REPORT ASSESSMENT OF TECHNICAL SOLAR ROOFTOP PV POTENTIAL IN VIETNAM May 2018 This report was prepared by the firm Effigis with inputs from their local partner, the Centre for Environmental Fluid Dynamics (CEFD) at the University of Vietnam, under contract to the World Bank. It is Solar Power Scale-Up Technical Assistance Project: Vietnam [Project ID: P162510]. The work was funded by the Korean Green Growth Trust Fund (KGGTF), and benefited from staff time and support provided by the Energy Sector Management Assistance Program (ESMAP). Due to the technical nature, the content of this report has not undergone World Bank peer review. Users are therefore advised to exercise caution when utilizing the information and data contained. The data generated under this deliverable is temporarily available on a visualization website hosted by Effigis, but is likely to be transferred to the ENERGYDATA platform in the future. Copyright © 2018 THE WORLD BANK Washington DC 20433 Telephone: +1-202-473-1000 Internet: www.worldbank.org This work is a product of consultants of the World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of the World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work and accept no responsibility for any consequence of their use. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of the World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because the World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for non- commercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: +1-202-522-2625; e-mail: pubrights@worldbank.org. Furthermore, the KGGTF Program Manager would appreciate receiving a copy of the publication that uses this publication for its source sent in care of the address above. All images remain the sole property of their source and may not be used for any purpose without written permission from the source. Attribution Please cite the work as follows: World Bank. 2018. Assessment of Technical Solar Rooftop PV Potential in Vietnam. Washington, DC: World Bank. World Bank Assessment of Technical Rooftop Solar PV Potential in Vietnam Final report Selection # 1231720 Submitted by : Effigis Geo-Solutions Inc. 30 May 2018 Selection No. 1231720 Final Report TABLE OF CONTENTS 1. EXECUTIVE SUMMARY .................................................................................................................................. 1 2. BACKGROUND .............................................................................................................................................. 2 3. AREA OF INTEREST AND SATELLITE IMAGERY ............................................................................................... 5 4. APPLICATION OF THE METHODOLOGY.......................................................................................................... 9 4.1. SOLAR RADIATION FROM THE GLOBAL SOLAR ATLAS .............................................................................................10 4.2. IDENTIFICATION AND CHARACTERIZATION OF ROOFTOPS ........................................................................................11 4.3. SUITABLE ROOFTOP SURFACE AREAS FOR PV SYSTEM INSTALLATION .........................................................................20 4.4. TECHNICAL ROOFTOP SOLAR PV POTENTIAL ASSESSMENT FOR HCMC AND DA NANG .................................................22 4.5. EXTRAPOLATION TO THE ENTIRE COUNTRY ..........................................................................................................25 4.6. WEB-BASED PLATFORM FOR THE DISSEMINATION OF TECHNICAL ROOFTOP SOLAR PV POTENTIAL MAPS ..........................30 5. ASSESSMENT OF TECHNICAL ROOFTOP SOLAR PV POTENTIAL RESULTS ..................................................... 33 5.1. TECHNICAL ROOFTOP SOLAR PV POTENTIAL FOR HCMC .......................................................................................33 5.2. TECHNICAL ROOFTOP SOLAR PV POTENTIAL FOR DA NANG ....................................................................................41 5.3. TECHNICAL ROOFTOP SOLAR PV POTENTIAL FOR VIETNAM .....................................................................................48 6. IDENTIFICATION OF ROOFTOP BATCH FOR SOLAR PV DEVELOPMENT ........................................................ 51 6.1. DATA COLLECTION .........................................................................................................................................51 6.2. DATA PROCESSING .........................................................................................................................................51 6.3. SURVEY RESULTS............................................................................................................................................52 Ho Chi Minh City .................................................................................................................................................52 Da Nang .............................................................................................................................................................53 6.4. CHARACTERISTICS OF SURVEYED ROOFTOPS .........................................................................................................54 6.5. SOLAR PV POTENTIAL .....................................................................................................................................58 6.6. SURVEY CHALLENGES ......................................................................................................................................59 7. CONCLUSIONS AND FUTURE PERSPECTIVES ................................................................................................ 60 7.1. FUTURE PERSPECTIVES ....................................................................................................................................61 Annex 1 Web Platform Annex 2 CEFD Survey Report – HCMC and Da Nang (separate document) May 2018 iv Selection No. 1231720 Final Report LIST OF FIGURES Figure 1: Acquisition dates of WorldView-3 images used for Ho Chi Minh City (total surface area of 369 km2)................................................................................................................................. 5 Figure 2: Administrative boundaries (districts & sub-districts) of HCMC area of interest ............. 6 Figure 3: Acquisitions of WorldView-3 images used for Da Nang (total surface area of 175 km2)6 Figure 4: Administrative boundaries (districts & sub-districts) of Da Nang area of interest ......... 7 Figure 5: Ground extent of the Landsat-8 images used for the Vietnam mosaic ......................... 8 Figure 6: Methodological approach for rooftop solar PV technical potential assessment ............ 9 Figure 7: Global Solar Atlas and GTI data extraction for HCMC ................................................10 Figure 8: Global Solar Atlas and GTI data extraction for Da Nang ............................................11 Figure 9: Digital surface model (DSM) and digital terrain model (DTM) for HCMC ....................12 Figure 10: Digital surface model (DSM) and digital terrain model (DTM) for Da Nang ...............12 Figure 11: DHM correction with roads and hydrography............................................................13 Figure 12: Examples of road and hydrography vector errors that required updating for HCMC .13 Figure 13: Examples of road and hydrography vector errors that required updating for Da Nang .................................................................................................................................................14 Figure 14: Example of results of rooftop detection steps using an O-O classification approach 15 Figure 15: Rooftops detected in HCMC (each rooftop identified by a yellow dot) ......................16 Figure 16: Rooftops detected in Da Nang (each rooftop identified by a yellow dot) ...................16 Figure 17: Correspondence between rooftop shape and DHM used for slope estimation..........17 Figure 18: Rooftop type extracted by the deep learning algorithm .............................................18 Figure 19: Shifting of rooftop polygons to overlay with the DSM ................................................19 Figure 20: Evolution of building-related shading for certain times of day ...................................19 Figure 21: Examples of information elements extracted for each rooftop for HCMC ..................20 Figure 22: Examples of suitable (yellow) and non-suitable (red) rooftops ..................................22 Figure 23: Examples of total yearly PV energy produced by HCMC rooftops in MWh ...............23 Figure 24: Hourly exoatmospheric solar radiation power received at HCMC .............................24 Figure 25: Daily and yearly total of exoatmospheric solar irradiation received at HCMC ...........25 Figure 26: Landsat-8 OLI mosaic of Vietnam (bands 5-6-4) ......................................................25 Figure 27: Landsat-8 OLI mosaic (bands 5-6-4) - part of HCMC ...............................................27 Figure 28: Building density map - part of HCMC .......................................................................27 Figure 29: Landsat-8 OLI mosaic (bands 5-6-4) - part of Da Nang ............................................28 Figure 30: Building density map - part of Da Nang ....................................................................28 Figure 31: Rooftop suitable area calculated using WorldView-3 imagery and estimated from Landsat-8 imagery ....................................................................................................................29 Figure 32: Screenshot of the Web application after a selection is made by the user ................31 Figure 33: Focusing on the commune level for Ho Chi Minh City layer .....................................32 Figure 34: Ratio of built-up area over total sub-district area ......................................................35 Figure 35: Number of suitable rooftops per sub-district .............................................................36 Figure 36: Ratio of number of suitable rooftops to total number of rooftops ...............................36 Figure 37: Ratio of suitable rooftop surface area to total rooftop surface area ...........................37 Figure 38: Rooftop solar PV capacity ........................................................................................37 Figure 39: Rooftop solar PV technical potential .........................................................................38 Figure 40: Distribution of rooftop surface areas for HCMC ........................................................39 May 2018 v Selection No. 1231720 Final Report Figure 41: Distribution of suitable rooftops surface areas for HCMC .........................................39 Figure 42: Distribution of yearly totals of global horizontal irradiation for HCMC rooftops ..........40 Figure 43: Distribution of yearly totals of global irradiation received by tilted-surface (according to rooftop geometry) for HCMC rooftops ...................................................................................40 Figure 44: Distribution of yearly totals of global irradiation received by optimal surface (latitude- inclined and south-oriented) for HCMC rooftops........................................................................40 Figure 45: Ratio of built-up area over total sub-district area – Da Nang ....................................43 Figure 46: Number of suitable rooftops per sub-district – Da Nang ...........................................44 Figure 47: Ratio of number of suitable rooftops to total number of subdistrict rooftops – Da Nang .................................................................................................................................................44 Figure 48: Ratio of suitable rooftop surface area to total rooftop surface area per subdistrict – Da Nang....................................................................................................................................45 