WAVES Technical Report July 2017 Valuing Protective Services of Mangroves in the Philippines Technical Report Wealth Accounting and the Valuation of Ecosystem Services i www.wavespartnership.org WAVES Team at the Institute of Hydraulics at the University of Cantabria ÍÑIGO J. LOSADA RODRÍGUEZ PELAYO MENÉNDEZ FERNÁNDEZ ANTONIO ESPEJO HERMOSA SAÚL TORRES ORTEGA PEDRO DÍAZ SIMAL FELIPE FERNÁNDEZ PÉREZ SHEILA ABAD HERRERO NICOLÁS RIPOLL CABARGA JAVIER GARCÍA ALBA Team at The Nature Conservancy MICHAEL W. BECK SIDDHARTH NARAYAN DANIA TRESPALACIOS ANGELA QUIROZ This Technical Report was primarily funded by World Bank WAVES program. Additional funding was provided by the International Climate Initiative (IKI) of the German Federal Ministry for the Environment, Nature Conservation, Building and Nuclear Safety (BMUB), and the Lloyd’s Tercentenary Research Foundation. This Technical Report is accompanied by a Policy Brief for policy makers. Many thanks to Stefanie Sieber, Glenn-Marie Lange, Maya Gabriela Q. Villaluz, Gem Castillo, Rosalyn Sontillanosa, Marnel Ratio and to the attendees of the Valuation training the week of Dec 5, 2016. Suggested Citation: Losada, I.J., M. Beck, P. Menéndez, A. Espejo, S. Torres, P. Díaz-Simal, F. Fernández, S. Abad, N. Ripoll, J. García, S. Narayan, D. Trespalacios. 2017. Valuation of the Coastal Protection Services of Mangroves in the Philippines. World Bank, Washington, DC. Points of contact: Michael W. Beck, mbeck@tnc.org Íñigo J. Losada Rodríguez, inigo.losada@unican.es WAVES - Global Partnership for Wealth Accounting and the Valuation of Ecosystem Services Wealth Accounting and the Valuation of Ecosystem Services (WAVES) is a global partnership led by the World Bank that aims to promote sustainable development by mainstreaming natural capital in development planning and national economic accounting systems, based on the System of Environmental-Economic Accounting (SEEA). The WAVES global partnership (www.wavespartnership.org) brings together a broad coalition of governments, United Nations agencies, nongovernment organizations and academics for this purpose. WAVES core implementing countries include developing countries—Botswana, Colombia, Costa Rica, Guatemala, Indonesia, Madagascar, the Philippines and Rwanda—all working to establish natural capital accounts. WAVES also partners with UN agencies— UNEP, UNDP, and the UN Statistical Commission—that are helping to implement natural capital accounting. WAVES is funded by a multi-donor trust fund and is overseen by a steering committee. WAVES donors include—Denmark, the European Commission, France, Germany, Japan, The Netherlands, Norway, Switzerland, and the United Kingdom. ii Table of Contents Page EXECUTIVE SUMMARY v INTRODUCTION 1 Section Overview 1 Coasts at Risk 2 The Role of Mangroves in Coastal Protection 2 Taking Nature into Account 3 Mangroves in the Philippines 4 Measuring the Protective Services of Mangroves in the Philippines 5 Methods at a Glance 6 National and Local Scales 6 Assessing Flooding Under Two Conditions 6 Scenarios for Mangrove Cover 7 People & Property Flooded 8 DATA SOURCES 9 Section Overview 9 Coastline Data 9 Bathymetry and Topography Data 9 Mangrove Cover 9 Coral Reef Cover 10 Climate Data 10 Wave Climate and Sea Level 10 Tropical Cyclones 12 Tidal Gauges 14 Exposure 14 Population 14 Population Below Poverty 15 Residential Stock 15 Industrial Stock 17 Roads Network 17 CONSTRUCTING THE COASTAL PROFILES 18 Section Overview 18 Constructing the Coastal Profiles 18 Profile Classification 19 MODELING COASTAL HABITATS: NUMERICAL MODEL DELFT 3D 21 Section Overview 21 Numerical Set-up for 1D Simulations 21 Modelling Coastal Habitats: Mangroves and Coral Reefs 21 Validation and Sensitivity Analysis 21 Offshore Validation 21 Nearshore Validation 22 Sensitivity Analysis: Storm Surge Intensity and Duration 23 REGULAR WAVE CLIMATE 24 Section Overview 24 Offshore Dynamics 24 Wave climate selection process 26 Habitat Pathway: 1D propagation 27 iii Flood Height Reconstruction 27 Comparing Flood Height for Different Scenarios 28 TROPICAL CYCLONES 30 Section Overview 30 Modelling Tropical Cyclones 30 Offshore Dynamics 32 National Scale Nearshore Dynamics: Historical Tropical Cyclones 39 Classification of Tropical Cyclones 39 Habitat Pathway 40 Flood Height Reconstruction 40 Comparing Flood Height for Different Scenarios 41 Regional Scale Nearshore Dynamics: Historical and Synthetic Tropical Cyclones 42 Synthetic Tropical Cyclones Generation 42 Flood Height Reconstruction 43 COASTAL IMPACTS: FLOODING MASK CALCULATION IN THE PHILIPPINES 45 Section Overview 45 National Scale Results (30m resolution) 45 Local Scale Results in Pagbilao and Busuanga (5m resolution) 45 ASSESSING THE BENEFITS OF HIGH RESOLUTION DATA AND FLOODING MODELS 49 Section Overview 49 Comparing the Effects of Higher and Lower Resolution Elevation Data 49 Regular Wave Climate 49 Cyclone Data 50 Flooding Methods: Hydraulically Connected Bathtub (5m) vs RFSM (5m) 50 DAMAGES AND BENEFITS 53 Section Overview 53 Exposure, Damage Curves and the Estimation of Risk Probability 53 Exposure of Assets: People and Stock 53 Damage Curves 53 Assessing Risk - Combining Spatial Results 54 National Scale Results 55 Damages vs Return Period 57 Expected Damages: Annual and by Return Period 57 Expected Benefits: Annual and by Return Period 58 Annual Expected Benefits per Ha of Mangroves 58 Catastrophic Benefits (100 years return period event) 59 National Maps of Annual Expected Benefits 60 Local Scale Results: Pagbilao 62 Damages vs Return Period 63 Annual Expected Damage 64 Annual Expected Benefits 65 Annual Expected Benefits per Ha of Mangroves 65 Benefits for catastrophic events (100 year return period event) 66 CONCLUSIONS 67 ANNEX 1: Figures 68 ANNEX 2: Physics and Governing Equations 83 ANNEX 3: Computational Cost and Hard Disk Memory Required 85 REFERENCES CITED 86 iv EXECUTIVE SUMMARY Mangroves and other coastal ecosystems act as natural defenses that protect people and property from storms, floods, erosion, and other coastal hazards, reducing coastal risk. Mangroves protect coastlines by decreasing the risk of flooding and erosion. The roots of mangroves retain sediments and prevent erosion, while the prop roots, trunks and canopy reduce the force of incoming wind and waves and reduce flooding. Yet the value of these ecosystems is often not fully accounted for in policy and management decisions, and thus they continue to be lost at alarming rates, increasing the risk faced by coastal communities. Between 1980-2005, the world lost 19% of its mangroves. The Philippines has lost hundreds of thousands of hectares of mangroves in the last century. When mangroves are degraded or destroyed, the coast line becomes more exposed to the destructive impacts of waves and storm surge, and coastal communities have greater risks from the impacts of storms, floods, and sea level rise. The Philippines is at high risk from coastal hazards and natural defenses can help reduce these risks. Between 2005 to 2015, 2,754 natural hazards affected the Philippines: 56% of property damage was caused by typhoons and storms, and another 29% was caused by floods. Due to a recognition of these increasing risks, and of the potential role of natural defenses to reduce these risks, the Government of the Philippines has committed to restoring mangroves as part of its risk reduction strategy, and the Philippines WAVES program on natural capital accounting is helping the Philippines incorporate the value of mangroves into their national accounts. This Technical Report, and its accompanying Policy Brief, provide a social and economic valuation of the flood protection benefits from mangroves in the Philippines. This work aims to support decisions across development, aid, risk reduction and conservation sectors as they seek to identify sustainable and cost-effective approaches for risk reduction. This Technical Report applies the Expected Damage Function approach recommended by the World Bank to quantify the risk reduction benefits from mangroves in the Philippines. Using high-resolution flooding models, the Report examines the flooding that would occur with and without mangroves under different storm conditions throughout the Philippines, and estimates the annual expected benefits of mangroves for protecting people and property in social and economic terms. The report examines flooding under regular storms and under extreme conditions (e.g. typhoon) and compares the people and property damaged under 3 different scenarios of mangrove cover: historical mangrove cover (1950), current mangrove cover (2010), and no mangrove cover. The protection services of mangroves are valued nationally across the Philippines using national models. The values obtained from these national models are compared to values from sensitivity analyses in a few locations (i.e., Pagbilao and Busuanga) calculated with very high-resolution data and models, so that the results of the national models may be verified. In summary, the key findings are: • In the Philippines, if the current mangroves (data from 2010) were lost, 24% more people would be flooded annually, i.e., an additional 613,000 more people many of whom live in poverty. • Damages to residential and industrial property would increase by 28% to more than US $1 billion annually; and 766 km of roads would be flooded. • One hectare of mangroves in the Philippines provides on average more than US $3200/ year of direct flood reduction benefits. v • Based on the Philippines’s current population, the mangroves lost between 1950 and 2010 have resulted in increases in flooding to more than 267,000 people every year. Restoring these mangroves would bring more than US $450 million/year in flood protection benefits. • Mangroves provide the most protection for frequent lower intensity storms (for example, 1-in-10 year storm events). For more catastrophic events, such as the 1-in-25 year storm, they provide more than US $1.6 billion in averted damages throughout the Philippines. When combined with built infrastructure, mangroves provide an effective defense against storms and coastal flooding. • The results are presented in maps that show the spatial variation in the flood reduction benefits provided by mangroves to identify the places where mangrove management may yield the greatest returns. Incorporating the value of these ecosystem services into a country’s system of natural capital accounts can ensure that these ecosystems are accounted for in policy and management decisions. Currently, only a subset of the benefits provided by ecosystems is valued, usually extractive services such as fish and timber harvests. Many critical services that rely on keeping ecosystems intact, such as flood protection and climate mitigation, are rarely valued. This encourages short-term over-exploitation and reduces the quantity and quality of the goods and services provided by natural capital. Better valuations of the protection services of coastal habitats may halt the loss of our natural capital and ensure the provision of ecosystem services- what gets measured, gets managed. Mangrove conservation and restoration can be an important part of the solution for reducing coastal risks. By valuing these coastal protection benefits in terms used by finance and development decision-makers (e.g., annual expected benefits), these results can be readily used alongside common metrics of national economic accounting, and can inform risk reduction, development and environmental conservation decisions in the Philippines. vi Valuing the Protective Services of Mangroves in the Philippines 1 | Introduction 1.0 Section Overview frameworks to ensure better management, and in green investments for risk reduction. This section provides the context and background for the modelling of the To assess the coastal protection services of protective services of mangroves in the mangroves, this Report follows a five-step Philippines. This Technical Report was methodology recommended by the World commissioned by the World Bank Wealth Bank (World Bank 2016). The five steps Accounting and the Valuation of Ecosystem involve: estimation of offshore dynamics; Services (WAVES) program in the estimation of nearshore dynamics; the Philippines. The program is intended to influence of habitats; estimation of coastal support the Philippine government strategy impacts with and without habitats; and for incorporating the value of ecosystem estimation of the resulting flood damages to services, including coastal protection, into people and property with and without their natural capital accounting system. To habitats. (More detail will be found in support the government’s strategy, the Section 1.6.) The methodology evaluates the Philippines WAVES program on natural protective services of the habitats – in this capital accounting includes a component on case, mangroves –in terms of avoided flood the ecosystem services of mangroves, damages to people and property. This Report including the coastal protection services of uses two national scale analyses and one mangroves. The objective of this Technical local scale analysis. The national scale Report is to help the Philippine WAVES analyses look at flooding from waves and program and the Philippine government from historical tropical cyclones using a construct these mangrove accounts by simplified one dimensional (1-D) numerical providing a methodology for quantifying the model. The local scale analysis uses a higher protective role of mangroves. resolution, two dimensional (2-D) model to look at flooding from an extensive synthetic The Philippines is among the most at-risk database of tropical cyclones. All the countries in the world (Beck 2014, World Risk analyses assess habitat values under three Report 2016). Typhoons, storms and floods scenarios: historic mangrove cover (i.e. account for around 80% of the total losses mangrove cover in 1950); current mangrove from disasters, with estimates of annual cover (i.e., mangrove cover in 2010); and an average losses totaling nearly US $3 billion ‘extreme’ future scenario where all (NEDA 2017, UNISDR 2015a). More than 60% mangroves are lost. of the country’s 101 million people live on the coastline1 and are heavily dependent on its The report examines the role of mangroves in natural ecosystems for resources and reducing the flooding risks from ‘regular’ livelihoods. The Philippines has lost climate conditions (including daily ocean approximately half of their mangrove habitat waves and sea level conditions) and from over the past century, thereby losing the extreme conditions (including local and protective benefits of these coastal specific extreme events and tropical ecosystems. Realizing the risk reduction role cyclones). It measures the protective services of mangroves, the Government of the that mangroves offer today under existing Philippines has committed to restoring mangrove cover, and the protection services mangroves as part of its coastal protection that have been lost over the past half century strategy. There is now broad interest in due to mangrove degradation and loss. It incorporating the value of ecosystem calculates the people and property that benefits into natural capital accounting would be affected by flooding under these 1 http://sdwebx.worldbank.org/climateportalb/home.cfm?page=country_profile&CCode=PHL&ThisTab=Dashboard 1 Valuing the Protective Services of Mangroves in the Philippines scenarios, including the people under The Philippines has extensive experience poverty affected by flooding. The protection with disaster risk reduction. Government services of mangroves are valued nationally policies mandate assistance to local across the Philippines, and locally in communities that conduct vulnerability and Pagbilao; the local study serves to validate risk assessments. Between 2006-2011, the the accuracy of the national results. Philippine government launched the READY Project, a multi-agency initiative led by the 1.1 Coasts at Risk National Disaster Coordinating Council (NDCC) to address disaster risk management The 2011 Global Assessment Report (GAR) at the local level in 27 high risk Philippine on Disaster Risk Reduction highlights that provinces. Building on the outputs of the the risk of economic loss due to tropical READY Project, the National Operational cyclones, storm surge and floods is growing Assessment of Hazards (NOAH) Program, led as the exposure of economic assets increases by the DOST, provided enhancements to and the health of coastal ecosystems existing geo-hazard vulnerability degrades. Already, the proportion of the assessments and maps through world’s GDP annually exposed to tropical sophisticated, scenario-based mapping cyclones has increased from 3.6 % in the which integrate probabilistic climate 1970s to 4.3 % in the first decade of the modelling and simulation. 2000s (UNISDR 2011). In 2011, insured losses from natural disasters (especially coastal and The Philippine Development Plan of riverine hazards) reached an all-time high. 2017-2022 includes strategies to rehabilitate Erosion, flooding, and extreme weather and restore degraded natural resources, events affect hundreds of millions of protect fragile ecosystems and improve the vulnerable people, important infrastructure, welfare of resource-dependent communities and economic activity, and cause significant (NEDA 2017). However, efforts to integrate losses to national economies. The impacts of disaster risk reduction and climate change coastal hazards such as tropical cyclones can adaptation in to comprehensive land use and be devastating to coastal economies. These development plans are hampered by many impacts will continue to worsen with factors, including the limited availability of continued climate change. appropriately-scaled probabilistic hazard maps and the lack of capacity to use this The Philippines is one of the most at risk geospatial information when it is available. nations to the impacts of coastal storms. It Funding for adaptation and hazard ranks as the nation with the third highest mitigation is limited particularly as much of number of recorded landfalls of tropical the available resources are needed for relief cyclones and the second most landfalls over and recovery. the past 5 decades2 . The Philippines are also socially vulnerable to storm exposure: they 1.2 The Role of Mangroves in Coastal rank third, after Vanuatu (1st) and Tonga (2nd), for countries with the highest disaster risk in Protection the world (World Risk Report 2016). The Coastal and marine habitats, particularly National Economic and Development coral reefs and mangroves, can substantially Authority (NEDA) identifies that coastal reduce vulnerability and risk, providing hazards contribute significantly to the natural protection from waves, wind and population’s vulnerability, and that storms storm surge. Both mangroves and reefs are such as Super Typhoon Yolanda (Haiyan) now regularly cited in both conservation and have slowed progress in poverty alleviation development literature for their role in (DRR Platform 2014). natural coastal protection; i.e. for their value in reducing the impacts of coastal erosion 2 http://www.aoml.noaa.gov/hrd/tcfaq/E25.html 2 Valuing the Protective Services of Mangroves in the Philippines and inundation during storms, as well as at risk from the impacts of storms, floods, providing important co-benefits for fisheries and sea level rise. production, tourism, and in the case of mangroves, carbon sequestration. Despite the myriad of ecosystem services that they provide, the value of mangroves as Seagrasses were not included in this study ‘green infrastructure’ is still not fully because data on the national distribution of recognized, and they continue to be lost and seagrasses in the Philippines does not exist, degraded. Global losses of coastal habitats and because this team currently lacks an are high: 30-50% of the world’s wetlands operational model for the flood reduction have been lost (Zedler and Kercher 2005), benefits of sea grasses. It is also true that this 19% of mangroves were lost just between team has not prioritized these models 1980-2005 (Spalding et al. 2010), and 75% of because the flood reduction benefits of the world’s coral reefs are rated as seagrasses are much lower than those of threatened (Burke et al. 2011). Often, the loss mangroves and reefs. of these habitats is greatest around large populations- the places were the impacts of Mangroves are particularly effective at coastal degradation are greatest, and where providing coastal protection to people and the most people stand to benefit from property. The aerial roots of mangroves coastal ecosystems. Sixty percent of the retain sediments, and prevent erosion. The world population is expected to live in urban mangrove’s roots, trunks and canopies areas by 2030, with greater concentration significantly reduce the drag force from around coastal areas3 . This means that rates incoming wind and waves. The entire of coastal development will be increasing structure can reduce the force of wind waves with heavy investments in coastal and flood waters. If mangroves are degraded infrastructure and potential of loss of more or destroyed, the loss of their aerial roots coastal habitats. leads to erosion, coastal regression, soil destruction and increasing water depth. The 1.3 Taking Nature into Account more exposed coastline is more vulnerable to the destructive impacts of waves and storm There is a growing need for policies that surge, and is at higher risk of coastal flooding encourage the conservation and restoration and erosion. As mangroves are degraded and of habitats that provide coastal protection, in lost, more people and property are directly the places that yield the greatest protection benefits and are most cost effective Figure 1 A conceptual representation of the distribution of natural infrastructure along a beach profile. Coral reefs, sea grasses, mangroves and dunes provide coastal protection services. 