ASSESSMENT OF BIOMASS AND CARBON OF MANGROVES IN WEST AFRICA

ASSESSMENT OF BIOMASS AND CARBON OF MANGROVES IN WEST AFRICA Final Report 08/15/2014 This report was made possible with support from the American people delivered through the U.S. Agency for International Development (USAID). The contents are the responsibility of the authors of this report and do not necessarily reflect the opinion of USAID or the U.S. Government.

ASSESSMENT OF BIOMASS AND CARBON OF MANGROVES IN WEST AFRICA Final Report Prepared by: Wenwu Tang, Wenpeng Feng, Meijuan Jia, Huifang Zuo Center for Applied Geographic Information Science Department of Geography and Earth Sciences University of North Carolina at Charlotte Carl C. Trettin, Ph.D. Center for Forested Wetlands Research U.S. Forest Service

LIST OF TABLES Table 1. Area of mangroves reported in the literature... 20 Table 2. Summary of original datasets used for biomass estimation in West Africa... 21 Table 3. Summary of area of mangroves in each country for USGS and NASA datasets... 22 Table 4. Statistics of canopy heights of mangroves for each country (data source: NASA)... 23 Table 5. Distribution of canopy heights in the study region... 24 Table 6. Above- and below-ground biomass and carbon for each canopy height class in the study region... 25 Table 7. Mean biomass and carbon density for each country in the study region... 26 Table 8. Mean biomass and carbon density of each canopy height class for each country in the study region... 27 Table 9. Total biomass and carbon of each canopy height class for each country in the study region... 28 Table 10. Comparison of biomass and carbon estimated from datasets at coarse and fine spatial resolutions... 29 Table 11. Summary of the shortest distance from mangroves to nearest city for each country in the study area... 30

LIST OF FIGURES Figure 1. Map of canopy heights in Africa... 31 Figure 2. Map of canopy height of mangroves in West Africa... 32 Figure 3. Histogram of canopy heights for the entire study area... 33 Figure 4. Histogram of canopy heights for the entire study area... 34 Figure 5. Map of canopy height classes in Guinea... 35 Figure 6. Histogram of canopy heights in Guinea... 36 Figure 7. Histogram of canopy height classes in Guinea... 36 Figure 8. Map of canopy height classes in Sierra Leone... 37 Figure 9. Histogram of canopy heights in Sierra Leone... 38 Figure 10. Histogram of canopy height classes in Sierra Leone... 38 Figure 11. Map of canopy height classes in Liberia... 39 Figure 12. Histogram of canopy heights in Liberia... 40 Figure 13. Histogram of canopy height classes in Liberia... 40 Figure 14. Map of canopy height classes in Cote d Ivoire... 41 Figure 15. Histogram of canopy heights in Cote d Ivoire... 42 Figure 16. Histogram of canopy height classes in Cote d Ivoire... 42 Figure 17. Map of canopy height classes in Ghana... 43 Figure 18. Histogram of canopy heights in Ghana... 44 Figure 19. Histogram of canopy height classes in Ghana... 44 Figure 20. Map of canopy height classes in Togo... 45 Figure 21. Histogram of canopy heights in Togo... 46 Figure 22. Histogram of canopy height classes in Togo... 46 Figure 23. Map of canopy height classes in Benin... 47 Figure 24. Histogram of canopy heights in Benin... 48 Figure 25. Histogram of canopy height classes in Benin... 48 Figure 26. Map of canopy height classes in Nigeria... 49 Figure 27. Histogram of canopy heights in Nigeria... 50 Figure 28. Histogram of canopy height classes in Nigeria... 50 Figure 29. Map of canopy height classes in Cameroon... 51 Figure 30. Histogram of canopy heights in Cameroon... 52 Figure 31. Histogram of canopy height classes in Cameroon... 52 Figure 32. Map of the mean biomass of mangroves for the study area in West Africa... 53 Figure 33. Map of the mean biomass of mangroves in Guinea... 54 Figure 34. Map of the mean biomass of mangroves in Sierra Leone... 55 Figure 35. Map of the mean biomass of mangroves in Liberia... 56 Figure 36. Map of the mean biomass of mangroves in Cote d'ivoire... 57 Figure 37. Map of the mean biomass of mangroves in Ghana... 58 Figure 38. Map of the mean biomass of mangroves in Togo... 59 Figure 39. Map of the mean biomass of mangroves in Benin... 60 Figure 40. Map of the mean biomass of mangroves in Nigeria... 61 Figure 41. Map of the mean biomass of mangroves in Cameroon... 62

Figure 42. Map of canopy height classes in Guinea (spatial resolution: 1 km 1 km)... 63 Figure 43. Map of canopy height classes in Sierra Leone (spatial resolution: 1 km 1 km)... 64 Figure 44. Map of canopy height classes in Liberia (spatial resolution: 1 km 1 km)... 65 Figure 45. Map of canopy height classes in Cote d Ivoire (spatial resolution: 1 km 1 km)... 66 Figure 46. Map of canopy height classes in Ghana (spatial resolution: 1 km 1 km)... 67 Figure 47. Map of canopy height classes in Togo (spatial resolution: 1 km 1 km)... 68 Figure 48. Map of canopy height classes in Benin (spatial resolution: 1 km 1 km)... 69 Figure 49. Map of canopy height classes in Nigeria (spatial resolution: 1 km 1 km)... 70 Figure 50. Map of canopy height classes in Cameroon (spatial resolution: 1 km 1 km)... 71 Figure 51. Map of distance from mangroves to nearest city... 72 Figure 52. Map of differences of mangrove area in Guinea for different data sources... 73 Figure 53. Map of differences of mangrove area in Sierra Leone for different data sources... 74 Figure 54. Map of differences of mangrove area in Liberia for different data sources... 75 Figure 55. Map of differences of mangrove area in Cote d Ivoire for different data sources... 76 Figure 56. Map of differences of mangrove area in Ghana for different data sources... 77 Figure 57. Map of differences of mangrove area in Togo for different data sources... 78 Figure 58. Map of differences of mangrove area in Benin for different data sources... 79 Figure 59. Map of differences of mangrove area in Nigeria for different data sources... 80 Figure 60. Map of differences of mangrove area in Cameroon for different data sources... 81