Figure 49: Rooftop solar PV capacity in MW – Da Nang ...........................................................45 Figure 50: Rooftop solar PV technical potential – Da Nang .......................................................46 Figure 51: Distribution of rooftop surface areas for Da Nang .....................................................47 Figure 52: Distribution of suitable rooftops surface areas for Da Nang ......................................47 Figure 53: Distribution of yearly totals of global horizontal irradiation for Da Nang rooftops ......47 Figure 54: Distribution of yearly totals of global irradiation received by tilted-surface (according to rooftop geometry) for Da Nang rooftops ................................................................................48 Figure 55: Distribution of yearly totals of global irradiation received by optimal surface (latitude- inclined and south-oriented) for Da Nang rooftops ....................................................................48 Figure 56: Distribution of suitable surface areas and yearly totals of PV potential (on horizontal surfaces) for rooftop areas of Vietnam ......................................................................................50 Figure 57: Rooftop survey activities ..........................................................................................51 Figure 58: Image obtained from Flycam and corresponding 3D rendering ................................52 Figure 59: Distribution of 65 surveyed buildings per category - HCMC ......................................55 Figure 60: Distribution of 108 surveyed buildings per category – Da Nang................................57 Figure 61: Web platform for Map technical rooftop solar PV potential in Vietnam......................63 Figure 62: Shift between rooftop footprint at ground level and top-of-building level ...................68 LIST OF TABLES Table 1: Parameters of the extrapolation relationship................................................................29 Table 2: Web site layer specifications .......................................................................................30 Table 3: Solar capacity according to scale of potential rooftop installation for commercial and industrial buildings in HCMC .....................................................................................................38 Table 4: Solar capacity according to scale of potential rooftop installation for commercial and industrial buildings in Da Nang ..................................................................................................46 Table 5: Rooftop survey/inspection success rate for HCMC ......................................................52 Table 6: Rooftop survey/inspection success rate for Da Nang ..................................................53 May 2018 vi Selection No. 1231720 Final Report LIST OF ACRONYMS AOI Area of interest CEFD Centre for Environmental Fluid Dynamics (University of Hanoi) DHM Digital height model DSM Digital surface model GHI Global horizontal irradiation GOPTA Global optimal irradiation (surface with slope = latitude and azimuth = south) GTI Global tilted irradiation (slope and azimuth correspond to rooftop geometry) HCMC Ho Chi Minh City HR High resolution Kt Clearness index, ratio between ground and exoatmospheric solar radiation MW Megawatt MWh Megawatt-hour PV Photovoltaic SVM Support vector machine TIN Triangulated irregular network YAS Yearly weighted average of building shading May 2018 vii Selection No. 1231720 Final Report 1. Executive summary To address the impact of climate change and continue the growth trajectory in a sustainable way, Vietnam has announced ambitious targets for the development of renewable energy electricity generation in the country. The revised National Power Development Plan VII (Revised PDP 7) for the period 2016-2020 with a vision to 2030, approved by Vietnam’s Prime Minister in 2016, specifies a target of 6.5 % of electricity generation from renewable energy sources (excluding large-scale hydropower) by 2020 and 10.7% by 2030. The Revised PDP 7 also stipulates a target of 12 GW by 2030. To achieve its ambitious deployment targets, it is key for the Government of Vietnam to promote solar and wind power through a clear and sustainable strategy while ensuring that their deployment will not impede economic development by imposing additional costs. Solar power is an increasingly attractive electricity generating option for the country thanks to recent cost reductions, quick construction rates, and the contribution it can make to ensuring energy security and environmental sustainability. While large and dense cities in countries like Vietnam have limited vacant ground for the development of solar power plants, they do have thousands of large buildings with rooftops (≥ 500 m2) that are suitable for the installation of industrial-scale solar PV systems in addition to hundreds of thousands of residential buildings that, put together, can receive a huge quantity of solar panels. This project was undertaken to develop and apply a cost-effective methodology based on satellite imagery to assess the rooftop solar photovoltaic (PV) technical potential for two cities in Vietnam: Ho Chi Minh City (HCMC) and Da Nang. To achieve this objective, satellite imagery with the highest spatial resolution that is commercially available (30 cm) was acquired over some 370 km2 in HCMC and 175 km2 in Da Nang. It was processed using conventional (photogrammetry, classification) and advanced (deep learning) image processing techniques to build a database containing information such as building footprint, building height, land use, as well as several rooftop characteristics including surface area suitable for solar PV system installation, shape, slope, orientation, and shadows. This information was combined with solar radiation data from the Global Solar Atlas to assess the corresponding total capacity (power in MW) and the total potential generation (energy in MWh). A web platform was designed to disseminate the project results. City authorities can use it to view the results at the commune or building levels for both HCMC and Da Nang as well as modulate technical parameters to recalculate PV potential for individual rooftops. Finally, it can also be used to view solar PV potential extrapolated at the commune level for the entire country. Nearly one million rooftops were characterized for HCMC and 600 thousand in Da Nang. The development of the rooftop database and web platform was complemented by a ground survey (interviews with building owners and various measurements) in both cities. The survey allowed collecting detailed information including name of building owner, electricity consumption and electricity-related specifications, owner willingness to have solar panels installed, rooftop material, etc. for more than 200 buildings that were pre-identified as the most suitable for solar PV installation. May 2018 1 Selection No. 1231720 Final Report Solar PV capacity results show that built-up areas in Vietnam offer a huge potential for producing PV electric energy from its rooftops. Vietnamese decision-makers can make use of the database and web platform to plan and develop the solar energy sector at their respective levels. For instance, the results show that HCMC could develop up to 6.4 GW solar PV capacity. Some 25 % of this corresponds to public and industrial buildings, the most suitable in terms of rooftop size and owner ability/willingness to install PV systems. In the case of Da Nang, the total solar PV capacity is 1.1 GW of which 41 % corresponds to public and industrial buildings. Assuming that PV systems were to be installed on only 5 % of all suitable rooftops, this could generate up to 900 GWh in HCMC and 160 GWh in Da Nang. This represents 6.6 % of HCMC’s needs (population of 8.6 million with per capita electric energy consumption of 1,565 kWh) and 6.9 % of Da Nang’s needs (population of 1.45 million with per capita energy consumption of 1,565 kWh) covered by solar PV sources. These proportions slightly exceed the governmental target of 6.5 % of renewable energy source by 2020. This project not only confirms, for both HCMC and Da Nang, the relevance of using large rooftops for solar PV system installation, but also quantifies the rooftop PV potential based on the characterization of nearly all rooftops detected in these two cities on satellite imagery. It allowed evaluating the total rooftop area available for PV system installation and, thus, calculating the total rooftop PV capacity. Knowing the surface area of each individual rooftop also allowed evaluating (1) the capacity of large surface area rooftops (mostly industrial/commercial and public) that can be targeted by national policies; and (2) the capacity of smaller rooftops (mostly residential) the owners of which could be encouraged to consider solar PV system installation. As an example, the analysis allowed determining that there are 25 commercial/industrial rooftops in HCMC that have peak capacity (yearly average) greater than 1MW and for which the total capacity reaches approximately 10 MW. In the case of Da Nang, 33 commercial/ industrial rooftops were found to have peak capacity (yearly average) greater than 1 MW and for which the total capacity reaches 50 MW. 2. Background Vietnam has witnessed impressive economic growth and poverty reduction in the past 25 years. The country’s GDP has grown from about US$34 billion in 2000 to $194 billion in 2015. The development of the energy sector has been a key factor in the recent industrialization process, creating jobs and increasing shared prosperity. Over the past decades, Vietnam has transformed from a primarily agricultural economy with a rural population to a mixed economy with substantial development of commercial and industrial activities and an urbanizing population that has increasing access to modern energy. Access to electricity services - which was below 10 percent in 1986 - has grown to 99 percent in 2012, contributing to reducing poverty and boosting shared prosperity. Expanded grid electrification of rural households was mirrored by a sustained increase in the value of gross domestic product per capita. Rapid urbanization, fast and sustained increase in energy consumption driven by the success of the electricity access program (leading to growing demand and productive uses in rural areas, which has contributed to poverty reduction), improvements in living standards and growing industrialization are at May 2018 2 Selection No. 1231720 Final Report the core of Vietnam's development challenges in achieving energy security and sustainable growth in the power sector. While the energy access agenda drove electricity demand expansion in the 1990s, the industrial sector has taken the lead in the past decade. In the decade from 2005-2014, industrial electricity consumption grew at an annual rate of 11.8 per cent, more than any other sector. Employment by the industry sector has also grown substantially by 7 percent per year. Partly due to the massive increase in electricity demand, greenhouse gas (GHG) emissions have more than doubled over the past decade, and industry, power and transport sectors are projected to account for the bulk of future increases in GHG emissions. Climate change represents a significant threat to economic and human development globally and Vietnam is particularly vulnerable due to its high exposure to extreme weather events. To address the impact of climate change and continue the growth trajectory in a sustainable way, Vietnam has announced ambitious targets for the development of renewable energy electricity generation in the country. The revised National Power Development Plan VII (Revised PDP 7) for the period 2016-2020 with a vision to 2030, approved by Vietnam’s Prime Minister in 2016, specifies a target of 6.5% of electricity generation from renewable energy sources (excluding large-scale hydropower) by 2020 and 10.7% by 2030. The Revised PDP 7 also