3 http://www.who.int/gho/urban_health/en/ 3 Valuing the Protective Services of Mangroves in the Philippines compared to other coastal protection has changed throughout the whole country. strategies. Better valuations of the protection Figure 2 illustrates these two trends of services from coastal habitats could inform mangrove loss and redistribution. decisions to meet multiple objectives in risk reduction and environmental management. Many mangrove areas have been converted One important pathway through which these to aquaculture ponds or development services may be considered is in national despite the fact that these lands at low economic accounts. The United Nations has elevation are the most at risk to coastal identified a general approach for assessing hazards. The restoration of mangroves ecosystem services in these accounts. To through conversion of abandoned fish ponds assist the development of these policies, the has been problematic due to land tenure World Bank WAVES Policy and Technical issues as well as biogeochemical changes Experts Committee commissioned including acid sulfide build-up in the “Managing coasts with natural solutions: sediments (Primavera 1991). Some Guidelines for measuring and valuing communities have left a small but critical mangroves and coral reefs” (World Bank strip of mangroves on the coastline, in part to 2016). These guidelines show how to assess help reduce erosion and to provide barriers and value the coastal protection services of for aquaculture ponds, in other parts to mangroves, seagrasses and coral reefs. The encourage sediment retention that may lead Guidelines recommend using process-based to new land. For example, Figure 1.1 (see approaches, in particular the Expected Annex 1) shows land cover and land use Damage Function (EDF) approach for change in a part of Pagbilao Bay. spatially explicit valuation of the coastal protection services from mangroves. The The Philippines have a history of planting EDF is adapted from approaches commonly coastal mangrove trees in densely populated, used in engineering and insurance to assess highly degraded coastal watersheds. risks and benefits. Research has shown that socioeconomic 1.4 Mangroves in the Philippines factors were more important than ecological factors in determining the relative success of Mangroves in the Philippines provide a restoration efforts (Walters 1997). One variety of ecosystem services to adjacent example is the Talabong Mangrove Sanctuary coastal populations, including food, timber, in Bais Bay, which has 200 hectares of other livelihood activities, and coastal nationally recognized nursery habitat for protection. Traditionally, rural Filipinos wildlife and fish. The Bais local government intensively used mangroves for fuelwood, unit spearheaded restoration and construction, shellfish collection, fishing, and rehabilitation efforts through a ‘household settlement (Walters 1997). Mangroves have planting’ program, where coastal significant economic importance for the poor communities were encouraged to plant and the most vulnerable coastal inhabitants, multi-species mangrove trees. Residents particularly the landless and women (Walters started small, private mangrove plantations 1997). to provide wood and other products In the Philippines, rates of mangrove (Walters, 1997). In another example, the deforestation have been among the highest Philippines Department of Environment and in the world (Myers 1998) (Hamilton et al. Natural Resources (DENR) encouraged 1989, Primavera 1991, 1995), and many of the planting and conservation of mangroves by remaining mangroves are highly degraded local residents through 25-year private leases (Walters 1997). From 1950 to 2010, around on intertidal land to encourage mangrove 50,000 ha of mangroves were lost. In stewardship (Walters 1997). addition, the distribution of mangrove cover 4 Valuing the Protective Services of Mangroves in the Philippines Among the greatest challenges for mangrove To support the government’s strategy, the conservation and restoration in the Philippines WAVES program on natural Philippines are population growth and capital accounting includes a component on coastal development. Between 2017 and the ecosystem services of mangroves, 2022, there will be an additional 8.3 million including carbon sequestration, ecotourism, Filipinos. Some regions, such as Metro and coastal protection services. The Manila, will become denser, and increased development of mangrove accounts, and concentrations will encourage land their inclusion into the Philippines System of conversion. It is difficult to effectively enforce National Accounts, will enable the environmental laws. There is a lack of government to consider the protection sustainable financing, and limited access to services of mangroves, and will ultimately funding for climate change adaptation, inform decisions surrounding disaster risk disaster and risk reduction and insurance management, coastal zone management, and products for local governments. climate change adaptation. Furthermore, there is a lack in capacity to use science-based information. The private To help the Philippine WAVES program and sector, which could complement government the Philippine government construct these efforts by providing risk transfer mangrove accounts, this Technical Report mechanisms, has so far been minimally provides a methodology for quantifying the involved. protective role of mangroves in the Philippines. It examines the role of mangroves in reducing the flooding risks 1.5 Measuring the Protective Services from ‘regular’ climate conditions (including of Mangroves in the Philippines daily ocean waves and sea level conditions) and from extreme conditions (including local In the wake of several destructive typhoons, and specific extreme events and tropical the Government of the Philippines has cyclones). It measures the protective services committed to restoring mangroves as part of that mangroves offer today under existing its coastal protection strategy under the mangrove cover, and the protection services National Greening Program. A recently issued that have been lost over the past half century Executive Order “Expanding the Coverage of due to mangrove degradation and loss. It the National Greening Program” (EO 193 s. calculates the people and property that 2015) further identified the critical role of would be affected by flooding under these forests including mangroves. scenarios, including the people under poverty affected by flooding. The protection Figure 2 An illustration of mangrove loss and redistribution. The image on left shows the loss of mangrove cover in Roxas. The image on the right shows the redistribution of mangrove cover between 1950 and 2010 in Samar Island. 5 Valuing the Protective Services of Mangroves in the Philippines services of mangroves are valued nationally the highest resolution models. The results of across the Philippines, and locally in the national and local scales were compared Pagbilao; the local study serves to validate to ascertain the accuracy of the national the accuracy of the national results. models with the higher resolution local models. 1.6 Methods at a Glance The methods used in this study for 1.6.2 Assessing Flooding Under Two evaluating the protection services of habitats Conditions against coastal flooding follow the recommendations in the Guidelines for the A critical part of this work is the assessment Valuation of Natural Coastal Protection of the extent of flooding from regular storms (World Bank 2016). Figure 3 visually and tropical cyclones (also referred to as TC). summarizes the methods (see Annex 1 Figure To estimate this flood risk, we considered 1.2 for more detail). First, an understanding of flooding data from two types of events, offshore dynamics is constructed from the which we refer to as ‘regular conditions’, and historic databases generated by the Institute ‘extreme conditions’ or ‘tropical cyclones’. of Hydraulics at the University of Cantabria, The first data set, for regular conditions, which include historical time series of waves, considers 30+ years of wave and water level astronomical tide, storm surge, mean sea data from the Philippines, which captures level and wind. Then, using hydrodynamic significant events of waves and storm surge, models (SWAN and DELFT 3D), waves and but does not capture the most extreme water levels are propagated from offshore to events produced by tropical cyclones. The nearshore, and over habitats (mangroves and second dataset considers the effects of coral reefs). In the fourth step, waves and tropical cyclones, the most extreme events, water level are propagated on shore, and the across the Philippines. flooding impacts are calculated. In the final We use the following data for regular and fifth step, the coastal assets damaged by extreme wave and surge conditions at local flooding under different habitat conditions and national scales: (i.e with and without mangroves) are 1- Regular Conditions: compared, and the benefits provided by mangroves are calculated. a. National Scale: Historical wave climate and Sea Level (from 1992 to In this report, we do not consider freshwater 2015), 30m resolution of DTM and flooding. We only analyze coastal flooding Hydraulically connected bathtub due to ocean events and the role of mangroves in reducing these events. The approach for coastal flooding using models for freshwater catchment flooding GIS algorithm. are very different from the coastal flooding b. Local Scale 1 (Pagbilao): Historical models. In the future it would be useful to wave climate and Sea Level (from combine models on freshwater and coastal 1992 to 2015), 5m resolution of DTM flooding. and Hydraulically connected bathtub approach for coastal flooding using 1.6.1 National and Local Scales GIS algorithm. We examined flood risks and benefits at two 2- Tropical Cyclones: different scales. At the national scale, we a. National Scale: Historical Tropical applied high resolution models across the Cyclones (from 1951 to 2015), 30m country. For key test locations (e.g., resolution of DTM and Hydraulically Pagbilao) where better data (particularly bathymetric data) was available, we applied connected bathtub approach for coastal flooding using GIS algorithm. 6 Valuing the Protective Services of Mangroves in the Philippines Figure 3 Methodology to evaluate coastal protection services of ecosystems like coral reefs and mangroves. 1: Oceanographic data are combined to assess offshore sea states. Stage 2: Waves are modified by nearshore hydrodynamics. Stage 3: Effects of habitat on wave run-up and surge are estimated. Stage 4: Flood heights are extended inland along profiles (every 200 m) for four locally generated, storm events (10, 25, 50, 100-yr events) with and without mangroves. Stage 5: The land, people and built stock damaged under the flooded areas are estimated (see World Bank 2016). b. Local Scale 1 (Pagbilao): Historic 1. Mangrove cover in 1950 (Defense Tropical Cyclones (from 1951 to Mapping Agency DMA): The first 2015), 5m resolution of DTM and ‘Historical Mangroves’ scenario Hydraulically connected bathtub considers the mangrove cover that approach for coastal flooding using existed in 1950, when approximately GIS algorithm 360,000 ha of the Philippines were covered by dense mangrove forests. c. Local Scale 2 (Pagbilao): Historic This is the earliest comprehensive and Synthetic Tropical Cyclones cover data available. (5000 years), 5m resolution of DTM and high resolution model for 2. Mangrove cover in 2010 (DENR): The coastal flooding (RFSM-EDA). second ‘current mangroves’ scenario considers the mangrove cover that Additional details may be found in the existed in the Philippines in 2010. This Annex. is the most recent comprehensive 1.6.3 Scenarios for Mangrove Cover mangrove cover data available. Since 1950, there has been an evident loss of We analyze flooding in the Philippines under mangrove cover, and significant three different scenarios of mangrove cover reduction in mangrove density. in the Philippines: Mangrove cover in the Philippines 7 Valuing the Protective Services of Mangroves in the Philippines decreased from 360,000 ha in 1950 to agglomerations. The variables included in the 310,000 ha in 2010. database are number of residents, and economic value of residential, commercial 3. Without mangroves: The third and industrial buildings (De Bono et al., scenario, ‘No Mangroves’, assumes a 2015). The GAR15 database follows a top- hypothetical situation in which all down approach using geographic mangrove forests in the Philippines distribution of population and gross have been completely destroyed. domestic product (GDP) as proxies to distribute the rest of socio-economic Figure 4 represents the three different variables (population, income, education, mangrove cover scenarios in Pagbilao. health, building types) where statistical Pagbilao, located on the northern shore of information including socio-economic, Tayabas Bay in Quenzon province, building type, and capital stock at a national exemplifies the rapid mangrove lost that has level are transposed onto the grids of 5x5 or characterized much of the Philippines: the 1x1 using geographic distribution of loss of mangrove extent is evident between population data and gross domestic product scenarios 1 and 2. The images also show a (GDP) as proxies (UNISDR, 2015c). redistribution of mangrove cover, particularly visible in the middle of the bay. Our estimation of benefits from mangroves considers only the direct effects of mangroves on flood reduction, it does not 1.6.4 People & Property Flooded consider the many other benefits from The analysis calculates the people and mangroves (e.g., fisheries, timber, livelihoods) property affected by flooding, including the and indirect impacts (e.g., business total population flooded, and the number of disruption from storms) on the local people below poverty affected. This study economies. uses data from the 2015 Global Assessment Report on Disaster Risk Reduction (GAR15, UNISDR 2015a) on the economic value of residential and industrial stock. The GAR15 provides a global exposure database with a standard 5 km spatial resolution and a 1 km detailed spatial resolution on coastal areas, estimating the economic value of the exposed assets, as well as their physical characteristics in urban and rural Figure 4 Mangroves scenarios in Pagbilao region: (1) Historical, (2) Current and (3) No mangroves scenario. 8 Valuing the Protective Services of Mangroves in the Philippines 2 | Data Sources 2.0 Section Overview database, and SEAWIFS 1km resolution of coral reefs bathymetry worldwide. The The following section describes the data and global ETOPO bathymetry database is the sources of data used in this study. Most commonly used in regional and global of the hydrodynamic and coastline analyses flooding analyses. In tropical countries such use the latest, freely available, global data of as the Philippines, the bathymetry of shallow, the best possible resolution for climate and nearshore coral reefs is critical for predicting sea-level projections, tropical cyclone tracks, flooding, because coral reefs play a critical bathymetry, coastlines, topography and role in wave energy dissipation by reducing mangrove and coral reef cover. Additionally, waves reaching mangrove shorelines. the local scale analysis use a locally available However this type of bathymetry is currently 5 meter resolution IFSAR DTM for not accounted for in ETOPO data or in other topography and a synthetic analysis of national and global flooding models (Beck et cyclone tracks based on historical tropical al. In review). We approached this problem cyclones in the region. Exposure data on the by combining the ETOPO data with data number of people, property and roads were from SeaWiFS project (NASA) that includes obtained from a mix of global databases and information about water depth over the coral data provided by local authorities. The reefs. With a spatial resolution of 1km, economic values of residential and industrial SeaWiFS bathymetry is the most accurate stock nationwide were estimated based on database that may account for a coral reef´s these datasets. location and depth. 2.1 Coastline Data For topography, we used ETOPO 1: 1.6 km resolution (1 arc min) worldwide topo-bathy The coastline is obtained from the NOAA database, and SRTM 30 PLUS: 30 m across database GSHH (Global Self-consistent, The Philippines. Adequate flooding analyses Hierarchical, and High-resolution Geography require high resolution topography data, or Database). Out of the 5 different resolutions digital terrain model (DTM). For the national provided in this database (i.e. full resolution level analyses, the 30x30m horizontal (0.1 km), high resolution (0.2 km), resolution DTM elevation SRTM30M-PLUS intermediate resolution (1 km), low resolution (Shuttle Radar Topography Mission) was the (5 km) and coarse resolution (25 km)), this best available data. For regional scale study uses the full resolution. Islands smaller analysis using higher resolution flooding than 2 km in perimeter are not considered. modelling approaches, IFSAR topography This study considers 32,859 km of coastline (available in Pagbilao) was used. It provides a spread across 1,311 islands ranging in DTM with a resolution of 5m. Figure 2.1 (see perimeter from 4,763 km to 1.98 km. annex 1) illustrates the resolution of some of 2.2 Bathymetry and Topography Data the different bathy-topo data that we used locally and nationally. Good bathymetry and topography are critical for these flooding analyses (Beck et al. In 2.3 Mangrove Cover review, World Bank 2016). The availability This study used two different datasets for and quality of bathymetry and topography mangrove cover: datasets varied greatly across the Philippines- we used the best available data 1. Historical Mangroves (1950): at both local and national scales. Topographic Maps at 1:50,000 Scale For bathymetry, we used ETOPO 1:1.6 km originally published by the US Army resolution (1 arc min) global topo-bathy Service and compiled from aerial 9 Valuing the Protective Services of Mangroves in the Philippines photographs taken from 1947 to 1953 of water reaching coastlines. To adequately (http://www.namria.gov.ph/ measure the water reaching mangroves, the download.php). This is the earliest effect of coral reefs must be taken into comprehensive mangrove cover data account. Our wave propagation model takes available. into account both the friction and wave breaking provided by coral reefs in the 2.Current Mangroves (2010): The most Philippines (see Figure 5). recent mangrove cover data available is from the 2010 Land Cover Mapping This study uses the 2010 Millennium Reef Project, which used high resolution Map Project, released by the United Nations Environmental Programme World satellite imageries such as the Conservation Monitoring Center (UNEP- Advanced Very Near Infra-Red (AVNIR), WCMC), to obtain a global spatial Panchromatic Remote Sensing for distribution of tropical and subtropical coral Stereo Mapping (PRISM) and Satellite reefs. The data draws from multiple sources, Pour l’Observation de la Terre (SPOT 5) including University of South Florida's to generate the land cover data for the Millennium Coral Reef Mapping Project Philippines. A total of 245 AVNIR Seascape database and the World Fish images with 10 m resolution were used. Centre in collaboration with WRI and TNC. For areas without AVNIR images, SPOT images with 10 m resolution were used. 2.5 Climate Data 2.4 Coral Reef Cover 2.5.1 Wave Climate and Sea Level In the Philippines, as in many tropical environments, coral reefs often exist Historical wave climate and sea level time alongside mangroves. Coral reefs are series are required to evaluate the coastal submerged natural structures which break protection role provided by mangrove forests waves and dissipate wave energy through in the Philippines. Most of the existing friction, thus reducing the volume and force databases are hourly or 6-hourly averaged, Figure 5 The coral reef cover (green) and mangrove cover (red) in Pagbilao, Luzon, in 2010. 