CONTENTS 1. EXECUTIVE SUMMARY...1 2. STUDY AREA...2 3. LITERATURE REVIEW...2 3.1 ESTIMATION OF MANGROVE COVERAGE... 2 3.2 ESTIMATION OF CANOPY HEIGHT OF MANGROVES... 3 3.3 ESTIMATION OF BIOMASS FROM MANGROVES... 4 4. DATA PROCESSING AND ANALYSIS...5 4.1 DATA... 5 4.2 ASSESSMENT OF MANGROVE AREA... 6 4.3 CLASSIFICATION OF CANOPY HEIGHTS... 6 4.4 ESTIMATION OF BIOMASS AND CARBON BASED ON CANOPY HEIGHT BY COUNTRY... 7 4.5 REGIONAL-LEVEL ESTIMATION BIOMASS AND CARBON USING DATA AVAILABLE ONLINE... 9 4.6 PROXIMITY OF MANGROVES TO HUMAN SETTLEMENTS... 10 5. DISCUSSION...11 5.1 COMPARISON OF MANGROVE AREA BETWEEN THOSE REPORTED IN THE LITERATURE AND OUR ESTIMATES... 11 5.2 MANGROVE AREA, CANOPY HEIGHTS, BIOMASS AND CARBON ESTIMATION IN THE ENTIRE STUDY AREA... 12 5.3 COMPARISON OF MANGROVE AREA, CANOPY HEIGHTS, AND BIOMASS ESTIMATION BY COUNTRY... 12 6. CONCLUSION...15 7. ACKNOWLEDGEMENT...17 8. REFERENCES...18 9. TABLES...20 10. FIGURES...31

1. Executive Summary Quantifying carbon stocks in mangrove forests in Africa is difficult because of their remote location, challenging site conditions, and uncertainties with the conceptual framework that has been developed primarily in Asia. Recent advances in remote sensing data and associated analytical tools suggest that they may be used effectively to provide a first approximation of carbon pools, and serve as a useful basis for designing field-based inventories that are intended to quantify carbon stocks for REDD+ as well as other markets. Our objective in this project is to characterize the spatial distribution of biomass and carbon stocks in mangroves in West Africa, specifically along the Atlantic coast from Guinea to Cameroon, using available spatial data obtained from remote sensing and GIS (Geographic Information Systems) technologies. In this project, we conduct GIS-based data processing and spatial analysis to evaluate the coverage, canopy height, biomass, and carbon of mangroves in West Africa. We first introduce study area, covering 9 countries in West Africa. We then conduct literature review on the existing work of mangrove study regarding coverage, canopy height and biomass. In Section 4, we report how we process and analyze mangrove data in the study area. The coverage of mangroves is estimated using data from NASA and USGS and relevant comparison given. Canopy heights of mangroves are derived from NASA dataset and classified into five classes to facilitate further analysis. Aboveground biomass of mangroves is estimated using empirical allometric equation which is a function of canopy height. We estimate the mean above- and belowground biomass of mangroves in our study area at two levels: regional and country. We compare and discuss the estimated mean and total biomass of mangrove forests in our study region. Further, we propose an approach that combines online available data of mangrove coverage and canopy heights to estimate biomass at regional or higher levels (in Section 4.5) and investigate possible anthropogenic impacts on mangrove forests. In the Discussion section, we provide detailed discussion on the area, canopy height, and biomass of mangroves in and across different countries in our study area. In the conclusion section, we summarize findings in this project. Highlights of our findings are: The total area of mangrove forests in our study area is estimated as 13,897.78 km 2. The total mangrove biomass and carbon (above- and below-ground) are 272,562,453 and 136,281,227 Mg for the entire study area. At the country level, the largest area of mangrove forests is observed in Nigeria, comprising 65% of the mangrove forests in the entire study area, while Togo has the smallest area (2.29 km 2 ) among the 9 countries in West African. Report, data, and maps related to this project can be accessed through the following web links: Report of this project is available at (30 MB): http://cybergis.uncc.edu/westafrica/report.zip Data of this project is available at (2.12 GB; password: santee): http://cybergis.uncc.edu/westafrica/data.zip Maps of this project is available at (92MB; password: santee): Center for Applied Geographic Information Science, UNC Charlotte Page 1

