modelling assumptions setbacks in the flood prone area. Given the amount and the quality of available data, and The scenarios are a combination of interventions that the large scale of application of the mitigation strategies, either reduce exposure to flooding, reduce flood risk, the analyses results discussed in the following are or a combination of both (Table 5.1). By removing developed at the level of a preliminary design. In such a people from flood prone areas within riparian setback framework, several simplifications are made regarding buffers the number of people and structures exposed the implementation of the mitigation strategies in the to flooding is reduced. By implementing GUD and hydrologic/hydraulic model. additional storage interventions the flood hydrograph is lowered and flood risk is reduced. The cost of each intervention includes the potential costs associated 5.0.1 Riparian setback in lower floodplain with resettlement. Resettlement costs are described in This intervention would not have a measurable effect detail in Appendix 4 for the riparian buffer zone in the on the storm hydrograph. However, it involves the flood prone areas and in Section 4.5.3.3 for the GUD resettlement of households from a defined development interventions in the catchment area. setback zone along the rivers, and therefore changes the number of buildings exposed to flooding. Table 5.1 Scenarios 1-5 and their estimated costs Reduce exposure  No interventions People and structures in flood prone areas removed from 60m buffer in flood prone areas Scenario 1 No interventions in catchment $62.6 million Reduce flood risk Scenario 2 Scenario 3 ↓ GUD interventions in catchment1 $84 million $138.5 million2 Scenario 4 Scenario 5 GUD with additional storage $124 million $178.5 million 1 GUD: (a) restoration of forests in upper catchment, (b) rehabilitated and enhanced riparian and floodplain areas in middle catchment, (d) river cleaning in middle catchment, (c) floodplain rehabilitation in lower catchment, (e) swales in flood prone areas. 2 This is less than the sum of 1 and 2 since the number of buildings at risk in the buffer is reduced, and so a reduced number of structures have to be relocated                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 67 5.0.2 Combined GUD interventions All the beneficial effects induced by this combination In our analysis, the effect of the GUD measures on flood of interventions were lumped together and expressed flows (the hydrograph) was simulated in the hydrological as a change of the input hydrograph, based on the model through a change in the AMC. This approach assumption that the set of GUD interventions (Scenario is not new in the literature; for example, Jalayer et 2) has the equivalent effect of an improved capacity al. (2013) modelled the increase of urbanization (i.e. of the soil to infiltrate the storm water as would be the increased impermeability of the soil due to the achieved if the AMC was changed from AMCIII to AMCII, variation of land-use) in a lumped manner by increasing or if the soil hydrologic class was changed from soil B the AMC class. In this study, the same approach was to Soil A which has a similar effect (Figure 5.1). Such followed but in the opposite direction: the hydrograph variations are consistent with what has been observed associated with the new scenario is obtained varying in the literature; in fact, increasing the soil permeability, the AMC, from AMC III to AMCII. A more detailed the peak reduction varies from 40% (Drake, 2014; explanation on this point is provided through the Klingner, 2014) up to 90% (De Paola et al. 2013; statistical analyses presented in section 3. Given the 2015; Kowalik and Walega 2015; Aceves and Fuamba results presented in Figure 5.1, it would have been 2016). In this study, the extent of GUD interventions justified to use AMC I. However, using AMC II is more was estimated to meet this level of reduction in the conservative (in engineering terms, i.e. the potential hydrograph for a 1:10 return interval flood. damage is increased) and therefore more appropriate for risk assessment purposes. The resultant changes are shown in Figure 3.16. For a 1:10-year flood, modelled reductions in the flood peak ranged from about 45% in sub-catchment 3 to about 60% in sub-catchment 1. Figure 5.1 Two different ways to implement the mitigation strategies in the input hydrograph for different return periods                                                  Page 68 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT 5.0.3 Additional storage in Scenarios 4 and 5 where k is the storage constant of the floodplain storage In the lower part of the main Msimbazi River, there is a and is equal to: floodplain area that has no construction on it. This area was identified as a potential area for the creation of an in-line floodplain storage area. The effect of the storage basin was modelled according to the procedure described Substituting 5.4 in 5.1 it is possible to obtain the above. With an area of 50 ha and a design depth of 6 m, differential equation of the on line floodplain storage the maximum volume that such storage can absorb is 3 which indicates the relation between inflow and outflow: million cubic meters. The detention basin was modeled as a physical separation of the discharge between the storage and the urban area. There are two possible methods to implement such measures in the inundation code. The This equation is generally solved by means of finite first is to implement it as a floodable area that can differences schemes. Knowing the initial hydrograph, be reached by the flow through some hydraulic links. the previous equation allows one to evaluate the Such modeling requires a detailed description of all the outflow hydrograph to consider for the two-dimensional design detail and generally cannot be implemented in propagation. a preliminary analysis. The second approach involves removing the discharge that is accumulated in the floodplain storage from the flow. This method is 5.0.4 Caveats facilitated by the design equation presented below. All the presented hypotheses are valid at a preliminary design phase for two main reasons: (a) not all the The continuity equation for a flow is: mitigation strategies are defined and spatially allocated in a definitive manner (e.g. the application of swales in the urban area); (b) the full implementation of the mitigation strategies that can be obtained with a more where W is the volume of the storage, qin and qout sophisticated procedure requires more refined data the incoming hydrograph and the outflow hydrograph (necessary to calibrate all the involved coefficients) respectively. Let is S the total surface of the floodplain and a more detailed knowledge of the territory and of storage and h is the water depth inside the storage; then the hydrographic characteristics (that can be acquired W can be expressed as below: only through bespoke surveys). Therefore such an “aggregate” effect can still be considered informative in a context of lack/absence of detailed information. Moreover, the limitations due to the lack of more spatially refined data, such as spatially distributed river gauge and precipitation data and a map of the existing The outflow hydrograph can be expressed through the urban drainage or sewage systems (if existing) does not outlet formula allow a more rigorous procedure. It must also be stressed that this analysis has focused on flood mitigation, because water quality amelioration capacity of these interventions will not have any where  is the discharge coefficient (generally assumed measurable benefit under the current levels of investment equal to 0.61, Daugherty and Franzini, 1965);  is the in sanitation and sewage systems, due to the overload of outlet area, that is variable with the return period pollutants into the drainage system. Once these systems considered in order to fill up the entire capacity of the are in place, these stormwater management interventions storage; g is the gravity acceleration (i.e. 9.81 m/s2). will also be able to provide economic benefits in terms of ‘polishing’ of water quality. Deriving h from 5.3 and substituting it in 7.1 a new expression of the total flood volume is obtained:                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 69 Costs of the interventions introducing a good solid waste management program for existing and new drainage, floodplain reconnection in the lower river, a river cleaning program in the lower 5.0.1 Riparian setback in lower floodplain catchment, and the rehabilitation of river buffers and The number of buildings at risk in the setback zone is reforestation in the mid to upper catchment of the 2422, or 19% of the entire portfolio of 12 744 structures Msimbazi RIver. Along the mid to upper catchment at risk. The structures identified are a mix of unplanned areas of the Lubango, Ubungo and Sinza Rivers mixed- residential, commercial, industrial and utility buildings. use enhanced riparian and floodplain areas were also Each of these has a different replacement cost and these considered. These areas are currently occupied by costings per m2 of area can be found in Appendix 2. If approximately 4800 dwellings. In order to guarantee required (i.e. if the setback policy has not already been the functionality of the proposed mitigation strategies, enforced by the time of project implementation), the operations of inspections and maintenance are relocation of these buildings and people could require required as well. Such costs are assumed to be 1% an initial cost of an estimated US$62.6 million. This and 5% of the total initial cost per year, for inspection includes the value of the buildings, an 8% disturbance and maintenance, respectively. Moreover, a further allowance, and an additional 10% cost for demolition community-based river cleaning programme of US$1 and removal of debris. million per year is also considered. The total cost associated with the combined GUD interventions, and Note: It is understood that in December 2015 an including maximum resettlement costs, is estimated to estimated 700 unplanned structures were removed be $84 million. Cost details can be found in Section 4.5.6 from the setback zone in the lower catchment. This is and Table 4.7. approximately 30% of the total number of structures identified during this assessment, which if taken into account would lower the cost to $41 million. However the 5.0.3 Additional storage in Scenarios 4 and 5 resettlement of households was not successful, as plots were resold or people returned soon after evacuation To estimate the construction costs of additional because they had been moved to areas far from their storage, three main elements have been considered: social networks and places of work (M. Bitekerezo, World purchase of 50 ha of land, excavation of 3 million m3 Bank, pers. comm.). of soil, 20,000 m3 of reinforced concrete works (about 7‰ of the entire flood plain volume, rough estimation of works for protection of the site, potential artificial 5.0.2 Combined GUD interventions levee in some areas, and all associated flow control structures), and a team of 100 foremen and labourers The proposed mitigation strategies affect about 125 working for two years. The final value is then multiplied ha and 20 ha in catchments 2 (Ubungo) and 3 (Sinza), by a factor of 1.36 which represents the additional respectively, and about 1200 ha in catchment 1 costs as a percentage of the total cost, i.e. indirect, (Msimbazi). This scenario, described in more detail in preparation, administration and contingency costs (see Section 4.5, included improvement of the drainage in Error! Reference source not found. for details). The cost the lower river flood-prone areas by building swales and breakdown is presented in Table 5.2. Table 5.2 Cost breakdown of detention basin ID Unit Unit cost ($) Cost (US$ millions) Purchase of land 50 ha 1000 $/ha 0.05 Excavation (90% of total volume) 2.7x106 m3 8.6 $/m3 23 Concrete works 20x103 m 3 255 $/m 3 5.1 Labor force (730 days) 100 people 8.48 $/day 0.62 Total cost (1) ≈ 29 Total cost (2) = 1.36 x Total cost (1) ≈ 40                                                  Page 70 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Scenario evaluation approach The costs of the mitigation strategy (SI) were taken to The different scenarios were evaluated in terms of their include not only initial implementation cost C0 but also return on investment (ROI). Defined in general terms, ongoing inspection (CI) and maintenance (CM) costs for ROI analysis (Reilly & Brown 2011) compares the desired the implemented interventions. The total investment target outcomes (i.e. benefits) an investment yields with associated with the generic mitigation strategy SI is then the costs of that investment. equal to: ROI analysis is routinely applied in both the private and If the expected number of inspections and/or public sectors to evaluate the performance of competing maintenance events is known, it is possible to estimate financial investment opportunities, programs or projects, the mean expected annual cost of inspection (Ci) and and is equally applicable to conservation projects (Boyd maintenance (Cm). Here we follow common practice and et al. 2012). If benefits are expressed in physical units assume that Ci and Cm are a percentage of the initial (e.g. reduction in the number of buildings damaged by intervention cost. flooding), ROI analysis is equivalent to cost-effectiveness analysis; if benefits are monetized (e.g. value of avoided It is now possible to rewrite the ROI as: flood damages to buildings), ROI analysis is equivalent to fully-monetized benefit-cost analysis. Especially for projects that affect multiple outcomes of interest (e.g. damages to various types of infrastructure, human morbidity and mortality, agricultural production etc.) monetized ROI analysis is preferable as it allows for an The ROI time (TROI) or payback period represents the easier, integrated comparison of the differential impacts time necessary for the ROI of the intervention to reach of alternative investments (i.e. projects or programs) on unity (1). The ROI of the intervention is less than unity if those outcomes. total implementation costs of the intervention exceed the reduction in expected annual losses it produces. The EAL is the key input to a benefit-cost analysis of intervention scenarios. Specifically, the benefit of the In addition to ROI, the net present value (NPV) and intervention scenario is the difference in the present internal rate of return (IRR) were also used to assess the values (PV) of the expected annual (t) losses experienced viability of the different scenarios. For a project to be in the baseline (S0) and intervention (SI) scenarios, considered viable, the NPV must be positive. This places respectively: greater weight on values occurring closer to the present, which means that the future benefits of restoration projects will be down-weighted compared with the upfront investment costs, and have to be substantial in order for a project to be viewed positively. Projects can also be evaluated by estimating the IRR, which is the where EAL is the expected annual expenditures incurred discount rate at which the total net present value of the to repair or replace damaged structures during their project falls to zero. reference lifetime (n), and δ is the relevant interest rate on capital (= discount rate). Therefore, the benefit is the PV of the avoided expected annual losses due to Results the implementation of the mitigation strategy Si. All the All of the scenarios resulted in a significant impact on scenarios were compared with the baseline scenario, for EAL associated with flooding in the flood prone areas which an EAL of US$47.30 million was estimated. of the lower Msimbazi catchment, resulting in average annual cost savings ranging from $10 million to $26 million, or from 21% to 54% of present EAL (Table 5.3). The hydraulic results and the resultant hazard curves are presented in Appendices 8 and 9.