10 Valuing the Protective Services of Mangroves in the Philippines i.e., regular wave climate, and do not capture wave direction. The latest version of Global peak extreme events like Tropical Cyclones. Ocean Waves improves the previous existing datasets with improved spatial resolution We used GOW 2.0 (Global Ocean Waves), from 1.5º to 0.25º; and better captured local GOT (Global Ocean Tides) and DAC (Global extreme events (see Table 2). Storm Surge) to obtain datasets for waves and sea level (See Table 1 and 2). We The DAC (Dynamic Atmospheric Correction) combined these three datasets to build a 36- dataset is the worldwide water surface year time series (1979-2015), which yielded elevation induced by a pressure gradient and results in 315,360 1-hourly measures of sea wind in the period 1992-2014 (Carrère and states. Lyard 2003). The model is forced by the pressure and wind speeds at 10m altitude The GOW 2.0 (Perez et al. 2017) database provided by the European Centre for comes from CFS (http://cfs.ncep.noaa.gov/ Medium-Range Weather Forecasts (http:// cfsr/) reanalysis which provides reliable time www.ecmwf.int/en/research/climate- series of atmospheric pressure and the reanalysis) reanalysis. The storm surge induced wind field worldwide with 0.25º database was recently extended for the resolution from 1979 to 2015 (http:// period 1871-2010 (Cid et al. 2014, Cid Carrera ihpedia.ihcantabria.com/wiki/IH_DATA). The 2015) by using the 20th Century Reanalysis main data for the Philippines from GOW 2.0 ensemble (Compo et al. 2011) as a predictor database was wave height, peak period and to reconstruct global 20th century surge Forcing Spatial Spatial resolution Time Time interval Method coverage (latitude by longitude) resolution Waves (GOW 2.0) CFS Global 0.25º 1h 1979 - 2015 Astronomical Tide TPX07+T- 0.25º (GOT) Global 1h 1900-2099 Tide Storm Surge (DAC) ECMWF Global 0.25º 6h 1979-2015 Table 1 Datasets for waves and sea level. Spatial resolution Forcing Spatial Time (latitude by Time interval Method coverage resolution longitude) Waves  (GOW  2.0) CFS Global 0.25º 1h 1979  -­‐  2015 Astronomical  Tide   0.25º (GOT) TPX07+T-­‐Tide Global 1h 1900-­‐2099 Storm  Surge  (DAC) ECMWF Global 0.25º 6h 1979-­‐2015 Table 2 Oceanographic datasets used as input for the propagation model and their spatial and temporal resolution and historical time series. 11 Valuing the Protective Services of Mangroves in the Philippines levels. The Storm Surge time series (DAC) (maximum 1-minute surface wind speeds in does not detect Tropical Cyclones events knots and minimum central pressures in because storm surge is averaged every 6 milibars) for all Tropical Storms and Cyclones hours, infra-estimating the peak of sea level observed from 1951 to date. Despite global during a TC event. satellite based observations started in 1966, the IBTrACS database covers from 1950 to For the Mean Sea Level dataset, we assumed 2014, thus existing some uncertainties and a constant sea level across the Philippines. non-homogeneities before the 60s. In Figure 6, the historical tropical cyclone tracks 2.5.2 Tropical Cyclones making landfall in the Philippines from 1951 to 2014 are shown. 2.5.2.1 Historical Tropical Cyclone Tracks Historical tropical cyclone occurrence rates This study used the International Best Track coincide with those obtained in previous Archive for Climate Stewardship (IBTrACS) works such as Cinco et al., 2016, with an v03r08 (Knapp et al., 2010) provided by average of 19.8 TCs/year in the Philippines NOAA to characterize the tropical cyclone Area of Responsibility and only 7.8 TCs/year climate (typhoons) in the Philippines. This making landfall. file contains ensemble mean data from observations performed by different A climatology frequency analysis indicates institutions using various methods. Data that tropical cyclones in the Philippines can contains 6-hourly information about tropical occur in every calendar month, with the cyclone center location (latitude and greatest tropical cyclone activity longitude in tenths of degrees) and intensity concentrated in July, September, October Figure 6 Historical Tropical Cyclone tracks (gray lines) making landfall in the Philippines from 1951 to 2014. The green polygon represents the coastal buffer used to identify land falling events. 12 Valuing the Protective Services of Mangroves in the Philippines and November (see Figure 7 and Figure 2.2 with the other months of the year. Typhoons in the annex for more details). This is true for usually make landfall between December and all categories of typhoons, form Category 1 February, when the northeast monsoon is tropical storms to violent Category 5 active. typhoons. Figure 10, also shows the spatial distribution of tropical cyclone activity, which 2.5.2.2 Synthetic Tropical Cyclone Tracks indicates a strong latitudinal gradient with increasing number of tropical cyclones in The amount of damage caused by tropical north Luzon (nearly 1 tropical cyclone/year). cyclones depends not just on the intensity of the cyclone, but also on its track. Therefore, Typhoon activities in the Philippines are greatly influenced by monsoons and sea observation data is not sufficient to properly surface temperature in the Southern Pacific define return periods for different associated ocean. Many typhoons enter the Philippine hazards. To adequately capture the possible Area of Responsibility (PAR) during the number of tropical cyclones that could months of July to November when sea impact a particular region, we use a surface temperature is warmer compared stochastic method. Stochastic methods are Figure 7 Spatial distribution of the tropical cyclone activity. The legend marks the number of cyclones per year passing through each grid cell, from 0 (light blue) to >1.2 (purple .The axes mark latitude and longitude. 13 Valuing the Protective Services of Mangroves in the Philippines based on Monte Carlo simulations in which 2.6 Exposure the sequential development of tropical The assets susceptible to damage were cyclones is calculated statistically from given statistical parameters of tropical cyclone classified into five categories: data. The model used in this work is based 1. Population on the work of Nakajo et al., 2014. Three 2. Population below poverty tropical cyclone parameters are stochastically modeled: translation direction, 3. Residential stock speed, and minimum sea level pressure (see 4. Industrial stock Figure 2.3 in Annex 1). 5. Roads network 2.5.3 Tidal Gauges 2.6.1 Population To calibrate and validate the hydrodynamic The number and distribution of people in The model, data from six tide gauges was used, Philippines was obtained from the WorldPop downloaded from the Global Sea level database (http://www.worldpop.org.uk/), Observing System (GLOSS, http:// which provides, globally, the number and www.gloss-sealevel.org) (see Figure 2.4 in location of people per hectare (100mx100m) Annex 1). residing in low and middle income country. Figure 8 The spatial distribution of number of people living below poverty. Higher values are in darker shades. 14 Valuing the Protective Services of Mangroves in the Philippines Figure 9 Residential stock distribution (US $ millions). Darker shades indicates higher values. This project used the WorldPop 2010 (previous section) to obtain the number of population distribution layer. In addition to person per grid below poverty with the same mapping population counts, WorldPop 100m grid resolution of the original produces high resolution estimates of population layer. population demographics and characteristics which cover a range of factors, including age 2.6.3 Residential Stock and sex structures, births, pregnancies and This study uses data from GAR15 (UNISDR poverty. An example of the WorldPop 2015b), which includes the economic value of database in The Philippines is shown in residential, commercial and industrial Figure 2.5 (see Annex 1). buildings, as well as hospitals and schools, the number of residents and the type of 2.6.2 Population Below Poverty labor activities. GAR15 provides this data at The data for the percentage of population 5km spatial (see Figure 9). GAR15 combines below poverty was provided by local different sets of data to obtain its socio- authorities. This information was economic information: population disaggregated at the municipal level (see distribution (LandScan), night time light Figure 8). The data were transformed from intensities (Visible Infrared Imaging shape layer to raster, then the resulting raster Radiometer Suite, VIIRS), capital stock was multiplied by the total population layer (Perpetual Inventory Method (PIM) and 15 Valuing the Protective Services of Mangroves in the Philippines Figure 10 Industrial stock distribution (US $ millions). Darker blue indicates higher values. historical Gross Capital Formation (GCF) fields were summed: high, medium high, data from World Bank), Gross Regional medium low and low income for both Product (GRP) distribution (from several rural and urban residential stock. sources) and different socio economic 3. In each point of GAR layer, residential indicators (economic level, commercial, stock per capita was calculated by industrial, public, education and health data) dividing residential stock and adjusted as proxies to estimate the use of the building population. stock. (UNISDR, 2015c) 4. A raster layer was created for residential The study downscaled residential stock data stock per capita. Inverse distance in the following process: weighted interpolation was used for the creation of this raster. 1. For each point of GAR layer, the total population was calculated. Eight fields 5. Finally, using the population raster (from were summed: high, medium high, WorldPop, 100m resolution) the medium low and low income for both residential raster layer was calculated by rural and urban population. GAR data is multiplying residential stock per capita referenced to 2014, so an adjustment to and population. A scale verification was 2015 WorldPop estimates was performed. done, checking that sum of residential stock from GAR layer was the same that 2. In each point of GAR layer, total the sum of residential stock raster layer residential stock was calculated. Eight created. 16 Valuing the Protective Services of Mangroves in the Philippines Figure 11 Road network distribution. 2.6.4 Industrial Stock industrial stock was cut with the administrative national borders. Mirroring the process used for the residential stock, the study used GAR data to calculate 5. A scale verification was done, checking industrial stock. that sum of industrial stock from GAR 1. In each point of GAR layer, total industrial layer was the same that the sum of stock was calculated. Two fields were industrial stock raster layer created. (see summed: rural and urban industrial stock. Figure 10.) 2. A distance to roads network raster layer 2.6.5 Roads Network was created with a 100m resolution. This layer shows for each cell the distance in The study used OpenStreetMap to meters to the nearest road. characterize the roads network for the Philippines (see https:// 3. A kriging technique was used to obtain a www.openstreetmap.org/export). This data raster layer analyzing the relationship includes categories from motorways to between industrial stock (from GAR footways. We used data only for motorways, points), population and distance to roads network. trunk, primary and secondary roads. This information was then transformed into a 4. The previous step can place some stock in raster layer with 100m resolution. (see Figure places where there should not be (lakes, 11.) offshore), so the obtained distribution of 17 Valuing the Protective Services of Mangroves in the Philippines 3 | Constructing the Coastal Profiles 3.0 Section Overview for offshore dynamics, bathymetry and habitat cover. To reduce computational effort This section describes the construction of in estimating flooding along all these profiles, cross-shore coastal profiles, used to estimate the profiles were grouped into 250 the propagation of waves and surge levels representative ‘families’ and the flooding was from the ocean to the inland extent of the then estimated for each of these families. floodplain. A profile was created at every 200 m along the Philippines’ ~30,000 km 3.1 Constructing the Coastal Profiles coastline. Multiple steps were taken to ensure To measure the flooding that occurs that the profiles were accurate, ran parallel to throughout the Philippine coast, we divide the bathymetric gradient (i.e. wave direction) the coastline into equal sections, or ‘profiles’, and were limited to depths of less than 50 every 200m. The resolution of the profiles meters. The profiles were then paired with chosen depends on the scale of this the nearest, most representative data points particular project. For example, a 2 km Figure 12 Coastline for Lingayen Gulf, derived from NOAA data. The left shows a 200m resolution, the top right shows a high resolution of 200m, the bottom right shows a low resolution of 2km 18 Valuing the Protective Services of Mangroves in the Philippines resolution is reasonable for global scale 1. Sector method: We generated an projects, however for national scale projects average value of the gradients within a the resolution may be fine-tuned to 200 m. 60 degree range on either side of the The advantage of having one profile for seaward transect perpendicular to the every 200 m is that the potential errors coastline of each centroid. resulting from a two dimensional wave 2. Circular method: We generated the propagation process are reduced, and that average value of the gradients within a more realistic values of coastal Total Water circle (diameter=10km) around each Levels may be generated. The disadvantage centroid. of using such a fine scale is that it requires a huge computational effort and the use of The sector method reduces the number of statistical tools to simplify a huge amount of incorrect profiles (that is, profiles that do not profiles into representative clusters (see follow the wave direction) by 40% when Section 4: Profile classification). An example compared to the circular method. of two different resolutions of coastal profiles Furthermore, on coastlines dominated by (2 km vs 200 m) is shown in Figure 12 in small islands and non-linear configurations, Lingayen Gulf (Luzon). like those of the Philippines, the errors generated with circular method are greater The high resolution coastline obtained from because of the chaotic distribution of the NOAA database GSHH (see section 2.1) bathymetry gradients in these areas. allows the analysis of the Philippines coast at We applied a few further corrective steps to a 200 m resolution. The accuracy of these the profiles. First, we eliminated profiles that data will determine the accuracy of the started on land. Second, we eliminated hydrodynamic transformation of waves and profiles that followed the sequence ‘sea- sea level (astronomical tide + storm surge). land-sea’ from the coastline, seaward. These Profiles are traced starting from offshore and profiles generally occur in a bay or estuary moving towards the coastline. Care is taken and correspond to areas protected from to trace the profiles parallel to the waves. Thus, we assumed that no wave- bathymetric gradient, so that each profile induced flooding would occur at these tracks the main direction of waves as closely coastlines, and eliminated the corresponding as possible. Since we are tackling a two transects. dimensional problem using multiple one dimensional solutions, we omit a few The final step was to limit the profiles to important processes of wave transformation water depths of 50 m or less: at depths associated with wave direction (refraction greater than 50 m, waves propagate and diffraction, for instance). To reduce the differently, and the propagation model is no loss of information resulting from this longer valid. omission, we orient profiles parallel to the expected wave´s direction (perpendicular to 3.2 Profile Classification the contour lines) and with the highest In total, 171,888 profiles were drawn along resolution possible (every 200 m). 32,859 km of Philippine coastline. After To obtain mean bathymetry gradients for creating profiles for every 200 m of each profile, we tested two methodologies. coastline, and then refining them, we 19 Valuing the Protective Services of Mangroves in the Philippines intersected each profile with three data layers: 1. Bathymetry: SeaWifs (NASA) with 1 km resolution + ETOPO 1 (1.6km resolution) 2. Mangrove coverage in 1950 (DMA) and 2010 (LandCover) scenarios. 3. Coral reef coverage (Millenium Reef Map 2010, UNEP-WCMC) The high computational effort required to analyze the propagation of waves over the ecosystems of such a huge number of profiles forced us to reduce the number of profiles to a few representative ‘families’ of profiles. We used a clustering technique, K- MEANS, to group or classify the profiles into 250 representative families of profiles in the Philippines. To build these families, we considered the water depth along the whole profile for every kilometer (resulting in 20 water depth values), and the type of bottom cover for every kilometer for three types of bottoms: sand, coral reef, and mangrove (resulting in 20 bottom cover values). K-MEANS is a clustering method which aims to partition ‘n’ observations (171,888 profiles defined by their water depth and bottom type) into ‘k’ clusters (250 representative profiles, also defined by their water depth and bottom cover) in which each observation belongs to the cluster with the nearest mean. Although this is a challenging computational problem, there are efficient heuristic algorithms that converge quickly to a local optimum via an iterative refinement approach. Three different K-MEANS classifications are performed, for the 3 mangrove coverage scenarios, resulting in 750 families of coastal transects profiles (3 scenarios x 250 profiles/ scenario). 20 Valuing the Protective Services of Mangroves in the Philippines 4 | Modeling Coastal Habitats: Numerical Model DELFT 3D 4.0 Section Overview 4.3 Validation and Sensitivity Analysis This section describes the setup and validation Two validations and two sensitivity analysis of the Delft 3D numerical model suite used for were implemented before applying DELFT these analyses. The suite comprises models 3D model for waves and storm surge simulating flow and wave conditions. The propagation: models use information on offshore 1- Offshore validation. A national scale hydrodynamics, bathymetry, topography and mesh grid was tested with the aim of land cover and accounts for various physical validating the capacity of the model of processes. This section also describes the deep water sea level propagations. setup of the model and cross-shore profiles, 2- Nearshore validation. One specific event inclusion of habitats as land-cover type inputs, (Tropical Cyclone in August 1987) was model validation for offshore and nearshore simulated and propagated over different regions, and some analyses of sensitivity to mangrove forests typologies 2D and 1D. hydrodynamic inputs. More information on the 3- Sensitivity analysis of DELFT 3D against physics behind the models, including the storm surge intensity and duration in relevant equations, may be found in Annex 2. presence of mangroves (1D profile approach). 