http://cybergis.uncc.edu/westafrica/maps.zip 2. Study area Our study area covers 9 countries on the Atlantic coast of West Africa, including Guinea, Sierra Leone, Liberia, Cote d Ivoire, Ghana, Togo, Benin, Nigeria, and Cameroon (see Figure 1 and 2). These countries form a long and narrow shape from the northwest region of Guinea (16 W, 12 N) to the southernmost part of Cameroon (16 E, 1 N). West Africa has distinguished wet and dry seasons due to the interplay between hot-dry continental air mass and moisture-maritime air mass in tropical regions. The annual average temperature of our study area is around 26 C. Affected by ocean currents, our study area has an annual precipitation ranging from 1,000 to 2,000 mm (Saenger and Bellan 1995). Reported by Giri et al. (2011), Nigeria (the largest country in our study area) accounts for 4.7 percent of the total area of global mangrove forests. Tropical climate and saline water environments provide favorable conditions and habitats for mangrove forests in West Africa, substantially contributing to the diversity of mangroves in our study area. Seven local species of mangroves exist: Acrostichum aureum, Avicennia germinans, Conocarpus erectus, Conocarpus erectus, Rhizophora harrisonii, R. mangle, and R. racemosa. In addition, an introduced species, Nypa fruticans, has been observed in West Africa (see Fatoyinbo and Simard 2013). 3. Literature review To better protect mangroves from natural or anthropogenic threats, a strong focus should be paid on the monitoring of mangrove forests with respect to, for example, coverage, structure and biomass. In this section, we conduct literature review by focusing on the estimation of coverage, canopy height, and biomass of mangroves. 3.1 Estimation of mangrove coverage The habitats of most mangrove forests are usually located in harsh and complex environments, which make it extremely challenging and difficult to access and conduct field-based assessments (Fatoyinbo and Simard 2013). With advancement in remote sensing and computer technology, global distribution and coverage of mangrove forests have been increasingly estimated using remote-sensing techniques (Dahdouh-Guebas et al. 2000; Kovacs, Wang, and Blanco-Correa 2001; Dahdouh-Guebas et al. 2002). The total area of worldwide mangrove forests reported in the literature as summarized by Giri et al. (2011) is from 110,000 to 240,000 km 2 (Wilkie and Fortuna 2003; FAO 2007). Specifically, Giri et al. (2011) reported the total area of global mangrove forests as 137,760 km 2. Global Land Survey data (http://landsat.usgs.gov/science_gls.php) data from US Geological Survey (USGS) were applied to map distributions of mangrove forests at a spatial resolution of 30 m 30 m. Furthermore, Giri et al. (2011) developed a hybrid approach that combines supervised and unsupervised classifications based on Landsat imagery scenes. Meanwhile, Giri et al. (2011) compared estimations of global mangrove area from existing literatures. Center for Applied Geographic Information Science, UNC Charlotte Page 2

Besides estimations of mangrove area at the global scale, researchers have dedicated their efforts to the estimation and mapping of spatial distribution of mangroves in different regions (see Saenger and Bellan 1995; De Boer 2002; Adams, Colloty, and Bate 2004; Dahdouh-Guebas et al. 2004; Fatoyinbo et al. 2008; Kovacs et al. 2010; Fatoyinbo and Simard 2013; also see Table 1). Kovacs et al. (2010) conducted a study to map mangroves in Mabala and Yelitono mangrove islands in Guinea, Africa. Kovacs et al. s work is based on the use of four classes (including tall, medium, dwarf red, and black mangrove) within 10,442-ha mangrove forests (identified using field and remote sensing data). Kovacs et al. used an unsupervised classification approach ISODATA to identify mangroves based on satellite imagery from IKONOS and QuickBird. At the continental scale, Fatoyinbo and Simard (2013) estimated the total area of mangroves in Africa as 25,960 km 2. This estimation is achieved using an unsupervised Iterative Selforganizing Data Analysis (ISODATA) method and Landsat TM GeoCover data. 3.2 Estimation of canopy height of mangroves Mangrove forests are one of the most productive ecosystems in terms of carbon cycling and storages (Jennerjahn and Ittekkot, 2002). Thus, it is of critical importance to estimate the structural and functional attributes of mangrove forests in order to better investigate and understand global carbon cycling and storages. Recently, three dimensional structure and biomass of mangrove forests have been assessed and mapped using methods based on remote sensing technology (Simard et al. 2006; Simard et al. 2008; Lucas et al. 2007; Fatoyinbo et al. 2008; Fatoyinbo and Simard 2013). In existing literature, LiDAR and Interferometric Synthetic Aperture Radar (InSAR) have been applied for estimation and measurement because of their high spatial resolution (Fatoyinbo and Simard 2013). At the global scale, two InSAR data sources are available for the estimation of mangrove structure and biomass, including SRTM (Shuttle Radar Topography Mission; http://www2.jpl.nasa.gov/srtm/) and ICESat/GLAS (Ice, Cloud, and land Elevation Satellite/Geoscience Laser Altimeter System; http://icesat.gsfc.nasa.gov/). Specifically, the Shuttle Radar Topography Mission (SRTM) dataset consists of global-scale elevation data with a high spatial resolution (30 m 30 m for USA and 90 m 90 m at global level), collected by a space shuttle in February, 2000. The SRTM dataset has four versions so far. The first version of the SRTM data (version 1) was original digital elevation model (DEM) from the SRTM project. To cope with spurious data issue in such areas as water bodies, the National Geospatial-Intelligence Agency (NGA) processed version 1 SRTM data, leading to version 2 with well-defined water bodies and coastlines. Moreover, the SRTM Water Body Data (SWBD) including a vector-based coastline mask was generated in version 2. Then, the SWBD was used to clip the coastlines to generate version 3, while auxiliary DEMs were applied to fill the missing data in version 2. The version 3 shifted a half grid pixel from version 2. In the latest version of SRTM data (version 4), a set of interpolation methods and extra auxiliary DEMs were used to further refine elevation data at the global level. Canopy height information is included in SRTM Center for Applied Geographic Information Science, UNC Charlotte Page 3