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 71 Table 5.3 Impacts of Scenarios 1 to 5 on expected annual losses (EAL), and the percentage change in EAL Reduce exposure  No interventions People and structures in flood prone areas removed from 60m buffer in flood prone areas Baseline Scenario 1 No interventions in catchment US$47.30 million US$37.24 million (-21%) Reduce flood risk Scenario 2 Scenario 3 ↓ GUD interventions in catchment US$28.87 million US$23.16 million (-39%) (-51%) Scenario 4 Scenario 5 GUD with additional storage US$27.78 million US$ 21.64 million (-41%) (-54%) Creation of a development setback zone within 60 m storage (Scenario 4) did not have very much additional of rivers in the flood prone area reduces the number of effect on EAL when compared to GUD interventions buildings at risk from 12 744 to 10 047. Implemented alone (Scenario 2). The main effect was the reduction alone, this intervention would reduce damage costs by of buildings at risk by 21.2% (i.e. 10 047 vs. 12 744). This 21%. When implemented in conjunction with catchment is not so different to Scenario 2 because the floodplain interventions, the damage costs are further reduced. Note storage affects mainly the last part of the first sub- that the 60 m buffer only covers a portion of the flood domain and the fourth and fifth subdomains which prone area. While flood exposure could theoretically be have a low density of buildings (see Figure 3.17 for the eliminated by removing people from the entire flood numerical identification of the subdomains in the case prone area, it is likely to be impractical to achieve more study area). Additional storage resulted in a reduced than the legally-defined 60 m setback area since this area discharge in the last part of the first subdomain. falls within Dar es Salaam’s most built up area. The combination of all interventions in Scenario 5 does Taken alone, interventions to reduce flood risk, which have the highest overall effect, as would be expected. are mainly implemented in the catchment area, can also Since the total number of buildings at risk is reduced, the have a significant effect on EAL (Table 5.3). The GUD number of buildings at risk in the buffer area is reduced interventions described in Scenario 2 were designed as well to 18.4% (i.e. 1977 vs. 2422), and the EAL was to have a significant cumulative effect on the flood reduced by 54%. hydrograph. This led to an estimated 19.6% reduction in the number of buildings at risk (i.e. 10 253 vs. 12 744) The initial costs, NPV, IRR and ROI of the different and a 39% decrease in EAL. Scenarios are summarized in Table 5.4 and presented graphically below (Figure 5.2). Initial costs are seen to When GUD strategies were combined with the floodplain increase from Scenario 1 to 5. It should be noted that setback zone (Scenario 3), this resulted in both reduction resettlement costs account for a large portion of the of the exposure (due to relocation of buildings from initial costs, especially for scenarios 2 and 3. In spite a setback zone within the flood prone area) and of high costs, all the options considered had positive reduction of the flood intensity (due to reduction in the outcomes. hydrograph). The creation of a setback zone involves relocation of buildings and resettlement of households Net present value was highest for Scenarios 2 and from these areas, plus the total number of buildings 3. However, return on investment was highest for at risk is reduced (due to the lower flood intensity). Scenarios 1 and 2. A similar pattern is observed for IRR, Therefore the overall number of buildings at risk in the which appears to exceed hurdle rates in most cases. flood prone area was reduced to 15.5% (i.e. 2046 vs. The results suggest that the investment should initially 2422). This led to an even greater decrease in EAL in the be targeted at implementation of GUD measures in order of 51%. the catchment areas, and that if sufficient funds are available, these should be used to extend the investment The combination of GUD interventions with additional to include resettlement from a setback zone as well (i.e. scenario 3). Factors such as the availability of financial resources, the ROI, the impact on the environment and society, should also be taken into account at a definitive design stage.                                                  Page 72 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Table 5.4 Comparison of scenarios Scenario 1 2 3 4 5 C0 (US$ millions) 62.6 84 138.5 124 178.5 NPV (6%, 35 yrs., US$ millions) 88 125 115 72 69 IRR 19 21 14 12 10 ROI 10 Years 1.18 1.30 0.87 0.93 0.70 ROI 50 Years 2.53 2.34 1.63 1.75 1.33 Figure 5.2 Graphical representation of the results shown in Table 5.4                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 73 This page intentionally blank.                                                  Page 74 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT VI. SUMMARY AND CONCLUSIONS Throughout Africa urbanisation has been taking place It is important to note that this analysis did not at a rate that often outpaces the capacity and planning capture all the costs and benefits associated with the structures of cities to provide the necessary services implementation of GUD interventions. On the positive and regulation. This has led to deterioration of the side, these include the amenity benefits associated environment, living conditions and quality of life within with the creation of green open space areas along the cities, a loss of values associated with green open space riparian zones as well as improvement in biodiversity. areas, and rising government costs associated with A green urban development path would offer a variety reducing risks to people that result from environmental of opportunities for enhancing the livability of the problems, such as flooding. The notion of green urban city. On the negative side, it should be acknowledged development is therefore highly attractive, as it allows that relocation of people away from the setback areas cities to grow in a way that maintains their resilience could generate psychological suffering and anxiety in and standards of living. However, few studies have the affected individuals that is difficult to quantify or investigated what following a more sustainable, green compensate in monetary terms. urban development path will actually cost, and whether these costs can be justified. Moreover, what green Whilst conventional conveyance measures were urban development should look like is also not well not considered during this study it is important to defined, in terms of the degree to which it includes the acknowledge that solving the flooding and water conservation of river buffers and other natural areas, quality problems in Dar es Salaam will likely require the mimicking of natural processes through innovative a combination of conventional and green urban engineering design or the protection of downstream development measures. Within the Msimbazi catchment areas through conventional measures. a number of conveyance measures have been designed as part of the Dar es Salaam Metropolitan Development In this study we investigated the potential feasibility Project. These include the lining of 8.5 km of the of investing in green urban development interventions main drainage channel of the Sinza River and 5.4 km to alleviate flooding problems in Dar es Salaam by of secondary drainage sections along the Msimbazi analysing a range of stormwater management scenarios River. These engineering solutions have been designed that considered measures that either reduced exposure for a 1:25 year flood on the Msimbazi River and for a to flooding, reduced flood risk, or a combination 1:50 flood on the Sinza River. The unit cost of this is of these. The three types of measures considered estimated to be $1500 - 2500 per m, with a total cost of - implementation of restoration and rehabilitation $29 million. This is similar to the cost of the main GUD measures in the catchment, storage basins, moving intervention included in this study; the rehabilitation people away from flood prone areas – all led to and enhancement of middle catchment riparian and decreases in the damage costs of flooding. Absolute floodplain areas which cover 488 ha along the Msimbazi, benefits therefore increase as more measures are Sinza and Ubungo Rivers. combined, but so do costs. Taken alone, catchment rehabilitation measures provided higher net benefit than moving people from the flood prone areas, and also yielded the highest rates of return. The addition of a storage basin added least value, but largely because opportunities for the location of such an intervention were too low down in the catchment to be particularly effective. The results suggest that investment should be secured for the implementation of a combination of rehabilitation measures in the catchment that are specifically designed to attenuate flows and improve drainage, including formal solid waste management and community-based river cleaning programs, reforestation in the upper catchment, the rehabilitation of river buffers in the middle catchment and the reconnection of floodplains in the lower reaches. This could be part of an even broader catchment-to-coast rehabilitation programme for the Msimbazi River system which also aims to address water quality problems and the need for green open space within the rapidly-growing city.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 75 The role of catchment riparian and floodplain areas in Due to the limited availability of data, this study by biodiversity conservation must be emphasised as these necessity utilized simple models and assumptions. areas are considered critical for maintaining ecological While the results strongly suggest that catchment connectivity between terrestrial systems, rivers and rehabilitation interventions would yield a positive estuaries. These areas also include opportunities for outcome in economic terms, the figures presented other beneficial uses, such as sports fields and parks, here are preliminary and warrant further investigation and are more likely to reduce the chances of informal and refinement. The results, do however, provide a resettlement of the floodplain. Community-based river useful step towards informing policies and contributing cleaning programmes also provide important co-benefits to Dar es Salaam’s green urban development path. It including education, social awareness and community is recommended that investment is made into the development as evidenced by the effective operation of development of better hydrological data, through the Mlalakua River Restoration Project in Dar es Salaam. establishment of flow and additional rainfall gauges, as However the success of such programmes depends on well as development of detailed spatial datasets on soils, active support and diversified and resilient funding. land cover, the built environment and the city’s drainage These green urban development interventions, while systems. Moving forward these datasets can then be designed to control flooding impacts, also contribute to used to construct a more definitive analysis. water quality enhancement and present opportunities for generating amenity value, other ecosystem services, and uplift communities. Many of the investments required in the Msimbazi catchment do involve costly rehabilitation (catchment land cover and river-floodplain connection) and relocation of unplanned settlements from river margins, demonstrating that better historic protection of both catchment and floodplain areas would have been a far more efficient development path. This is important for the city to bear in mind as it prepares for rapid expansion, especially toward the south.                                                  Page 76 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT VII. REFERENCES Aceves, M. C. & Fuamba, M. 2016. Methodology for Selecting Best Management Practices Integrating Multiple Stakeholders and Criteria. Part 2: Case Study. Water 8(2): 56. AECOM 2015. Promoting green urban development in African Cities. Dar es Salaam, Tanzania Urban Environmental Profile. Prepared for the World Bank. AECOM Technical Services, Inc. April 2015 Allamano, P., Claps, P., Laio, F. & Thea, C. 2009. A data-based assessment of the dependence of short-duration precipitation on elevation. Physics and Chemistry of the Earth, Parts A/B/C, 34(10), 635-641. Apel, H., Aronica, G.T., Kreibich, H., Thieken, A.H. 2009. Flood risk analyses—how detailed do we need to be? Nat. Hazard, 49(1) 79–98. Armitage, N., Vice, M., Fisher-Jeffes, L., Winter, K., Spiegel, A., & Dunstan, J. 2013. Alternative Technology for Stormwater Management. The South African Guidelines for Sustainable Drainage Systems. WRC Report No. TT 558/13. Available at: http://www.wrc.org.za/Knowledge%20Hub%20Documents/Research%20Reports TT%20558-13.pdf. Accessed 1 March 2012 Aronica, G., Tucciarelli, T., & Nasello, C. 1998. 2D Multilevel Model for Flood Wave Propagation in Flood-Affected Areas. J. Water Resour. Plann. Manage. 124(4), 210–217. Bates, P.D. & De Roo, A.P.J. 2000. A simple raster-based model for flood inundation simulation. Journal of hydrology, 236(1), 54-77. Bates, P.D., Horritt, M.S., Aronica, G. & Beven, K. 2004. Bayesian updating of flood inundation likelihoods conditioned on flood extent data. Hydrological Processes, 18(17), 3347-3370. Biscarini, C., Di Francesco, S., Nardi, F. & Manciola, P. 2013. Detailed simulation of complex hydraulic problems with macroscopic and mesoscopic mathematical methods. Mathematical Problems in Engineering. vol. 2013, Article ID 928309, doi:10.1155/2013/928309. Boyd, J., Epanchin-Niell, R. S., & Siikamäki, J. 2012. Conservation return on investment analysis: a review of results, methods, and new directions. Resources for the Future Discussion Paper, (12-01). Brown, W. & Schueler, T.R. 1997. The Economics of Stormwater BMPs in the Mid-Atlantic Region: Final Report: an Examination of the Real Cost of Providing Stormwater Control. Center for Watershed Protection, pp17. Büchele, B., Kreibich, H., Kron, A., Thieken, A., Ihringer, J., Oberle, P. & Nestmann, F. 2006. Flood-risk mapping: contributions towards an enhanced assessment of extreme events and associated risks. Natural Hazards and Earth System Science, 6(4), 485-503. BWM, Bureau of Watershed Management. 2010. Riparian Forest Buffer Guidance. Report available at: http://www. elibrary.dep.state.pa.us/dsweb/Get/Document-82308/394-5600-001.pdf. Accessed 1 March 2012. Calvi, G. M. 2013. Choices and criteria for seismic strengthening. Journal of Earthquake Engineering, 17(6), 769-802. Carozza, S., De Risi, R., Jalayer, F. 2013. Guideline to decreasing physical vulnerability in the three considered cities. CLUVA project deliverable D2.5. Available at http://www.cluva.eu/deliverables/CLUVA_D2.5.pdf. Accessed 1 March 2016. Cavan, G., Lindley, S., Yeshitela, K., Nebebe, A., Woldegerima, T., Shemdoe, R., Kibassa, D., Pauleit, S., Renner, R., Printz, A., Buchta, K., Coly, A., Sall, F., Ndour, N.M., Ouédraogo, Y., Samari, B.S., Sankara, B.T., Feumba, R.A., Ngapgue, J.N., Ngoumo, M.T., Tsalefac, M., Tonye, E. 2012. Green infrastructure maps for selected case studies and a report with an urban green infrastructure mapping methodology adapted to African cities  CLUVA project deliverable D2.7. http://www.cluva.eu/deliverables/CLUVA_D2.7.pdf. Accessed 1 March 2012 CCIAM 2013. Analysis of the impacts or urban land use and climate change on coastal forest ecosystem and management. Written by Sylivester Chaisamba and Pamela Levira. Tanzania Meteorological Agency. June 2013                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 77 Chang, S. & Shinozuka, M. 1996. Life-Cycle Cost Analysis with Natural Hazard Risk. Journal of Infrastructure Systems 2(3): 118–126. Chang, L.F., Kang, J.L. & Su, M.D. 2009. Depth-Damage Curve for Flood Damage Assessments Industrial and Commercial Sectors. Proceedings of the 4th IASME / WSEAS Int. Conference on water resources, hydraulics & hydrology. Charvet, I., Suppasri, A., Kimura, H., Sugawara, D. & Imamura, F. 2015. Fragility estimations for Kesennuma City following the 2011 Great East Japan tsunami based on maximum flow depths, velocities and debris impact, with evaluation of the ordinal model’s predictive accuracy. Nat Hazards 79(3):2073–2099 Cobby, D.M., Mason, D.C., Horritt, M.S. & Bates, P.D. 2003. Two‐dimensional hydraulic flood modelling using a finite‐ element mesh decomposed according to vegetation and topographic features derived from airborne scanning laser altimetry. Hydrological Processes, 17(10): 1979-2000. Construction Industry Research and Information Association (CIRIA) 2007. The SUDS Manual (CIRIA C697). London. Cornell, C.A. & Krawinkler, H., 2000. Progress and challenges in seismic performance assessment, PEER Center News 3, 4 p. Daily, G. C. & Matson, P. A. 2008. Ecosystem services: From theory to implementation. Proceedings of the National Academy of Sciences, 105(28): 9455-9456 Daudi, L., Lugomela C., Nduku Uku, J. and De Trochal, M. 2012. Effect of nutrient enrichment on seagrass associated meiofauna in Tanzania. Marine Environmental Research 82: 49-58. Daugherty R. L. & Franzini J. B. 1965. Fluid Mechanics, 6th ed. McGraw-Hill, New York. pp. 338-349. Department of Environmental Protection. 