4.1 Numerical Set-Up for 1D Simulations 4- Sensitivity analysis of DELFT 3D against For the study case, waves and flow were coral reef and mangrove presence or propagated over 1D profiles. To simulate 1D absence (1D profile approach). propagations with Delft 3D, a 2D mesh with 3 These are further described in the following cells in Y-direction was created. X-direction is sections. assumed to be perpendicular to the coast and was divided in 10 m spaced cells. Profiles are 4.3.1 Offshore Validation 20 km long and they extend 10 km shoreward A numerical mesh of 5 km resolution was and 10 km landward. In total the numerical created. It covers all the Philippines (302x352 mesh will have 2000 x 3 cells (X and Y cells). ETOPO bathymetry was used in this directions). For an explanation of the physics first validation. Total Water Level and storm and governing equations see Annex 2. surge are the output variable validated with 4.2 Modelling Coastal Habitats: historical instrumental data. To validate the Total Water Level induced by Mangroves and Coral Reefs the previous mentioned forcing methods, six Coastal habitats like coral reefs or mangroves locations with bouys were chosen: Manila, are modeled by means of introducing a Legaspi, Davao, Subic Bay, Curmao and roughness value based on the corresponding Lubang. Additionally, two tropical cyclone Manning coefficient. Different values of events were simulated and validated: Manning coefficient were adopted: Typhoon Haiyan (November 2013), and Tropical Cyclone Nesat (September 2011). - Sand soil: n=0.02 (Zhang et al 2012) - Mangroves: n=0.15 (Zhang et al 2012) The following forcing methods were tested: a) Astronomical Tide - Coral reefs: n=0.05 (Prager 1991) 21 Valuing the Protective Services of Mangroves in the Philippines b) Wind contrast, SWAN boundaries were divided into 5 parts of equal length to account for changes c) Astronomical Tide + wind in wave height between the right and left side. d) Astronomical Tide + wind + waves (swell conditions) To simulate bottom induced friction, we e) Astronomical Tide + wind + waves (wind considered 3 scenarios with spatially variable conditions) roughness: (1) without mangrove, (2) with the current mangrove extension and (3) with the f) Wind + waves (wind conditions) historic mangrove extension. Manning's Two validation cases are shown as an example coefficients adopted the following values of the whole generated tests. It should be according to the soil type: noted that no significant differences were - Landward bottom type: n=0.033 observed between “wind” and “wind+waves” cases. Also, swell waves do not modify the - Seaward bottom type: n=0.02 Total Water Level due to the low intensity of - Mangroves: n=0.15 the swell component in The Philippines with When comparing the 1D and 2D simulations, respect to wind components. In Annex 1 we found that the model can not show the Figure 4.1 and Figure 4.2 show the high effect of the mangroves due to the short capacity of the model to reproduce offshore length of the mangrove profile and the wave sea level induced by tropical cyclones. period. In the 2D model, the total water level 4.3.2 Nearshore Validation slightly increases without mangroves, resulting in lower flood speed but larger flood extents. To carry out this analysis a simulation was run in the large-scale mesh of the Philippines for In conclusion, in order to capture the effect of the tropical cyclone of August 1987. The mangroves when using the DELFT 3D model, simulation began one day after the start of the greater mangrove cover and higher model cyclone to stabilize the model. The simulations resolution (i.e. decreasing cell size 100 m to 10 modeled the period from August 6, 1987 to m) is required. In other words, the smaller cell August 15, 1987. The model was forced with the size, the better the model will be able to boundary conditions mentioned in the simulate wave and sea level propagation in previous section. Results of the coarse mesh mangrove areas. were stored at points with a time resolution of 4.3.3 Sensitivity Analysis: Storm Surge 1 minute for the coast of Pagbilao. Intensity and Duration We performed 1D and 2D simulations of the We performed a set of theoretical cyclone. The 2D mesh has 468x234 cells of 100 simulations of a storm-surge event on a 1D meters of side. The time step is 30 seconds grid of 5001 cells of 5 m sides, and we and the turbulent viscosity has been analyzed the following parameters: considered constant at 0.4 m2/ s. In this mesh, we assumed that all boundaries are closed - Length of area covered by mangrove: 0, 2, 4, 6, 8, 10, 12 and 15km. except the offshore boundary that has been modeled as a level condition. Based on the sea - Storm-surge duration: 2, 4, 6 and 8 hours. levels obtained at the boundary points in the coarse mesh, we assumed a constant sea level throughout the nearshore model domain. In 22 Valuing the Protective Services of Mangroves in the Philippines - Maximum level (m) reached by storm- 2. A profile only with mangroves surge: 0.5, 1, 2 and 3m. All storm-surge have been generated as a Gaussian pulse. 3. A profile only with coral reefs Mangrove cover was assigned a Manning 4. A profile without mangroves and without coral reefs coefficient of 0.15. And any other bottom type is assigned a coefficient of 0.02 (i.e., in The Total Water Level in the coast was essence assumed to be bare bottom). obtained for each case. Results show that the The coastline is located at x=12.5 km. presence of both coral reefs and mangroves Mangroves extend seaward from x=12.5km provide more than a 149% reduction in flood until the corresponding mangrove length. height as compared to case 4, with no mangroves and no coral reef. We also find The following conclusions are derived from that mangroves alone provide more than a the analysis: 102% reduction in flood height (relative to - For the same storm surge, longer pulse the no habitat case), while coral reefs provide durations and lower friction result in more an additional reduction in flood height of 8%. flooding. Figure 13 shows the results of the sensitivity - Storm surge duration is the most critical analysis for storm surge duration of 2 and 8 variable affecting flooding level. hours. Annex 3 provides the computational - Larger mangroves decrease water level costs for these DELFT 3D simulations. (dissipation) and, consequently, the flooding extension. In conclusion, mangroves contribute significantly to storm surge reduction; reefs To ascertain how the DELFT 3D model is able do not contribute greatly to storm surge to provide the Total Water Level for different reduction. Reefs however contribute bottom types, four cases were run in a 1D primarily to flood reduction through wave numerical mesh: (not surge) attenuation. 1. A profile with mangroves and with coral reefs Figure 13 Sensitivity analysis of the effects of habitat on the reduction of flood height in DELFT 3D. Four cases are examined considering water level or flood height (z) with and without reefs mangroves. The black lines represents the elevation profile above (z>0) and below water. The green and red lines indicate where reefs and mangroves occur along the profile respectively. The red dashed line indicates the water level (flood height) across the profile. 23 Valuing the Protective Services of Mangroves in the Philippines 5 | Regular Wave Climate 5.0 Section Overview coastline over the profiles using the Delft 3D modelling suite, for each of the three This section describes the process for mangrove scenarios (historic, current and estimating the coastal flooding that occurs total loss). The final total water levels at the under ‘regular conditions’. First, data on shoreline determine the ‘flood height’ for hourly offshore wave conditions are each scenario. The difference between these associated with each cross-shore profile. results indicates the protective capacity of Next, after excluding any wave conditions mangroves for wave-induced flooding. that occur due to tropical cyclones to avoid double counting (see Section 6), the 5.1 Offshore Dynamics maximum values of specific wave parameters The total water level due to regular hourly are selected. The final selection of offshore wave conditions is estimated for 171,888 waves is then grouped into representative profiles perpendicular to the shore across the families of wave-climate to reduce entire country. Each profile is associated with computational effort. The waves from each a coastline point (see Figure 14). First, of these families are propagated to the Figure 14 Spatial distribution of offshore data points. To calculate offshore ocean dynamic, a global ocean waves database was used to obtain data on for tides, storm surge and wind for each of the red points. Mostly, the distribution of the points is constant. (See Annex 1 Figure 5.1 for the distribution of data points for the other offshore databases). 24 Valuing the Protective Services of Mangroves in the Philippines offshore hydrodynamic measurements are waves, astronomical tide, storm surge and assigned for each coastline point based on wind. Sea level rise projections for an RCP global data, and are used to estimate 8.5 scenario for the end of the century are nearshore wave heights. Based on these also included (Slangen 2014) to indicate measurements, the total water level at each areas that could be vulnerable future climate coastal point is obtained. Then, the effect of change (see Figure 15). vegetation on the total water level is assessed for 3 mangrove scenarios (see The nearest offshore measurements of waves Section 4). (GOW 2.0), astronomical tide (GOT) and storm surge (DAC) are identified and Offshore wave climate and sea level statistics assigned to each coastline point in the are obtained for all of Philippines to have an Philippines. The mesh resolution of the overview of the national distribution of national model at the coastline is 200 m. Figure 15 Statistics of Ocean Dynamics and Sea Level: (1) Significant wave height exceeded 12h/year, (2) Mean Peak Period, (3)Mean Wave Direction, (4) Astronomical Tide exceeded 1% of the time, (5) Storm Surge exceeded 5% of the time and (6) Sea Level Rise in 2090 according to RCP 8.5 (Slangen 2014) 25 Valuing the Protective Services of Mangroves in the Philippines However, the resolution of the global climates, and to avoid double-counting of offshore datasets is 25 km (0.25o). Therefore, extreme conditions in subsequent analyses of every 25 km of the coastline uses the same tropical cyclones, the time-series is filtered to offshore measurements. Figure 14 and Figure remove extreme sea-state measurements 5.1 in Annex 1 shows the spatial distribution that represent tropical cyclones (tropical of the offshore database points in The cyclone induced flooding is calculated in Philippines. Section 6). 207 tropical cyclones (TCs) are detected between 1979 and 2015 and filtered To reduce errors in translating the offshore out of the record using information from the data into nearshore values, two conditions cyclone occurrence databases (see Figure are followed in assigning these points: (1) the 16). The filtered sea states are then grouped offshore point must be inside the influence into 120 families by applying a K-MEANS area of the coastal point, which is defined by algorithm. Each family has information on the a triangle oriented +/-30o seaward; (2) where following parameters: there are multiple points within a triangle, the - Significant wave height (Hs) nearest point to the coastline point is chosen. This method minimizes errors in choice of - Peak Period (Tp) appropriate offshore points that can be - Wave´s direction (θ) critical in island regions where the directionality of waves is highly conditioned - Sea Level (SL), which is the by which side of the island is being summation of the Storm Surge and considered (see Annex 1 Figure 5.2). Astronomical Tide In total, there are 3,225 offshore points that 5.2 Wave climate selection process contain hourly time series data for the Each of the 171,888 coastline point and parameters Hs, Tp, θ and SL. The closest sea profiles is associated with a 36-year time states to the coastline, as defined in the series (1979-2015) of hourly sea state (wave national numerical mesh (5km X 5km), are and sea level) measurements, making a total used as inputs to the 1D Delft-3D model to of 315,360 hourly sea states (see Section simulate the propagation of waves over 2.5). To assess flooding from regular wave mangroves and coral reefs. Figure 16 Example of the process for identifying tropical cyclones (red dots) within regular wave climate (Hs, Tp), Sea Level and Wind speed time series 26 Valuing the Protective Services of Mangroves in the Philippines 5.3 Habitat pathway: 1D propagation reconstruct the Total Water Level in the whole coast of The Philippines. To do this, we Next, we simulate wave propagation from first generated a look-up table to estimate offshore to the shoreline, over vegetation. the Total Water Level for each representative This is done for 120 representative sea states profile for each mangrove scenario. The at each of the 250 theoretic profiles (see interpolation tables allow estimation of Total Section 4), and for each of the 3 mangrove Water Level based on the 4 sea-state scenarios, making a total of 90,000 parameters for each of the 120 simulations. These simulations are performed representative sea states. Using these tables, in Delft-3D. The Flow and Wave modules are a Total Water Level is obtained for every coupled (i.e. run simultaneously) for a profile across the country for the entire simulation time length of 60 minutes. The period (hourly from 1979-2015). numerical boundary conditions assume a non-stationary process with a triangular Extreme values of Total Water Level for the time-evolution of Hs and a constant SL within average wave climate simulations are each sea state. obtained using a Peak-Over-Threshold method with the threshold set at 98% (i.e. The output of each simulation provides the the top 2% of all values are defined as Total Water Level time series for every 10 m. extreme). To ensure time-independence of However, we are only interested in the the selected data points, values that occur maximum Total Water Level at the shoreline within 3 days of a previous value are (henceforth, referred to as “Flood Height”). excluded. A Pareto-Poisson distribution is From the outputs of the 90,000 simulations, then applied to the selected values to obtain 750 “look-up tables” or “interpolation tables” a return period distribution for the extreme of Flood Height are created. Each one of the Total Water Levels. Since the collected data 750 profile families thus have their own only span 36 years, the maximum return interpolation table with 120 sea-state period should not exceed this order of parameter combinations of Hs, Tp, SL and θ magnitude. Here, we assumed that 36 years and the associated output, Flood Height. of data allows us to obtain 50 years return period events. The Flood Height is thus These look-up tables can now be used to estimated every 200 m across the entire quickly estimate wave and surge dissipation country’s coastline for four return periods of by mangrove forests. The input variables 1, 10, 25 and 50 years and three mangrove needed to make these estimates would be scenarios. An example of a 25 years return the sea state parameters (i.e. Hs, Tp, SL and period Flood Height for regular wave climate θ) and the vegetation characteristics (i.e. is shown in Figure 17. The increase in Flood mangrove length and average water depth in Height is more significant under the no the mangroves) mangrove scenario, particularly when compared with the increase in Flood Height 5.4 Flood height reconstruction under the current mangrove scenario. In The simulations of wave and sea level other words, a greater loss of protection propagation over the 1D representative benefits occurs when moving from current profiles result in 90,000 theoretical values of mangroves to no mangroves, compared to Flood Height. However, the goal is to 27 Valuing the Protective Services of Mangroves in the Philippines Figure 17 Flood height (m) at the shore for a 1 in 25 year event in the regular wave climate data set, under three scenarios: historic mangrove cover in 1950, current mangrove cover in 2010, and a hypothetical no mangrove cover. The increase in Flood Height is more significant under the no mangrove scenario, particularly when compared with the increase in Flood Height under the current mangrove scenario. the loss that occurs when moving from - For events with a return period greater historical mangroves to current mangroves. than 10 years, mangrove protection does not increase with increasing 5.5 Comparing Flood Height for return periods. This implies that different scenarios mangroves are more efficient at protecting the coast for less intense The loss in protection from flooding due to storms. mangrove degradation over the last half century and the potential future loss in - In general, the effect of losing protection in case of complete destruction of mangroves produced the same this ecosystem are shown in the following percent increment in Flood Heights figures for different return period events (see across the country, except in few Figure 18 and Figure 19). The figures show critical areas like the South Islands or the differences in Flood Height to north Palawan Island which demonstrate the relevance and potential of experienced greater increments in mangroves in reducing flooding. Flood Height due to mangrove loss. Several conclusions can be drawn from these results: - The complete loss of mangroves will result in a loss of protection greater than what has already been lost due to degradation since 1950. 28 Valuing the Protective Services of Mangroves in the Philippines Figure 18 Difference in Flood Height (m) with (current) and without mangroves. For a 1 in 10 year return period event. Areas in yellow represent where mangroves have the most significant effects on flood height. Figure 19 Increment of Flood Height (m) (1) Between 1950 and 2010 and (2) Between now and a theoretical case of no mangroves scenario. 