stipulates a target of 850 MW of solar photovoltaic (PV) installed capacity by 2020, 4 GW by 2025 and 12 GW by 2030. To achieve its ambitious deployment targets, it is key for the Government of Vietnam to promote solar and wind power (together variable renewable energy (VRE)) through a clear and sustainable strategy while ensuring that their deployment will not impede economic development by imposing additional costs. Solar power is an increasingly attractive electricity generating option for the country thanks to recent cost reductions, quick construction rates, and the contribution it can make to ensuring energy security and environmental sustainability. To realize this opportunity and meet its ambitious target of 12 GW of solar power capacity installed by 2030, the Government of Vietnam needs to develop a clear deployment strategy that builds experience, lowers costs, and maximizes the economic benefits for the country. In this connection, the Government has requested the support of The World Bank (WB) for several solar PV activities, including the development of a Solar PV Strategy and the scaling-up of rooftop solar PV in two of Vietnam’s largest urban areas – Da Nang and Ho Chi Minh City. Under the rooftop solar PV engagement, the World Bank is assessing the technical rooftop solar PV potential, identifying city-level benefits and supporting both cities in developing a commercially viable rooftop solar PV program. One of its recommendations is to favor large rooftop PV installations, including portfolios consisting of multiple sites. While large and dense cities in countries like Vietnam have limited vacant ground for the development of solar power plants, they do have thousands of large buildings (over 500 m2) with rooftops that are suitable for the installation of industrial-scale solar PV systems as well as hundreds of thousands of residential buildings that, put together, can receive a huge quantity of solar panels. This project not only confirms, for HCMC and Da Nang, the relevance of using rooftops for solar PV system installation, but also quantifies the rooftop PV potential based on the characterization of nearly all rooftops detected in May 2018 3 Selection No. 1231720 Final Report these two cities on satellite imagery. The study allowed evaluating the total rooftop area available for PV system installation and, thus, calculating the total rooftop PV capacity. Knowing the surface area of each individual rooftop also allowed evaluating (1) the capacity of large surface area rooftops (mostly industrial and public) that can be targeted by national policies; and (2) the capacity of smaller rooftops (mostly residential) the owners of which could be encouraged to consider solar PV system installation. May 2018 4 Selection No. 1231720 Final Report 3. Area of interest and satellite imagery The area of interest (AOI) for Ho Chi Minh City includes the most densely populated area of the municipality. Archive imagery (acquired in 2015) and new acquisitions were required to cover the entire area (369 km2). Figure 1 shows the acquisition dates of the WorldView-3 images (30-cm resolution) that were used. Figure 2 shows the boundaries of administrative districts and sub-districts, the latter being used to produce aggregated maps and statistics of solar PV technical potential. The area of interest (AOI) for Da Nang also includes the most densely populated area of the municipality. As suitable archive imagery was unavailable, two new images were acquired to cover the entire area (175 km2). Figure 3 shows the acquisition dates of the WorldView-3 images (30-cm resolution) that were used, while Figure 4 shows the boundaries of administrative districts and sub-districts. Figure 1: Acquisition dates of WorldView-3 images used for Ho Chi Minh City (total surface area of 369 km2) May 2018 5 Selection No. 1231720 Final Report Figure 2: Administrative boundaries (districts & sub-districts) of HCMC area of interest Figure 3: Acquisitions of WorldView-3 images used for Da Nang (total surface area of 175 km2) May 2018 6 Selection No. 1231720 Final Report Figure 4: Administrative boundaries (districts & sub-districts) of Da Nang area of interest Medium-resolution Landsat-8 imagery (30 m) was also downloaded from USGS’ web site to cover the entire country as it was required for extrapolating the results from the high-resolution imagery. The most recent cloud-free Landsat-8 images (total of 37) were selected and downloaded to cover the country. The footprints (ground coverage) of these images are provided in Figure 5. They were acquired between 2013 and 2016. May 2018 7 Selection No. 1231720 Final Report Footprints of Landsat-8 images used for the mosaic Figure 5: Ground extent of the Landsat-8 images used for the Vietnam mosaic May 2018 8 Selection No. 1231720 Final Report 4. Application of the methodology The approach and methodology that were adopted for the three phases of the project are summarized in Figure 6. Figure 6: Methodological approach for rooftop solar PV technical potential assessment Phase 1 comprises (1) the estimation of solar radiation using the Global Solar Atlas; (2) the processing of the high-resolution satellite imagery resulting in the identification and characterization of rooftops; (3) the determination of rooftop suitability; and (4) calculation of solar PV potential for each suitable rooftop. Phase 2 involves carrying out a questionnaire survey and rooftop inspection of a set of buildings in each of the two cities in an effort to validate the results obtained during Phase 1. May 2018 9 Selection No. 1231720 Final Report Finally, Phase 3 includes (1) the extrapolation of the rooftop solar PV potential obtained for HCMC and Da Nang to the entire country, (2) the design and implementation of a Web-based platform to disseminate the results and (3) presentation of the results to HCMC and Da Nang stakeholders and related training. 4.1. Solar radiation from the Global Solar Atlas The Global Solar Atlas (http://globalsolaratlas.info) was used as the source of solar irradiation data at ground level (Figures 7 and 8). Three parameters were extracted from the Atlas: (1) global solar irradiation received on a horizontal surface (GHI); (2) global solar irradiation on a surface with an inclination angle equal to the latitude and oriented to the south (optimal surface, GOPTA); and (3) solar irradiation on a tilted surface according to rooftop slope and azimuth (GTI). For GTI, look-up tables were produced by varying the “azimuth” and “inclination” parameters via an automatic programmed tool. This allowed calculating the solar irradiation received by each rooftop according to its slope and aspect, in addition to the solar irradiation received by equivalent horizontal and optimal surfaces. Figure 7: Global Solar Atlas and GTI data extraction for HCMC May 2018 10 Selection No. 1231720 Final Report Figure 8: Global Solar Atlas and GTI data extraction for Da Nang 4.2. Identification and characterization of rooftops a. High-resolution imagery preprocessing The preprocessing of WorldView-3 images included the following operations: orthorectification, pansharpening, rescaling from 11-bits to 8-bits and enhancement. b. Rooftop height extraction Rooftop height was extracted from the digital height model (DHM), which corresponds to the difference between the digital surface model (DSM with 1-m resolution) that was extracted from the WorldView-3 stereoscopic pairs and the digital terrain model (DTM with 10-m resolution), which represents ground level elevation (see Figures 9 and 10). The production of the DSM involved (1) importing the stereoscopic pairs into the image processing software; (2) aerotriangulation with tie points; (3) automatic correlation; and (4) TIN1 to raster conversion. 1 TIN : Triangulated irregular network May 2018 11 Selection No. 1231720 Final Report DSM at 1-m resolution (from WV-3 stereo DTM at 10-m resolution pairs) Figure 9: Digital surface model (DSM) and digital terrain model (DTM) for HCMC DSM at 1-m resolution (from WV-3 stereo pairs) DTM at 10-m resolution Figure 10: Digital surface model (DSM) and digital terrain model (DTM) for Da Nang When the DHM is calculated (DHM = DSM - DTM), the pixels corresponding to roads and hydrography are set to “0” (Figure 11). In some places, the ancillary data (roads and hydrography) needed to be updated using the WorldView-3 imagery (Figures 12 and 13) to take into account changes that happened (for example new roads) after the production of the ancillary data. May 2018 12 Selection No. 1231720 Final Report DHM before correction DHM after correction Figure 11: DHM correction with roads and hydrography Road vector errors (red) on top of WV-3 image Hydrography errors (red) on top of WV-3 image Figure 12: Examples of road and hydrography vector errors that required updating for HCMC May 2018 13 Selection No. 1231720 Final Report Road vector errors (red) on top of WV-3 image Hydrography errors (red) on top of WV-3 image Figure 13: Examples of road and hydrography vector errors that required updating for Da Nang c. Extraction of rooftop footprints The detection of rooftops is based on an object-oriented (O-O) classification approach comprising the following steps: ▪ Multi-level segmentation (Figure 14.1); ▪ Elimination of small and elongated objects; ▪ Detection of rooftops (classification and fusion) according to spectral signature, shape and DHM (Figures 14.2 and 14.3). Manual validation (quality check) is then required to make the following corrections (Figure 14.4): ▪ Aggregation or fusion of rooftops sub-parts (when rooftop color is not homogeneous); ▪ Elimination of false alarms (sides of buildings when acquisition angle is high; roads when DHM is not available due to clouds and their shadows, etc.); ▪ Addition of the undetected rooftops (hidden by clouds or their shadows); ▪ Completion of partial rooftops (hidden by trees). May 2018 14 Selection No. 1231720 Final Report 1: Segmentation 2: Classification 3: Fusion 4: Manual quality check © DigitalGlobe 2017 Figure 14: Example of results of rooftop detection steps using an O-O classification approach This process led to the detection of about one million rooftops inside the Ho Chi Minh City AOI and about 600 k rooftops inside Da Nang AOI (Figures 15 and 16). May 2018 15 Selection No. 1231720 Final Report Figure 15: Rooftops detected in HCMC (each rooftop identified by a yellow dot) Figure 16: Rooftops detected in Da Nang (each rooftop identified by a yellow dot) May 2018 16 Selection No. 1231720 Final Report d. Characterization of rooftops Characterizing the rooftops is carried out to determine the rooftop area that is suitable for PV system installation and to calculate solar radiation received by this area. ▪ Surface area and other dimensions Standard GIS functions were used to extract the following information from the rooftop polygons: surface area, width (W) and height (Hmin and Hmax of DHM pixels) of each rooftop. ▪ Slope and aspect Rooftop aspect (orientation) is derived using a standard GIS function. The slope is calculated as atan((Hmax-Hmin)/(W/2)). Small variations in rooftop height indicate a flat rooftop. Figure 17 shows examples of the good correspondence between rooftop shape (on WV-3 image) and the DHM. WorldView-3 image DHM used for slope estimation Figure 17: Correspondence between rooftop shape and DHM used for slope estimation ▪ Determination of rooftop types using deep learning A deep learning approach was used to classify the rooftop types (flat vs. tilted, the number of sides and the level of obstruction). Visual assessment of the WV-3 images led to the identification of nine types of rooftops (Figure 18): • Flat without obstructions • Four-sided • Flat with 0 to 10 % obstructions • Complex • Flat with 10 to 30 % obstructions • Curved (semi-cylindrical) • Flat with ≥ 30% obstructions • Circular (conical) • Two-sided May 2018 17 Selection No. 1231720 Final Report For each type of rooftop, roughly one hundred samples were selected to “train” the deep learning algorithm. The process of the deep learning