2010. Riparian Forest Buffer Guidance. Bureau of Watershed Management, the Pennsylvania Clean Streams Law. Document 394-5600-00. Department of Co-operative governance and Traditional Affairs (DoCGTA) 2010. Municipal Infrastructure: an industry guide to infrastructure service delivery levels and unit costs. Republic of South Africa. De Martino, G., De Paola, F., Fontana, N., Marini, G., & Ranucci, A. 2012. Experimental assessment of level pool routing in preliminary design of floodplain storage. Science of the Total Environment, 416, 142-147. De Paola, F., Ranucci, A., & Feo, A. 2013. Antecedent moisture condition (SCS) frequency assessment: a case study in southern Italy. Irrigation and Drainage, 62(S2), 61-71. De Paola, F., Giugni, M., Topa, M. E., & Bucchignani, E. 2014. Intensity-Duration-Frequency (IDF) rainfall curves, for data series and climate projection in African cities. Springer Plus, 3(1), 133. De Paola, F., Galdiero, E., Giugni, M., & Pugliese, F. 2015. Sustainable development of storm-water systems in African cities considering climate change. Procedia Engineering, 119, 1181-1191. De Risi, R. 2013b. A probabilistic bi-scale frame work for urban flood risk assessment. PhD dissertation. Department of structures for Engineering and Architecture, University of Naples Federico II, Naples De Risi, R., Goda, K., Mori, N., & Yasuda, T. (2016). Bayesian tsunami fragility modeling considering input data uncertainty. Stochastic Environmental Research and Risk Assessment, 1-17. De Risi, R., Jalayer, F., Iervolino, I., Kyessi, A., Mbuya, E., Yeshitela, K., Yonas, N. 2012. Guidelines for vulnerability assessment and reinforcement measures of adobe houses. CLUVA project deliverable D2.4. Available at http:// www.cluva.eu/deliverables/CLUVA_D2.4.pdf. Accessed 1 March 2016. De Risi, R., Jalayer, F., De Paola, F., Iervolino, I., Giugni, M., Topa, M.E., Mbuya, E., Kyessi, A., Manfredi, G. & Gasparini,  P. 2013a. Flood Risk Assessment for Informal Settlements, Natural Hazards, 69(1), 1003-1032.                                                  Page 78 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT De Risi, R., Jalayer, F., Iervolino, I., Manfredi, G., & Carozza, S. 2013c. VISK: a GIS-compatible platform for micro-scale assessment of flooding risk in urban areas. In COMPDYN, 4th ECCOMAS Thematic conference on computational methods in structural dynamics and earthquake engineering. Kos Island, Greece. De Risi, R., Jalayer, F., De Paola, F., & Giugni, M. 2014. Probabilistic delineation of flood-prone areas based on a digital elevation model and the extent of historical flooding: The case of Ouagadougou. Boletín Geológico Minero, 125, 329-340. De Risi, R., Jalayer, F., & De Paola, F. 2015. Meso-scale hazard zoning of potentially flood prone areas. Journal of Hydrology, 527, 316-325. De Wolf, H., Ulomi, S.A., Backeljau, T. and Blus, R. 2001. Heavy metal levels in the sediments of four Dar es Salaam mangroves accumulation in, and effect on the morphology of the periwinkle, Littoraria scabra (Mollusca: Gastropoda). Environment International 26: 243-249. Di Baldassarre GD, Schumann G, & Bates P 2009. Near real time satellite imagery to support and verify timely flood modelling. Hydrological Processes, 23(5), 799-803. Dym, C. L., Little, P., Orwin, E. J., & Spjut, R. E. 2009. The Design Process. Engineering design: a projectbased introduction, 17-32. Dodov, B.A. & Foufoula-Georgiou, E. 2006. Floodplain morphometry extraction from a high-resolution digital elevation model: a simple algorithm for regional analysis studies. Geoscience and Remote Sensing Letters, IEEE, 3(3): 410-413. Drake, C. W. 2014. Assessment of flood mitigation strategies for reducing peak discharges in the Upper Cedar River watershed. MS (Master of Science) thesis, University of Iowa. Available at: http://ir.uiowa.edu/etd/4617. Accessed 1 March 2016. Dutta, D. & Tingsanchali, T. 2003. Development of loss functions for urban flood risk analysis in Bangkok, in 2nd International Conference on New Technologies for Urban Safety in Mega Cities of Asia. University of Tokyo, Japan. p. 229-238. Ellingwood, B. R. & Wen, Y.K. 2005. Risk-benefit-based design decisions for low-probability/high consequence earthquake events in Mid-America. Progress in Structural Engineering Material, 7:56–70. El‐Nasr, A. A., Arnold, J. G., Feyen, J., & Berlamont, J. 2005. Modelling the hydrology of a catchment using a distributed and a semi‐distributed model. Hydrological processes, 19(3), 573-587. Environment Agency, 2007. Cost-benefit of SUDS retrofit in urban areas. Science Report – SC060024. Ertas, A., & Jones, J. C. 1996. The engineering design process. New York: Wiley. Faber, M.H. & Rackwitz, R. 2004. Sustainable decision making in Civil Engineering. Structural Engineering International, 3:237–242. Fabio, P., Aronica, G.T. & Apel, H. 2010. Towards automatic calibration of 2-D flood propagation models, Hydrology and Earth System Sciences, 14(6), 911-924. FEMA P-58-1. 2012. Seismic Performance Assessment of Buildings. Volume-1 – Mythology. Fletcher, T., Duncan, H., Lloyd, S, & Poelsma, P. 2003. Stormwater Flow and Quality and the Effectiveness of Non- proprietary Stormwater Treatment Measures. Report for the NSW EPA. April 2003. Cooperative Research Centre for Catchment Hydrology, Melbourne, Victoria. Fletcher, T.D., Shuster, W.F., Hunt, W.F., Ashley, R., Butler, D., Arthur, S., Trowsdale, S., Barraud, S., Semadeni-Davies, A., Bertrand-Krajewski, J-L., Mikkelsen, P.S., Rivard, G., Uhl, M., Dagenais, D. & Viklander, M. 2015. SUDS, LID, BMPs, WSUD and more – The evolution and application of terminology surrounding urban drainage. Urban Water Journal 12:7. 525-542, DOI: 10.1080/1573062X.2014.916314.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 79 FLO-2D S.I., FLO-2D® User’s Manual. 2004: Nutrioso, Arizona. Food and Agrciultural Organisation of the United Nations (FAO). 2011. Forests beneath the grass. Edited by Durst, P.B., Sajise, P. & Leslie, R.N. Proceedings of the Regional Workshop on Advancing the Application of Assisted Natural Regeneration for Effective Low-Cost Restoration. Frangopol, D.M., Lin, K.Y. & Estes, A.C. 1997. Life-Cycle Cost Design of Deteriorating Structure. ASCE Journal of Structural Engineering, 123(10): 1390-1401. Frangopol, D.M., Kallen, M.J. & Noortwijk, J.M. 2004. Probabilistic models for life-cycle performance of deteriorating structures: review and future directions. Progress in Structural Engineering Material, 6:197–212. Gallant, J.C. & Dowling, T.I. 2003. A multiresolution index of valley bottom flatness for mapping depositional areas. Water Resources Research, 39(12). Gencturk, B. & Elnashai, A. S. 2012. Life cycle cost considerations in seismic design optimization of structures. Structural Seismic Design Optimization and Earthquake Engineering: Formulations and Applications, 1-22. Giandotti, M. 1940. Previsione empirica delle piene in base alle precipitazioni meteoriche, alle caratteristiche fisiche morfologiche dei bacini; Applicazione del metodo ad alcuni bacini dell’appennino ligure. Pubbl, 2, 5-13. (In Italian). Gill, S.E., Handley, J.F., Ennos, A.R., Pauleit, S., Theuray, N. & Lindley, S.J. 2008. Characterising the urban environment of UK cities and towns: a template for landscape planning in a changing climate. Landscape and Urban Planning 87: 210–222. Gissing, A., & Blong, R. 2004. Accounting for variability in commercial flood damage estimation. Australian Geographer, 35(2), 209-222. Giugni, M., De Paola, F., Topa M.E. 2012. Guidelines on engineering design and management of storm water systems. CLUVA project deliverable D2.2. Available at http://www.cluva.eu/deliverables/CLUVA_D2.2.pdf. Accessed 1 March 2016. Goda, K., & Hong, H. P. (2008). Estimation of Seismic Loss for Spatially Distributed Buildings. Earthquake Spectra, 24(4), 889-910. Guha-Sapir, D., Below, R., Hoyois, P.H. 2015. EM-DAT: international disaster database. Available viawww.emdat.be. Accessed 1 March 2016. Hallegatte, S. & Przyluski, V. 2010. The Economics of Natural Disasters: Concepts and Methods. World Bank Policy Research Working Paper 5507. Washington DC: The World Bank. https://openknowledge.worldbank.org/ bitstream/handle/10986/3991/WPS5507.pdf Hamilton, R.A. 2015. Available at: http://www.treeimprovement.org/sites/default/files/reforestation_investment_ does_it_pay.pdf. Accessed 1 March 2012. Hagen, E., & Lu, X. X. (2011). Let us create flood hazard maps for developing countries. Natural hazards, 58(3), 841- 843. Herath, S., Dutta, D., & Musiake, K. (1999). Flood damage estimation of an urban catchment using remote sensing and GIS. In Proc. the Eighth International Conference on Urban Storm Drainage. Hill, A. & Linder, C. 2010. Modelling informal urban growth under rapid urbanisation. Prepared for the TU Dortmund University. June 2010 Hull City Council, 2015. Hull City Council Local Flood Risk Management Strategy. United Kingdom ISO 15686-5. 2006. Buildings – Service life planning – Part 5: Life cycle costing. International Organization for Standardization.                                                  Page 80 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT International Union for the Conservation of Nature (IUCN), World Wide Fund for Nature (WWF). 2002. Forest landscape restoration; the opportunities and challenges for Eastern Africa. A synthesis report of a regional workshop in Mombasa, Kenya. Jiang, Y., Yuan, Y. & Piza, H. 2015. A review of applicability and effectiveness of low impact development/ green infrastructure practices in arid/semi-arid United States. Environments 2: 221–249; doi:10.3390/ environments2020221 Jacobi P. 1997. Importance of vegetable production systems in Dar es Salaam, Tanzania. Dar es Salaam: Urban Vegetable Promotion Project. Jalayer, F., De Risi, R., Manfredi, G., De Paola, F., Topa, M.E., Giugni, M., Bucchignani, E., Mbuya, E., Kyessi, A. & Gasparini, P. 2013a. From climate predictions to flood risk assessment for a portfolio of structures, 4881- 4888, 11th International Conference on Structural Safety & Reliability (ICOSSAR 2013), Columbia University, New York, NY, June 16-20, 2013. Jalayer, F., De Risi, R., Herslund, L., Garcia-Aristizabal, A. & Carozza, S. 2013b. Report describing the developed methods and its possible relevance for planning and decision-making. CLUVA project deliverable D2.5. Available at http://www.cluva.eu/deliverables/CLUVA_D2.17.pdf. Accessed 1 March 2016. Jalayer, F., De Risi, R., De Paola, F., Giugni, M., Manfredi, G., Gasparini, P., ... & Cavan, G. 2014. Probabilistic GIS-based method for delineation of urban flooding risk hotspots. Natural hazards, 73(2), 975-1001. Jalayer, F., De Risi, R., Kyessi, A., Mbuya, E., & Yonas, N. 2015. Vulnerability of Built Environment to Flooding in African Cities. Book chapter in Urban Vulnerability and Climate Change in Africa, Future City 4, S. Pauleit et al. (eds.). (pp. 77-106). Springer International Publishing, Switzerland. Jalayer, F., Carozza, S., De Risi, R., Manfredi, G., & Mbuya, E. 2016. Performance-based flood safety-checking for non- engineered masonry structures. Engineering Structures, 106, 109-123. Joksimovic, D. & Alam, Z. 2014. Cost efficiency of low impact development (LID) stormwater management practices. Procedia Engineering 89: 734 – 741. Jonkman, S.N., Vrijling, J.K. & Vrouwenvelder, A.C.W.M. 2008a. Methods for the estimation of loss of life due to floods: a literature review and a proposal for a new method, Nat. Hazard, 46:353–389 Jonkman, S. N., Bočkarjova, M., Kok, M., & Bernardini, P. 2008b. Integrated hydrodynamic and economic modelling of flood damage in the Netherlands. Ecological economics, 66(1), 77-90. Kang, J.L., Su, M.D. & Chang, L.F. 2005. Loss functions and framework for regional flood damage estimation in residential area. Journal of Marine Science and Technology, 13(3). 193-199. Kassenga, G. R. & Mbuligwe, S. E. 2009. Impacts of a solid waste disposal site on soil, surface water and groundwater quality in Dar es Salaam City, Tanzania. Journal of Sustainable Development in Africa 10(4): 73-94. Khan S. & Kelman I (2012). Progressive climate change and disasters: connections and metrics. Natural hazards, 61(3), 1477-1481. Kihampa, C. 2013. Environmental exposure and public health concerns of municipal solid waste disposal in Dar es Salaam, Tanzania. Journal of Sustainable Development in Africa 15(3): 198-208. Kirpich, Z. P. 1940. Time of concentration of small agricultural watersheds. Civil Engineering, 10(6), 362. Klingner, W. 2014. The effects of increased infiltration and distributed storage on reducing peak discharges in an agricultural Iowa watershed: the Middle Raccoon River. MS (Master of Science) thesis, University of Iowa. Available at: http://ir.uiowa.edu/etd/4664. Accessed 1 March 2016. Kong, J. & Frangopol, D. 2003. Evaluation of Expected Life-Cycle Maintenance Cost of Deteriorating Structures. Journal of Structural Engineering, 129(5), 682–691.