50 years return period event 29 Valuing the Protective Services of Mangroves in the Philippines 6 | Tropical Cyclones 6.0 Section Overview difference between these results indicates the protective capacity of mangroves for This section describes the process for surge-induced flooding in Pagbilao. estimating the coastal flooding that occurs due to tropical cyclones. First, data on 6.1 Modelling Tropical Cyclones historical tropical cyclones in the region are used to reconstruct the offshore dynamics To determine the role of the mangroves in relevant to cyclone-induced storm surge. This attenuating storm surge at both the national and the local scale, a set of processes of involves reconstructing wind and sea-level pressure fields for historical tropical cyclones, different spatial scales must be tackled. setting up the numerical model to simulate There are several large scale factors the resulting waves and sea levels at the concerning tropical cyclone characteristics national scale, and validating the model and the general shape of the coast that contribute to the amount of storm induced based on historically available observations of sea level during extreme events. The surge at a given location: tropical cyclone conditions are grouped into - Central pressure: lower the pressure 548 representative families. Next, these higher the surge waves and sea levels are propagated to the - Storm intensity: stronger winds will coastline over nearshore bathymetry and produce a higher surge habitats, for the three mangrove scenarios (historic, current and total loss) using the - Storm size: larger the storm, higher the Delft 3D model. From this, Total Water Levels surge (or Flood Heights) are obtained all along the - Storm forward speed: faster storms nation’s coastline, for multiple return periods increase surges on open coasts, slower (10, 25 50 and 100 years). Comparing the storms increase surges in bays Flood Heights for each scenario indicates the protective capacity of mangroves for storm - Landfall angle and approach: storms surge-induced flooding. approaching perpendicular to the coast are more likely to produce This analysis is then repeated using an higher surges extended database of synthetic tropical - Shape of the coastline: concave cyclones for the higher resolution local scale coastlines will experience higher analyses in Pagbilao. Here, 1,462 synthetic surges tropical cyclones nationwide are generated, of which 456 are specific to the Pagbilao - Tide amplitude and phase: concurrent region. These events are then used to high tides will increase storm surges estimate Flood Heights at the Pagbilao Storm surge is also highly dependent on coastline for multiple return periods (7 to local features, such as coral reef barriers, 200 years) using the same process used for wetlands or mangrove forests that will affect the national model. The final Total Water the flow of water. Moreover, it is at the local Levels at the shoreline determine the ‘Flood scale where nonlinearities between waves, Height’ for each mangrove scenario. The sea levels and currents have a greater effect 30 Valuing the Protective Services of Mangroves in the Philippines on the Total Water Level. Storm surge solving the local scale processes where reduction by mangroves is expected to mangrove extent, coral reef presence and depend on a number of mangrove forest seafloor bathymetry must be accounted for. characteristics and on the flood process Due to the reduced depths at the local scale, itself. These factors include: nonlinear interactions are produced between waves and sea level. Dingemans et al. (1987) - Mangrove width: the rate of flood demonstrated that the wave radiation stress reduction of mangroves appears to contributes to wave-driven flow in shallow range from 5-15 cm/km (Krauss et al., waters where wave dissipation due to 2009) to 50 cm/km (Zhang et al. bottom friction and wave breaking take 2012). place. To account for these interactions, a coupled modeling approach is adopted in - Mangrove vegetation characteristics: which the modification of the wave field due the density of the mangrove to variations in sea level during a given vegetation and the diameter of aerial tropical cyclone are considered roots and stems are expected to affect simultaneously with wave setup the mangroves’ capacity to reduce contributions to the Total Water Level. storm surge. However, few data are available to support this assumption. The coastal risk assessment will be - Storm surge height and storm forward conducted at the national scale and at a local speed: depending on the height of the study site in Pagbilao, in south Luzon. surge, it will interact with different Existing frameworks to assess risk from parts of the mangroves (aerial roots, tropical cyclone hazards can be broadly trunks and leaves). Consequently, the summarized in three categories: flow will experiment different friction - Worst case scenario: the goal of this rates, and thus different attenuations approach is to find the maximum of the water level. Depending on the possible flood extent. A set of worst forward speed, the surge can occur for case tropical cyclones are proposed anywhere between a few hours to (generally category 3 or higher) that more than a day. Numerical make landfall with different angles and simulations (Zhang et al. 2012) at different distances respective to the indicate that mangroves are more study site. Proposed scenarios are effective at attenuating faster surges based on expert judgment or historical than slower ones. knowledge. Due to the different processes involved, a - Based on historical best track data: two-step methodology has been adopted to even though the worst possible event evaluate the role of mangroves in attenuating has not been recorded, if there are storm surge at the national scale. The first enough numbers of tropical cyclones step consists on determining offshore that have crossed within a distance of dynamics (waves and sea levels) produced the study site, it is possible to obtain by all the historical available records of realistic estimations of the hazard in tropical cyclones that have impacted the term of probabilities or return periods. country. The second step is focused on 31 Valuing the Protective Services of Mangroves in the Philippines - Based on synthetic tropical cyclone al., 2010), FVCOM (Weisberg and Zeng, tracks: this approach is based on the 2008) or Delft3D (Veeramony et al., 2014) idea that storm surge damage is have been used to model surge inundation in sensitive to both the intensity and the coastal areas due to tropical cyclones, track of the tropical cyclone. Based on obtaining good estimations of both flood the available historical information depth and extent. Figure 6.1 in Annex 1 about tropical cyclone activity in the shows the general scheme of the area, a stochastic Monte Carlo methodology used to obtain offshore total simulation is used to generate water level estimations. thousands of synthetic cyclones that Regardless of what tropical cyclone track are in statistical agreement with data is used in the risk assessment (historical observations. A large set of events (i.e. or synthetic), the statistical storm surge spanning several years) is then estimation can be summarized in six steps: available for the extreme value analysis, thus reducing uncertainties. 1. Selecting tropical cyclone events to be simulated In the present work, the national scale assessment is based on historic storm track 2. Obtaining tropical cyclone wind and sea level pressure fields information, and the local scale assessment at the Pagbilao site is based on modeling a 3. Grid design and models setup large number of synthetic events. A fully 4. Calibration and validation probabilistic coastal flood risk assessment approach is unaffordable at the national 5. Running the models scale, which encompasses more than 7,000 6. Extreme value analysis islands and islets and more than 30,000 km of coastline. For this reason and due to the Even though not all tropical cyclone events large number of land fall events found in the will produce significant surges, all historical historical tropical cyclone record (548 tropical cyclones that made land fall in the events), the national assessment is based Philippines between 1951 and 2014 have been solely on historical track information. At the considered to determine annual expected local scale in Pagbilao, the risk assessment benefits in terms of the reduction of flooding has been achieved by modelling a large by mangroves at the national scale. In number of events (1,462 synthetic tropical Pagbilao, both historical and synthetic cyclones whose tracks cross less than 300 tropical cyclones have been used to obtain km from Pagbilao) that represent 5,000 the offshore dynamics and statistics in order years of plausible tropical cyclone activity. to test uncertainties derived from a limited number of events. In all cases, three mangrove scenarios have been considered: mangrove extent in 1950, mangrove extent in 2010, and a hypothetical 6.2 Offshore Dynamics scenario where all mangroves are lost. The first step to run the wave and Modeling waves and sea levels is usually an hydrodynamic models is to obtain the wind efficient and reliable method for estimating fields corresponding to each of the events to risks in coastal areas. A number of storm be simulated. Generally, there are three basic surge modeling systems such ADCIRC (Lin et 32 Valuing the Protective Services of Mangroves in the Philippines approaches to reconstruct tropical cyclone estimated from the parametric model of wind fields: measurements, high resolution Holland (1980): atmospheric models, and parametric/ analytical simplified models (Holland et al., 1980, 2010; Emanuel and Rotunno, 2011; etc.). (7.2) In this study, for each selected storm, the surface axisymmetric wind field is estimated where Pc is the core pressure (minimum by calculating the wind velocity at the pressure), Pn the pressure at infinite radius, gradient level with the analytical wind profile and B the Holland parameter: of Emanuel and Rotunno (2011) which has yielded relatively good results (Lin and Chanvas, 2012). The wind speed (V) is determined as follows: (7.3) As an example, Figure 20 shows the wind footprint of the Super Typhoon Haiyan which (7.1) devastated the city of Tacloban in November 2013, including a time step of the parametric where r is the radius, f the Coriolis parameter, wind model for November 8th at 20:00 UTC. Rm the radius of maximum winds, and Vm the The hourly wind and sea level pressure fields maximum wind speed. To force the wave and generated with the parametric model are the hydrodynamic models, a reduction factor of forcing of the Delft 3D model (https:// 0.9 is used. The asymmetry of the wind field oss.deltares.nl/web/delft3d ), in which the is accounted for by adding 60% of the storm processes of tide, wind setup, inverse translation velocity. The surface pressure P is barometers and wave setup are simulated Figure 20 Wind footprint of the Super typhoon Haiyan which devastated the city of Tacloban in November 2013 (left), parametric wind field for the November 8th at 20:00 UTC (left). Wind speeds in legend are in km/h. 33 Valuing the Protective Services of Mangroves in the Philippines together, conserving the nonlinear and this grid is chosen for the baseline study. interaction between them. The Delft3D Increased model resolution should produce modeling suite is composed of several more realistic surges and waves near the modules of which this study utilizes the coast though this will increase the Delt3d-FLOW and Delft3d-WAVE modules computational cost. As an example, Figure 21 (see section 5). shows the differences of the simulated Haiyan storm surge in Tacloban using a 5 km The computational domain extends from or 2 km grid resolutions. Both grids produce 111.5-130.5° E and 4-21.5° N which has been similar storm surge patterns, with the 2 km found to be sufficiently wide to model the grid doing a better job of capturing small sea states generated by tropical cyclones scale processes such the long wave traveling long distances from the east. This amplification in the Gulf of Leyte. achieves a compromise between the quality of the desired results and the required Figure 6.2 in Annex 1 displays the grid used computational effort and allows the model to for the baseline storm surge study together capture distant waves and the interactions with the coastal points where waves and sea between extreme water levels and waves levels are obtained. Seafloor bathymetry and closer to the coastline. elevation data are extracted from the ETOPO 1 database which is a 1 arc-minute global The resolution of the numerical grid is set at relief model of Earth's surface that integrates 5 km X 5 km as an effective compromise land topography and ocean bathymetry from between computational cost and the ability numerous global and regional data sets to capture surge patterns. Simulated sea (Amante and Eakins 2009). level validations of different tropical cyclones indicate that the 5 km grid reasonably Tropical cyclone simulations are performed in reproduces storm surges in most locations, 2D mode with a time step of 30s. For Figure 21 Effects of grid size on surge predictions. Differences between the maximum simulated surges in Tacloban generated by the Super Typhoon Haiyan using 5 km (left) and 2 km grids. Similar patterns are obtained with both grids with the difference of the slightly larger storm surge levels captured with 2 km grid 34 Valuing the Protective Services of Mangroves in the Philippines Figure 22 Validation of the simulated storm surge (in meters) of Typhoon Rammasun in Legaspi and Subic Bay tide gauges. The upper panel represent the cyclone track with the minimum pressure in hPa. Figure 23 Validation of the simulated storm surge (in meters) of Typhoon Nesat in Manila tide gauge. The upper panel represent the cyclone track with the minimum pressure in hPa. validation, boundary conditions have been interactions will be considered in the next defined throughout with harmonic step of the methodology in which a 1D profile constituents obtained from the TPXO7.2 WAVE-FLOW coupling is performed. Global Tide Model (Egbert and Erofeeva 2002). However, since we are more Finally, the simulations of offshore dynamics interested in the statistical distribution of the are validated against observations of historic residuals, serial tropical cyclone simulations storm events. Although Super Typhoon have been carried out using a Neumann type Haiyan remains the most severe event in term boundary condition in which a water level of storm surges and destruction in the slope is defined rather than absolute water Philippines, there are no available tide level. Tests showed that, at the selected mesh gauges in areas where this typhoon resolution, coupling the FLOW and WAVE impacted to validate our model results. For modules did not result in any appreciable this reason Typhoon Rammasun (July 2014, increase in model skill. Existing non- see Figure 22) and Typhoon Nesat linearities in wave-current-sea level (September 2011, see Figure 23) have been 35 Valuing the Protective Services of Mangroves in the Philippines chosen to validate the simulation of offshore worldwide (http://www.aviso.altimetry.fr/en/ dynamics. data.html). Altimetry data corresponding to November 8th, 2013, when Super Typhoon We find that our model results align with Haiyan made land fall, are shown in Figure 6. observations, with only a few discrepancies 4 (Annex 1). As can be seen, due to of a few tens of centimeters found locally. It instrument limitations, there are no is unclear whether these differences are due measurements in the eye of the tropical to the model itself, the forcing data (track, cyclone, nevertheless waves up to 12 m were intensity, radius of maximum winds, etc.), or observed at a short distance to the north. local features of the observations. As an Measured values reasonably agree with those example, Figure 6.3 (Annex 1) shows the obtained throughout the simulation. results of the maximum storm surge, significant wave height, and mean wave Figure 24 shows the expected storm surge period produced by Super typhoon Haiyan. height for the one in 50 year event across the Philippines. Most of the flood heights are Higher resolutions might be necessary in some areas to account for local processes below 1 m (note blue colors in Figure 24). that can contribute significantly to the However water levels are above 2.5 m in experienced surges. Due to the large extent areas around Lamon (Calabarzon) and San analyzed, local processes will be solved in Miguel (Bicolandia) Bays (note redder colors in Figure 24). This coastline is one of the the next steps using a 1D profile wave-sea level coupled approach. most exposed to tropical cyclones in the Even though no wave measurement is Philippines, with a tropical cyclone occurring available for validating the simulated waves, every 2 years on average, and the presence there is a set of satellite missions (TOPEX, of pronounced bays and flat-gently-slopes contributes to storm surge amplification. ERS, GFO, etc.) that provide altimetry measurements of the wave heights Figure 6.5 (Annex 1) shows the water levels Figure 24 Storm surge (in meters above the mean sea level) due to wind set-up and the inverse barometer effect for 50 years return period. 36 Valuing the Protective Services of Mangroves in the Philippines for the other storm events including the 5, 10, and exceeding 20 m heights in northeastern 50 and 100 years return periods. Although Luzon. the 100 year return period has higher water Once the tropical cyclone dynamics at the levels, there is only a slight increase in values national scale are simulated, we then with respect to the 50 year map, indicating propagate these waves and sea levels that there are limits on the highest storm towards the coast to assess the influence of surge levels. mangrove forests in attenuating these storm surges. This step involves the coupling of Figure 25 and Figure 6.6 (in Annex 1) show high resolution wave and sea level models the same maps as above but for significant with a modification of the Manning’s wave heights produced by tropical cyclones coefficient as a function of the bottom cover following a similar extreme analysis. Waves type, depending on the mangrove scenario up to 10 m are observed in the most exposed being modelled. By coupling the wave and coasts of the northeastern region, where no flow models, we translate wave radiation continental shelves exist to attenuate waves stress into a contribution to the Total Water by breaking or bottom friction, for every 5 Level at the coast, and we consider the storm years. On the other hand, in the Subuyan, surge duration as a factor in the attenuation Visaya interior seas waves do not exceed 4 m potential of mangroves. due to fetch and depth limitations. For a return period of 100 years, tropical cyclones At the Pagbilao site, we apply a fully produce waves higher than 4 m in all the probabilistic approach by simulating 1,462 Philippines excluding the southernmost synthetic tropical cyclones, which represent region. The most exposed coast in the east 5,000 years of tropical cyclone activity (see can experience waves up to 14 m in Siargao Figure 6.7 in Annex 1). The stochastic tropical Island, increasing further towards the north cyclone track modeling is able to completely Figure 25 Significant wave heights (in meters) for 50 years return periods 37 Valuing the Protective Services of Mangroves in the Philippines cover the selected study area, thus water surface all along the Tablas strait, accounting for a wide range of possibilities causing the water to finally be piled up over of tropical cyclone approach angles, land fall the Tayabas Bay. locations and intensities. To compare statistical results from the Despite the large number of simulated historical and synthetic tracks, a general tropical cyclones, the maximum observed extreme value distribution was fitted to the surge in the 5,000 year time period storm surge data. As can be seen, both (considering only wind and inverse distributions are fairly similar, with the barometer effect) does not exceed 1.6 m, synthetic tracks being slightly higher. Results even for the most surge-prone synthetic indicate little difference below the 100 year track. As an example, Figure 26 shows a return period between the two distributions. synthetic tropical cyclones that generates one of the highest surges. This fast moving After defining offshore dynamics generated severe typhoon has a south-north track when by each tropical cyclone, we propagate crossing Pagbilao. With minimum pressures waves and sea levels over the previously of 890 hPa before reaching the Philippines, defined 1D profiles. Two different scales will the tropical cyclone weakens when first be analyzed with different inputs: making landfall in Moro Gulf, and then further intensifies to 930 hPa and 220 Km/h 10- min- 1. National scale: historical tropical averaged winds when crossing the warm cyclones. Jolo Sea towards Pagbilao. This tropical 2. Local scale: historical tropical cyclones cyclone generates high surges in Pagbilao, + synthetic tropical cyclones and it is able to transfer momentum to the Figure 26 Storm surge generated by the synthetic tropical cyclone number 1221. 38 Valuing the Protective Services of Mangroves in the Philippines 6.3 National Scale Nearshore hourly data of Significant Wave Height, Peak Dynamics: Historical Tropical Period and Storm Surge. These are reduced using the following approach: Cyclones Offshore dynamics from 63 years of historical 1. First, each of the 548 historical tropical cyclones (from 1951 to 2014) were tropical cyclone time series are obtained for the entire Philippines coast (548 reduced to the maximum values of 4 events, 8.6 events per year). Waves and sea variables over the storm duration: level are forced into a 5 km x 5 km mesh grid Maximum Significant Wave Height; in the whole country with SWAN (for waves) Peak Period associated to the and DELFT3D (for storm surge). maximum Significant Wave Height; Maximum Sea Level (i.e. Surge + Tide); 6.3.1 Classification of Tropical Cyclones and tropical cyclone duration. This The closest points to the coastline of the reduces the data into 3,225 points x coarse national numerical mesh (5x5km) are 548 tropical cyclone x 1 h/ tropical going to be used to feed the 1D numerical cyclone = 1,767,300 combinations of simulations over the vegetation fields (coral Hsmax, Tp, Sea Level and tropical reefs and mangroves). In total, there are cyclone duration. 