classification comprises: (1) extracting a sub-image corresponding to the rooftop; (2) computing deep features; and (3) applying an SVM2 classifier to the deep features. Flat without obstructions Flat with 0 - 10 % obstructions Flat with 10 - 30 % obstructions Flat with > 30 % obstructions Two-sided Four-sided Complex Curved Circular Figure 18: Rooftop type extracted by the deep learning algorithm 2 SVM: Support Vector Machine: supervised learning models with associated learning algorithms that analyze data used for classification and regression analysis May 2018 18 Selection No. 1231720 Final Report ▪ Shading Shading that comes from surrounding buildings, which varies according to the time of day, is calculated using a GIS function based on the DSM (not DHM) and the sun’s position. Thus, for each building, shading is calculated for each month and varying times of day between sunrise and sunset. To do this, the rooftop footprints were shifted to their actual position (their footprint at ground level) so as to enable correct overlay with the DSM (Figure 19). Rooftop footprint on the image DSM before shift DSM after shift Figure 19: Shifting of rooftop polygons to overlay with the DSM Figure 20 shows an example of building-generated shading simulated for the month of February at intervals of 1/6 of day duration (dd) from sunrise to sunset. Shading appears on the western side of the buildings in the morning hours and on the eastern side in the afternoon hours. They are largest for the taller buildings at the times close to sunset and sunrise. DSM used for shading generation Shading calculated from DSM at several times of day (solar time) – February 12h - 1/3 dd 12h - 1/6 dd 12h 12h +1/6 dd 12h + 1/3 dd Figure 20: Evolution of building-related shading for certain times of day May 2018 19 Selection No. 1231720 Final Report e. Integration of land use and sub-district identification Information on land use (LU) is available at a fine resolution3 and appears to correctly reflect the type of building in terms of the category: industrial, residential, public, etc. This information is taken into account when applying the area suitability criteria (minimum of 10 m2 for residential rooftops and 100 m2 for industrial/commercial/public rooftops). Similarly, by taking into account the district and sub-district boundaries, it is possible to obtain relevant statistics of the results (m2, MW, and MWh) for the desired administrative unit(s). Figure 21 shows a set of key information elements that were extracted at building scale. Figure 21: Examples of information elements extracted for each rooftop for HCMC 4.3. Suitable rooftop surface areas for PV system installation One of the most important suitability criteria concerns the surface area available for PV system installation. The total rooftop area (A), as measured from the rooftop footprint, needs to be reduced to take into account the perimeter that is necessary for maintenance work. This is simply done using a buffer, which was fixed at 1 m. The obstructions and average shading also need to be taken into account in the assessment of the suitable surface area. 3 A land use map of HCMC was provided by CEFD. May 2018 20 Selection No. 1231720 Final Report The rooftop buffer-corrected surface area is calculated for each of the different types of rooftops considered as indicated below. The quantities involved in these calculations are: A: area of the detected rooftop B: buffer (1 m) Acb: buffer-corrected area ROA: ratio of obstructed area L: rooftop length RSA: ratio of shaded area W: rooftop width Buffer-corrected surface areas for different types (shapes) of rooftops Correction for shading and obstructions The correction for obstructions and shading is the same for all buildings. Note that obstructions are significant only for flat rooftops. The rooftop surface area that is suitable for PV system installation (Asuit) is defined as follows Asuit = Acb . (1-ROA) . (1-RSA) May 2018 21 Selection No. 1231720 Final Report Examples of suitable and non-suitable rooftops Figure 22 shows two examples of industrial rooftops that are rejected because their suitable area is less than 100 m2, owing to the maintenance buffer and presence of obstructions. Numbers correspond to suitable surface area in m2 Figure 22: Examples of suitable (yellow) and non-suitable (red) rooftops 4.4. Technical rooftop solar PV potential assessment for HCMC and Da Nang Global solar irradiation (in kWh) received by a rooftop during a given year is simply the product of the global solar radiation extracted from the Global Solar Atlas (in kWh/m2 per year) by the suitable area of the rooftop (Asuit (m2)). The application of this rule is simple for global horizontal irradiation (GHI) and global irradiation received by latitude inclined surfaces (OPTA), regardless of rooftop type. For flat rooftops, global tilted irradiation is equal to global horizontal irradiation and therefore simple to calculate. However, the assessment of the global tilted irradiation received by tilted rooftops needs to take into account the rooftop type as well as slope and aspect, in addition to suitable surface area Asuit and the GTI value extracted from the Global Solar Atlas for the corresponding slope and azimuth. a. Solar radiation-related criteria for determining rooftop suitability for PV system A rooftop is considered unsuitable if the ratio of GTI/A (kWh/m2) is less than 50 % of the maximum reachable. In other words, a rooftop that produces less than 50 % of its maximum capacity due to all losses (buffer, shadows, slope and orientation and obstructions) is rejected. May 2018 22 Selection No. 1231720 Final Report b. PV energy assessment: total potential generation in MWh The PV energy (PVE in MWh) is simply the product of the solar radiation received by the suitable area of the rooftop by the PV panel efficiency (PVeff). PVEHoriz = PVeff * GHIrooftop PVELat = PVeff * GLatrooftop PVETilt = PVeff * GTIrooftop The calculations presented in this report (see results in section 4) reflect a PVeff value of 20 percent. In the Web platform that was developed during the project, PVeff is a variable parameter the value of which can be selected by the user inside a range between 0.16 and 0.26. Figure 23 shows the total yearly PV energy (PVETilt) potentially produced by a few suitable rooftops. Figure 23: Examples of total yearly PV energy produced by HCMC rooftops in MWh c. Method for PV power (total installed capacity MW) assessment Since solar radiation power varies substantially during the day, the “peak PV capacity” is set as the maximum power received in a day (at noon) on an optimal surface (slope = latitude; azimuth = south). Figure 24 shows that yearly average of exoatmospheric solar radiation power (Gopt0) calculated for HCMC is equal to 1,287 W/m2 at solar noon. For Da Nang, that yearly average of exoatmospheric solar radiation power is equal to 1,260 W/m2 at solar noon. These values need to be multiplied by the clearness index (Kt=GHI/GHI0) to retrieve the average power received at ground level in order to take into account average yearly cloud cover. May 2018 23 Selection No. 1231720 Final Report Figure 24: Hourly exoatmospheric solar radiation power received at HCMC The yearly average of the clearness index can be assessed by dividing the ground level solar irradiation provided by the Global Solar Atlas for HCMC (GHI = 1,775 kWh/m2 per year) by its exoatmospheric value calculated by summing daily solar irradiation; the latter is equal to 3,536 kWh/m2 per year (Figure 25). Hence: Kt = 1,775/3,536 = 0.5. Thus, at ground level, the yearly average solar radiation power received at solar noon is equal to 0.5 x 1,287 W/m2 = 643 W/m2. The peak PV power output (yearly average capacity at solar noon on optimal surface) for HCMC is equal to: Peak capacity = PVeff x 643 (W/m2) x Asuit (m2) Using the same reasoning, the peak PV power output (yearly average capacity at solar noon on optimal surface) for Da Nang is equal to: Peak capacity = PVeff x 630 (W/m2) x Asuit (m2) This means that, for example, a rooftop of 15 m2 (10 PV panels of 1.5 m2) in HCMC will have a yearly average capacity of 0.2 x 643 x 15 ≈ 1.9 kW. It should be noted that in case of perfectly clear sky conditions (Kt ≈ 1), this value will be nearly double that amount (≈ 3.8 kW). May 2018 24 Selection No. 1231720 Final Report Figure 25: Daily and yearly total of exoatmospheric solar irradiation received at HCMC 4.5. Extrapolation to the entire country The first step of the extrapolation process consists in calculating ground reflectance of the Landsat-8 imagery. This allows radiometric homogenizing and mosaicking the images to classify them in a single block (Figure 26). Figure 26: Landsat-8 OLI mosaic of Vietnam (bands 5-6-4) May 2018 25 Selection No. 1231720 Final Report The next step involved mapping building density using an object-oriented classification approach taking into account both spectral and morphological characteristics. The scale used for building density evaluation was that of city blocks delineated by roads and hydrography. Large city blocks (> 0.1 km2) were re-segmented because their density was in many cases not uniform. Building density (Figures 27 & 28 for HCMC and Figures 29 & 30 for Da Nang) was classified into the following categories: • High density of large buildings; • High density of small buildings; • medium density of large buildings; • medium density of small buildings; • Low building density; • Absence of buildings. A visual revision of the building density classification results was carried out to eliminate errors and harmonize the map of the entire country. May 2018 26 Selection No. 1231720 Final Report Figure 27: Landsat-8 OLI mosaic (bands 5-6-4) - part of HCMC Figure 28: Building density map - part of HCMC May 2018 27 Selection No. 1231720 Final Report Figure 29: Landsat-8 OLI mosaic (bands 5-6-4) - part of Da Nang Figure 30: Building density map - part of Da Nang May 2018 28 Selection No. 1231720 Final Report Establishing the relationship between building density and rooftop photovoltaic potential required estimating the built-up surface area that corresponds to each density class. The “useful” (suitable) built-up area of each class derived from the Landsat-8 imagery was then calculated by multiplying its total surface area by the “built-up ratio” and by “suitability ratio” that corresponds to the respective classes. Suitable surface area = Polygon surface area X Built-up ratio X Suitability ratio The “built-up” and “suitability” ratios calculated for each built-up density class were derived from the Landsat-8 images for HCMC and Da Nang (see Table 1 and Figure 31). These coefficients were used to extrapolate the PV potential to the entire country. In Figure 31, each point represents an object (city block or sub-block) extracted from Landsat-8. Table 1: Parameters of the extrapolation relationship Built-up density class Built-up ratio Suitability ratio No buildings 0 0 Low built-up density 0.20 0.50 Medium density of large buildings 0.27 0.72 Medium density of small buildings 0.35 0.59 High density of large buildings 0.49 0.78 High density of small buildings 0.57 0.52 Figure 31: Rooftop suitable area calculated using WorldView-3 imagery and estimated from Landsat-8 imagery May 2018 29 Selection No. 1231720 Final Report The results related to suitable surface area (Asuit) obtained for classified objects (city blocks or homogeneous parts of a block) were integrated at the commune level, which is the spatial scale used in the platform. Solar PV potential parameters were calculated using the estimated Asuit. Global irradiation (GHI, OPTA and GTI for variable tilts and azimuths) were obtained from the Global Solar Atlas using an automatic application that enters the centroid commune coordinates (and variable tilts and azimuths for GTI) and reads GHI, OPTA (GLatitude) and GTI. 4.6. Web-based platform for the dissemination of technical rooftop solar PV potential maps The Web platform can be used to see the results of the project maps (and associated values) for HCMC and Da Nang