data, which allows for estimating the canopy height of mangroves based on the assumption that mangroves are observed at sea level (Fatoyinbo et al. 2008). Simard et al. (2006) mapped the mean canopy height of mangroves and estimated the biomass in the Everglades National Park (ENP) in south Florida, a tidally influenced tropical area. High spatial resolution (30 m 30 m) SRTM data were used to measure the mean canopy height of mangrove forests, calibrated by LiDAR and DEM data from USGS. Field data, collected both in Everglades National Park and Biscayne Bay, were applied in a linear regression model to explore relationships between mean canopy height of mangroves and biomass. Moreover, Simard et al. (2006) reported that intermediate-height mangroves with a height around 8 m hold the most amount of standing biomass in the Everglades National Park. For the same objective, Simard et al. (2008) developed a systematic method to assess the canopy height and aboveground biomass with support from SRTM data, ICEsat/GLAS waveforms and field data in Cienage Grande de Santa marta (CGSM) in Colombia. SRTM data (version 2) with a spatial resolution of 90 m 90 m were used, calibrated independently by field and ICEsat/GLAS datasets, to estimate the canopy height of mangrove forests. Meanwhile, a relationship between biomass and mangrove height was derived based on field data and published allometric equations in three studies, including Day Jr et al. (1987), Fromard et al. (1998), and Smith III and Whelan (2006). Furthermore, Simard et al. (2008) evaluated the biomass loss of mangroves in their study area. 3.3 Estimation of biomass from mangroves The structure and biomass of mangroves in Africa have been investigated and reported in the literature (see Fatoyinbo et al. 2008; Fatoyinbo and Simard 2013). The study area of Fatoyinbo et al. (2008) is Mozambique with the third largest area of mangroves in Africa. The spatial distribution of mangrove heights was estimated and mapped with Landsat Enhanced Thematic Mapper Plus (ETM+) scenes (spatial resolution: 30 m 30 m) and SRTM data (version 3). The SRTM data were calibrated to obtain the canopy height of mangroves using the height calibration equation from Caribbean (Simard et al. 2006). Based on previously published allometric equation, field data were collected to develop a linear height-biomass relationship to estimate the aboveground biomass of mangroves. Results from Fatoyinbo et al. (2008) illustrated that the total amount of aboveground biomass of mangroves in Mozambique was 23.6 million tons. Likewise, Fatoyinbo and Simard (2013) conducted biomass estimation from the canopy height of mangroves for the entire Africa. In their work, SRTM data (version 4) were applied to generate a single DEM for the study area. Meanwhile, ICESat/GLAS data were used to estimate canopy height. A linear regression model, estabilishing the relationship between relative canopy height of GLAS point (rh 75 ; unit: m) and SRTM DEM height (H SRTM ; unit: m), was used in Fatoyinbo and Simard (2013) as follows: (1) Center for Applied Geographic Information Science, UNC Charlotte Page 4

where a and b are the slope and intercept correspondingly. In terms of the calculation of aboveground biomass (noted as biomass; unit: Mgha -1 ), Fatoyinbo and Simard (2013) used a global allometric equation developed by Saenger and Snedaker (1993) (same notation used as in Fatoyinbo and Simard (2013), which is a function of canopy height: where B H is mean biomass (unit: Mg/ha); H is canopy height (unit: m). (2) 4. Data processing and analysis 4.1 Data Our estimation on mangrove biomass in West Africa is based on a collection of datasets, including canopy height, mangrove coverage, and country boundary. Table 2 summarizes these datasets and their source. The original canopy height dataset that we used in this work is obtained from NASA, including 9 Tagged Image File (TIF) files covering 9 countries in our study region (see Figure 2). The value of canopy height ranges from 0 to 255 m in original data (see Figure 3). A small number of cells with height values of 6 or 21 m exist in the original dataset (Figure 4), possibly due to rounding operation by NASA. To facilitate the GIS-based data processing, we merged these image-based datasets together into a single raster dataset to represent the canopy height of the entire study area (Mosaic to new raster in ArcGIS). Suggested by NASA (through communication with Dr. Marc Simard), there exist about 10 percent errors in the original dataset of canopy height. We applied 1% as a cut-off threshold (corresponding to 32 m) to further process the canopy height data: cells with height values less than or equal to 32 m remain unchanged, while for these cells with height values above 32 m their heights were replaced with values estimated using spatial interpolation. Regarding spatial interpolation, we applied the Inverse Distance Weighting (IDW) algorithm (Bolstad 2005) to produce a canopy height surface of our study area (see Equation 3). (3) where Z j is the canopy height value estimated for cell j; Z i is the canopy height value in the neighborhood cell i. d ij is the distance between cell i and j, and α is an exponent specified by users (α=2 in this study). This interpolated height surface allows us to estimate height values for those cells, in which original values are higher than 32 m. Our spatial interpolation procedures include the following steps. First, we converted those cells, height values of which are less than and equal to 32 m into control points for spatial interpolation. For the area without sample points in our study region, Center for Applied Geographic Information Science, UNC Charlotte Page 5