3,225 points which contain the following 2. Then, since not all areas of the country hourly data time series: will experience these surges, maximum Storm Surges below 10 cm - Significant wave height time series are excluded. The tropical cyclones within each tropical cyclone producing significant storm surges - Peak period time series within each comprise only 1% of the total dataset, tropical cyclone or 17,673 combinations of Hsmax, Tp, Sea Level and tropical cyclone - Storm surge time series within duration. each tropical cyclone 3. Finally, as with the profiles, statistical However, storm surge is not the only tools are applied to reduce the component of Total Water Level offshore. number of hydrodynamic cases to be Sea Level is obtained by adding the simulated. A clustering technique, maximum Astronomical Tide to the storm Maximum Dissimilarity Algorithm surge time series at each location. It is crucial (MDA), was chosen for this purpose. In to consider the worst-case scenario in terms our clustering tests, MDA performed of Sea Level because of the sensitivity of the better than K-means inidentifying the numerical model to water depth when minimum number of clusters required propagating waves over coral reefs and to capture the maximum wave height mangroves. and storm surge. Note that the average tropical cyclone duration is 8 hours, but the longest events The coastal profiles were classified into 250 could be 50 hours long. families for each scenario with K-MEANS Thus we have 3,225 points x 548 tropical algorithm, or 750 theoretical profiles, cyclones x 8 h/tropical cyclone = 14,138,400 representing the entire Philippine coast. 39 Valuing the Protective Services of Mangroves in the Philippines 6.3.2 Habitat Pathway model, we have created a “pick-up table” or “interpolation table” of 37,500 different The 50 selected tropical cyclones are combinations which allows us to interpolate propagated over the 750 theoretical profiles the Flood Height for any scenario proposed. using DELFT 3D. In total, we have 50 x 750 = 37,500 simulations covering all bathymetry, 6.3.2 Flood Height Reconstruction bottom type and hydrodynamic combinations for the entire country. We take the maximum Total Water Level at the coast (i.e. Flood Height) for each of the The 37,500 1D simulations all consider the 37,500 simulations with the aim of following: reconstructing the Flood Height along the entire Philippine coastline, with the following - Flow and Waves modules are run steps: simultaneously (coupling model): DELFT 3D propagates storm surge - First we generate the interpolation and astronomical tide induced flow tables with 5 columns (Hs, Tp, Sea and SWAN module propagates waves Level and tropical cyclone duration as over the updated water depth. predictor variables and Flood Height as predicted variable) and 50 rows - The computational time length is two (the same as the number of times the tropical cyclone duration. representative tropical cyclones - Wave characteristics (Hs and Tp) are selected with MAXDISS clustering constant within the whole tropical technique). We have one table per cyclone scenario and profile (3 scenarios x 250 profiles/scenario= 750 tables). Figure - Total Water Level discharge has 6.11 in annex 1 shows a scheme of the triangular shape within the tropical interpolation datasets generated in the cyclone event, with a peak equal to project. the maximum Sea Level (Sea Level= Storm Surge + Astronomical Tide 99%, - Next, for each profile we select the corresponding to the worst case corresponding interpolation table for scenario) occurring at the mid-point of the 3 mangrove scenarios. Note that the tropical cyclone duration. we had previously grouped each of the 171,888 profiles into 250 - Profile discretization: 10m resolution representative families. (2000 nodes/profile) - Then, the 548 historical tropical From these simulations, a final value of Total cyclones are reconstructed at each Water Level at the shoreline (i.e. Flood profile by interpolating in the Height) is obtained for the entire Philippines mentioned tables. The interpolation coastline. With the 37,500 input variables technique is based on the Radial Basis combinations (Hs, Tp, Sea Level and tropical Functions (Camus et al. 2011a). The cyclone duration) from the hydrodynamic methodology has been tested in case point of view, and mangrove length and studies in different papers (Camus et mangrove average depth from the coastal al. 2011b, Nunes and Pawlak 2008). habitat point of view, and the resulting The methodology is not dependent on 37,500 Flood Height values calculated by the location, and is applicable globally 40 Valuing the Protective Services of Mangroves in the Philippines since the selection algorithm and obtained at each profile (for examples reconstruction technique are solely see Figure 6.11 in annex 1) dependent on the quality of the These tables are calculated for each profile databases to be downscaled to the (171,888 profiles for the entire Philippine coast and the number of cases coast) and three scenarios (1950, 2010 and selected using the MDA. no mangroves), providing the Flood Height - Once the 548 tropical cyclones at each for every 200 m of coastal line for 15 profile are reconstructed, the extreme different return periods. Figure 27 shows the values analysis is performed. For that Flood Height for a 25 year return period purpose, Peak Over Threshold under three scenarios as an example. selection method is applied to obtain the extreme regime of Flood Height. 6.3.4 Comparing Flood Height for We fixed a threshold of 98%, Different Scenarios corresponding to the strongest 2% To understand the effect of mangrove forests tropical cyclones events over the total on the propagation of Total Water Levels 548 registered between 1951 and 2014. (and thus the resulting Flood Height), the The 2% most destructive events extreme regime of Total Water Level offshore correspond to 11 tropical cyclones, is compared with the resulting Total Water Level on the coast after passing through the resulting in 1 extreme tropical cyclone mangroves. The selected profile contains a every 5 year, and consequently a mangrove extension of 1 km. Two scenarios minimum return period of 5 years. are shown: first, the current scenario of mangrove cover, and second, the - Three tables (one for each scenario) hypothetical scenario of no mangrove cover with 5, 6, 7, 8, 9, 10, 15, 20, 25, 30, 40, (see Figure 6. 9 in Annex 1). 50, 100, 150 and 200 years Return Period events of Flood Height are Figure 27 Flood height (m) in The Philippines for 25 year return period under tropical cyclone conditions, for scenarios with historical (1950) current (2010) and no mangrove cover. 41 Valuing the Protective Services of Mangroves in the Philippines Comparing scenarios allows us to evaluate 6.4 Regional Scale Nearshore the effect of mangroves in Total Water Level dissipation. Two comparisons can be done: Dynamics: Historical and Synthetic - Loss of protection between 1950 and Tropical Cyclones in Pagbilao 2010: How much did the Total Water Level increase from 1950 to 2010 due For a high resolution analysis in a local area, to mangrove loss? the historical tropical cyclones database is not enough to statistically study the extreme - Current protection of mangroves: How regime of these events. This is because the much would the Total Water Level number of “real” tropical cyclones in a small increase if we lose mangroves now? area is too small. Figure 28 shows both comparisons for the case of 50 year return period tropical 6.4.1 Synthetic Tropical Cyclones cyclone event. Generation Of the total 548 tropical cyclones registered in The Philippines, only 37 tracks pass Figure 28 Current protection: Increase of Flood Height (m) for a 50 year return period under tropical cyclone conditions between current mangrove cover and a theoretical case of no mangroves cover 42 Valuing the Protective Services of Mangroves in the Philippines through the Pagbilao study site, one event synthetic tropical cyclones for Pagbilao. This every two years. This is insufficient for is done as follows: statistical analysis, so we developed a new - The Flood height is calculated at each and larger database based on the available profile (1,162 profiles in Pagbilao) by historical information of the activity in the interpolating from the tables created with area. We used a stochastic Monte Carlo the 37,500 Delft3D model simulations for simulation to generate 1,462 synthetic historical tropical cyclones. The input cyclones in the whole country that are in variables are (Hs, Tp, Sea Level max, statistical agreement with observations and tropical cyclone duration). Note that the previously generated interpolation tables represent 5,000 years of storm activity. Out are “profile-specific”. Thus, we only have of the 1,462 tropical cyclones simulated to directly associate each profile to the throughout the Philippines, 456 tropical corresponding table and interpolating the cyclones pass through the Pagbilao study Flood Height accordingly. However, to site. extend the methodology to other sites, two other variables, defining habitat Offshore dynamics for these 5,000 years characteristics, should be considered: (456 tropical cyclones) of synthetic tropical mangroves length and mangrove average water depth. cyclones are obtained along the Pagbilao coast. As was done with the historical - Once the 456 tropical cyclones are tropical cyclone at national scale, waves and reconstructed at each profile, the extreme sea level are forced in 5 km x 5 km mesh grid values analysis is performed, similar to the historical tropical cyclone analyses, using in the whole country with SWAN (for waves) a Peak Over Threshold method with a and DELFT3D (for storm surge). The 98% threshold. For Pagbilao, the 2% most numerical simulations generated 546 destructive events correspond to 9 combinations of waves (significant wave tropical cyclones, resulting in 1 extreme height, peak period and wave direction) and tropical cyclone every 550 years. This Storm Surge intensity and duration. occurrence rate is so small that a new hypothesis is proposed to accomplish the 6.4.2 Flood Height Reconstruction extreme analysis: The 456 simulated (synthetic) tropical cyclones occur in the We evaluate the capacity of mangroves to same period of time as the historical attenuate waves and sea level, including the dataset, i.e. 1951-2014 (not 5,000 years, as effect of different scenarios of mangrove originally modelled). Consequently, the cover on Flood Height, at the local scale. The occurrence rate is increased to one national scale model already generated tropical cyclone event every 7 years. This 37,500 cases of Flood Height covering a allows a minimum return period of 7 wide range of input combinations of wave years. height, peak period, sea level and tropical - Three tables (one for each scenario) with cyclone duration. With the aim of reducing 7, 8, 9, 10, 15, 20, 25, 30, 40, 50, 100, 150 the computational cost, these interpolation and 200 years Return Period events of tables (for example see Figure 6.11 in Annex Flood Height are obtained at each profile, 1) for historical tropical cyclones were used shown in Figure 6.11 in annex 1. These to interpolate the Flood Height in Pagbilao. tables are calculated for each profile (1,162 profiles in Pagbilao) and each The study site covers 232 km of coastline and scenario (historical, current, and no includes 1,162 profiles for which the Flood mangrove cover), providing the Flood Height will be reconstructed for 456 Height level for every 200 m of coastal 43 Valuing the Protective Services of Mangroves in the Philippines line for 13 different return periods. The 25 and 50-years return period Flood Height are plotted as an example along the Pagbilao coastline for the three scenarios in Figure 29 and 30. Figure 29 Flood height in Pagbilao for 25 year return period event under tropical cyclone conditions for the three scenarios of historical, current and no mangroves. The values are water or flood height at the coast ranging from ~1m (yellow) up to ~5m (blue). Figure 30 Flood height in Pagbilao for 50 year return period event under tropical cyclone conditions for the three scenarios of historical, current and no mangroves. The values are water or flood height at the coast ranging from ~1m (yellow) up to ~5m (blue). 44 Valuing the Protective Services of Mangroves in the Philippines 7 | Coastal Impacts: Flooding Mask Calculation in the Philippines 7.0 Section Overview 7.1 National Scale Results (30 m This section describes how we translated resolution) flood heights at the coastline to inland flood At the national level, we used a flooding extents, at both the national scale and at the model with a 30 m resolution. In Figures 33 local scale in Pagbilao and compares the to 37, we show the results of this model at results at these different scales. The flood specific sites in South Manila and Roxas for a extent estimates are done nationwide for 50 year return period event under water ‘regular’ waves and historic tropical cyclones, levels predicted under ‘regular conditions’ and in Pagbilao for historic and synthetic and ‘extreme conditions’ (or tropical cyclone tropical cyclones. Flood extents are conditions). estimated for three mangrove scenarios- historic, current and total loss. At the national scale, the flood extents are estimated using a 7.2 Local Scale Results in Pagbilao simple hydraulically-connected bath-tub model and a 30 m elevation database. In and Busuanga (5m resolution) Pagbilao, a much higher resolution database At the local scale, in the Pagbilao and at 5 m resolution allows for the use of a more Busuanga sites, high resolution elevation sophisticated flooding model (the Rapid data is available (IFSAR 5m), which allows for Flood Spreading Methodology), which more detailed flooding analyses. The following figures show the mangrove cover provides better estimates of inland flood at each location (Pagbilao in Figure 36 and extents. MANILA Figure 31 Flooding in South Manila for 50 year return period event under regular conditions. Blue and grey colors indicate flooding extent under current mangroves (2010) and no mangroves. 45 Valuing the Protective Services of Mangroves in the Philippines Figure 32 Flooding in South Manila for 50 year return period under tropical cyclone conditions. The light and dark blue polygons indicate flooding extent under the different scenarios of current mangroves (2010) and no mangroves. ROXAS Figure 33 Roxas study site: mangrove cover in 2010 appears in green on the right. Figure 34 Flooding in Roxas for 50 year return period under regular conditions. The light and dark blue polygons indicate flooding extent under the different scenarios of current mangroves (2010) and no mangroves. 46 Valuing the Protective Services of Mangroves in the Philippines Figure 35 Flooding mask in Roxas for 50 years return period event in the Tropical Cyclones data. The light and the dark blue polygons in the figure on the right represent water and flooding of land under the two different scenarios. PAGBILAO Figure 36 The left image shows the location of Pagbilao in the Philippines. The right image shows the mangrove cover in 1950 and 2010. Figure 37 Flood extent and water depth in Pagbilao for 1 in 10 and 1 in 50 years return period event under tropical cyclone conditions under three different mangrove scenarios, Historical (1950), Current (2010) and No Mangroves. 47 Valuing the Protective Services of Mangroves in the Philippines BUSUANGA Figure 38 The left image shows the location of Busuanga in the Philippines. The right image shows the mangrove cover in 1950 and 2010. Figure 39 Flood extent and water depth in Busuanga for 1 in 10 and 1 in 50 year return period event under tropical cyclone conditions under three different mangrove scenarios: Historical (1950), Current (2010) and No Mangroves. Busuanga in Figure 38) and the flooding extent and water depth for three scenarios (historical, current and no-mangroves) and two storm return periods, 10 and 50 years (Pagbilao in Figure 37 and Busuanga in Figure 39). These results are from flooding levels predicted from extreme, or tropical cyclone, conditions. 48 Valuing the Protective Services of Mangroves in the Philippines 8 | Assessing the Benefits of High Resolution Data and Flooding Models 8.0 Section Overview 8.1 Comparing the Effects of Higher In this section we assess the potential of high and Lower Resolution Elevation Data resolution data and models for assessing flooding and flood protection benefits. There 8.1.1 Regular Wave Climate are 2 key messages from this assessment for For regular wave climate, we compared two the present and future work. Digital Terrain Model (DTM) databases: National DTM SRTM30 PLUS, with 30m 1. Flood estimates are greatly improved resolution (available across the Philippines) with higher resolution elevation data. and the regional DTM IFSAR with, 5 m At the national level, the best available resolution (only available in Pagbilao). The data was of low resolution. Our two DTM databases are compared by using comparisons with higher resolution the same flooding method, i.e., a site-based data suggest that we hydraulically-connected bathtub algorithm, underestimate the actual flooding which consists of connecting cells below the levels. Coastal Total Water Level. Other more sophisticated flooding methods, like RFSM- 2. With high resolution elevation data, EDA (Rapid Flood Spreading Model-Explicit flood estimates can be improved by Diffusion waves with Acceleration), cannot using better Rapid Flood Spreading be applied with such coarse elevation data. Models (RFSM) instead of hydraulically This comparison shows the high sensitivity of connected bathtub flooding models. the flooding model to the DTM resolution. The RFSM models provide lower site- The 30 m elevation data provides lower based flooding envelopes. However, quality results and underestimates coastal even with a more accurate model, the flooding, while the 5 m resolution database low resolution data available still lead provides more detailed flooding maps. (see us to underestimate flooding in our Figure 40 and 41.) national analyses. Figure 40 Flood Height in Pagbilao for 50 year return period event under regular conditions with current mangrove cover. The left image shows the SRTM 30m, the right image shows IFSAR 5m. 49 Valuing the Protective Services of Mangroves in the Philippines Figure 41 Flood Height in the most populated area of Pagbilao for 50 year return period under regular conditions for current mangrove cover. The left image shows the SRTM 30m, the right image shows IFSAR 5m. 8.1.2 Cyclone Data 8.2 Flooding Methods: Hydraulically The same comparison explained in Section Connected Bathtub (5m) vs RFSM 8.1.1 was performed for tropical cyclones (5m) events. Both DTM resolutions, SRTM30 PLUS The two flooding methodologies were and IFSAR 5 m, are tested for tropical compared at the local scale in Pagbilao using cyclones, and the same conclusions are the flood levels from the tropical cyclone reached: a minimum of 5 m DTM resolution is data. The DTM is IFSAR 5 m resolution, required if we want an accurate Flood Height which clearly improves the SRTM30 PLUS map. The 30 m DTM is too coarse for (30 m resolution). The hydraulically regional or local studies, but it is the highest connected bathtub methodology shows an resolution available at the national (or larger) overestimation of the flooding extent with scale. (See Figure 42 and 43.) respect to RFSM-EDA model. This could be due to the differences in the flooding method: - The hydraulically-connected bathtub algorithm is based on the hydraulic Figure 42 Flood Height in Pagbilao for 50 year return period event under tropical cyclone conditions for current mangrove cover. The left image shows the SRTM 30m, the right image shows IFSAR 5m. 50 Valuing the Protective Services of Mangroves in the Philippines Figure 43 Flood Height in the most populated area of Pagbilao for 50 year return period event under tropical cyclone conditions for current mangrove cover. The left image shows the SRTM 30m, the right image shows IFSAR connectivity between cells and assumes assumed with the peak of the flow equal that coastal flooding is stationary. The to the Flood Height associated to a given flooding mask given by this method is the return period. Furthermore, the model maximum envelope within a sea state allows to include water inlets (rainfalls or considering a constant landward flow. river flows), and outlets (water infiltration, Additionally, no soil friction or porosity is evapotranspiration). considered (no water loss by infiltration) In summary, the hydraulically connected and all connected cells below the Flood bathtub model provides the envelope of a Height are filled with water. constant flooding within the whole TC - RFSM-EDA is a high resolution hydraulic duration and, consequently, it overestimates model which considers the flow rate the flooding mask (which may be useful from within the tropical cyclone event. In this the perspective of flood warnings where it case, a triangular water discharge is can be better to over estimate flood levels). Figure 44 Flooding Height in Pagbilao region for 50 year return period event under tropical cyclone conditions for current mangrove cover. The left image shows the Hydraulically-connected Bathtub method, the right image shows the RFSM-EDA method. 