at the commune or building levels. The results can be viewed at the commune level for the entire country. The following paragraphs explain the technology behind this platform. Server Side The Web platform that was developed to present the project results is accessible at http://rooftoppvpotential.effigis.com/. It was designed and built exclusively with open source software. The data Five different layers are available on the platform. Table 2 presents the different layers and their specifications. Table 2: Web site layer specifications Format Feature Number of Coordinate Layer level Data type used in type features System Web site Country Polygons Communes 10,805 WMS EPSG4326:WGS84 Ho Chi Minh City Polygons Communes 248 GeoJSON EPSG4326:WGS84 Ho Chi Minh City Points Rooftops 330,090 WMS EPSG4326:WGS84 Da Nang Polygons Communes 49 GeoJSON EPSG4326:WGS84 Da Nang Points Rooftops 148,844 WMS EPSG4326:WGS84 Because of the very large number of rooftops in each of the two cities, the rooftop layer is only viewable at a certain zoom level to ensure the fluidity of the navigation on the map. Showing individual rooftops at a lower zoom level is not useful since there are too many and they would seem to overlap each other. The rooftop layer for each city and the commune layer at the country level are loaded as WMS (web map service). May 2018 30 Selection No. 1231720 Final Report Client side On the client side, different tools and libraries are used to create a user-friendly cartographic application shown in Figure 32. The available layers depend on the area of interest selected in the left sidebar menu. Figure 32: Screenshot of the Web application after a selection is made by the user Information on how to use the platform is available in "The Platform" menu (http://rooftoppvpotential.effigis.com/platform.html#howtouseit). Figure 33 illustrates the dropdown menu for the selection of the area of interest. If a user selects Ho Chi Minh City in the menu, the platform zooms on the city to show the commune layer. The appropriate legend is created and added to the map. May 2018 31 Selection No. 1231720 Final Report Figure 33: Focusing on the commune level for Ho Chi Minh City layer May 2018 32 Selection No. 1231720 Final Report 5. Assessment of technical rooftop solar PV potential results This section presents some statistics on parameters representing technical rooftop solar PV potential at the commune level (for HCMC, Da Nang and the entire country) and also at the building level for the two cities. The results of the project are provided in the web platform http://rooftoppvpotential.effigis.com/ described in Section 4.4. The following sections provide some statistics of technical rooftop PV potential. 5.1. Technical rooftop solar PV potential for HCMC The results comprise statistical data and maps on technical rooftop solar PV potential at the city level, district level and according to building type and use. At the city level (entire HCMC area of interest): • Number of rooftops Total number of detected rooftops 955,411 Total number of suitable rooftops 316,535 Suitable rooftops as percentage of total rooftops 33 % Number of suitable flat rooftops without obstructions 112,723 Number of suitable flat rooftops with 0-10% obstructions 61,123 Number of suitable flat rooftops with 10-30% obstructions 27,688 Number of suitable flat rooftops with >30% obstructions 1,877 Number of suitable two-sided rooftops 94,006 Number of suitable four-sided rooftops 3,194 Number of other suitable (complex, curved, circular) rooftops 15,924 • Surface areas Total surface area of the AOI 366,000,000 m2 (366 km2) Total surface area of the detected rooftops 116,286 941 m2 Total surface area of suitable rooftops 50,391 567 m2 Surface area of suitable rooftops as percentage of total rooftops 43 % Surface area of suitable flat rooftops without obstructions 7,466,801 m2 Surface area of suitable flat rooftops with 0-10% obstructions 4,788,418 m2 Surface area of suitable flat rooftops with 10-30% obstructions 2,863,791 m2 Surface area of suitable flat rooftops with >30% obstructions 323,769 m2 Surface area of suitable two-sided rooftops 27,581,573 m2 Surface area of suitable four-sided rooftops 900,799 m2 Surface area of suitable other (complex, curved, circular) rooftops 6,466,415 m2 May 2018 33 Selection No. 1231720 Final Report • PV power (yearly average capacity at solar noon on horizontal surfaces) Total estimated roof PV capacity 6,379 MW Estimated PV capacity on flat rooftops without obstructions 945 MW Estimated PV capacity on flat rooftops with 0-10% obstructions 606 MW Estimated PV capacity on flat rooftops with 10-30% obstructions 362 MW Estimated PV capacity on flat rooftops with >30% obstructions 40 MW Estimated PV capacity on two-sided rooftops 3,491 MW Estimated PV capacity on four-sided rooftops 114 MW Estimated PV capacity on other (complex, curved, circular) rooftops 818 MW • PV energy (potential) Total potential annual electricity on horizontal surface 17,889,006 MWh Total potential annual electricity on latitude-tilted (optimal) surface 18,201,434 MWh Total potential annual electricity on tilted (rooftop slope) surface 17,231,337 MWh Total potential annual electricity on horizontal surface per rooftop type Flat without obstructions 2,650,714 MWh Flat with 0-10% obstructions 1,699,888 MWh Flat with 10-30% obstructions 1,016,645 MWh Flat with >30% obstructions 114,938 MWh Two-sided rooftops 9,791,458 MWh Four-sided rooftops 319,783 MWh Other (complex, curved, circular) rooftops 2,295,577 MWh ▪ Total potential annual electricity on horizontal surface per building use ▪ Public 'Government Organization office land' (1,166 roofs) 131,145 MWh 'Education - Training' (2,704 roofs) 269,390 MWh 'Land for Security services' (3,676 roofs) 619,199 MWh 'Health Services land' (571 roofs) 71,248 MWh 'Market' (334 roofs) 65,423 MWh 'Sports Facilities land' (166 roofs) 31,296 MWh 'National Defence land' (126 roofs) 13,542 MWh May 2018 34 Selection No. 1231720 Final Report ▪ Cultural/religious 'Croyance use Land' (128 roofs) 9,874 MWh 'Cultural Services land' (457 roofs) 66,937 MWh 'Religious use Land' (1,083 roofs) 108,740 MWh 'Monuments and famous landscapes land' (19 roofs) 3,133 MWh ▪ Industrial: 'Industrial Parkland' (1,031 roofs) 417,826 MWh 'Production and Services land' (12360 roofs) 2,630,800 MWh 'Production of Ceramic and Porcelain construction material' (108 roofs) 18,682 MWh ▪ Residential: 'Urban residential land' (240,350 roofs) 9,525,333 MWh 'Rural residential land' (36,655 roofs) 2,149,469 MWh At the sub-district level: The rooftop PV technical potential at the sub-district level is illustrated in the maps found on the following pages. Figure 34: Ratio of built-up area over total sub-district area May 2018 35 Selection No. 1231720 Final Report Figure 35: Number of suitable rooftops per sub-district Figure 36: Ratio of number of suitable rooftops to total number of rooftops May 2018 36 Selection No. 1231720 Final Report Figure 37: Ratio of suitable rooftop surface area to total rooftop surface area Figure 38: Rooftop solar PV capacity May 2018 37 Selection No. 1231720 Final Report Figure 39: Rooftop solar PV technical potential At the building level: Table 3 shows the number of suitable commercial and industrial rooftops along with their total solar capacity for different categories of rooftops clustered by their size/individual installation capacity. This information is useful for investors wishing to target large industrial buildings and maximize economic and environmental benefits. Table 3: Solar capacity according to scale of potential rooftop installation for commercial and industrial buildings in HCMC Commercial and Industrial Buildings in Ho Chi Minh City Scale of each potential Number of suitable rooftops Total potential solar rooftop rooftop installation identified capacity (MW peak) < 200 kW peak 12,722 689.8 MW 200 kW to 500 kW peak 943 278.4 MW 500 kW to 1 MW peak 143 95.4 MW 1 MW to 2 MW peak 23 28.1 MW May 2018 38 Selection No. 1231720 Final Report Commercial and Industrial Buildings in Ho Chi Minh City 2 MW to 3 MW peak 1 2.1 MW > 3 MW peak 1 5.7 MW Total 13,833 1,099.5 MW The frequency distribution of rooftop surface areas extracted from the WorldView-3 images and corresponding suitable surface areas and solar irradiation for HCMC rooftops are provided in Figures 40 to 44. Detailed information for each individual rooftop is easily accessible on the Web platform. Figure 40: Distribution of rooftop surface areas for HCMC Figure 41: Distribution of suitable rooftops surface areas for HCMC May 2018 39 Selection No. 1231720 Final Report Figure 42: Distribution of yearly totals of global horizontal irradiation for HCMC rooftops Figure 43: Distribution of yearly totals of global irradiation received by tilted-surface (according to rooftop geometry) for HCMC rooftops Figure 44: Distribution of yearly totals of global irradiation received by optimal surface (latitude-inclined and south-oriented) for HCMC rooftops May 2018 40 Selection No. 1231720 Final Report 5.2. Technical rooftop solar PV potential for Da Nang At the city level (entire Da Nang area of interest): • Number of rooftops Total number of detected rooftops 646,958 Total number of suitable rooftops 148,882 Suitable rooftops as percentage of total rooftops 23 % Number of suitable flat rooftops without obstructions 107,936 Number of suitable flat rooftops with 0-10% obstructions 18,798 Number of suitable flat rooftops with 10-30% obstructions 10,085 Number of suitable flat rooftops with >30% obstructions 83 Number of suitable two-sided rooftops 10,860 Number of suitable four-sided rooftops 522 Number of other suitable (complex, curved, circular) rooftops 598 • Surface areas Total surface area of the AOI 174,690,000 m2 (175 km2) Total surface area of the detected rooftops 28,613,785 m2 Total surface area of suitable rooftops 9,145,406 m2 Surface area of suitable rooftops as percentage of total rooftops 32 % Surface area of suitable flat rooftops without obstructions 3,967,929 m2 Surface area of suitable flat rooftops with 0-10% obstructions 818,634 m2 Surface area of suitable flat rooftops with 10-30% obstructions 376,657 m2 Surface area of suitable flat rooftops with >30% obstructions 7,733 m2 Surface area of suitable two-sided rooftops 3,681,278 m2 Surface area of suitable four-sided rooftops 88,515 m2 Surface area of suitable other (complex, curved, circular) rooftops 204,662 m2 • PV power (yearly average capacity at solar noon on horizontal surfaces) Total estimated roof PV capacity 1,140 MW Estimated PV capacity on flat rooftops without obstructions 494 MW Estimated PV capacity on flat rooftops with 0-10% obstructions 102 MW Estimated PV capacity on flat rooftops with 10-30% obstructions 47 MW Estimated PV capacity on flat rooftops with >30% obstructions 0.96 MW Estimated PV capacity on two-sided rooftops 458 MW May 2018 41 Selection No. 1231720 Final Report Estimated PV capacity on four-sided rooftops 11 MW Estimated PV capacity on other (complex, curved, circular) rooftops 25 MW • PV energy (potential / year) Total potential annual electricity on horizontal surface 3,189,917 MWh Total potential annual electricity on latitude-tilted (optimal) surface 3,231,986 MWh Total potential annual electricity on tilted (rooftop slope) surface 3,088,135 MWh Total potential annual electricity on horizontal surface per rooftop type Flat without obstructions 1,384,013 MWh Flat with 0-10% obstructions 285,539 MWh Flat with 10-30% obstructions 131,377 MWh Flat with >30% obstructions 2,697 MWh Two-sided rooftops 1,284,029 MWh Four-sided rooftops 30,874 MWh Other (complex, curved, circular) rooftops 71,386 MWh ▪ Total potential annual electricity on horizontal surface per building use ▪ Public Government, Organization' (489 roofs) 17,688 MWh 'Education - Training' (1,997 roofs) 74,475 MWh 'Security' (203 roofs) 5,879 MWh 'Medical facilities' (526 roofs) 19,571 MWh 'Market' (177 roofs) 5,396 MWh 'Sport facilities' (541 roofs) 21,229 MWh 'National Defence' (5083 roofs) 230,595 MWh ▪ Cultural/religious 'Beliefs' (1 