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 81 Kowalik, T. & Walega, A. 2015. Estimation of CN parameter for small agricultural watersheds using asymptotic functions. Water, 7(3), 939-955. Kundzewicz, Z. W. 2002. Non-structural flood protection and sustainability. Water International, 27(1), 3-13. Leader, T. & Wallingford, H.R. 2009. Language of risk. Deliverable D32.2. FLOODsite project. http://floodsite.net/ html/partner_area/project_docs/T32_04_01_FLOODsite_Language_of_Risk_D32_2_v5_2_P1.pdf. Accessed 01 March 2016 Liu, Y., Bralts, V.F. & Engel, B.A. 2015. Evaluating the effectiveness of management practices on hydrology and water  quality at watershed scale with a rainfall-runoff model. Science of the Total Environment 511: 298–308. Lynch, L. & Tjaden, R. 2000. When a Landowner Adopts a Riparian Buffer – Benefits and Costs. University of Maryland, Department of Environmental Protection. Report available at: https://extension.umd.edu/sites/ default/files/_docs/programs/riparianbuffers/FS774.pdf. Accessed 01 March 2016 Machiwa, J.F. 1992. Heavy metal content in coastal sediments of Dar es Salaam, Tanzania. Environment International 18(4):409-415. Maione, U. & Mosiello, U. 1993. Elementi di statistica per l’idrologia. (In Italian). Marco, J.B. 1994. Flood risk mapping. Chapter 20 in Coping with Floods. Rossi G, Harmancioğlu N & Yevjevich V (eds.) Springer Netherlands; 257, 353-373. McPherson, G., Simpson, J.R., Peper, P.J. & Xiao, Q. 1999. Benefit-Cost analysis of Modesto’s Municipal urban forest. Journal of Arboriculture 25(5): 235-248. Merz, B., Thieken, A.H. & Gocht, M. 2007 Flood Risk Mapping At The Local Scale: Concepts and Challenges. Chapter 13 of Flood Risk Management in Europe. Begum S, Stive MF, Hall J, (eds.) Springer Netherlands; 25, 231-51. Mockus, V. 1972. National engineering handbook, Section 4, hydrology. US Soil Conservation Service, Washington, DC. Morales Torres, A., Perales Momparler, S., Jefferies, C. & Domenech, I.A. 2015. Report on Stormwater Management. E2STORMED PROJECT, Improvement of energy efficiency in the water cycle by the use of innovative storm water management in smart Mediterranean cities, www.e2stormed.eu Michie, R. 2010. Cost Estimate to Restore Riparian Forest Buffers and Improve Stream Habitat in the Willamette Basin, Oregon. State of Oregon, Department of Environmental Quality, Watershed Management Section. Report available at: http://www.deq.state.or.us/wq/tmdls/docs/WillametteRipCost030310.pdf. Accessed 01 March 2016. Mozambique-LNG. 2015. Mozambique Gas Development Project, Resettlement Action Plan. Anadarko Mozambique Area 1. Document number: EA-MZ-SR0000-RRG-U14-00006-29. Mremi, S.D. & Machiwa, J.F. 2003. Heavy metal contamination of mangrove sediments and the associated biota in Dar es Salaam, Tanzania. Tanzania Journal of Science 29(1): 61-76. Mrutu, A., Nkotagu, H. & Luilo, G. 2013. Spatial distribution of heavy metals in Msimbazi River mangrove sediments in Dar es Salaam coastal zone, Tanzania. International Journal of Environmental Sciences 3(5):1641-1655. Mtanga, A. & Machiwa, J. 2007. Assessment of heavy metal pollution in sediment and polychaete worms from the Mzinga Creek and Ras Dege mangrove ecosystems, Dar es Salaam, Tanzania. Western Indian Ocean Journal of Marine Science 6(2):125–135. Muzuka, A.N.N. 2007. Distribution of heavy metals in the coastal marine surficial sediments in the Msasani Bay-Dar es Salaam harbour area. Western Indian Ocean Journal of Marine Science 6(1):73–83. Mwegoha, W.J.S. & Kihampa, C. 2010. Heavy metal contamination in agricultural solid and water in Dar es Salaam city, Tanzania. African Journal of Environmental Science and Technology 4(11): 763-769                                                  Page 82 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Nadal NC, Zapata RE, Pagán I, López R and Agudelo J 2010. Building Damage due to Riverine and Coastal Floods. Journal of Water Resources Planning and Management, 136(3), 327-336. Natural Resources Conservation Service (NRSC) 2005. Economics and Survival of Hand-Planted Riparian Forest Buffers in West Central Maine. US Department of Agriculture, Maine. NEMC 2009. Dar es Salaam River Basin Management Project. Draft Project Concept. O’Brien, J.S., Julien, P.Y. & Fullerton, W.T. 1993. Two-dimensional water flood and mudflow simulation. Journal of hydraulic engineering, 119(2), 244-261. Paths for all. 2014. Estimating price guide for path projects. Scotland. Pauleit, S. & Duhme, F. 2000. Assessing the environmental performance of land cover types for urban planning. Landscape and Urban Planning, 52(1): 1-20. Pistrika, A. & Tsakiris, G. 2007. Flood Risk Assessment: A Methodological Framework, Water Resources Management: New Approaches and Technologies, European Water Resources Association, Chania, Crete - Greece, 14-16 June. Pistrika, A.K. 2010. Flood damage estimation based on flood simulation scenarios and a GIS platform, European Water, 30: 3-11. Pistrika, A.K. & Jonkman, S.N. 2010. Damage to residential buildings due to flooding of New Orleans after hurricane Katrina. Nat. Hazard, 54:413–434 Porter, K., Kennedy, R., & Bachman, R. 2007. Creating fragility functions for performance-based earthquake engineering. Earthquake Spectra, 23(2), 471-489. Prime Minister’s Office Regional Administration and Local Government (PMO-RALG) 2014a. Resettlement Action Plan (RAP) for the Proposed Mto Ng’ombe Sub-project, Kinondoni Municipality. Surface Water Drainage System Subproject under the Dar es Salaam Metropolitan Development Project (DMDP). Tanzania Strategic Cities Project (TSCP). Prime Minister’s Office Regional Administration and Local Government (PMO-RALG) 2014b. Environmental impact statement of the proposed rehabilitation/improvement of surface water drainage system in Ilala Municipality, Dar es Salaam Region. Surface Water Drainage System Subproject under the Dar es Salaam Metropolitan Development Project (DMDP). Tanzania Strategic Cities Project (TSCP). Reese, S., Bradley, B. A., Bind, J., Smart, G., Power, W., & Sturman, J. 2011. Empirical building fragilities from observed damage in the 2009 South Pacific tsunami. Earth-Science Reviews, 107(1), 156-173. Reilly, F. & Brown, K. 2011. Investment analysis and portfolio management. 10th Edition. Cengage Learning. Australia. pp 1058. Sakijege, T., Sartohadi, J., Marfai, M.A., Kassenga, G.R. & Kasala, S.E. 2014. Assessment of adaptation strategies to flooding: a comparative study between informal settlements of Keko Machungwa in Dar es Salaam, Tanzania and Sangkrah in Surakarta, Indonesia. Jàmbá:Journal of Disaster Risk Studies 6(1), 1-10. Scawthorn, C., Blais, N., Seligson, H., Tate, E., Mifflin, E., Thomas, W., Murphy, J. & Jones, C. 2006a. HAZUS-MH Flood Loss Estimation Methodology. I: Overview and Flood Hazard Characterization. Natural Hazards Review 7, SPECIAL ISSUE: Multihazards Loss Estimation and HAZUS, 60–71. Scawthorn, C., Flores, P., Blais, N., Seligson, H., Tate, E., Chang, S., Mifflin, E., Thomas, W., Murphy, J., Jones, C. & Lawrence, M. 2006b. HAZUS-MH Flood Loss Estimation Methodology. II. Damage and Loss Assessment. Natural Hazards Review 7, SPECIAL ISSUE: Multihazards Loss Estimation and HAZUS, 72–81. Schwarz, J. & Maiwald, H. 2008. Damage and loss prediction model based on the vulnerability of building types. 4th International symposium of Flood Defence, Toronto, Canada, May 6-8.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 83 Severn Trent Water, 2015. Severn Trent Environmental Protection Scheme (STEPS). United Kingdom Smith, D.I. 1994. Flood damage estimation - A review of urban stage-damage curvesand loss function. - Water SA, 20(3), 231-238. Soil Conservation Service, 1972. National engineering handbook, Section 4, Hydrology. U.S. Department of Agriculture, Washington D.C., 762 p. Soil Conservation Service, 1975. Urban hydrology for small watersheds, U.S. Department of Agriculture, Washington D. C., updated 1986. Suppasri, A., Mas, E., Charvet, I., Gunasekera, R., Imai, K., Fukutani, Y., ... & Imamura, F. 2013. Building damage characteristics based on surveyed data and fragility curves of the 2011 Great East Japan tsunami. Natural Hazards, (2), 319-341. Tanzania Meteorological Agency (TMA) 2011. Urban Poverty & Climate Change in Dar es Salaam, Tanzania: A Case Study. March 10, 2011.Matari et al. 2008 Tanzania Standards for Receiving Water, Effluents and soils (TZS 789:2003). Tanzania Bureau of Standards. TEEB. 2010. The Economics of Ecosystems and Biodiversity: Mainstreaming the Economics of Nature: A Synthesis of the Approach, Conclusions and Recommendations of TEEB. Thampapillai, D.J., & Musgrave, W.F. 1985. Flood damage mitigation: a review of structural and nonstructural measures and alternative decision frameworks. Water Resources Research, 21(4), 411-424. TNC, 2015. Upper Tana-Nairobi Water Fund: A Business Case. Available at: http://www.nature.org/ourinitiatives/ regions/africa/upper-tana-nairobi-water-fund-business-case.pdf. Accessed 1 March 2012. Topa, M. E., Giugni, M., & De Paola, F. 2014. Off-stream floodplain storage: Numerical modeling and experimental analysis. Journal of Irrigation and Drainage Engineering, 141(1), 04014040. Toronto and Region Conservation Authority (TRCA) 2013. Assessment of Life Cycle Costs for Low Impact Development Stormwater Management Practices. Sustainable Technologies Evaluation Program. Xiao, Q. & McPherson, G. 2002. Rainfall interception by Santa Monica’s municipal forest. Urban Ecosystems 6: 291- 302. UN Habitat 2010. Citywide action plan for upgrading unplanned and unserviced settlement in Dar es Salaam. United Nations Human Settlements Programme, Kenya. URS, 2003. Water Sensitive Urban Design Technical Guidelines for Western Sydney. Report prepared for the Upper Parramatta River Catchment Trust, Sydney, New South Wales. URT, 2014 TSCP. Tanzania Strategic Cities Project (TSCP) Environmental Impact Statement of the Proposed Rehabilitation/Improvement of Surface Water Drainage System in Kinondoni Municipality, Dar es Salaam Region. Prepared by COWI in association with PROCESL. November 12, 2014. Viparelli, C. 1963. Ricostruzione dell’idrogramma di piena. L’Energia Elettrica, 6, 421-428 (In Italian) Waidler, D., White, M., Steglich, E. Wang, S. Williams, J., Jones, C.A. & Srinivasan, R. 2009. Conservation practice modeling guide for SWAT and APEX. Draft - August 2009, provided by Spatial Sciences Laboratory, Texas A&M University, College Station, Texas Watkiss, P. Downing, T., Dyszynski, J., Pye, S. et al. 2011. The Economics of Climate Change in the United Republic of Tanzania. Report to Development Partners Group and the UK Department for International Development. Published January 2011.Kihampa 2013. World Health Organization (WHO). 2004. Guidelines for Drinking Water Quality. 3rd Ed., World Health Organization, ISBN: 92-4-154638-7, 516. Yang, T.H., Chen, Y.C., Chang, Y.C., Yang, S.C. & Ho, J.Y. 2015. Comparison of Different Grid Cell Ordering Approaches in a Simplified Inundation Model. Water, 7(2), 438-454.                                                  Page 84 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT APPENDIX I: BASELINE INUNDATION RESULTS First sub-domain                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 85                                                  Page 86 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Second sub-domain                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 87                                                  Page 88 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Third sub-domain Third sub-domain                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 89                                                  Page 90 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Fourth sub-domain                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 91                                                  Page 92 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Fifth sub-domain                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 93                                                  Page 94 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT APPENDIX II. BUILDING COST ESTIMATES In this appendix the costs used to estimate the total Table A2.1 Unit construction costs for different building typologies in Dar es Salaam potential losses for the buildings affected by flooding is presented. Building type Unit cost $/m2 Two main sources have been used: the National Swahili house 100 Construction Council of Tanzania (www.ncc.org.zm) Bungalow, corrugated iron sheet roofing 120 and the 26th edition of annual African Property and Bungalow, tiled roof 150 Construction Handbook (http://www.coolrooftoolkit. org/wp-content/uploads/2014/07/AEcom- Bungalow, slab roof 180 ConstructionHandbookFinal_v2.pdf) released by AECOM. Maisonette double storey, slab roof 170 Flats 150 Table A2.1 lists the unit construction cost for different building typologies in Dar es Salaam. Residential average multi-unit high-rise 667 Residential Luxury unit high rise 894 Residential Individual prestige house 964 Commercial Standard office high rise 823 Commercial Prestige office high rise 1041 Commercial Major shopping centre 765 Industrial light duty factory 616 Industrial heavy duty factory 1100 Car park 490 Table A2.2 Unit construction costs for different Hotels Dar es Salaam Building type Unit cost $/key Budget 90,000 Midmarket 210,000 Upscale 280,000                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 95 This page intentionally blank.                                                  Page 96 Value of Durban’s natural capital and role in Green Urban Development APPENDIX III. CHARACTERIZATION OF THE BUILDINGS IN THE FLOOD PRONE AREA The breakdown of the buildings in the flood prone area are shown in the following tables according to their characteristics. Table A3.1 Breakdown according to the OpenStreet information Table A3.2 Breakdown according to the UMT information (Generic) Number of buildings in Number of buildings in Type Type flood prone area flood prone area Apartments 2 Dwelling 11319 Commercial 219 Feeder Road 8 Commercial / Residential 357 Green Belt 745 Construction 177 Hazard Land 384 House 6 Horticulture 15 Industrial 4 Industrial 160 Mosque 1 Play Ground 8 Public 24 Primary School 4 Residential 8887 Religious 12 School 9 Residential/Commercial 80 Utility 1 Not available 9 Not available 3057 Not available 3057 Table A3.3 Breakdown according to the UMT information (Specific) Number of buildings in Type flood prone area Education & culture 4 Horticulture 15 Major road corridor 8 Mangrove 12 Manufacturing 169 Mixed 7655 Mud/wood/sand brick construction 2423 Other open space 8 Religion 12 Riverine 1565 Villa & single story stone/concrete 873 Not available 3057                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 97 This page intentionally blank.                                                  Page 98 Value of Durban’s natural capital and role in Green Urban Development APPENDIX IV. FLOODPLAIN STRUCTURAL IDENTIFICATION AND BUILDING COSTS In this appendix, the criterion adopted for the structural Three main structural typology are identified: IM (i.e. identification and for the cost assessment is reported. informal masonry), FM (i.e. formal masonry), and RCF Such criterion is based on the analysis of all the potential (i.e. reinforced concrete frame). These are used in the combination of the characteristics reported in Appendix vulnerability assessment. 2 and 3. Sixty-two potential combination are recognized and interpreted in terms of structural typology and Because of the degree of uncertainty on the typology building unit costs. identification, the types of buildings used to estimate the costs of resettlement were reduced to six types (Table A4.2). Table A4.1 Structural identification and cost assessment criterion Characteristics Structural Typology Unit Cost $/m2 Apartments + Green Belt + Riverine IM 120 Commercial + Dwelling + Mixed RCF 745 Commercial + Dwelling + Mud/wood/sand brick construction IM 471 Commercial + Dwelling + Villa & single storey stone/concrete FM 487 Commercial + Feeder + Road Major road corridor FM 487 Commercial + Green + Belt Riverine FM 487 Commercial + Hazard + Land Riverine IM 120 Commercial + Industrial + Manufacturing RCF 858 Commercial/residential + Dwelling + Mixed FM 487 Commercial/residential + Dwelling + Mud/wood/sand brick construction IM 471 Commercial/residential + Dwelling + Riverine IM 471 Commercial/residential + Dwelling + Villa & single storey stone/concrete FM 487 Commercial/residential + Feeder + Road Major road corridor FM 487 Commercial/residential + Green Belt + Riverine IM 471 Commercial/residential + Hazard + Land Riverine IM 471 Commercial/residential + Play Ground + Other open space FM 487 Commercial/residential + Primary School + Education & culture FM 487 Commercial/residential + Residential/Commercial + Mixed IM 471 Construction + Dwelling + Mixed IM 120 Construction + Dwelling + Mud/wood/sand brick construction IM 120 Construction + Dwelling + Riverine IM 120 Construction + Dwelling + Villa & single storey stone/concrete FM 150 Construction + Green Belt + Riverine IM 120 Construction + Hazard Land + Mangrove IM 120 Construction + Hazard Land + Riverine IM 120 Construction + Industrial + Manufacturing RCF 858 House + Dwelling + Mixed IM 120 House + Green Belt + Riverine IM 120 Industrial + Green Belt + Riverine RCF 858                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 99 Table A4.1 Structural identification and cost assessment criterion (continued) Characteristics Structural Typology Unit Cost Industrial + Industrial + Manufacturing RCF 858 Mosque + Dwelling + Mixed FM 150 Public + Dwelling + Mixed FM 150 Public + Dwelling + Mud/wood/sand brick construction IM 120 Public + Dwelling + Villa & single storey stone/concrete FM 150 Public + Hazard Land + Riverine IM 120 Residential + Dwelling + Mixed IM 120 Residential + Dwelling + Mud/wood/sand brick construction IM 120 Residential + Dwelling + Riverine IM 120 Residential + Dwelling + Villa & single storey stone/concrete FM 150 Residential + Feeder + Road Major road corridor IM 120 Residential + Green Belt + Riverine IM 120 Residential + Hazard Land + Mangrove IM 120 Residential + Hazard Land + Riverine IM 120 Residential + Industrial + Manufacturing RCF 858 Residential + Play Ground + Other open space IM 120 Residential + Primary School + Education & culture FM 150 Residential + Religious + Religion FM 150 Residential + Residential/Commercial + Mixed IM 471 School + Dwelling + Mixed IM 120 School + Dwelling + Mud/wood/sand brick construction IM 120 Utility + Dwelling + Mixed FM 150 Not Available + Not Available + Manufacturing RCF 858 Not Available + Not Available + Villa & single storey stone/concrete FM 150 Not Available + Dwelling + Mixed IM 120 Not Available + Dwelling + Riverine IM 120 Not Available + Dwelling + Villa & single storey stone/concrete FM 150 Not Available + Green Belt + Riverine IM 120 Not Available + Hazard Land + Riverine IM 120 Not Available + Horticulture + Horticulture IM 120 Not Available + Industrial + Manufacturing RCF 858 Not Available + Residential/Commercial + Mixed IM 471 Not Available + Residential/Commercial + Riverine IM 471                                                  Page 100 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Table A4.2 Unit construction costs for different building typologies in Dar es Salaam Building type Mean of following types in D-1 Unit cost $/m2 Informal Masonry residential Bungalow, corrugated iron sheet roofing 120 Formal Masonry residential Bungalow, tiled roof; Flats 150 Informal Masonry commercial Bungalow, corrugated iron sheet roofing; Commercial Standard 472 office Formal Masonry commercial Bungalow, tiled roof; 374 Flats; Commercial Standard office Reinforced Concrete Frame commercial or residential Residential average multi-unit high-rise; 745 Commercial Standard office high rise Reinforced Concrete Industrial Industrial light duty factory 858 Industrial heavy duty factory                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 101 This page intentionally blank.                                                  