51 Valuing the Protective Services of Mangroves in the Philippines Figure 45 Flood Height in the most populated area of Pagbilao for 50 year return period event under tropical cyclone conditions for current mangrove cover. The left image shows the Hydraulically-connected Bathtub method, the right image shows the RFSM-EDA method. In contrast, the RFSM-EDA model provides the punctual flooding associated to the peak of the Tropical Cyclone event. Figure 44 and Figure 45 show the differences in the flooding mask for both methods in Pagbilao, by using the same mangrove layer corresponding to current scenario. Fifty years return period flooding event has been plotted with flood heights reaching 4 meters in most of Pagbilao coast. The most affected area seems to be the West side of the bay where an additional zoom has been made, in order to notice the differences of both methodologies (see Figure 44 and Figure 45). 52 Valuing the Protective Services of Mangroves in the Philippines 9 | Damages and Benefits 9.0 Section Overview 9.1.1 Exposure of Assets: People and Stock This section describes the process for estimating flooding exposure, the resulting As discussed in section 2, we assessed the damages to people and property, and the consequences of flooding across five key estimation of risk and probability based on variables across the Philippines, which have event return periods, for each of the three total national values of: mangrove cover scenarios, at the national 1. Population: 100,234,428 people and local scales. Available global, national 2. Population below poverty level: and local databases are used to estimate the 20,371,701 people population, population below poverty level, 3. Residential Stock: $US residential stock, industrial stock, and length 129,506,000,000 of roads for the country. This information is combined with empirical damage curves for 4. Industrial Stock: $US 87,475,000,000 population, stock and roads to estimate the damage from flooding under different return 5. Roads: 217,456 km periods. Finally, the damages to population, 9.1.2 Damage Curves stock and roads are compared for ‘regular’ conditions, historic tropical cyclones and We followed existing approaches for synthetic tropical cyclones across the three assessing the damages to built capital as a mangrove cover scenarios, at the local and function of the level of flooding. We national scales. We did not apply discount calculated the percentage of built capital rates because we are only estimating current that has been damaged (D) for a given expected benefits; we are not, for example, flooding level and a certain coefficient that estimating the future flood reduction must be calibrated as D(h) = h/(h +k). This benefits from measures such as mangrove curve indicates that as flooding level restoration, for which a discount rate would increases, the percent of damages to built be applied. capital also increases. The differences in damages across the three These functions are different for each mangrove cover scenarios give the benefits category (i.e. population, stock, road) and of mangroves for risk reduction in terms of can be different within the same category annual expected monetary benefits and in (i.e. different types of residential stock). We terms of people protected. The section also were not able to access damage curve data describes the process of annualizing the risk that may exist for all portions of the reduction benefits so that these mangrove Philippines, so we used curves derived from values can ultimately be included in national the common database of damage functions ecosystem service accounts. in US HAZUS. HAZUS is a set of models and data developed by the Federal Emergency 9.1 Exposure, Damage Curves and the Management Agency of the United States of Estimation of Risk Probability America that considers potential losses from natural disasters such as earthquakes, floods, and hurricanes. 53 Valuing the Protective Services of Mangroves in the Philippines We tested the use of various damage curves The magnitude of risk is determined by the for population, residential and industrial damage probability distribution existing at stock and roads from HAZUS in the different areas and sectors. Assessing the Philippines, and we found that the results risk borne by society requires the were not significantly different from combination of all the variables presented approaches using simpler curves. into a set of synthetic indicators presenting the probabilistic distribution of impacts. To define case-specific semi-empiric damage functions for the Philippines we used a In summary, we combine the flooding damage function for all categories, i.e. information for different return periods with population and population below poverty the exposure and vulnerability of people and level, residential and industrial stocks, and property to obtain the damage associated road network (see Figure 9. 1 in Annex1). with different probabilities. 9.1.3 Assessing Risk – Combining Spatial In terms of the spatial summation of the risk Results results, for each asset (population, population below poverty, residential stock, We assessed flood heights along each industrial stock and road network) we did the coastal profile and then identified the area following: flooded within each coastal study unit. We 1. Raster layers from flooding and extended the flood heights inland by exposure were loaded. We ensuring hydraulic connectivity between homogenized their projection systems points at a 30m resolution, a significant and resolutions (if necessary). advancement over more common bathtub 2. Raster layers were transformed into approaches in earlier global flooding models. matrices. Zero and NA values were From the flooding levels and flooding extent, normalized to avoid possible errors in we calculated the total area of land affected following steps. and damages at each study unit. Flooding 3. For each matrix element, the damage maps were also intersected with population function was applied and the damage data after resampling from the original 100 m was obtained. resolution to the 30 m of the digital elevation 4. Matrices were then transformed to model. In addition to assessing risk and raster layers and saved. damages for particular events (e.g., 100 year storm event), we also examined average 5. The damage for the Philippines was obtained by summing all matrices or annual expected damages and benefits raster layers values. provided by mangroves. Due to technical and computational To estimate annual risk, we integrated the limitations, to perform the described values under the curve that compares built methodology, the Philippines was divided capital damaged by storm return period, i.e., into 71 sections. A buffer zone was defined the integration of the expected damage with for each section to ensure data continuity the probability of the storm events. This step and avoid contour problems. The processes helps us define the spatial and temporal described were executed for each of these 71 distribution of the risk level borne by society. sections, and then results were merged into single layers (see Figure 9. 2 Annex 1). 54 Valuing the Protective Services of Mangroves in the Philippines 9.2 National Scale Results flood damages avoided by them. We examine the benefits provided by current The results help understand the expected mangroves, and the additional benefits that benefits provided by mangroves for flood could be provided if mangroves were reduction to people and property annually restored to their 1950 distribution. We and for catastrophic events. The benefits provide results in absolute terms and terms provided by mangroves are assessed as the relative to the total economy, population and Figure 46 People affected in the Philippines. Reg. Climate (left) and Tropical Cyclones (right) Figure 47 Poor people affected in the Philippines. Reg. Climate (left) and Tropical Cyclones (right) Figure 48 Residential stock damaged in the Philippines. Reg. Climate (left) and Tropical Cyclones (right) 55 Valuing the Protective Services of Mangroves in the Philippines Figure 49 Industrial stock damaged in the Philippines. Reg. Climate (left) and Tropical Cyclones (right) Figure 50 Total stock (residential + industrial) damaged in the Philippines. Reg. Climate (left) and Tropical Cyclones (right) Figure 51: Roads network damaged in the Philippines. Reg. Climate (left) and Tropical Cyclones (right) 56 Valuing the Protective Services of Mangroves in the Philippines hectares of mangrove. We recognize that mangroves. For these results, we show the relative benefits vary spatially. The benefits curves for the two types of flood data from per hectare may be important to consider in the regular wave models and the cyclone cost effectiveness analyses for restoration. models. 9.2.1 Damages vs Return Period 9.2.2 Expected Damages: Annual and by The core results are summarized in Figure 46 Return Period to Figure 51, which show the consequences The expected damages from flooding for of flooding in terms of people and each scenario of mangrove cover are in Table infrastructure flooded for the different storm 3. They are given as annual expected return periods. The annual expected damages, and as damages per return period. damages, i.e., flooding of people and property, are calculated via the integration of each curve. The curves show expected damages by storm return period for the three different habitat scenarios of current mangroves, historical mangroves and no TOTAL DAMAGE (Annual Expected Damage) Historical 2,253,954 POPULATION Current 2,521,004 (nº people) No Mangrove 3,134,465 Historical 558,009 POPULATION BELOW POVERTY Current 619,488 (nº people) No Mangrove 761,915 Historical 1,816 RESIDENTIAL STOCK Current 2,073 (millions US $ 2014) No Mangrove 2,637 Historical 1,308 INDUSTRIAL STOCK Current 1,503 (millions US $ 2014) No Mangrove 1,940 Historical 3,124 TOTAL STOCK Current 3,577 No Mangrove 4,577 Historical 2,784 ROADS Current 2,990 (Km) No Mangrove 3,757 Table 3 Annual Expected Damage in the Philippines in terms of people, stock and km of roads 57 Valuing the Protective Services of Mangroves in the Philippines ANNUAL EXPECTED BENEFITS OF MANGROVES FOR FLOOD REDUCTION POPULATION Current Benefits +613,431 (nº people) Potential Restoration Benefits +267,050 POPULATION BELOW Current Benefits +142,428 POVERTY (nº people) Potential Restoration Benefits +61,479 RESIDENTIAL STOCK Current Benefits +564 (millions US $) Potential Restoration Benefits +257 INDUSTRIAL STOCK Current Benefits +437 (millions US $) Potential Restoration Benefits +195 TOTAL STOCK Current Benefits +1,001 (millions US $) Potential Restoration Benefits +452 ROADS Current Benefits +767 (km) Potential Restoration Benefits +206 Table 4 Annual Expected Benefits provided by mangroves in the Philippines 9.2.3 Expected Benefits: Annual and by areas that have lost mangroves may have Return Period been developed in ways that prevent the restoration of previously existing mangroves. The annual expected benefits provided by 9.2.4 Annual Expected Benefits per Ha mangroves are estimated by comparing scenarios. The annual expected benefits of mangroves received from current mangroves are the Given that there are currently 310,000 ha of difference in damages between the Current mangroves in the Philippines, we can provide (2010) and No Mangrove scenarios. (see a figure for the average current annual Table 4). benefits per hectare of mangrove. Across the Philippines, each 10 hectares of mangrove We can also estimate how many more flood reduces flooding for 20 people of which 5 reduction benefits could potentially be are below the poverty level, and provides gained by restoring mangroves to their more than US $32,000 in prevented distribution in 1950. This is the difference damages to built stock annually. If between expected benefits from Historic mangroves were restored to their 1950 (1950) and Current (2010) mangroves. distribution, each 10 hectares of mangroves Caution should be applied when making would reduce flooding for an additional 9 assumptions about the potential benefits of people, 2 of them below poverty level, and mangrove restoration, because many of the 58 Valuing the Protective Services of Mangroves in the Philippines ANNUAL EXPECTED BENEFITS OF MANGROVES FOR FLOOD REDUCTION (each 10 Ha of Mangroves) POPULATION Current Benefits +20 (nº people/10ha) Potential Restoration Benefits +9 POPULATION BELOW POVERTY Current Benefits +5 (nº people/10ha) Potential Restoration Benefits +2 RESIDENTIAL STOCK Current Benefits +18,161 (millions US $ 2014/10ha) Potential Restoration Benefits +8,290 INDUSTRIAL STOCK Current Benefits +14,097 (millions US $ 2014/10ha) Potential Restoration Benefits +6,323 Current Benefits +32,258 TOTAL STOCK Potential Restoration Benefits +14,613 ROADS Current Benefits +25 (Km/10ha) Potential Restoration Benefits +7 Table 5 Annual Expected Benefits provided by each 10 Ha of mangroves in the Philippines would avoid more than US $14,000 in catastrophic event), mangroves would damages to built stock annually. reduce flooding for 1,290,308 people; 277,157 of them below the poverty level, and would 9.2.5 Catastrophic Benefits (100 years prevent US $2,046 million in damages to return period event) built stock. If mangroves were restored to Analyzing the most extreme events gives us their 1950 distribution, each 10 hectares of information about the maximum current mangroves would reduce flooding for an benefits and the maximum potential additional 503,297 people, 108,516 of them restoration benefits provided by mangroves below poverty level, and would avoid US in the Philippines. Across the Philippines, for $866 million in damages to built stock a 100 year return period event (i.e., a annually. CATASTROPHIC BENEFITS (100 years return period event) POPULATION Current Benefits +1,290,308 (nº people) Potential Restoration Benefits +503,297 Current Benefits +277,157 POPULATION BELOW POVERTY (nº people) Potential Restoration Benefits +108,516 RESIDENTIAL STOCK Current Benefits +1,147 (millions US $ 2014) Potential Restoration Benefits +496 INDUSTRIAL STOCK Current Benefits +900 (millions US $ 2014) Potential Restoration Benefits +370 Current Benefits +2,046 TOTAL STOCK Potential Restoration Benefits +866 ROADS Current Benefits +1,566 (Km) Potential Restoration Benefits +498 Table 6 Catastrophic benefits provided by mangroves in the Philippines against 100 year return period event 59 Valuing the Protective Services of Mangroves in the Philippines 9.2.6 National Maps of Annual Expected Benefits CURRENT BENEFITS: POPULATION Figure 52 National distribution of the Annual Current Benefits provided by mangroves to people in the Philippines (25 km aggregation units) CURRENT BENEFITS: POPULATION BELOW POVERTY Figure 53 National distribution of the Annual Current Benefits provided by mangroves to people below poverty in the Philippines (25 km aggregation units) 60 Valuing the Protective Services of Mangroves in the Philippines CURRENT BENEFITS: TOTAL STOCK: INDUSTRIAL + RESIDENTIAL STOCK Figure 54 National distribution of the Annual Current Benefits provided by mangroves to the total stock (industrial + residential) in the Philippines (25 km aggregation units) CURRENT BENEFITS: KM OF ROADS Figure 55 National distribution of the Annual Current Benefits provided by mangroves to roads in the Philippines (25 km aggregation units) 61 Valuing the Protective Services of Mangroves in the Philippines Figure 56 People flooded in Pagbilao. Reg. Climate (left) and Tropical Cyclones (right) Figure 57 People below poverty flooded in Pagbilao. Reg. Climate (left) and Tropical Cyclones (right) Figure 58 Residential stock damage in Pagbilao. Reg. Climate (left) and Tropical Cyclones (right) 9.3 Local Scale Results: Pagbilao available for Pagbilao, which allowed us to compare flood mapping results for high and The spatial distribution of benefits differs lower resolution analyses. Below using the along the coastline of the Philippines. As highest resolution elevation data and flood noted above, high resolution elevation data is model, we provide high resolution estimates 62 Valuing the Protective Services of Mangroves in the Philippines Figure 59 Industrial stock damaged in Pagbilao. Reg. Climate (left) and Tropical Cyclones (right) Figure 60 Total stock (residential + industrial) damaged in Pagbilao. Reg. Climate (left) and Tropical Cyclones Figure 61 Roads damaged in Pagbilao. Reg. Climate (left) and Tropical Cyclones (right). Note that the curves for roads flooded under regular wave climate are overlapped for historical and current mangroves. of the flood protection benefits from 9.3.1 Damages vs Return Period mangroves. The core results are summarized in Figure 56 to Figure 61, which show the consequences of flooding in terms of damages to people 63 Valuing the Protective Services of Mangroves in the Philippines and infrastructure for the different storm return periods. The curves show expected damages for habitat distribution scenarios of current mangroves, historical mangroves and no mangroves. The annual expected damages- i.e., flooding consequences to people and property- are the integration of each curve. For these results we show the curves for the two types of flood data from the regular wave models and the cyclone models. 