roof) 4 MWh 'Cultural facilities' (380 roofs) 6,960 MWh 'Religious' (311 roofs) 8,466 MWh 'Relics, famous landscapes' (29 roofs) 670 MWh ▪ Industrial: 'Industries' (5,970 roofs) 667,355 MWh 'Production, business' (6,699 roofs) 304,975 MWh 'Production of Ceramic and Porcelain construction material' (51 roofs) 1,039 MWh ▪ Residential: May 2018 42 Selection No. 1231720 Final Report 'Urban' (110,609 roofs) 1,554,928 MWh 'Rural' (6,074 roofs) 82,018 MWh Traffic (2028 roofs) 28,231 MWh At the sub-district level: The rooftop PV technical potential at the sub-district level is illustrated in the maps found in the following pages. For consistency purposes, the map legends are identical to those used for HCMC. In some cases, some classes may not appear in the map. Figure 45: Ratio of built-up area over total sub-district area – Da Nang May 2018 43 Selection No. 1231720 Final Report Figure 46: Number of suitable rooftops per sub-district – Da Nang Figure 47: Ratio of number of suitable rooftops to total number of subdistrict rooftops – Da Nang May 2018 44 Selection No. 1231720 Final Report Figure 48: Ratio of suitable rooftop surface area to total rooftop surface area per subdistrict – Da Nang Figure 49: Rooftop solar PV capacity in MW – Da Nang May 2018 45 Selection No. 1231720 Final Report Figure 50: Rooftop solar PV technical potential – Da Nang At the building level: Table 4 shows the number of suitable commercial and industrial rooftops along with their total solar capacity for different categories of rooftops clustered by their size/individual installation capacity. As was the case for Ho Chi Minh City, this information is useful for investors wishing to target large industrial buildings and maximize economic and environmental benefits. Table 4: Solar capacity according to scale of potential rooftop installation for commercial and industrial buildings in Da Nang Commercial and Industrial Buildings in Da Nang Scale of each potential Number of suitable Total potential solar rooftop installation rooftops identified rooftop capacity (MW peak) < 200 kW peak 11,483 183.7 MW 200 kW to 500 kW peak 258 76.8 MW 500 kW to 1 MW peak 55 36.1 MW 1 MW to 2 MW peak 29 39.1 MW 2 MW to 3 MW peak 3 7.2 MW < 3 MW peak 1 3.9 MW Total 11,829 346.8 MW May 2018 46 Selection No. 1231720 Final Report The frequency distribution of rooftop surface areas extracted from the WorldView-3 images and corresponding suitable surface areas and solar irradiation for Da Nang rooftops are provided in Figures 51 to 55. Detailed information for each individual rooftop is easily accessible on the Web platform. Figure 51: Distribution of rooftop surface areas for Da Nang Figure 52: Distribution of suitable rooftops surface areas for Da Nang Figure 53: Distribution of yearly totals of global horizontal irradiation for Da Nang rooftops May 2018 47 Selection No. 1231720 Final Report Figure 54: Distribution of yearly totals of global irradiation received by tilted-surface (according to rooftop geometry) for Da Nang rooftops Figure 55: Distribution of yearly totals of global irradiation received by optimal surface (latitude-inclined and south-oriented) for Da Nang rooftops 5.3. Technical rooftop solar PV potential for Vietnam • Number of objects (blocks or homogeneous sub-blocks) Number of communes in all Vietnam 10,805 Number of objects (city blocks or homogeneous sub-blocks) in all Vietnam 818,721 Number of communes with significant built-up area in all Vietnam 9,196 Number of objects (city blocks or homogeneous sub-blocks) classified 122,648 Number of objects in the class ‘Large rooftops – High built-up density’ 3803 Number of objects in the class ‘Large rooftops – Medium built-up density’ 513 Number of objects in the class ‘Small rooftops – High built-up density’ 32,140 Number of objects in the class ‘Small rooftops – Medium built-up density’ 45,265 Number of objects in the class ‘Small rooftops – Low built-up density’ 40, 927 May 2018 48 Selection No. 1231720 Final Report • Surface areas Total surface area of the entire country (Vietnam) 330,972 km2 Total rooftop surface area estimated for the entire Vietnam 18,657 km2 Suitable rooftop surface area estimated for the entire Vietnam 3,156 km2 Suitable rooftop surface area for the class ‘Large rooftops – High built-up density’ 127 km2 Number of objects in the class ‘Large rooftops – Medium built-up density’ 10 km2 Suitable rooftop surface area for the class ‘Small rooftops – High built-up density’ 785 km2 Suitable rooftop surface area for the class ‘Small rooftops – Medium built-up density’ 1,300 km2 Suitable rooftop surface area for the class ‘Small rooftops – Low built-up density’ 932 km2 • PV potential on horizontal surfaces Total PV horizontal potential of the entire country (Vietnam) 1,036 106 MWh/year Suitable rooftop surface area for the class ‘Large rooftops – High built-up density’ 42 106 MWh/year Number of objects in the class ‘Large rooftops – Medium built-up density’ 3 106 MWh/year Suitable rooftop surface area for the class ‘Small rooftops – High built-up density’ 272 106 MWh/year Suitable rooftop surface area for the class ‘Small rooftops – Medium built-up density’ 423 106 MWh/year Suitable rooftop surface area for the class ‘Small rooftops – Low built-up density’ 296 106 MWh/year • PV power (yearly average capacity at solar noon on horizontal surfaces) Total estimated PV capacity of the entire country (Vietnam) 370 829 MW Estimated PV capacity for the class ‘Large rooftops – High built-up density’ 14 887 MW Estimated PV capacity for the class ‘Large rooftops – Medium built-up density’ 1 059 MW Estimated PV capacity for the class ‘Small rooftops – High built-up density’ 97 496 MW Estimated PV capacity for the class ‘Small rooftops – Medium built-up density’ 151 517 MW Estimated PV capacity for the class ‘Small rooftops – Low built-up density’ 105 871 MW Figure 56 shows, for all classified objects (a city block or a homogeneous sub-block), the distribution of suitable surface areas of PV potential for rooftop areas of Vietnam. May 2018 49 Selection No. 1231720 Final Report Figure 56: Distribution of suitable surface areas and yearly totals of PV potential (on horizontal surfaces) for rooftop areas of Vietnam May 2018 50 Selection No. 1231720 Final Report 6. Identification of rooftop batch for solar PV development The Vietnamese counterpart of the project, the Centre for Environmental Fluid Dynamics (CEFD), was assigned the task of carrying out the survey of rooftops, corresponding to phase 2 of the project, in both Ho Chi Minh City and Da Nang. The goals of the survey were to (1) validate the attributes derived from the satellite images for 150 pre-selected rooftops from which 100 would be identified for visual inspection; (2) complete the information relating to land use type of the buildings; and (3) collection the information relating to electricity consumption and electric-technical specifications of the buildings. In Ho Chi Minh City, this activity was carried out between 6 March and 1 April 2018. In Da Nang, it was carried out between 6 and 23 January 2018. 6.1. Data collection The field survey involved filling in the questionnaire that had previously been approved by the WB project team. Interviews were carried out with the buildings’ owner and/or manager to obtain the required information. In addition, a drone (Phantom-3 Professional), equipped with an ortho-imaging flycam was used to validate rooftop type, obstruction element/object, rooftop slope, and a laser device (BOSCH-250m GLM 250VF) to measure building height. The process is illustrated in Figure 57. Figure 57: Rooftop survey activities 6.2. Data processing Data collected in the field were processed to make them suitable for further analysis. This involved (1) compiling the information from the questionnaires in an Excel file for each rooftop along with height information obtained from the laser device; (2) processing May 2018 51 Selection No. 1231720 Final Report the building images collected by the Flycam device to derive 3D images using Agisoft PhotoScan software; and (3) deriving, from the 3D images, the rooftop characteristics of height, slope, obstruction area, and roof area. An example of an image obtained with the Flycam is provided in Figure 58. Figure 58: Image obtained from Flycam and corresponding 3D rendering 6.3. Survey results Ho Chi Minh City CEFD reviewed 500 rooftops in HCMC that had been selected by Effigis according to a set of technical criteria. Among these, 396 rooftops were identified for which addresses and the owners’ names could be obtained. Out of these 396 rooftops, CEFD managed to interview 118 owners/managers/responsible staff of the buildings who had answered the invitation. A little more than half of these (65) allowed CEFD to carry out visual inspection and further implementation steps of the project, as summarized in Table 5. Table 5: Rooftop survey/inspection success rate for HCMC Number of Rooftop category % rooftops Rooftops selected according to established technical criteria 500 100 % Selected rooftops for which addresses and owner names could be 396 79.2 % obtained Rooftops for which interview / visual inspection could not be carried out 382 76.4 % during the survey period Rooftops for which building owners were interviewed but did not wish 53 10.6 % to provide information or were not collaborative Rooftops for which building owners were willing to have solar panels 65 13.0 % installed if selected and have additional information May 2018 52 Selection No. 1231720 Final Report Additional data was collected using the flycam and laser device for 45 of these 65 buildings. For those buildings where the owners did not allow CEFD to make those measurements, as-built drawings were requested and obtained immediately for two buildings. In the case of the 18 other buildings, the owners promised to send the as-built drawings by e-mail. As indicated in Table 5, 382 buildings could not be surveyed. The reasons for this are listed below: • No response despite several contact attempts via email & telephone (181) • Vacant land, under city master plan, under construction, national defense zone, airport area, cultural and historical heritages, etc. (79) • Impossible to reach personnel with building knowledge (47) • Owners/managers refused to provide any information (41) • Address unknown (entrance not found) &/or inaccessible (barriers or high fences) (25) • Owners requiring higher-level authorization (5) • Ownership to change soon (new owners unknown) (4) Da Nang The survey and inspection activities in Da Nang resulted in 286 rooftops being surveyed, 108 of which could be and were visually inspected (Table 6). Table 6: Rooftop survey/inspection success rate for Da Nang Number of Rooftop category % rooftops Rooftops selected according to established technical criteria 500 100 % Selected rooftops for which addresses and owner names could be 286 57.2 % obtained Rooftops for which building owners were interviewed but did not wish to 178 35.6 % provide information or were not collaborative Rooftops for which building owners were willing to have solar panels 108 21.6 % installed if selected and have additional information Additional data was collected using the flycam and laser device for 88 of these 108 buildings. The owners of ten other buildings provided adequate technical information to the survey team. In the case of buildings where the owners did not allow CEFD to make those measurements, as-built drawings were obtained for six of them. Finally, four flat rooftop buildings were accessible and their characteristics could be measured during visual inspection. As indicated in Table 6, 178 buildings could not be surveyed. The reasons for this are listed below: May 2018 53 Selection No. 1231720 Final Report • Owners/managers refused to provide any information or were not interested in the project (84) • Rooftops inaccessible because of security guards or absence of authorized staff (65) • Rooftops under building clearance plan of the City (9) • Owners were interested in solar PV installation but did not provide information (9) • Owner/manager unknown or abandoned, under construction (8) • Rooftop old, to be renovated or replaced (3) . 