we used 12 closest surrounding sample points to predict height values of these cells. Then, the height values of cells higher than 32 m in the original dataset were extracted based on the interpolated surface of canopy height (extracted by mask in ArcGIS). Third, the raster grid data of canopy height were generated by merging cells with correct height values from original dataset and those with interpolated height values (rounded to the nearest integer to remain consistent with original data). 4.2 Assessment of mangrove area There are 9 countries in our study area: Guinea, Sierra Leone, Liberia, Cote d Ivoire, Ghana, Togo, Benin, Nigeria and Cameroon. The assessment of mangrove area in these countries is based on the data of mangrove coverages from USGS (see Table 2). Two sets of data are available: global mangrove coverage data in 2000 used in Giri et al. (2011), and unpublished USGS data by G. Tappin. Please note that for the Tappin s dataset, since complete files were not provided when the analysis was conducted, we only report the information of mangrove area in those countries where data are available (data of four countries are missing: Cameroon, Cote d Ivoire, Nigeria, and Togo). Table 3 summarizes the area of mangroves reported from different sources (discussed later in Section 5). The areas were calculated by the following processes. First, we converted the raster of canopy height to polygons (using Raster to Polygon in ArcGIS), and dissolved the polygons by height values (Dissolve tool in ArcGIS). We then calculated the total area of the mangroves in our study area (Calculate geometry in ArcGIS). To better understand the distribution of canopy height of mangroves in each country, the mean and standard deviation of canopy height (see Table 4) were calculated (weighted by area). The boundary data of each country were from Global Administrative Area (GADM) database (see http://www.gadm.org). We merged the boundaries together (tool Merge in ArcGIS), and then divided the total mangroves polygons to 9 countries (Split tool in ArcGIS). Table 4 shows the results of area-weighted mean and standard deviation of canopy heights for each country and the entire study area. Figure 6, 9, 12, 15, 18, 21, 24, 27, and 30 are histograms of canopy heights for the 9 countries in our study area. 4.3 Classification of canopy heights According to the statistical distribution of canopy heights in the entire study area, we classified canopy heights into five classes among which the total number of cells (i.e. area) is balanced. The five classes are summarized as follow (also see Table 5): Class 1: 1 to 4 m of canopy height, covering 15.89% of the entire study area. Class 2: 5 to 7 m with an area percentage of 24.80%. Class 3: 8 to 9 m, occupying 19.36% of the entire study area. Class 4: 10 to 13 m with a percentage of 19.33%. Class 5: 14 to 32 m, which has a percentage of 20.62%. Center for Applied Geographic Information Science, UNC Charlotte Page 6

Figure 5, 8, 11, 14, 17, 20, 23, 26, and 29 are maps of canopy heights for the 9 countries in our study area. Histograms of canopy height classes for each country are illustrated in Figure 7, 10, 13, 16, 19, 22, 25, 28 and 31. 4.4 Estimation of biomass and carbon based on canopy heights by country The estimation of mangrove biomass plays a very important role in the study of carbon stock. Basically, carbon can be estimated from biomass as 50% of the biomass estimate. Furthermore, belowground biomass of mangroves can be estimated based on the ratio of aboveground to belowground biomass that was calculated for a range of diameters typical of mangrove stand using the general mangrove biomass equations provided by Komiyama et al. (2008). The average ratio was 0.38, ranging from 0.54 for small trees (5 cm diameter) to 0.31 for large trees (50 cm diameter). In this study, the mean biomass of a certain height (i.e., aboveground biomass; unit: Mg/ha) was calculated by using the allometric equation as in Equation 2, same as that used in Fatoyinbo and Simard (2013). Figure 32 shows the spatial distribution of the mean aboveground biomass of mangroves in the entire study area. Figures 33-41 are maps of the mean aboveground biomass of mangroves for the nine countries in the study area. The mean aboveground biomass of a canopy height class in the entire study area was calculated by: (4) where is the mean aboveground biomass for a specific canopy height class (unit: Mg/ha); N H is the number of cells with the canopy height H; B H is the mean aboveground biomass for canopy height H calculated by Equation 2. Using Class 3 as an example (ID=3), there are two different heights: 8 and 9 m (see Table 5). In order to obtain the mean aboveground biomass for this class, we first need to calculate the mean aboveground biomass using Equation 2 (the mean aboveground biomass is 121.4 Mg/ha for 8 m and 132.2 Mg/ha for 9 m). Then, the mean aboveground biomass of Class 3 was calculated by the mean aboveground biomass of different heights in the canopy height class weighted by area (here we use the number of cells in Table 5) for each height value. The mean aboveground biomass of Class 3 is thus calculated as: Table 6 shows the mean aboveground biomass for each class in the entire study area. We will discuss the results later (in Discussion section). Center for Applied Geographic Information Science, UNC Charlotte Page 7 (5)