Page 102 Value of Durban’s natural capital and role in Green Urban Development APPENDIX V. URBAN STORMWATER MANAGEMENT OPTIONS Passive engineering measures to improve conveyance A5.1.2 Enlargement of river channel/canalisation/levees/ dredging These measures are designed to protect areas from flooding by avoiding or mitigating the water flow off Excavation of a river channel involves either deepening stream over the riverbanks, or accommodating the or widening the channel to increase flood control flood adjusting the riverbed carrying out channel capacity. A river can be made to carry larger discharges improvement. Therefore, these kinds of measures try by improving the hydraulic condition of the channel to constrain the inundation without modification of the through measures such as dredging. Similarly, levees hydrograph. Examples are levees, cleaning from debris (embankments) can be built to increase the conveyance or increasing of section of the riverbed, and hydraulic capacity of the channel. bypass, also known as waterways. They involve physical construction to reduce or avoid possible impacts of Levees are generally built as an embankment (i.e. hazards, or application of engineering techniques to earthwork). In the urban context, if there is not enough achieve hazard-resistance and resilience in structures or land area to build such earth structures, then they systems. These kind of measures alter the streamflow are constructed with reinforced concrete or masonry of rivers and channels, resulting in the reduction of the walls. The levees location is designed according to frequency and severity of floods. For example, reservoirs the inundation analyses; their scope is to prevent the reduce peak flows; levees and flood walls confine flows inundation of floodplain. Their height is designed to within predetermined channels; improvements to prevent the inundation associated to a specific return channels reduce the peak stages; and flood ways help period. Once the height is defined, their design will divert excess flow. follow geotechnical rules if they are made with earth, or structural rules if they are concrete walls. To analyse the efficiency of such structures, they are modelled in A5.1.1 Drains and swales the hydraulic routine as a modification of the digital elevation model. These convey flows from built-up areas via small channels, and can generally deal with small floods of 1-2 year return period. A5.1.3 Hydraulic bypass A hydraulic bypass is a new channel built to laminate the peak discharge crossing the floodable area in the urban context. The new channel takes part of the discharge and brings it to the final destination through an alternative path. Construction of a hydraulic bypass is very expensive and requires the identification of the alternative path for the new channel. Figure A5.1 Schematic representation of levees at two side of the Figure A5.2 Schematic representation of a hydraulic bypass watercourse                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 103 A5.2 Active engineering measures to retard runoff Permeable pavements generally do not remove litter The active structural measures are to modify the and other debris from stormwater runoff as it tends to hydrograph by reducing and delaying the maximum peak remain on the surface as the water infiltrates. Soluble discharge. Examples include floodplain storage (in-line pollutants, however, do pass through the permeable or off-line) that stores the flood volume temporarily layer and surfaces that have an aggregate sub-base can in an adequate upstream capacity, leading to flood provide good water quality treatment. Permeable paving attenuation as a result of the discharge being gradually can be used in a variety of locations, such as parking lots, released (Topa et al. 2014). When the discharge falls private and public roads, industrial storage and loading below the maximum allowable flow, the flood volume is areas, bike lanes, walkways and terraces (Armitage et released back to the river (De Martino et al. 2012). Off- al. 2013). The use of this paving is however restricted to stream floodplain storages are often used since they do slopes that are less than 5%, or ideally flat, as the high not interfere with the natural drainage pattern between velocity stormwater from steep slopes does not have the stream and the floodplain, and only an outlet sufficient time to infiltrate before being washed away. structure is needed to regulate the outflow discharge. To ensure the long term effectiveness of permeable pavements regular inspections and maintenance are recommended. Blockage of the fine stone aggregate can A5.2.1 Permeable pavements sometimes be an issue and requires cleaning or replacing Permeable pavements refer to pavements that are if this does occur. This fine aggregate in the joints and constructed in such a way that they promote the slots is known to trap the most pollutants, including heavy infiltration of stormwater runoff through the surface metals. While clogging may be a maintenance concern, into the sub-layers or underlying substrata (Armitage the often enormous infiltration capacity of permeable et al. 2013). Permeable paving provides a surface pavement systems means that considerable clogging can that is suitable for pedestrian and/or vehicular traffic be tolerated (Armitage et al. 2013). while allowing rainwater to infiltrate through the surface. There are a number of different alternatives Permeable pavements are relatively expensive to for the surface material, including brick pavers, porous construct and can have high maintenance costs. concrete, porous asphalt, stone chip, and permeable However, they are incredibly efficient at reducing peak concrete block pavers (Armitage et al. 2013). Permeable flows and reducing runoff volume as well as reducing paving is usually constructed on top of a coarse gravel pollutants. They remove approximately 60-95% of TSS, base which creates the temporary storage facilities 70-90% of hydrocarbons, 50-80% of total phosphorous, and allows stormwater runoff to infiltrate into the 65-80% of total nitrogen and 60-95% of heavy metals substratum, ultimately promoting the recharge of the (Armitage et al. 2013). Permeable pavements do not groundwater table. The stored rainwater can also be provide any amenity, social or ecological benefits. reused for a number of purposes such as watering gardens and lawns (Armitage et al. 2013). Figure A5.3 Permeable paving allows water to soak into the gravel sub-base, temporarily holding the water before it soaks into the ground, or passes to an outfall Source:  susdrain, www.susdrain.org                                                  Page 104 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Advantages Limitations Significantly reduce stormwater discharge rates and volumes from Cannot be used where large sediment loads may be washed or impervious areas carried onto the surface of the paving Reduce peak flows to watercourses reducing the risk of flooding The implementation is generally limited to sites with slopes less downstream and reduce the effects of pollution in runoff on the than 5% environment Flexible and tailored solution that can suit the proposed usage and Risk of long-term clogging and weed growth if poorly design life maintained Allows for dual use of space, so there is no additional land take. They Not normally suitable for high traffic volumes and speeds increase the ‘usable’ area by utilising roadways, driveways and parking greater than about 50 km/hr, or for usage by heavy vehicles lots as stormwater drainage areas and/or high point loads Good community acceptability The pollutant removal ability of permeable pavements is lower than most other SuDS options. Stormwater runoff that is stored can be used to recharge the groundwater table and also be used for several domestic purposes Lined permeable pavement systems can be utilised where foundation or soil conditions limit infiltration processes Advantages Limitations Increases stormwater infiltration and corresponding groundwater If situated in coarse soil strata, groundwater contamination is a recharge possibility Decrease the frequency and extent of flooding Restricted to areas with permeable soils Effective in removing suspended particulates from stormwater Not appropriate on unstable or uneven land, or on steep slopes Due to their relatively narrow cross section, they can be utilised in most Prone to failure if sediment, debris and/or other pollutants are urban areas able to clog the gravel surface and/or backfilled aggregate material The pollutant removal ability of permeable pavements is lower than most other SuDS options. Negligible visual impact as they are generally below ground Lined permeable pavement systems can be utilised where foundation or soil conditions limit infiltration processes A5.2.2 Infiltration trenches Infiltration trenches are excavated trenches which In the first year of construction maintenance is are lined with geotextile and filled with rock, or important, especially after the first large rainfall event. other granular materials, and are designed to receive The trench needs to be assessed for performance and stormwater runoff from contiguous properties in urban any sediment and debris build up which can cause areas (Armitage et al. 2013). They create temporary clogging (Armitage et al. 2013). Removal and cleaning of subsurface storage of stormwater runoff thereby stone may be necessary. enhancing the natural capacity of the ground to store and drain water. Infiltration trenches allow water to infiltrate into the surrounding soils from the bottoms and sides of the trench. They usually have a rectangular vertical cross-section and are designed to receive stormwater runoff from adjacent properties and transportation links such as asphalt roads and footpaths (Armitage et al. 2013). The amount of water that can be disposed of by an infiltration trench within a specified time is dependent on the infiltration potential of the surrounding soil, the size of the trench, and the bulk density of the fill material. Stormwater runoff is treated by physical filtration to remove solids, adsorption onto the material in the trench and biochemical reactions involving micro-organisms in the soil.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 105 The construction costs associated with infiltration The size of the soakaway is dependent on the porosity of trenches are not very high, making them one of the aggregate used to fill the excavated pit. The soakaway the more cost effective options in terms of their empties either by percolation of the stormwater directly ability to reduce runoff volume and treat pollutants. into the underlying soil or via perforated sub-drains Their maintenance costs can be higher than other installed within the pit. Soakaways are usually designed to interventions, especially in areas that have fine grained store the entire volume from a design storm and be able soils. Infiltration trenches remove approximately 70- to infiltrate at least half of this volume within 24 hours to 80% of TSS, 60-80% of total phosphorous, 25-60% of create further capacity for the runoff from subsequent total nitrogen, and 60-90% of heavy metals (Armitage rainfall events (Armitage et al. 2013). A single soakaway et al. 2013). Their amenity and conservation value is can serve an area of roughly 1000 m2 but groups of poor, however they are generally constructed under the soakaways can serve areas as large as 100 000 m2 ground and so the aesthetic impact is negligible. (Armitage et al. 2013). They range in depth from between 1 – 4 metres but are usually approximately 1.5 metres in depth when serving a single building. A5.2.3 Soakaways (sub-surface infiltration trenches) The basic construction costs include clearing and Soakaways usually comprise an underground storage removing of topsoil, surface bed preparation, pit area that is packed with course aggregate or other excavation, supplying and laying filter fabric or porous media that gradually discharges stormwater into geotextile, supplying and laying of aggregate fill or the surrounding soil (Armitage et al. 2013). Soakaways porous media, supplying and laying of building sand, are similar to infiltration trenches in their operation supplying and laying of slotted pipes, top soiling of and are also known as sub-surface infiltration beds or verged areas, and grassing of surface area. sub-surface infiltration trenches). They usually handle roof runoff from single buildings, such as large industrial The amount of water disposed of by soakaways depends buildings. Multiple soakaways can be linked to each on the infiltration potential of the surrounding soil, the other to drain larger areas such as parking lots or size of the pit and the bulk density of the fill material. major roadways. The type of aggregate material used The amount of water retained by a soakaway is based on determines the infiltration characteristics of the device. the roof area of the building and the peak rainfall event Modular geo-cellular structures provide relatively (mm) during the flood season. Soakaways are estimated high stormwater treatment and rates of groundwater to be retain 70-80% of TSS, 25-60% of total nitrogen, recharge (Armitage et al. 2013). 