6.4. Characteristics of surveyed rooftops Ho Chi Minh City On the basis of the 65 surveyed buildings in Ho Chi Minh City, it can be said that the majority of building rooftops for which the owners were willing to have solar panels belong to the following overall categories: • 74 % belong to industrial land use • 75 % are made of sheet metal • 77 % have two-sided shape • 52 % have a surface area between < 5,000 m2 and up to 10,000 m2 • 47 % have no or less than 10 % obstructions The distribution of each of these characteristics is illustrated in Figure 59. May 2018 54 Selection No. 1231720 Final Report Figure 59: Distribution of 65 surveyed buildings per category - HCMC May 2018 55 Selection No. 1231720 Final Report Figure 59 (cont’d): Distribution of 65 surveyed buildings per category - HCMC Da Nang On the basis of the 108 surveyed buildings in Da Nang, it can be said that the majority of building rooftops for which the owners were willing to have solar panels belong to the following overall categories: • 92 % belong to industrial land use (factories and warehouses) • 82 % are made of sheet metal • 85 % have two-sided shape • 90 % have a surface area between < 5,000 m2 and up to 10,000 m2 • 96 % have no or less than 10 % obstructions Obstacles conventionally observed are ridge ventilators, roof vents, air conditioners, fans, dormers and elevators. The distribution of each of these characteristics is illustrated in Figure 60. May 2018 56 Selection No. 1231720 Final Report Figure 60: Distribution of 108 surveyed buildings per category – Da Nang May 2018 57 Selection No. 1231720 Final Report Figure 60 (cont’d): Distribution of 108 surveyed buildings per category – Da Nang 6.5. Solar PV potential Ho Chi Minh City Among the 65 rooftops that were surveyed and inspected in Ho Chi Minh City, six have a solar PV potential of less than 1,000 MWh/Yr, while 44 have a potential between 1,000 and 3,000 MWh/Yr and 15 have a potential greater than 3,000 MWh/Yr. Da Nang Among the 108 rooftops that were surveyed and inspected in Da Nang, 49 have a solar PV potential of less than 1,000 MWh/Yr, another 49 have a potential between 1,000 and 3,000 MWh/Yr and 10 have a potential greater than 3,000 MWh/Yr. May 2018 58 Selection No. 1231720 Final Report 6.6. Survey challenges The CEFD team in charge of the survey in both cities met with several challenges that should be considered, when possible, when planning for similar projects in the future. The main ones are summarized below. • Buildings belong to clearance zone in city’s master plan Force majeure (HCMC) • Construction too recent (does not appear on WV-3 images) Lack of collaboration from Information submitted was partial, leading to incomplete building owners/managers questionnaires In order to obtain all required information, many stake-holders Necessity to interview needed to be interviewed: owner, sub-lessor, accountant, several people for any given head office, company branch, technical staff, managers, etc., building often in different locations, resulting in a unreasonably lengthy process Organization and Especially with big companies and related bureaucracy administrative issues Building owners requiring Information required about expected benefits of collaborating additional information to the survey • Very hot weather (HCMC) • Bad weather (Da Nang) making it impossible to use the Flycam Other reasons • Heavy traffic and large distance between buildings to be surveyed (HCMC) • Flycam compass affected by radio waves • Flycam prohibited in certain areas Finally, the information related to energy consumption could not be collected through the interview process as the respondents often refused to, were unaware or provided wrong information. After discussing with EVN, it was found that only information on annual electricity consumption could be provided if the consumer number, electricity contractor’s name and the building address were provided by CEFD. Furthermore, since these data are protected by the terms of privacy included in the electricity provision contract between EVN and the consumers, providing such data would require the permission of the city’s People’s Committee. May 2018 59 Selection No. 1231720 Final Report 7. Conclusions and future perspectives This project developed an innovative and efficient methodology to assess technical rooftop solar PV potential for the entire country of Vietnam, and for two cities in greater detail, using satellite imagery and machine learning techniques. A terrain campaign allowed collecting complementary information related to additional criteria for PV implementation that cannot be extracted from the imagery (ownership, rooftop material, interest for PV implementation, etc.). A total of only 65 buildings were inspected in Ho Chi Minh City while 108 were inspected in Da Nang. A drone was used to obtain rooftop characteristics (presence of obstructions, rooftop material, etc.) when the rooftops were inaccessible and a laser device was used to get height estimation. The estimated rooftop PV potential values were extrapolated to the entire country using a simple linear model that relates, for HCMC and Da Nang, the built-up density derived from Landsat-8 imagery and the suitable rooftop surface area extracted from WorldView- 3 imagery. Finally, a web platform was designed and implemented to disseminate the results of the project, i.e., the PV potential (energy in MWh) and capacity (power in MW) of each suitable rooftop for HCMC and Da Nang and each commune in Vietnam. The results of the project show that built-up areas in Vietnam offer a huge potential for producing PV electric energy from its rooftops. The information produced during the project can be advantageously used by Vietnamese decision-makers to plan and develop the renewable solar energy sector. In particular, HCMC can develop up to 6.4 GW of which 7 % correspond to public buildings, 18 % to industrial buildings and 65 % to residential buildings. In the case of Da Nang, the total capacity is 1.1 GW of which 11 % correspond to public buildings, 30 % to industrial and 52 % residential. If we consider that average electricity consumption per capita reached 1,565 kWh (end of 2015 – according to 2016 Vietnam Electricity Annual Report4) and that the population of HCMC is about 8.6 M inhabitants, the total electrical energy consumption reaches some 13,500 GWh per year for HCMC. According to the results of sections 5.1 and 5.2 of this report, the total rooftop PV potential for HCMC is around 18,000 GWh, which largely exceeds the city’s needs. Obviously, it is unrealistic to hope reaching this potential. However, if only 5% of all suitable rooftops were to be used for PV systems, 6.6 % (i.e. 18,000 GWh x 5 % / 13,500 GWh) of the city’s needs could be covered by solar PV sources. This proportion is very close to the governmental target of 6.5 % of renewable energy source by 2020. This can be said to be realistic as, based on the questionnaire-based survey and inspection, at least 50 % (65/118) of owners in HCMC of identified suitable rooftops showed interest and/or would be willing to have solar PV systems installed on their rooftops. In the case of Da Nang with a population of 1.45 million inhabitants, the total electrical energy consumption reaches roughly 2,300 GWh per year. The total rooftop PV potential for Da Nang is around 3,200 GWh. As for HCMC, this largely covers the city’s needs. If we consider again that only 5 % of all suitable rooftops were to be used for PV systems, 4 http://www.evn.com.vn/userfile/files/2017/3/AnnualReport2016.pdf May 2018 60 Selection No. 1231720 Final Report 6.9 % (3,200 GWh x 5 % / 2,300 GWh) of the city’s needs could be covered by solar PV sources. Industrial and public buildings are the most appropriate target from a technical perspective as rooftop surfaces are large. Their proportion, in terms of PV capacity, is about 25 % (of 6.4 GW) for HCMC and 40 % (of 1.1 GW) for Da Nang. Many small, residential unit roofs may not be commercially viable to develop, but at an aggregate level they constitute a large share of the available roof space. Their share in is 65 % (of 6.4 GW) and 52 % (of 1.1 GW) in HCMC and Da Nang, respectively. For large industrial and public buildings, policies and regulations should be put in place to allow this technical potential to be commercially developed. As an example, the analysis allowed determining that there are 25 commercial/industrial rooftops in HCMC that have peak capacity (yearly average) greater than 1MW and for which the total capacity reaches approximately 10 MW. In the case of Da Nang, 33 commercial/industrial rooftops were found to have peak capacity (yearly average) greater than 1 MW and for which the total capacity reaches 50 MW. In the case of smaller residential rooftops, measures could be put forward by the Government to encourage building owners to convert their electrical systems to solar PV. 7.1. Future perspectives Improvements that could be considered to further enhance the outcome of the project are presented below: 1) Application of the project methodology to other cities and/or countries Now that the methodology has been developed and tested with success, the technical rooftop solar PV potential could be assessed for other cities in Vietnam and countries and the current web platform be reproduced for these new study areas with a significantly reduced cost and time of execution. In addition, it is much more economical than existing methods based, for example, on lidar data. 2) Improvements to the web platform Interactive crowdsourcing functionalities could be added to collect complementary information from users (building owner or other user) through the platform. o The information collected in situ on rooftop characteristics (address, ownership, rooftop type and material, slope, obstructions, photos, owner’s concerns, etc.) that can help the assessment of the actual suitability of the rooftop and the interest of the owner for PV system installation could be recovered through the platform. This information could be used by WB or city authorities to monitor the actual PV implementation in the study area. o Extending the platform to support mobile applications. May 2018 61 Selection No. 1231720 Final Report 3) Increase in the precision of the solar energy potential assessment o Improvement in the temporal scale by providing solar radiation data at hourly scale and on a monthly basis; o Addition of solar radiation received by vertical surfaces of each building, which is interesting for energy efficiency applications (ex.: heating and cooling sizing). May 2018 62 Selection No. 1231720 Final Report Annex 1 – Web platform Web-based platform for technical rooftop solar PV potential dissemination A screen shot of the Web platform developed for disseminating technical rooftop solar PV potential is shown in Figure 61. The following paragraphs are included in ‘meta- information’ menus of the platform (‘About’, ‘Data’ and ‘The platform’). They are useful to learn about the origin of the project, it objectives, data used, the methodology and how to use this web platform. Figure 61: Web platform for Map technical rooftop solar PV potential in Vietnam May 2018 63 Selection No. 1231720 Final Report About The World Bank and the International Finance Corporation, collectively The World Bank Group, have provided this Web site on Technical Rooftop Solar PV Potential in Vietnam, to support the scale-up of solar power in that country. This work is funded by the Energy Sector Management Assistance Program (ESMAP), a multi-donor trust fund administered by The World Bank and supported by 13 official bilateral donors. It is part of a global ESMAP initiative on Renewable Energy Resource Mapping that includes biomass, small hydro, solar and wind. The World Bank Group has selected Effigis Geo-Solutions to develop a methodology using high-resolution satellite imagery to determine solar PV potential and produce this Web site. Purpose of this