Likewise, we calculated mean aboveground biomass for each country using the following formula: (6) where is the mean aboveground biomass of a country (unit: Mg/ha); is the number of cells in the country with canopy height H; B H is the mean aboveground biomass for canopy height H calculated by Equation 2. The results of the mean aboveground biomass for each country are shown in Table 7. The mean aboveground biomass of each class in each country was calculated by the following formula: ( ) (7) where is the mean aboveground biomass of a specific canopy class (class ID: ClassID) in a specific country (unit: Mg/ha); is the number of cells in the country with canopy height H; B H is the mean aboveground biomass for canopy height H calculated using Equation 2. The results of the mean aboveground biomass of each class for each country are shown in Table 8 (for detailed discussion, please refer to Section 5). The total aboveground biomass (unit: Mg) was calculated by the corresponding mean aboveground biomass (unit: Mg/ha) multiplied by the area (unit: ha). Since we used km 2 as the unit for the area, we need to convert the area to ha (1 km 2 =100 ha). The formula for calculating total aboveground biomass is: where TB is total aboveground biomass (unit: Mg). B is the corresponding mean aboveground biomass calculated by Equation 4, 6, or 7. A is area (unit: km 2 ). 100 is a scaling factor for unit conversion. Using class 3 as an example, the mean aboveground biomass is 126.1976 Mg/ha (see Equation 5). The area of the entire study area is 13,897.7783 km 2 (see Table 4), and the percentage of class 3 in the entire study area is 19.36% (see Table 5). The area of class 3 is 2,690.6099 km 2. Thus, the total aboveground biomass of canopy height class 3 is: The total aboveground biomass of each class in our study area is shown in Table 6. Table 9 summarizes the total aboveground biomass of each class in each country and the total aboveground biomass of each country. Given aboveground biomass, we estimated belowground biomass using the following equation: (8) (9) Center for Applied Geographic Information Science, UNC Charlotte Page 8

where r ba is the ratio of belowground biomass compared to aboveground biomass (38% in this study). B belowground and B aboveground are the belowground and aboveground biomass. Based on the relationship of carbon and biomass, we estimated the carbon (both below- and aboveground) by equation 11. where r ca is the ratio of above/below ground carbon compared to above/below ground biomass (50% in this study). C and B are carbon and biomass. 4.5 Regional-level estimation of biomass and carbon using data available online Besides using canopy height data with a spatial resolution of 90 m 90 m, we applied a coarserresolution canopy height data (available at global scale; see http://wwwradar.jpl.nasa.gov/coastal/) to estimate the mangrove biomass and carbon in our study area. The global canopy height has a spatial resolution of 1 km 1 km, reported in Simard et al. (2011), based on the spaceborne LiDAR data collected from Geoscience Laser Altimeter System (GLAS) aboard ICESat (Ice, Cloud, and land Elevation Satellite) in 2005. In Simard et al. (2011), SRTM elevation data (with a spatial resolution of 90 m 90 m) were used to produce a slope map for minimizing the bias of canopy height estimation. However, as Simard et al. (2011) reported, canopy height in their work is over-compensated due to the heterogeneous characteristics of mangroves. In order to identify mangroves in our study region, global mangrove coverage data from Giri et al. (2011) were used. While the spatial resolution of canopy height data is coarse (1km 1km), the combination of global canopy height data from NASA and mangrove coverage from USGS provides a mean of estimating mangrove biomass and carbon at a global scale, which may offer insight into regional- or continental-level mangrove studies. We used the following three steps to estimate mangrove biomass and carbon in our study area: First, we generated vector-based GIS data of mangrove coverage by clipping the global mangrove coverage data using the boundary of our study region. Second, the mangrove coverage data was used as a mask to extract canopy height information from the global canopy height data. As a result, the GIS data of canopy height of mangroves in our study area was generated at a spatial resolution of 1km 1km. Figure 42-50 show maps of canopy height classes of mangroves in 9 countries based on the GIS data of mangrove canopy height at a spatial resolution of 1 km 1 km. (10) (11) Center for Applied Geographic Information Science, UNC Charlotte Page 9

Third, a global allometric equation (see Equation 1) suggested by Saenger and Snedaker (1993) was applied to calculate aboveground biomass based on the canopy height of mangroves. Then we estimated belowground biomass using equation 10. The above- and belowground carbon were estimated using equation 11. Table 10 compares results of mangrove area, biomass (mean and total), and carbon (mean and total) estimated from data at two spatial resolutions: coarse (1 km 1 km) and fine (90 m 90 m). The total mangrove area for West Africa was 20.16 percent lower when using coarseresolution data, compared to fine-resolution data. With respect to mangrove area estimated by fine-resolution data, the largest difference between two datasets is in Liberia, as much as 57.06%, followed by Nigeria (41.6%) and Cote d Ivoire (37.74%). Unlike mangrove area, the mean biomass and carbon of mangroves estimated from the coarse-resolution data is lower than that from the fine-resolution data in each country, especially in Liberia (63.6%), Benin (60.18%), and Cote d Ivoire (52.69%). The total biomass and carbon for the entire study area estimated from the two datasets is close. At the country level, the total biomass and carbon of mangrove forests in Cote d Ivoire, Liberia, and Nigeria obtained from coarse-resolution data was underestimated with respect to that from fine-resolution data, while it is overestimated in other 6 countries. Our analysis results suggest that, in comparison with the fine-resolution data (90 m 90 m), using the coarse-resolution data (1 km 1 km) tends to overestimate the biomass (mean and total) and carbon (mean and total) of mangrove forests in our study area. Yet, the combination of coarse-resolution canopy height data (from NASA) and mangrove area (from USGS) provides a feasible and convenient approach for estimating biomass and carbon of mangroves at regional or global levels. 4.6 Proximity of mangroves to human settlements In this study, we conducted spatial analysis to evaluate the proximity of mangroves to human settlements (e.g., cities or towns) to gain insight into anthropogenic impacts on mangroves. GIS data of cities/towns used in this study was from ESRI (http://www.esri.com). We extracted cities (including national capitals, provincial capitals, major population centers, and landmark cities) for 9 countries in our study region, and converted them into a vector-based GIS dataset of cities. The locations of major cities in West Africa were represented in points. Then, based on the fineresolution canopy height data (90 m 90 m from NASA), a cost-distance surface was generated to derive the shortest Euclidean distance from mangrove forests to nearest city (see Figure 51). Table 11 summarizes the shortest distance from mangrove forests to nearest city for each country. As observed from Table 11, the longest Euclidean distance (22.93 km) from mangrove forests to nearest city is in Togo in our study region. In other words, mangroves in Togo may receive the least anthropogenic disturbance among 9 countries. This may explain increase in the area of mangrove forests in Togo (see Table 3) from Giri et al (2011) s USGS data (for the year of 2000) Center for Applied Geographic Information Science, UNC Charlotte Page 10