60-80% of total phosphorous, 60-90% of E.coli and 60-90% of heavy metals. Soakaways are relatively cost- effective in terms of runoff reduction as well as in terms of their ability to remove suspended solids. Their ability to remove dissolved nutrients is not as effective as some other interventions. Advantages Limitations Have reasonable design lives of up to 20 years if maintained properly and Usually limited to relatively small connected areas relatively easy to construct Significantly decrease stormwater runoff volume, peak flow and rate They do not function well when constructed on steep slopes or in unstable areas Particularly effective in removing particulate and suspended stormwater Sub-drain piping systems must be utilised runoff pollutants when soakaways are implemented in very fine silt and clay stratum because of the low infiltration rates Reduce downstream erosion and flooding Sedimentation within the collection chambers will cause a gradual reduction in the storage capacity Minimal net land take Ecological and amenity value is poor                                                  Page 106 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Figure A5.4 Soakaways are square or circular excavations either filled with rubble or other aggregate fill that are able to attenuate and treat significant amounts of stormwater. They can be grouped and linked together to drain large areas such as highways and industrial areas reducing the amount of runoff entering streams and rivers. A5.2.4 Green roofs There are three main types of green roofs, namely: Green roofs comprise a multi-layered system that covers extensive green roofs, intensive green roofs and simple the roof of a building with vegetative cover (Armitage intensive green roofs (Armitage et al. 2013). Extensive et al. 2013). The use of vegetative roof covers and roof green roofs generally incorporate low growing and low gardens is an important source control for stormwater maintenance vegetation that covers the whole roof runoff as they are designed to intercept and retain surface. The roof is only accessed for maintenance precipitation close to where it falls (i.e. at the source) purposes. Usually indigenous vegetation such as mosses, reducing the volume of runoff and attenuating peak herbs and grasses are used as they are relatively flows. Green roofs provide great benefits in densely self-sustaining. Intensive green roofs incorporate urbanised areas where there tends to be less space for planters and trees and tend to have a high level of some of the other BMP interventions. Green roofs are accessibility (Armitage et al. 2013). Rainwater harvesting usually constructed on flat or gently sloping roof tops no interventions are often used as the primary irrigation greater than 30 degrees. The vegetative layer sits upon source for intensive green roof flora. These roof systems a drainage layer which in turn lies upon a water proof require more intensive and frequent maintenance. membrane to prevent any leakage below (Armitage et al. Simple intensive green roofs are a combination of 2013). Green roofs that are constructed in this manner both extensive and intensive green roofs, having both typically have weights of between 40 – 60 kg per m2. larger plants as well as low lying ground cover. These The structural design of the green roof needs to account roofs generally require high levels of maintenance such for the additional weight of the green roof component as cutting, fertilizing and watering – which requires materials and expected water detention volumes increased accessibility (Armitage et al. 2013). (Armitage et al. 2013). Green roofs are particularly effective when constructed on roofs with large surface areas such as commercial or industrial buildings or large residential blocks. Irrigation may be required to keep the roof green during particularly dry periods. Figure A5.5 Green roofs achieve runoff treatment and infiltration through the construction of vegetative cover on roofs which increases storage, evapotranspiration and attenuation Source:  susdrain, www.susdrain.org                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 107 Advantages Limitations Good removal capability of atmospherically deposited urban pollutants More costly than conventional roof-runoff practices due to their added structural, vegetative and professional requirements (professionals are required to ensure implementation of the waterproofing and plant requirements Can be designed to closely mimic the pre-development state of buildings Opportunities for retrofitting may be limited by roof structure (size, strength etc.) Ecological, aesthetic and amenity benefits Not appropriate for steep roofs Can be constructed on both new and already existing buildings Detention of water within green roof storage layer may result in failure to the waterproofing membranes which in turn may cause leakage or cause roof collapse Help to insulate and regulate buildings against temperature extremes Plant varieties may be quite limited. Using indigenous vegetation is best Can be applied to high density urban areas May improve air quality No additional land take Maintenance of green roofs include irrigation during There are two types of stormwater collection and reuse establishment of vegetation, inspection for bare systems that are generally applicable to residential, patches, weeds and plants that require replacement. commercial and industrial uses; namely the pumped Leaf litter removal may be required for certain systems supply system and the gravity supply system. In Dar es and any possible stresses related to the roof and building Salaam the gravity supply system would be the most structure need to be checked. practical. The water collected in the tank from the rooftop is then gravity fed into specified application Green roofs are expensive to construct and are one points in and around the building. The harvesting system of the least cost-effective options in terms of the cost could just involve the collection of rainwater from per unit reduction of runoff volume or pollutant loads. rooftops via gutters into a storage tank where water Green roofs remove approximately 60-95% of TSS and can then be collected for use. One large tank could be 60-95% of heavy metals (Armitage et al. 2013). They connected to and supply a number of houses. provide a number of social and aesthetic benefits such as air quality improvement in urban areas, temperature control, and amenity value. A5.2.5 Rainwater Harvesting Rainwater harvesting systems collect and store rainfall from hardened surfaces for later use. With minimal treatment the water that is collected could be used to supplement the potable water supply and can be used for a number of activities such as toilet flushing and irrigating crops and gardens (Armitage et al. 2013). Storage of runoff from roofs and other elevated impervious surfaces is provided by rainwater tanks, barrels, cisterns or other storage structures until the water is required (Armitage et al. 2013). The utilisation of stormwater as a water source not only saves potable water but it also significantly reduces the stormwater discharge from roofs. Rainwater harvesting systems are known to be particularly useful during extreme rainfall events as they help to protect receiving streams and rivers by reducing the initial runoff volumes and the associated polluted (Armitage et al. 2013).                                                  Page 108 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT There are a number of different types of stormwater The initial construction costs associated with the collection and storage systems that are commercially rainwater harvesting system are relatively expensive available. An effective system will include strategically with the tank constituting the most significant cost. placed roof gutters and pipes, a filter sock to catch However, maintenance costs can be low and the water leaves/debris, a rainwater storage facility such as a that the tanks supply to households is an extremely tank or barrel, and a UV disinfection device. Storage important benefit, especially in areas where access to facilities that are child proof, insect and vector proof running water is limited. should be given preference during the selection process, especially if the systems are to be placed in residential areas (Armitage et al. 2013). The following water balance A5.2.6 Vegetated swales equation is often used to calculate the volume of usable Swales are shallow vegetated channels with flat and rainfall or the annual collectable rainfall: sloped sides that are designed to store and convey runoff as well as remove pollutants. Although swales are V = R x A x C x FE usually lined with grass, a variety of different types of vegetation can be used to suit the specific site (Armitage Where: et al. 2013). Swales serve as an alternative option to V = volume of usable rainwater (l) the more typical roadside kerb or gutter and generally have a larger stormwater storage capacity so they help R = average rainfall over a period (mm) to reduce runoff volumes and peak stormwater flows (Armitage et al. 2013). Their ability to store and convey A = Area contributing to runoff (m2) significant volumes means that they require relatively large surface areas in order to function effectively. C = runoff coefficient (0-1) FE = filter efficiency (0-1) For a standard flat roof the runoff coefficient is 0.4 and the filter efficiency is generally recommended to be 0.9 as a conservative estimate (Armitage et al. 2013). Figure A5.6 Diagram of a rainwater harvesting system. The first picture shows high stormwater runoff with none of the rain being collected whereas the second picture shows how rainfall is trapped and collected from the roofs in tanks and the amount of runoff entering streams and rivers is significantly reduced.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 109 Advantages Limitations Can significantly reduce potable water consumption or provides Water quality needs to be monitored and is generally such that significant amounts of water to those that have no access to potable the water can only be used for supplementary purposes water Reduces pollutant loads that enter nearby watercourses Rainwater reuse on a domestic scale is relatively expensive with the storage tanks constituting the most significant cost of the system Attenuates flood peaks Wide variety of storage containers available and generally easy to install Swales are commonly used in combination with other Vegetated swales have low capital costs and are cost- systems, such as buffers and bio-retention interventions, effective in their ability to reduce peak flows and runoff to form a treatment train. In doing so runoff is retained volumes and to reduce pollutants. They have medium and dissolved pollutants in stormwater runoff are also to good amenity potential in that they provide a green removed. The combination of infiltration and bio- alternative to grey infrastructure in urban environments. infiltration removes the dissolved pollutants and the Vegetated swales remove approximately 60-90% of TSS, larger particles are filtered by the vegetation (Armitage 70-90% of hydrocarbons, 25-50% of total phosphorous, et al. 2013). A swale that has been well designed should 30-90% of total nitrogen and 40-90% of heavy metals. provide reduction in impervious cover, pronouncement of the surrounding natural landscape and multiple aesthetic enhancements, and they should be designed A5.2.7 Filter strips to meet flow conveyance requirements and effective Filter strips are maintained grassed areas of land that stormwater pre-treatment (Armitage et al. 2013). They are used to manage shallow overland stormwater are usually suitable for road medians, verges, car parking runoff through several filtration processes in a very runoff areas, park and recreational edges. similar manner to buffer strips (Armitage et al. 2013). Filter strips are usually gently sloping and provide The effective design life of a swale is directly related to opportunities for slow conveyance and infiltration. They the standard of maintenance, particularly in the first two therefore help to attenuate floods peaks and retain years during the period of plant establishment which pollutants. They are commonly designed to accept often requires frequent weed control and replanting runoff from upstream development and are usually (Armitage et al. 2013). Swales have the potential to located between hard-surfaced areas and a receiving manage stormwater indefinitely if they are properly stream, surface water collection or treatment system. maintained. Maintenance activities tend to include They may also be used downstream of agricultural land regular mowing of grassed surfaces, weed control, re- to infiltrate and intercept runoff from these areas. seeding of bare ground, frequent clearing of litter and debris, and watering during extended dry periods. Advantages Limitations Usually less expensive and more aesthetically pleasing than kerbs and Usually require a larger land area than conventional kerb and their associated concrete- and stone-lined channels channel drainage systems Runoff from adjacent impermeable areas is often completely infiltrated Not suitable for steep areas or areas with roadside parking in-situ using swales Reduce stormwater runoff volumes and delay runoff peak flows Risks of blockages in connecting pipe work Retain particulate pollutants as close to the source as possible Limited removal capabilities for soluble pollutants and fine sediment Easy to incorporate into landscaping with low capital costs Standing water in swales has the potential to result in the breeding of mosquitoes and the generation of foul odours Pollution and blockages are visible and easily dealt with                                                  Page 110 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Figure A5.7 Swales are shallow grassed or vegetated channels used to collect and/or move water Source:  susdrain, www.susdrain.org Filter strips use vegetative filtering as a primary means Filter strips are designed specifically to control for of stormwater runoff pollutant removal and if properly nutrients and pollution more so than water quantity and designed are able to remove most sediment and other are therefore more efficient at trapping and reducing settleable solids such as hydrocarbons (Armitage et al. TSS and pollutants than they are at reducing stormwater 2013). Soluble nutrients and heavy metals, however, runoff. Grass filter strips remove approximately 50- are often not adequately removed. The pollutant 85% of TSS, 70-90% of hydrocarbons, 10-20% of total removal and water retention characteristics of filter phosphorous, 10-20% of total nitrogen and 25-40% of strips is determined by the relationship between the heavy metals. length, width, slope and soil permeability compared to the stormwater runoff rate and velocity (Armitage et al. 2013).                