Web platform This Web platform was designed with the purpose of disseminating the parameters that represent rooftop solar PV technical potential in Vietnam. The parameters are provided (1) at the Commune level (for the entire country) and (2) at the rooftop level (for two cities: HCMC and Da Nang) and comprise: o Total yearly PV energy potentially produced on a horizontal surface (in MWh/year) o Total yearly PV energy potentially produced per square meter on a horizontal surface (in MWh/m2/year) o Total yearly PV energy potentially produced on a tilted surface, inclined and oriented according to rooftop geometry (in MWh/year) o Total yearly PV energy potentially produced by an optimal (latitude tilted - south oriented) surface (in MWh/year) o Total rooftop area of each rooftop (in m2) and each commune (in km2) o Suitable rooftop area of each rooftop (in m2) and each commune (in km2) o Percent of suitable area of each rooftop and each commune o Average of PV power potentially produced at noon on a horizontal surface (in MW) How it was done The technical rooftop solar photovoltaic potential was determined using the following steps: 1. Very high-resolution satellite imagery (WorldView-3 with 30-cm resolution) stereoscopic pairs were used to detect and characterize building rooftops for two cities in Vietnam: Ho Chi Minh City and Da Nang. May 2018 64 Selection No. 1231720 Final Report 2. The suitable area for the installation of PV systems was determined for each rooftop in these two cities and the solar radiation received on (1) horizontal, (2) tilted (according to the rooftop slope and orientation) and (3) optimal (when slope = latitude and orientation = south) surfaces were calculated. Global solar radiation data (horizontal, tilted and optimal) were obtained from the global solar atlas (http://globalsolaratlas.info/). 3. The technical rooftop solar PV potential calculated using steps 1 and 2 was extrapolated from these results to the entire country. A statistical relationship was established between the suitable area that was derived from the WorldView-3 images for the two cities and the built-up area that was derived from Landsat-8 images and then used to calculate the suitable area for solar PV system implementation for the entire country. How to use this platform Select Area of Interest: Select between: the results can be explored either at the Commune level (single values HCMC: for entire Commune or at the Rooftop level (values for each suitable rooftop identified by red dot when zoom is sufficient) the results can be explored either at the Commune level (single values Da Nang: for entire Commune or at the Rooftop level (values for each suitable rooftop identified by red dot when zoom is sufficient) Vietnam: the Commune box must be checked to explore the results Getting information for each Commune at country level 1. Select Vietnam 2. Click on a commune to obtain its individual values for the following variables: PV Horizontal; PV Horizontal/m2; PV Latitude; PV Tilted; Commune total Area; Suitable Area; % Suitable Area; Capacity (electric power) at Noon. Getting information for each Commune at City level 1. Select either Ho Chi Minh City or Da Nang 2. Hover over the individual communes to obtain their respective total Rooftop PV- suitable area in km2 and total PV horizontal in MWh/year. The result will dynamically appear in the top left corner info-box in the map. 3. Click on a Commune to obtain its individual values for the following variables: PV Horizontal; PV Horizontal/m2; PV Latitude; PV Tilted; Commune Total Area; Suitable Area; % Suitable Area; Capacity (electric power) at Noon. May 2018 65 Selection No. 1231720 Final Report Getting information for individual rooftops: 1. Select either Ho Chi Minh City or Da Nang 2. Zoom in sufficiently on area of interest to see caption: “Click on a rooftop to get info” under Rooftops box 3. Click on Rooftops box 4. Click on red dot to obtain values for an individual rooftop for the following: PV Horizontal; PV Horizontal/m2; PV Latitude; PV Tilted; Rooftop area; Suitable Area; % Suitable area. 5. Click on PV Calculator to determine total energy in Wh/year. You are asked to select the efficiency of the solar panel (standard values between 0.16 and 0.26) and determine the PV type (Horizontal, Latitude or Tilted). Data Description WorldView-3 imagery (http://worldview3.digitalglobe.com/): WorldView-3 is a commercial Earth observation satellite owned by DigitalGlobe (acquired by Maxar in 2017). It was launched on 13 August 2014. WorldView-3 collects panchromatic imagery of 0.31 m resolution, eight-band multispectral imagery with 1.24 m resolution, shortwave infrared imagery at 3.7 m resolution, and CAVIS (Clouds, Aerosols, Vapors, Ice, and Snow) data at 30 m resolution. It has an average revisit time of < 1 day and is capable of collecting up to 680,000 km2 per day. This satellite was used because the images it collects currently have the highest spatial resolution (31 cm) that is commercially available. The following rooftop information, required to assess technical solar PV, was extracted from the WorldView-3 stereoscopic image pairs of the two cities (Ho Chi Minh City and Da Nang): outline or footprint; height; surface area, slope and orientation; type (flat, two- sided, four-sided, with or without obstructions, etc.); and shading (from surrounding buildings). Rooftop category (residential, public and industrial/commercial) was obtained from a local government-produced land use map for the individual cities. Landsat-8 imagery (https://landsat.usgs.gov/landsat-8): The Landsat 8 satellite, launched in February 2013, images the entire Earth every 16 days. It is the latest in a continuous series of land remote sensing satellites that began in 1972. Data collected by the instruments onboard the satellite are available to download at no charge from EarthExplorer, GloVis, or the LandsatLook Viewer within 24 hours of acquisition. Landsat-8 provides moderate-resolution imagery, from 15 meters to 100 meters, of Earth’s land surface and polar regions. Landsat 8 operates in the visible, near-infrared, short wave infrared and thermal infrared spectrums. It captures more than 700 scenes a day. May 2018 66 Selection No. 1231720 Final Report Global Solar Atlas Data: The Global Solar Atlas (http://globalsolaratlas.info), provided by the World Bank Group, was used as the source of solar irradiation data at ground level. Three parameters were extracted from the Atlas: global solar irradiation received on a horizontal surface (GHI); global solar irradiation on a surface with an inclination angle equal to the latitude and oriented to the south (optimal surface, GOPTA); and solar irradiation on a tilted surface according to rooftop slope and azimuth (GTI). This allowed calculating the solar irradiation received by each rooftop (or part of it) according to its slope and aspect, in addition to the solar irradiation received by equivalent horizontal and optimal surfaces. The base layer The basemap tiles are requested through the Google maps JavaScript API. These tiles are used in agreement with the Google Maps/Google Earth APIs Terms of Service (https://developers.google.com/maps/terms). IMPORTANT NOTE: The different base layers available in this application are used ONLY for visualization. The imagery one sees inside Google Earth is different from the images that were used for this project and might be older. This means that some buildings might not appear in the imagery provided by Google Earth since they were built after the image acquisition, but they do appear in the WorldView-3 imagery used in this project. That explains why some Rooftop markers appear at a location where it seems like there are no buildings. Similarly, buildings might also have been destroyed since the acquisition of the Google Earth imagery and no rooftop marker appears on them since they no longer exist in the more recent imagery. Spatial accuracy The rooftops were detected on orthorectified WorldView-3 imagery and their footprints brought back to ground level. Thus, the rooftop polygon and its centroid will appear shifted (as illustrated in Figure 62) relative to its position on the image (Google Maps or even WorldView-3). The shift is proportional to the building’s height and can be calculated by: Shift = Building Height X tan(Viewing Angle) May 2018 67 Selection No. 1231720 Final Report Example 1 WorldView-3 Google Maps Example 2 WorldView-3 Google Maps Figure 62: Shift between rooftop footprint at ground level and top-of-building level Definitions Yearly amount of energy, converted by a PV system into electricity, that is PV Horizontal expected to be generated at a given site (a rooftop or a Commune) on a horizontal surface (unit = MWh/year). Yearly amount of energy, converted by a PV system into electricity, that is PV expected to be generated at a given site on a horizontal surface, per square meter Horizontal/m2 (unit = kWh/m2/year). This value is useful for determining the size of the PV systems to be installed. Yearly amount of energy, converted by a PV system into electricity, that is PV Latitude expected to be generated at a given site (a rooftop or a Commune) on a south- oriented surface with a slope equal to latitude (unit = MWh/year). Yearly amount of energy, converted by a PV system into electricity, that is PV Tilted expected to be generated at a given site (a rooftop or a Commune) on a surface having the same slope and orientation as the rooftops (unit = MWh/year). May 2018 68 Selection No. 1231720 Final Report Total surface area of a rooftop, as calculated from the footprint that was extracted Rooftop Area from the WorldView-3 imagery. Helps eliminate rooftops with surface area that is under the minimum threshold. Subtracts from the Rooftop Area the surface area corresponding to shading, obstructions, and a 1-m maintenance buffer around the PV systems. At the Suitable Area Commune and country levels, it corresponds to the sum of all suitable rooftop areas (unit = km2). % Suitable area Corresponds to : ( Suitable Area / Rooftop Area ) x 100 Yearly average of PV power that is expected to be generated, at solar noon, at a Capacity Noon given site (a rooftop or a Commune) on a horizontal surface (unit = MW). The efficiency of a photovoltaic system is the ratio of conversion of the received Panel solar energy into electrical energy. Values of PV panel efficiency range between efficiency 0.16 and 0.26 (https://news.energysage.com/what-are-the-most-efficient-solar- panels-on-the-market/). A photovoltaic system, also PV system or solar power system, is a power system designed to supply usable solar power by means of photovoltaics. It consists of an arrangement of several components, including solar panels to PV system absorb and convert sunlight into electricity, a solar inverter to change the electric current from DC to AC, as well as mounting, cabling and other electrical accessories to set up a working system (https://en.wikipedia.org/wiki/Photovoltaic_system). Calculates total amount of energy (Wh) over an entire year on the basis of PV calculator selected roof characteristics (surface area, type of PV system (horizontal, latitude or tilted) to be installed. Determining the suitable area of a rooftop for PV system installation Several criteria need to be examined to determine the suitable area of a rooftop for solar PV system installation. These include: surface area, amount of shading, presence of obstructions, and orientation. Rooftops with suitable surface areas (after taking into account shading, obstructions and PV system perimeter buffer for maintenance work) less than 10 m2 for residential buildings or 100 m2 for industrial/commercial/public buildings are automatically eliminated. A rooftop is also rejected if its value for “PV Tilted” is less than 50 % of that of “PV Horizontal”. This means that the rooftop has a slope and/or an orientation that make it unsuitable for PV energy production. May 2018 69 Selection No. 1231720 Final Report Annex 2 - CEFD Survey Reports (combined in separate document) May 2018 70