to Tappin s unpublished data in 2014. At another extreme, mangrove area in Guinea decreases from 2,343.97 to 2,191.08 km 2 from Giri et al. (2011) s data to Tappin s data, which may be attributed to the close proximity of mangrove forests to city. 5. Discussion 5.1 Comparison of mangrove area between those reported in the literature and our estimates Table 3 compares data from different sources, including global-level mangrove coverage from USGS (USGS1 in the table) and canopy height data from NASA (NASA2). As we could observe from this Table, estimates of mangrove area were different because they are based on alternative data sources and generated using different classification algorithms (see literature review in Section 3.1; also see Giri et al. 2011; Fatoyinbo and Simard 2013). Among 9 countries in our study region, Nigeria is the one with the largest discrepancy of mangrove area between the two datasets; up to 2,929 km 2 (our estimate is 47.40% more than that based on the USGS dataset by Giri et al. (2011)). The country with the second largest difference of mangrove area is Cameroon, 546 km 2 (25.69% less in our estimate than that from Giri et al. (2011) s USGS dataset) of difference (much smaller than that in Nigeria). While the area difference in Togo is the smallest (about 4 km 2 ), it is the second largest percentage difference (64.45% of decrease from Giri et al. (2011) s dataset to our estimate based on NASA dataset). Nigeria and Liberia are the two countries in which the area of mangroves estimated from the USGS dataset is smaller than our estimate. The mangrove coverage for Liberia is 99.83 km 2 from the USGS dataset, but it is 202.60 km 2 from the NASA dataset from our estimation (102.95% higher). Please note that the relationship of the mangrove area among countries remains the same despite the source, leading to the fact that regional ranking of these countries with respect to mangrove area does not change based on different data sources. Figures 52-60 show maps of discrepancies in spatial distributions of mangroves from the two data sources. The main difference of spatial distributions of mangroves is located in the southern part of coastline in Guinea and Cameroon (see Figure 52 and 60). Conversely, in Sierra Leone, the main difference occurs in its northern coastline (see Figure 53). Although Nigeria has the largest estimation difference of mangrove area, locations that contribute to this difference is dispersed in its entire coastal region instead of being clustered in certain area (see Figure 59). The area of mangroves in our study is higher than those from previous work by NASA (Fatoyinbo and Simard 2013) for all countries but Guinea. The differences of these two datasets (i.e., from Fatoyinbo and Simard (2013) and our estimates) are much smaller than that between estimates from Giri et al. (2011) s USGS dataset and our estimates. Regarding the country with the largest difference of area estimation, two sets of comparisons (estimates based on Giri et al. (2011) s dataset and our estimates; Fatoyinbo and Simard (2013) and our estimates) yield consistent results Nigeria. However, the difference is much smaller between Fatoyinbo and Center for Applied Geographic Information Science, UNC Charlotte Page 11

Simard (2013) and our estimates (536 km 2 ) than that between estimates from Giri et al. (2011) s USGS dataset and our estimates based on NASA dataset (2,929 km 2 ). Similarly, the smallest difference in mangrove area is in Togo: 0.29 km 2 between Fatoyinbo and Simard (2013) and our estimate as well as 4.15 km 2 between the estimate from Giri et al. (2011) s USGS dataset and our estimate. 5.2 Mangrove area, canopy heights, biomass, and carbon estimation in the entire study area The total area of mangrove forests in West Africa (including 9 countries) is estimated as 13, 897.78 km 2 (see Table 4). The largest area of mangrove forests is located in Nigeria with an area of 9,109.29 km 2 (65.54% of the total area of mangrove forests of our entire study region), followed by Guinea (12.98%), Cameroon (11.36%), Sierra Leone (7.37%), Liberia (1.46%), Ghana (0.58%), Cote d Ivoire (0.25%) and Benin (0.14%). The smallest area of mangroves at country level is found in Togo, in which there are only 2.29 km 2 mangrove forests occupying 0.02% of the total area of mangroves in our study region. With respect to the area of mangroves, large variations among countries might be attributed to human activities (e.g., converting mangrove area to farmland) and spatial characteristics of mangrove habitats, such as low latitude, river delta area, and so on. The statistical distribution of mean biomass and carbon among 9 countries is more even (see Table 7), because mean biomass in each country is mainly associated with canopy height of mangroves (see Equation 2). 5.3 Comparison of mangrove area, canopy heights, biomass, and carbon estimation by country In this sub-section, we discuss the estimation of mangroves in each country (from northwest to southeast of West Africa) by focusing on area, spatial distribution, canopy height, biomass, and carbon. The first country that we discuss is Guinea. Guinea has the second largest area of mangrove forests among these 9 countries, with an area of 1,804.6 km 2 (see Table 4), spreading evenly along with its entire coastline (see Figure 5). Reported by FAO (2007), 9% of mangroves in the entire Africa is located in Guinea. Histogram of canopy heights in Guinea (see Figure 6) shows that the distribution of canopy heights in Guinea is highly similar to that of the entire study region (see Figure 4). In terms of five canopy height classes defined in section 4.3 (as illustrated in Figure 7), canopy heights of mangroves in Guinea concentrates in class 2 (5-7 m) and class 3 (8-9 m), while the area of mangroves in class 5 (14-32 m ) is the smallest. The average height of mangroves in Guinea is 7.85 m with a standard deviation of 4.24 m. With respect to spatial patterns (see Figure 5), the locations of mangroves falling within height class 5 mainly concentrate in the island area in the north and the southernmost coastal region of Guinea. According to the canopy height of mangroves, mean aboveground biomass and carbon of mangroves in Guinea are 119.22 Mg/ha and 59.61 Mg/ha (see Table 7), while total aboveground biomass and carbon are 21,515,191 Mg and 10,757,596 Mg in Guinea (the third largest total biomass and carbon in West Africa; see Table 9). Center for Applied Geographic Information Science, UNC Charlotte Page 12