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 111 Advantages Limitations Installation and maintenance costs are relatively low and layout and Clogging of subsurface drainage media can occur if maintenance design is flexible is poor Significant removal of suspended solids and hydrocarbons. They trap the Limited potential for the removal of fine sediments and pollutants close to source dissolved pollutants Infiltration of stormwater runoff helps to attenuate flood peaks Stormwater runoff needs to be spread out in order for the strips to operate optimally Integrate well within the natural landscape and can provide open space Minimal stormwater storage capacity and not good at treating areas for recreation as well as amenity value high velocity flows. They are not suitable for steep slopes. A5.2.8 Sand filters There are many different forms of sand filters. They usually comprise of a sedimentation chamber that is linked to an underground filtration chamber comprising sand or other media through which stormwater runoff can pass (Armitage et al. 2013). The sedimentation chamber facilitates the removal of suspended particulates and heavy metals, whilst the filtration chamber removes smaller particulate pollutants. The removal mechanism is partly through filtration by the sand bed and partly through microbial action within the media (Armitage et al. 2013). Sand filters tend to be installed for use in impervious areas that are less than 8000m2 but may be designed to manage runoff from larger areas too. Figure A5.8 Swales are shallow grassed or vegetated channels used to collect and/or move water Sand filters are similar to bio-retention areas and other Source:  susdrain, www.susdrain.org bio-retention systems, with the only difference being that stormwater runoff passes through a linear filter medium without vegetation (Armitage et al. 2013). The primary objective for sand filters is water quality improvement and they are particularly effective in the removal of hydrocarbons. They are also used extensively to remove sediment and other particulate pollutants from the first flush (Armitage et al. 2013). Sand filters can be expensive to construct and often require regular maintenace, making them a less cost- effective option. They are highly efficient at removing suspended solids and pollutants. They remove approximately 80-90% of TSS, 50-80% of hydrocarbons, 50-80% of total phosphorous, 25-40% of total nitrogen, 40-50% of E.coli and 50-80% of heavy metals from stormwater runoff (Armitage et al. 2013). Figure A5.9 Bio-retention areas are landscaped depressions employed to manage runoff by passing it through several natural processes. Rain gardens are an example of a bio- retention area. Source:  susdrain, www.susdrain.org                                                  Page 112 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Advantages Limitations Particularly effective in removing suspended solids (TSS) Generally ineffective in controlling stormwater peak discharges Efficient stormwater management technologies in areas with limited Limited potential for the removal of fine sediments and space as they can be implemented dissolved pollutants beneath impervious surfaces Premature clogging is likely to occur in sand filters if they receive excessive sediment carrying runoff, especially from construction sites and areas with open soil patches They manage stormwater runoff effectively on relatively flat terrains Large sand filters are not generally attractive, especially if they with high ground water tables where bio-retention systems are are not covered with grass or other vegetation inappropriate The filtered effluent can be reused for most non-potable domestic water Sand filters are expensive to implement and maintain relative to uses including: toilet flushing, dish washing and garden watering; and most options technologies May be retrofitted with relative ease into existing impervious If designed and/or implemented incorrectly, they may fail, developments, constrained urban locations or in series with resulting in standing pools of water which have the potential to conventional stormwater management systems attract nuisances such as mosquitoes and midges. Advantages Limitations Reduces runoff volumes and rates, and attenuates flood peaks effectively Not suited to areas where the water table is shallower than 1.8m Flexible application means these areas are easily incorporated into a Requires frequent landscaping and maintenance to remain wide variety of landscapes aesthetically pleasing Very effective at the removal of most stormwater runoff pollutants Susceptible to clogging if surrounding landscape is not managed Well-suited for installation in highly impervious areas, provided the Not suitable for areas with steep slope system is well-engineered and adequate space is made available Good retrofit capability Construction costs can be high Aesthetically pleasing A5.2.9 Bio-retention areas Routine inspections and maintenance are required to Bio-retention areas, sometimes referred to as ‘rain ensure that bio-retention areas function effectively. The gardens’ are landscaped depressions which are typically design life of these areas, as with most interventions, under drained and rely on engineered soils, enhanced is directly related to the quality and frequency in vegetation and filtration to remove pollution and reduce maintenance (Armitage et al. 2013). Maintenance runoff downstream (Armitage et al. 2013). They are includes regular inspections, litter and debris removal, usually employed to manage the runoff from the first replacement of mulch areas, vegetation management 25mm of rainfall by passing runoff through a number and sediment removal. of natural processes such as filtration, absorption, Bio-retention areas can have high initial construction biological uptake, sedimentation, infiltration and costs, making them less cost-effective in terms of cost detention. These areas tend to include a number of per unit reduction of runoff volumes and pollutant loads. different smaller stormwater interventions such as filter They remove approximately 50-80% of TSS, 5-80% of strips, temporary pond areas, sand beds, mulch layers hydrocarbons, 50-60% of total phosphorous, 40-50% and a wide variety of vegetation (Armitage et al. 2013). total nitrogen and 50-90% of heavy metals (Armitage et They are particularly effective at managing stormwater al. 2013). Their amenity potential is good. runoff from minor and more frequent rainfall events. Bio-retention areas can manage stormwater runoff on a number of sites, such as between residential plots, alongside parking lots, adjoining roadways and within large landscaped impervious areas. The engineered soil media and the different varieties of vegetation are managed to capture and treat a specified water quality volume of stormwater runoff and in doing so they reduce runoff quantities and rates whilst improving the quality of stormwater entering watercourses further downstream (Armitage et al. 2013).                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 113 Advantages Limitations Able to temporarily store large volumes of stormwater thus attenuating Not very good at removing dissolved pollutants and fine downstream flood peaks material Relatively inexpensive to construct and easy to maintain Generally not as effective in removing pathogens as constructed wetlands Serve multiple purposes during drier seasons, particularly as sports Siltation can be a problem and the floors of detention ponds fields, play parks or commons can become swampy for some time after major rainfall If managed regularly, they can add aesthetic value to adjoining Not very suitable in areas with a relatively high water table, or residential properties as well as presenting fewer safety hazards than where the soil is very coarse and there is a risk of groundwater wet ponds due to the absence of a permanent pool of water. contamination A5.2.10 Detention basins The hydraulic and pollution removal performance Detention basins or detention ponds are temporary of detention basins depends on good maintenance. storage facilities that are usually dry but are designed so Regular inspections are needed to check if the clearing that they are able to store stormwater runoff for short of accumulated sediment is necessary, especially if the periods after high rainfall events (Armitage et al. 2013). basin is being used as a field or common (Armitage et The captured stormwater either infiltrates into the al. 2013). Other maintenance includes the management underlying soil layers or is drained into the downstream of vegetation, inspections after high rainfall events, and watercourse at a predetermined rate. Therefore they possible de-silting. are effective at regulating the flow in downstream Detention basins are relatively inexpensive to construct watercourses. Generally detention basins are designed and have low maintenance costs, making them cost- to temporarily store as much water as possible for 24 effective options for control runoff. Detention basins – 72 hours whilst aiming to provide a safe and secure remove approximately 45-90% of suspended solids, 30- public environment (Armitage et al. 2013). 60% of hydrocarbons, 20-70% total phosphorous, 20- Detention basins are typically lined with grass and are 60% total nitrogen, 50-70% E.coli and 40-90% of heavy designed to be multifunctional in that they provide metals (Armitage et al. 2013). access to recreational area when dry. They are surface The strategic positioning of such storage areas in urban storage basins that provide flow control through the areas can enrich the urban environment and facilitate attenuation of stormwater runoff and also facilitate maintenance operations. In fact, such areas, given their some settling of particular pollutants. Detention basins dimensions, can be easily used as social and recreation tend to be located towards the end of the stormwater areas, such as play grounds or football fields, or for management train so are used if the extended treatment agriculture. There is a good example of this in San Paolo, of runoff is required. The pollutant removal capability Brazil, where floodplain storage has been applied to of a detention basin can be improved through the mitigate the flood risk from the Tamanduateí River, as construction of a sediment trap at the entrance to the shown in Error! Reference source not found.b (Giugni et basin (Armitage et al. 2013). al. 2012). Figure A5.10 (a) Lamination effect due to the flood plain storage and (b) Example of flood plain storage in San Paulo, Brazil Source:  Giugni et al. 2012                                                  Page 114 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT A5.2.11 Constructed treatment wetlands ƒƒ The inlet zone which includes a sediment forebay for Wetlands are generally marshy areas of shallow water the removal of the more coarse sediments and litter that are either partially or completely covered in entering the system; aquatic vegetation. Wetlands provide habitat for a wide ƒƒ The macrophyte zone which is usually shallow and variety of fauna and flora and provide aesthetic appeal, heavily vegetated and facilitates the removal of finer especially in urban areas where green open space is particles and the uptake of soluble nutrients such as limited. Constructed wetlands are man-made systems nitrogen and phosphorous; that are designed to mimic the natural wetland systems in areas where they were not previously found (Armitage ƒƒ The macrophyte outlet zone which channels cleaner et al. 2013). stormwater runoff downstream; and They are able to serve larger catchment areas and are ƒƒ The high flow bypass channel which protects the inlet, very useful at removing nutrients and suspended solids outlet and vegetative zones from damage and scour from stormwater runoff from residential areas. The most during abnormally high flow events. common stormwater pollutant treatment processes that wetlands provide are sedimentation, fine particulate Other considerations include litter traps or trash racks filtration and biological nutrient and pathogen removal at the inlet to the wetland which prevents litter, debris, (Armitage et al. 2013). The percentage removal of course sediment and other pollutants from entering pathogens and nutrients depends largely on the the macrophyte zone and from being carried further pollution concentration of the inflow, the rate at which downstream. The selection of the vegetation to be used the water is flowing through the wetland, the pollution in the wetland is important and a number of selection saturation level of the wetland and the degree to which criteria should be considered, such as the speed at which the nutrients and pathogens adhere to other particles the vegetation establishes itself and grows, the disease and sediments (Armitage et al. 2013). or weed risk associated with vegetation, the suitability of the vegetation for the local climate, the tolerance of Constructed wetlands usually include four distinct zones vegetation to becoming water-logged and the pollutant (Armitage et al. 2013): removal capacity of the various vegetation types (Armitage et al. 2013). Advantages Limitations Highly efficient at removing pollutants from stormwater runoff Wetlands could potentially attract mosquitos and birds whose faeces can increase the amount of phosphorous in the water May attenuate peak stormwater flows depending on location and design Limited to relatively flat land of wetland Good community acceptability and provides amenity value in urban Limited depth range for flow attenuation and little reduction in environments run volume Flooding of the wetland may result in water logging of the plants which may result in die off and a loss in treatment efficiency Figure A5.11 Constructed treatment wetlands are man-made systems designed to mimic natural wetland systems Source:  susdrain, www.susdrain.org                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 115 Inspection and maintenance of constructed wetlands A5.3.2 River cleaning and stewardship can be frequent and costly, however can be reduced One approach to keeping rivers clear of litter and through effective pre-treatment such as litter traps, debris and maintaining a healthy river system is to trash racks and sediment forebays at the inlet to the involve communities that live alongside rivers and wetland (Armitage et al. 2013). Maintaining healthy streams. Community involvement projects can have vegetation and adequate flow conditions is essential multi-sectoral impacts as they generate employment to the efficient functioning of the constructed wetland opportunities, provide awareness, safeguard and this requires harvesting of the vegetation, such as communities and provide city-wide services such as papyrus or reeds. Once harvested the vegetation can be functioning river systems that are clean and clear of composted and re-used. litter. Sections of rivers or streams are maintained by cooperatives which are responsible for removing Wetland construction costs can be high when compared alien vegetation, rubble and any solid waste blocking to other interventions, however their ability and the free flow of water down the stream or river. They efficiency in removing nutrients and pollutants makes are also responsible for maintaining the grass and them relatively cost-effective. They also have the added other vegetation along the banks of the waterway. benefit of providing amenity value. Construction costs The cooperatives generally consist of members of the per hectare of wetland are exponential, meaning the community that are unemployed and vulnerable and cost per hectare decreases the larger the wetland. the project focuses on raising awareness and generating Constructed wetlands are estimated to remove employment. Two examples of such projects include the approximately 80-90% of suspended solids, 50-80% Mlalakua River Restoration Project in Dar es Salaam and of hydrocarbons, 30-40% of total phosphorous, 30- the Sihlanzimvelo Stream Cleaning Project in Durban: 60% total nitrogen, 50-70% E.coli and 50-60% of heavy metals (Armitage et al. 2013). ƒƒ In Dar es Salaam, the Mlalakua River Restoration Project was initiated in 2012 and is a multi-stakeholder partnership that has focused on implementing A5.3 Non-structural interventions measures that enhance healthy living conditions of the Non-structural measures do not involve physical riverine communities, and prevent further pollution on construction but use knowledge, practice or agreement a sustained basis. The Mlalakua River originates from to reduce risks and impacts, in particular through the Mzinga and Kizinga Rivers and drains into Msasani policies and laws, public awareness raising, training Bay in Kinondoni Municipality. The restoration project and education (Kundzewicz 2002). These include forms part of the International Water Stewardship flood warning systems, land use regulations such Programme (IWaSP), an international programme for as development setbacks which identify where water security managed by the Deutsche Gesellschaft development can and cannot occur, or to what elevation fur Internationale Zusammenarbeit (GIZ). Project structures should locate their lowest habitable floor to; activities include physical clean-up of the Mlalakua flood proofing and retrofitting of buildings may increase River, the establishment of sustainable solid waste and the strength against flood actions; elevation of buildings wastewater management systems, such as introducing may avoid completely the inundation. Flood insurance private waste collectors and developing new recycling and relocations also belong to this typology of measure. centres, building capacity of service providers, raising Some of these measures are described in more detail awareness in communities, improving household below. sanitation, and implementing effective law enforcement measures. Project partners include the Wami River Basin Water Board (WRBWB), National A5.3.1 Sweeping and solid waste management Environment Management Council (NEMC), the local Interventions such as street sweeping and proper Kinondoni Municipal Council (KMC), Coca-Cola removal and disposal of solid waste help to reduced Kwanza, Nabaki Africa, Nipe Fagio, the Bremen sediment (and hence pollution) loads entering the Overseas Research and Development Association drainage system, and help to prevent solid waste from (BORDA), and GIZ. Donor funding for the initial phase blocking culverts and reducing the efficiency of the of the project was approximately EUR 400 000. In April conveyance system. 