9.3.2 Annual Expected Damage Three scenarios were analyzed: mangrove level at 1950, mangrove level at 2010 and no mangroves. The annual expected damage for Pagbilao for each scenario was obtained. TOTAL DAMAGE (Annual Expected Damage) Historical 57,814 POPULATION Current 59,145 (nº people) No Mangrove 60,414 Historical 7,285 POPULATION BELOW POVERTY Current 7,462 (nº people) No Mangrove 7,655 Historical 26,41 RESIDENTIAL STOCK Current 27,26 (millions US $ 2014) No Mangrove 27,94 Historical 19,79 INDUSTRIAL STOCK Current 20,09 (millions US $ 2014) No Mangrove 20,37 Historical 46,21 TOTAL STOCK Current 47,35 No Mangrove 48,31 Historical 82,34 ROADS Current 89,55 (km) No Mangrove 96,79 Table 7 Annual Expected Damage in Pagbilao in terms of people, stock and km of roads 64 Valuing the Protective Services of Mangroves in the Philippines 9.3.3 Annual Expected Benefits 9.3.4 Annual Expected Benefits per 10 The annual expected benefits received from Ha of mangroves current mangroves is the difference in There are currently 4,560 ha of mangroves in damages between the Current and No Pagbilao. Each 10 hectares of mangroves Mangrove scenarios. We can also estimate reduce flooding for 2.92 people, 42% of them how many more flood reduction benefits below poverty level, and avoid more than US could potentially be gained by restoring $2,107 in damages to built stock annually. In mangroves to their distribution in 1950. ANNUAL EXPECTED BENEFITS OF MANGROVES FOR FLOOD REDUCTION POPULATION Current Benefits +1,269 (nº people) Potential Restoration Benefits +1,332 POPULATION BELOW POVERTY Current Benefits +193 (nº people) Potential Restoration Benefits +178 RESIDENTIAL STOCK Current Benefits +0.681 (millions US $ 2014) Potential Restoration Benefits +0.847 INDUSTRIAL STOCK Current Benefits +0.280 (millions US $ 2014) Potential Restoration Benefits +0.298 Current Benefits +0.961 TOTAL STOCK Potential Restoration Benefits +1.145 ROADS Current Benefits +7,241 (Km) Potential Restoration Benefits +7,211 Table 8: Annual Expected Benefits provided by mangroves in Pagbilao ANNUAL EXPECTED BENEFITS OF MANGROVES FOR FLOOD REDUCTION (each 10 Ha of Mangroves) POPULATION Current Benefits +2.92 (nº people/10ha) Potential Restoration Benefits +2.78 POPULATION BELOW POVERTY Current Benefits +0.42 (nº people/10ha) Potential Restoration Benefits +0.39 RESIDENTIAL STOCK Current Benefits +1,493 ((millions US $ 2014)/10ha) Potential Restoration Benefits +1,857 INDUSTRIAL STOCK Current Benefits +614 ((millions US $ 2014)/10ha) Potential Restoration Benefits +654 Current Benefits +2,107 TOTAL STOCK Potential Restoration Benefits +2,511 ROADS Current Benefits +15.88 (Km/10ha) Potential Restoration Benefits +15.81 Table 9 Annual Expected Benefits provided by each 10 Ha of mangroves in Pagbilao 65 Valuing the Protective Services of Mangroves in the Philippines CATASTROPHIC BENEFITS (100 years return period event) POPULATION Current Benefits +1,741 (nº people) Potential Restoration Benefits +1,135 POPULATION BELOW POVERTY Current Benefits +286 (nº people) Potential Restoration Benefits +164 RESIDENTIAL STOCK Current Benefits +0.643 (US $ millions 2014) Potential Restoration Benefits +0.642 INDUSTRIAL STOCK Current Benefits +0.313 (mill. US$ 2014) Potential Restoration Benefits +0.252 TOTAL STOCK Current Benefits +0.955 ($US Millions) Potential Restoration Benefits +0.895 ROADS Current Benefits +6,728 (Km) Potential Restoration Benefits +7,532 Table 10 Catastrophic Benefits provided by mangroves in Pagbilao against 100 years return period event 1950, there were 6,652 ha of mangroves. If US $895,000 million in damages to built mangroves were restored to their 1950 stock annually. distribution, each 10 hectares of mangroves in Pagbilao would reduce flooding for an additional 2.78 people, 39% of them below poverty level, and would avoid more than US $2,511 million in damages to built stock annually. 9.3.5 Benefits for catastrophic events (100 year return period event) Analyzing the most extreme events gives us information about the maximum current benefits and the maximum potential restoration benefits provided by mangroves in Pagbilao. Across Pagbilao and for a 100 year return period event (i.e., a catastrophic event), mangroves would reduce flooding for 1,741 people, 286 of them below poverty level, and avoid US $955,000 of damages to in built stock.If mangroves were restored to their 1950 distribution, they would reduce flooding for an additional 1,135 people, 164 of them below poverty level, and would avoid 66 Valuing the Protective Services of Mangroves in the Philippines 10 | Conclusions Mangrove conservation and restoration can Comprehensive Land Use Plans of local be an important part of the solution for governments. reducing coastal risks. This Report provides a • PAGASA10 and Local Government Units social and economic valuation of mangroves may use these results to inform and that can inform the policy and practice of improve their risk assessment and flood many Philippine agencies, businesses and risk mapping. organizations across development, aid, risk reduction and conservation sectors as they • These results can be considered in seek to identify sustainable and cost- insurance industry risk models, which may effective approaches for risk reduction. potentially influence insurance premiums By showing the spatial variation of the flood in the Philippines.The results may inform reduction benefits provided by mangroves, the development of innovative finance these results can identify the places where mechanisms, including catastrophic mangrove management may yield the hazard bonds, resilience bonds, and blue greatest returns. By valuing these coastal bonds, which could better account for the protection benefits in terms used by finance value and potential premium reductions and development decision-makers (e.g., associated with mangrove conservation annual expected benefits), these results can and restoration. be readily used alongside common metrics • In the past nature-based measures for of national economic accounting, and can coastal protection, such as mangrove inform risk reduction, development and restoration, were not assessed for their environmental conservation decisions in the cost effectiveness for risk reduction, Philippines. because rigorous values of their coastal protection benefits were missing. In the Philippines, many opportunities exist Now we can rigorously value these for the application of these results: services, and we can inform cost-benefit • analyses and comparisons of different These results can help identify priority coastal protection options, including sites for mangrove conservation and natural defenses, built defenses and restoration for coastal protection, either hybrid approaches. as ‘stand-alone’ solutions, or part of hybrid approaches that combine natural defenses, like mangroves, with built infrastructure. Numerous programs can incorporate these results into their plans and analysis, including: the National Greening Program; Integrated Area Development, Risk Resilience and Sustainability Program; Green Climate Fund and People Survival Fund; and the 67 Valuing the Protective Services of Mangroves in the Philippines Annex 1 | Figures Figure 1.1 Example of land cover and land use change in Pagbilao Bay. Previously existing mangrove forests have been converted to agriculture Mangroves continue to propagate on the new coastline. Figure 1.2 General view of the methodology to evaluate coastal protection services of ecosystems like coral reefs and mangroves. 68 Valuing the Protective Services of Mangroves in the Philippines Figure 2.1 Topography IFSAR 5m resolution (colored area) and the ensemble Topography (ETOPO) and bathymetry (SEAWIFS) with 1km resolution (red dots) in Pagbilao. Figure 2.2 Monthly climatology of tropical cyclone activity. 69 Valuing the Protective Services of Mangroves in the Philippines Figure 2.3 Tropical Cyclone tracks in the 5,000 year synthetic dataset. Figure 2.4 Location of the six tide gauges used in this study. 70 Valuing the Protective Services of Mangroves in the Philippines Figure 2.5 Example of WorldPop layer in Pagbilao and Manila areas in The Philippines in 2010 (People per hectare). Figure 3.1 Specific methodology to evaluate the coastal protection provided by mangroves against regular wave climate and Tropical Cyclones events in The Philippines. Step 0 has been additionally included to explain the pre- processing work in coastal segmentation. This step is specific of this projects and it has not been included in the general methodology (Figure 3 and Figure 1. 2 in the Annex) 71 Valuing the Protective Services of Mangroves in the Philippines Figure 4.1 Validation of TWL generated by Haiyan Tropical Cyclone forced with Astronomical Tide+Wind+Waves (left). Validation of Storm Surge generated by Haiyan Tropical Cyclone forced with Wind (right). Red lines are the numerically simulated TWL and blue line are the field measurements of the TWL Figure 4.2 Validation of TWL generated by Nesat Tropical Cyclone forced with Astronomical Tide+Wind+Waves (left). Validation of Storm Surge generated by Nesat Tropical Cyclone forced with Wind (right). Red lines are the numerically simulated TWL and blue line are the field measurements of the TWL 72 Valuing the Protective Services of Mangroves in the Philippines Figure 5.1 Offshore Ocean Dynamics database points: (1) Waves, (2) Astronomical Tide, (3) Storm Surge and (4) Wind Figure 5.2 Offshore dynamics datasets selection method. Example of one point in the coast of Pagbilao 73 Valuing the Protective Services of Mangroves in the Philippines Figure 5.3 Interpolation table for Flood Height estimation. Values obtained from the maximum Total Water Level in coast provided by DELFT 3D runs. Note – to be completed later. Figure 6.1: General scheme of the methodology used to obtain offshore total water level estimations. 74 Valuing the Protective Services of Mangroves in the Philippines Figure 6.2 The 5 Km grid used for the baseline storm surge study, coastal points where waves and sea levels are obtained are highlighted in cyan. Figure 6.3 Maximum simulated storm surge produced by the Super Typhoon Haiyan (upper panel), significant wave height (bottom left panel) and wave mean period (bottom right panel). 75 Valuing the Protective Services of Mangroves in the Philippines Figure 6.4 Altimeter significant wave heights (in meters) measured on November 8th, 2013 Figure 6.5 Storm surge (in meters above the mean sea level) due to wind set-up and the inverse barometer effect for 5, 10, 50 and 100 years return period. 76 Valuing the Protective Services of Mangroves in the Philippines Figure 6.6 Significant wave heights (in meters) for 5, 10, 50 and 100 years return periods Figure 6.7 Comparison of the historical available tropical cyclone tracks against synthetic in Pagbilao 77 Valuing the Protective Services of Mangroves in the Philippines Figure 6.8 Comparison the generalized extreme value distributions in Pagbilao fitted to the historical data (blue dots and line) and synthetic (red dots and line). Figure 6.9 Total water level WL offshore versus Total Water Level inshore in one 1D profile of Pagbilao 78 Valuing the Protective Services of Mangroves in the Philippines Figure 6.10 Predicted increase in Flood Height (1) Between Historical and Current Mangrove distribution and (2) Between Current and a No Mangroves scenario for a 25 year return period event considering tropical cyclone events. Figure 6.11 Interpolation table for Flood Height estimation. Values obtained from the maximum Total Water Level in coast provided by DELFT 3D runs. NOTE – this table to be completed later. Figure 9.1 Damage functions of people and people below poverty level (left), residential and industrial stock (middle) and road network (right). They represent the percentage of damage at each flooding height level 79 Valuing the Protective Services of Mangroves in the Philippines Figure 9.2 The Philippines sections in which the country was divided for optimize the computation cost Figure 9.3 National distribution of the Annual Potential Benefits provided by mangroves to people in the Philippines (25 km aggregation units) 80 Valuing the Protective Services of Mangroves in the Philippines Figure 9.4 National distribution of the Annual Potential Benefits provided by mangroves to people below poverty in the Philippines (25 km aggregation units) Figure 9.5 National distribution of the Annual Potential Benefits provided by mangroves to the total stock (industrial + residential) in the Philippines (25 km aggregation units) 81 Valuing the Protective Services of Mangroves in the Philippines Figure 9.6 National distribution of the Annual Potential Benefits provided by mangroves to roads in the Philippines (25 km aggregation units) 82 Valuing the Protective Services of Mangroves in the Philippines Annex 2 | Physics and Governing Equations Delft 3D model has been used in this project to propagate waves and storm surge induced by Tropical Cyclones and regular storms in the Philippines. The Delft3D modeling suite is composed of several modules of which this study utilizes the Delt3d-FLOW and Delft3d-WAVE modules. The FLOW module has been implemented to calculate the contribution of the storm surge and astronomical tide to the Total Water Level. It has been externally forced with the offshore water level calculated at the beginning of each profile (offshore water level is the linear summation of offshore storm surge and astronomical tide). The FLOW module assumes shallow water Boussinesq approach to solve the Navier Stokes equations for incompressible fluid (depth average continuity equations and momentum equations are simultaneously solved in 2D mode). WAVE module has been implemented to calculate wave set-up contribution to the Total Water Level in the coast. Wave radiation stresses and their gradients are computed and shared to the hydro dynamic model on the same mesh used in the hydrodynamic simulation. A wide range of validation cases (explained in the following sections), support Delft3D model. In this project, flow and waves are simultaneously simulated by coupling SWAN model (waves) with DELFT 3D (flow). The wave computation uses flow characteristics from a completed Delft3D-FLOW computation, so that the effect of flow on waves is accounted for. Coupled Delft 3D model (flow+wave) account for the following physics: - Wave refraction over a bottom of variable depth and/or a spatially varying ambient current - Depth and current-induced shoaling - Wave generation by wind - Wave dissipation by white capping - Dissipation by depth-induced breaking - Dissipation due to bottom friction (three different formulations) - Nonlinear wave-wave interactions (both quadruplets and triads) - Wave blocking by flow - Transmission through, blockage by or reflection against obstacles - Diffraction The governing three-dimensional equations describing free-surface flows can be derived from the Navier-Stokes equations after averaging over turbulence time-scales (Reynolds-averaged Navier-Stokes equations). Such equations express the physical principle of conservation of volume, mass and momentum. In this section, we describe the shallow water equations, for which the depth is assumed to be much smaller than the horizontal length scales of flow and bathymetry. Under the further assumption that the vertical accelerations are small compared to the horizontal ones the shallow water assumption is valid. This means that the vertical momentum equation is reduced to the hydrostatic pressure relation. The equations in case of a non-hydrostatic pressure are described in a separate section. In the latter case, non-hydrostatic pressure terms are added to the shallow water equations, which make the equations practically equivalent to the 83 Valuing the Protective Services of Mangroves in the Philippines incompressible Navier-Stokes equations. This means that in Delft3D-FLOW the user has the possibility to either apply a hydrostatic or a non-hydrostatic pressure model. The three-dimensional hydrostatic shallow water equations, which for convenience of presentation are given in Cartesian rectangular coordinates in the horizontal and V-coordinates in the vertical, are described by: (5.1) (5.2) (5.3) Where are the three directional components of velocity, is the local water depth, is the water level oscillation with respect the mean water level, is the coriolis coefficient, is the total water depth (), are the local source and sink per unit volume and is the hydrostatic pressure. The vertical velocities Z in the V-coordinate system are computed from the continuity equation, where the global source or sink per unit area and are the depth-average velociety in x-direction and y-direction. (5.4) by integrating in the vertical from the bottom to a level. The (comparatively small) vertical velocity w in the x-y-z Cartesian coordinate system can be expressed in the horizontal velocities, water depths, water levels and vertical velocities according to: (5.6) 84 Valuing the Protective Services of Mangroves in the Philippines Annex 3 | Computational Cost and Hard Disk Memory Required An additional analysis of DELFT 3D was carried out, consisting in measuring the computational cost and the memory required to save results under different forcing conditions. The test case characteristics are the following: - 4 days length Tropical Cyclone (Haiyan) - 56 control points with time records every minute - Time records every hour in the whole mesh (302x352 =106304 cells) Case Computational cost (min) Memory (MB) Astronomical Tide 5 793 Wind 7 1200 Wind+ Astronomical Tide 7 1200 Wind+Astronomical Tide +Waves(swell) 104 1600 Wind+Astronomical Tide +Waves(wind) 104 1600 Wind+Waves(wind) 104 1600 Table 12 Computational cost and memory required for Natianal scale 2D simulations with Delft 3D under different forcing methods 85 Valuing the Protective Services of Mangroves in the Philippines References Cited Amante C, Eakins BW. 2009. ETOPO1 1 arc-minute global relief model: procedures, data sources and analysis. National Oceanic and Atmospheric Administration. Beck MW. 2014. Coasts at Risk: An assessment of coastal risks and the role of environmental solutions. Narrangansett. Beck MW, Losada I, Menendez O, Reguero BG, Diaz Simal P, Fernandez F. In review. The global flood protection savings provided by coral reefs. Nature Communications. Burke L, Reytar K, Spalding M, Perry A. 2011. Reefs at risk revisited. WRI. Camus P, Mendez FJ, Medina R. 2011a. A hybrid efficient method to downscale wave climate to coastal areas. Coastal Engineering 58:851-862. Camus P, Mendez FJ, Medina R, Cofiño AS. 2011b. Analysis of clustering and selection algorithms for the study of multivariate wave climate. Coastal Engineering 58:453-462. Carrère L, Lyard F. 2003. Modeling the barotropic response of the global ocean to atmospheric wind and pressure forcing - comparisons with observations. Geophysical Research Letters 30:DOI: 10.1029/2002GL016473. Cid A, Castanedo S, Abascal AJ, Menéndez M, Medina R. 2014. A high resolution hindcast of the meteorological sea level component for Southern Europe: the GOS dataset. Climate Dynamics 43:1-18. Cid Carrera A. 2015. Modelado Numérico y Estadístico de la Componente Meteorológica del Nivel del Mar a Escala Regional y Global. Numerical and Statistical Modelling of the Atmospheric Surge at Regional and Global Scale. De Bono, Andrea, Bruno Chatenoux. 2015. A Global Exposure Model for GAR 2015, UNEP/Grid, Geneva. Background Paper prepared for the 2015 Global Assessment Report on Disaster Risk Reduction. Geneva, Switzerland: UNISDR. DRR Platform. 2014. Disaster Risk Reduction and Management in the Philippines. A desk study conducted by the Swiss NGO DRR Platform following Typhoon Haiyan, Luzerne, July 2014. Available at http://www.drrplatform.org/images/DocPub/RiskAssessment.pdf Egbert GD, Erofeeva SY. 2002. Efficient inverse modeling of barotropic ocean tides. Journal of Atmospheric and Oceanic Technology 19:183-204. Hamilton LS, Dixon JA, Miller GO. 1989. Mangrove forests: an undervalued resource of land and the sea. Pages 254-288 in Borgese EM, Ginsburg N, Morgan JR, eds. Ocean Yearbook 8, University of Chicago Press,. Myers N. 1998. Environmental degradation and some economic consequences in the Philippines. Environmental Conservation 15:205-214. NEDA. 2017. Philippine Develpoment Plan 2017-2022. Pasig City. Report no. Nunes V, Pawlak G. 2008. Observations of bed roughness of a coral reef. Journal of Coastal Research 24:39-50. Perez, J., Menendez, M., Losada, I.J. 2017 GOW2: A global wave hindcast for coastal applications. Coastal Engineering, 124: 1-11. Primavera JH. 1991. Intensive prawn farming int he Philippines: ecological, social and economic implications. Ambio 20:28-33. ---. 1995. Mangroves and brackish water pond culture in the Philippines. Hydrobiology 295:303-309. Spalding MD, Kainuma M, Collins L. 2010. World mangrove atlas. Earthscan. UNISDR. 2011. Global assessment report on disaster risk reduction revealing risk, redefining development. UN. ---. 2015a. Annual Report 2014. United Nations Office for Disaster Risk Reduction. 86 Valuing the Protective Services of Mangroves in the Philippines ---. 2015b. Making development sustainable: the future of disaster risk management, global assessment report on disaster risk reduction. United Nations Office for Disaster Risk Reduction. ---.2015c Global assessment report on disaster risk reduction 2015. GAR Global Risk Assessment: Data, Methodology, Sources and Usage. Walters BB. 1997. Human ecological questions for tropical restoration: experiences from planting native upland trees and mangrove in the Philippines. Forest Ecology and Management 99:275-290. World Bank. 2016. Managing coasts with natural solutions: Guidelines for measuring and valuing the coastal protection services of mangroves and coral reefs. (eds M. W. Beck, G-M Lange). Washington, DC: World Bank. Report no. World Risk Report. 2016. World Risk Report: Logistics and infrastructure. Berlin. Report no. Zedler JB, Kercher S. 2005. Wetland resources: status, trends, ecosystem services, and restorability. Annual Review of Environment and Resources 30:39-74. Zhang K, Liu H, Li Y, Xu H, Shen J, Rhome J, Smith Iii TJ. 2012. The role of mangroves in attenuating storm surges. Estuarine, Coastal and Shelf Science 102–103:11-23. 87