2016 the multi-stakeholder project came to an end with the project being handed over to the Mlalakua Community Change groups which will continue on with improving the health of the river.                                                  Page 116 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT ƒƒ In Durban, the Sihlanzimvelo Stream Cleaning Project A5.3.3 Riparian buffers has been very successful in areas of the municipality A riparian buffer is a vegetated area, or buffer strip, where a number of rivers were considered critical in that is located adjacent to a stream or river channel and terms of health and functioning. Approximately 470km is usually forested, which helps to shade and partially of degraded river systems were identified and pilot protect the waterway from the impacts of adjacent study areas were initiated. Residents of the four land uses. Riparian buffers play an important role in communities formed part of the initial pilot study. improving water quality as well as providing stormwater They were employed to clean and maintain sections of infiltration benefits and conservation value. Riparian the river adjacent to where they live. This includes buffers are similar to filter strips but differ in that they unblocking of culverts and the removal of litter and are generally forested and always occur adjacent to river alien vegetation. Grass and vegetation along the channels. Filter strips tend to be located in urban areas riverbed is maintained to a certain height. The results adjacent to development. have been impressive and rivers have become cleaner, the risk of flooding has reduced through the removal Riparian buffers reduce excess amounts of sediments, of litter and debris and the communities feel safer as organic material, nutrients and pesticides in surface the areas became more accessible and crime has runoff and reduce excess nutrients and other chemicals decreased. Through the project, residents have in shallow ground water flow (Waidler et al. 2009). become more aware of the benefits that are derived They are also known to reduce pesticide drift entering from healthy river systems and have an incentive to the water body. With the use of suitable indigenous keep it clean. The Sihlanzimvelo Stream Cleaning vegetation, riparian buffers have the potential to provide Project is funded by the eThekwini Municipality and a habitat corridor for wildlife (Armitage et al. 2013). the South African government’s Expanded Public Works Programme (EPWP) and includes a contractor development component. The budget for the project is R45 million (approximately US$3 million). Over the course of the project a total of 732 job opportunities have since been created. Advantages Limitations Relatively low costs involved in planting and establishing buffer zones Relatively limited potential for the removal dissolved nutrients Significantly improve water quality of streams and rivers Infiltration of stormwater runoff helps to attenuate flood peaks Natural intervention that provides amenity and conservation value. Figure A5.12 Riparian buffers are located adjacent to streams and river channels. They can either be made up of grasses and smaller plants as in picture (a) or they can be densely vegetated with trees and bushes as in picture (b). They provide a buffer between adjacent land uses such as agriculture and residential areas and waterways.                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 117 Riparian buffers can be cost-effective in that they The capital costs involved in catchment reforestation are require no major engineering or construction. The costs relatively low when compared to other interventions. are associated with the purchasing of seedlings and This is because the intervention involves no engineering the labour required to plant them. Riparian buffers are or construction work and is based solely on the efficient at removing suspended solids, hydrocarbons planting of trees and shrubs. Costs include the buying and other pollutants. They are less effective at removing of seedlings and the labour involved in planting them. dissolved nutrients such as nitrogen and phosphorus. Catchment reforestation provides numerous benefits They contribute to the infiltration of stormwater runoff such as amenity and conservation value as well as and therefore attenuate flood peaks. contributes to providing clean water. A5.3.4 Catchment reforestation Catchment reforestation is an important intervention that does not differ much from the riparian buffer intervention. Catchment reforestation focuses on planting indigenous trees and shrubs within the greater catchment area, in particular in areas that were previously forested. By increasing the number of larger trees and shrubs in the catchment the amount of runoff entering streams and rivers in reduced through trapping and infiltration. Forested areas are well known for their ability to reduce runoff as well as reduce nutrient and pollutant loads entering waterways. Reforestation in the catchment also increases conservation value and amenity value. Advantages Limitations Relatively low costs involved in planting and re-establishing forested Relatively limited potential for the removal dissolved nutrients areas Significantly improve water quality of streams and rivers Infiltration of stormwater runoff helps to attenuate flood peaks Natural intervention that provides amenity and conservation value. Figure A5.13 Catchment reforestation will aid in runoff infiltration reducing the overall amount of stormwater reaching rivers and streams. Reforestation will also aid in removing sediments and nutrients.                                                  Page 118 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Trees absorb rainfall, slow down flow velocity, disperse River channels that are forested have a higher roughness surface runoff, offset water discharge, filter pollutants, which means that the flood arrives later and that the and reduce excess nutrient and sediment loads into peak flow is attenuated when compared to channels the rivers and streams (Rutherford et al. 2006, Ouyang cleared of vegetation. The response to larger floods et al. 2013, Opperman 2014). Therefore land cover generally differs from smaller floods with smaller change, such as deforestation, increases nutrient and attenuation of the peak observed in the case of the small sediment loads entering waterways, alters infiltration flood (Rutherford et al. 2006). Revegetating the riparian rates, elevates greenhouse gas emissions and leads to zone in the Murrumbidgee catchment in Australia had changes in regional and local hydrological cycles (Ouyang a considerable effect on the size and timing of the flood et al. 2013). The latter results in a significant reduction peak reaching different outlets (Error! Reference source in floodwater retention and an associated loss of flood not found.; Rutherford et al. 2006). At the upstream site control (Ouyang et al. 2013). Therefore reforestation (C) the peak is attenuated y 18% and at the larger outlet and the development of forested floodplain buffers (A) the peak is attenuated by 29% (Error! Reference in a catchment can reduce the water discharge and source not found.). sediment load into the rivers and streams and enhance flood attenuation based on catchment characteristics (Ouyang et al. 2013). Vegetation can have numerous impacts on the amount of rainfall that becomes runoff and can generally affect flooding in three specific ways: by affecting the size and shape of the stream channel (geomorphology), by altering the amount of water that reaches the stream channel (hydrology), and by altering the resistance to flow (hydraulics) (Rutherford et al. 2006, Opperman 2014). Figure A5.14 The effect of revegetation on discharge upstream and downstream of the Murrumbidgee in Australia Source:  Rutherford et al. 2006                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 119 This page intentionally blank.                                                  Page 120 Value of Durban’s natural capital and role in Green Urban Development APPENDIX VI. COST ESTIMATES FOR SELECTED GUD INTERVENTIONS Table A6.1 Unit cost estimates for GUD interventions extracted from stormwater management literature and updated to 2015 US$ costs Intervention Unit Cost Unit 2015 cost 2015 cost US$ Source Grassed swales 15 £ per m2 20 31 CIRIA 2007 12 £ per m2 16 24 EA 2007 4.5 Aus$ per m 2 8 6 Fletcher et al. 2003 9.5 Aus$ per m 2 17 13 Fletcher et al. 2003 10 Aus$ per m2 17 13 URS 2003 18 Aus$ per m 2 31 23 URS 2003 18 Aus$ per m 2 31 23 URS 2003 8 £ per m2 8.5 13 Paths for all, Scotland 2014 9 Can$ per m2 9.5 7 Toronto & Region Conservation Authority 2013 8 £ per m2 8 12 Severn Trent Water 2015 305 R per m (2010) 2 417 33 Armitage et al 2013 12 £ per m2 12 18 Hull City Council 2015 15 Euro per m 2 15 17 Morales Torres et al. 2015 18 Detention Basin 18 £ per m3 24 37 CIRIA 2007 per m stored 3 45 £ per m3 59 90 Stovin and Swan 2007 volume 18 £ per m3 24 37 SNIFFER 2007 20 £ per m3 20 31 Hull City Council 2015 22 Euro per m3 22 24 Morales Torres et al. 2015 50 Euro per m 3 51 57 Natural Water Retention Measures Project 2013 46 Retention pond 45 Euro per m 3 45 50 Morales Torres et al. 2015 (wet) per m3 25 £ per m3 33 50 CIRIA 2007 27 US$ per m 3 39 39 US EPA 1999 46 Constructed 40 Euro per m3 40 44 Morales Torres et al. 2015 Wetland per m3 treated 28 £ per m3 37 57 CIRIA 2007 volume 40 US$ per m3 46 46 UN-Habitat 2008 49 Floodplain 109 Rand per m 3 149 11.7 Armitage et al. 2013 restoration (gardens) per m3 8.6 US$ per m3 8.6 8.6 Tanzania National Construction Council (from DeRisi report) (excavation 102 Rand per m3 140 11.04 Department of Co-operative governance costs) and Traditional Affairs (DoCGTA). (2010). 10.5                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 121 Table A6.1 Unit cost estimates for GUD interventions extracted from stormwater management literature and updated to 2015 US$ costs (continued) Intervention Unit Cost Unit 2015 cost 2015 cost US$ Source Riparian Buffers 3627 US$ per ha (lower 4008 4008 Michie 2010 bound) per hectare 4906 US$ per ha (upper 5421 5421 Michie 2010 bound) 793 US$ per ha (lower 873 873 Dep. Environmental Protection 2010 bound) 1911 US$ per ha (upper 2103 2103 Dep. Environmental Protection 2010 bound) 640 US$ per ha (lower 802 802 NRCS Illinois 2005 bound) 836 US$ per ha (upper 1047 1047 NRCS Illinois 2005 bound) UPPER bound 2857 LOWER bound 1894 AVG 2376 Catchment 917 US$ per ha 917 917 TNC 2015. reforestation per hectare 1048 US$ per ha 1158 1158 FAO 2011 1195 US$ per ha 1195 1195 http://www.greentoscale.net/en/ green2scale-ratkaisut/afforestation-and- reforestation (2016) 1090                                                  Page 122 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT APPENDIX VII. COST ESTIMATE FOR DETENTION BASIN This appendix focuses on the costs of construction, Additional costs, that are a percentage of the total work such as the land purchase, earthwork, etc. The values costs, are generally added to the above costs. Table A7.4 reported below present the information collected from below lists all the additional costs (as percentage of the the Tanzania National Construction Council and from the total civil works cost) to add to the final value. study of different ongoing projects in Dar es Salaam. The transport to the waste treatment plant of the removed soil/derbies is assumed having a percentage of 10% of the total earthwork cost. Table A7.1 Unit costs for basic construction work item Work item Description Unit cost Purchase of land 1000 $/ha Excavation 8.6 $/m3 Earthwork Embankment 12.8 $/m3 Concrete work 255 $/m3 Table A7.2 Unit costs for basic construction materials Work item Description Unit cost Gasoline Liter 0.69 Diesel Liter 0.67 Portland cement 1000 kg 98.62 Reinforcement bar 1000 kg 495.62 Fine aggregate m3 10.39 Coarse Aggregate m3 21.82 Plywood m2 23.85 Timber m3 286.15 Wooden pile m 7.95 Wood m3 238.46 Table A7.3 Unit costs for basic construction materials Labor Unit cost Foreman 8.48 Skilled labor 4.43 Common labor 3.64 Unskilled labor 2.91 Operator for heavy equipment 8.02 Driver for light vehicle 7.64 Carpenter 6.75 Welder 9.35 Mechanic 9.70 Electrician 10.39                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 123 Table A7.4 Additional costs to consider as percentage of the total civil work cost Labor Percentage respect to the total cost of the civil works Preparation works 7% Contractor’s indirect costs 10% Engineering service 7% Contingency 10% Government administration cost 1% Carpenter 6.75 Welder 9.35 Mechanic 9.70 Electrician 10.39                                                  Page 124 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT APPENDIX VIII: INUNDATION RESULTS RELATED TO SCENARIO 2 First sub-domain                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 125                                                  Page 126 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Second sub-domain                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 127                                                  Page 128 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Third sub-domain                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 129                                                  Page 130 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Fourth sub-domain                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 131                                                  Page 132 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Fifth sub-domain                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 133 The hazard curves Figure A8.31 Hazard curves calculated for the mitigation strategy 2                                                  Page 134 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT APPENDIX IX: INUNDATION RESULTS RELATED TO SCENARIO 4 First sub-domain                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 135                                                  Page 136 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Fourth sub-domain                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 137                                                  Page 138 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Fifth sub-domain A9.4                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 139                                                  Page 140 AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT The hazard curves Figure A9.19 Hazard curves calculated for the mitigation strategies 4                                                  AMELIORATION OF FLOOD RISK IN THE MSIMBAZI RIVER CATCHMENT Page 141 This page intentionally blank.                                                  Page 142 Value of Durban’s natural capital and role in Green Urban Development