Accounting for 7.37% mangrove coverage in 9 countries, the area of mangroves in Sierra Leone is estimated as much as 1,024.6 km 2 (see Table 4), the fourth largest coverage of mangrove forests in West Africa. Similar to Guinea, the majority of mangrove forests in Sierra Leone concentrate in the northernmost and southern river delta area (see Figure 8). In terms of canopy height of mangrove forests, height class 1 (1-4 m) and class 2 (5-7 m) are dominant among five canopy height classes (see Figure 10). Compared to Guinea, even though the area of mangroves in Sierra Leone is much smaller than that in Guinea, Sierra Leone has larger mean biomass and carbon of mangroves (125.48 Mg/ha and 62.74 Mg/ha for aboveground; see Table 7). This is because the averaged canopy height of Sierra Leone is 8.39 m, higher than that in Guinea (see Table 4). Furthermore, the total aboveground biomass and carbon of mangrove forests are 12,856,853 Mg and 6,428,427 Mg, the fourth largest total biomass and carbon of mangroves in the entire study area. The total area of mangrove forests in Liberia is about 202.6 km 2 (see Table 4), around 1.46% of the total coverage of mangrove forests in West Africa. Figure 11 indicates that mangrove forests are mainly distributed in the northern coastline region of Liberia. Like the distribution of mangrove height in Sierra Leone, showed by Figure 13, canopy height of mangrove forests in Liberia is nested in canopy height class 1 (1-4 m) and class 2 (5-7 m), with a mean value of 8.59 m close to that in Sierra Leone (see Table 4). Therefore, the high averaged value of canopy height in Liberia results in high mean biomass and carbon values (125.8 Mg/ha and 62.9 Mg/ha for aboveground), the fourth largest mean biomass and carbon of mangroves in West Africa (see Table 7). The total aboveground biomass and carbon of mangroves in Liberia are 2,548,743 Mg and 1,274,372 Mg. Due to lack of estuarine complex and river delta, the area of mangrove forests in Cote d Ivoire is 35.33 km 2 (see Table 4), only taking the proportion of 0.25 percent of mangrove area in West Africa. Mangrove forests of Cote d Ivoire are observed in the middle of the shoreline (see Figure 14). The histogram of canopy heights in Cote d Ivoire illustrates that mangroves in this country are mainly clustered in low and high height classes, lacking medium height (see Figure 15 and 16). However, the mean value of canopy height (10.71 m) is the second largest in West Africa (see Table 4). Thus, Cote d Ivoire has the second largest mean biomass and carbon of mangrove forests: 143.16 Mg/ha and 71.58 Mg/ha for aboveground (see Table 7), leading to 505,845 Mg and 252,923 Mg of total aboveground biomass and carbon. Similar to Cote d Ivoire, most mangrove forests are located in the middle of coastline region in Ghana (see Figure 17). The area of mangrove forests in Ghana is 80.64 km 2, twice larger than that in Cote d Ivoire (see Table 4). In terms of the canopy height of mangroves, most of the mangrove forests in Ghana are very low, ranging from 2 m to 5 m (see Figure 18 and 19). The mean aboveground biomass and carbon of mangroves are 113.14 Mg/ha and 56.57 Mg/ha, while the total aboveground biomass and carbon are 912,341 Mg and 456,171 Mg (see Table 7 and 9). Center for Applied Geographic Information Science, UNC Charlotte Page 13

Togo is the country with the smallest area of mangrove forests: 2.29 km 2, only accounting for 0.02% of the total area of mangroves in West Africa (see Table 4). Mangrove forests grow in the most eastern and western coastline area in Togo (see Figure 20). Similar to the distribution of canopy height in Ghana, most mangroves in this area (see Figure 22) belong to height class 1 (1-4 m). Due to a low averaged canopy height of 4.69 m (see Table 4), the mean aboveground biomass and carbon of mangroves in Togo are as low as 85.89 Mg/ha and 42.94 Mg/ha. The total biomass and carbon of mangroves in Togo (about 19,669 Mg and 9,835 Mg for aboveground) are the smallest among 9 countries (see Table 9). Benin has the second smallest area of mangroves (19.16 km 2 ) in our study area (slightly larger than that in Togo; see Table 4). Most of the mangrove forests in Benin are located in river delta area in the western and middle coastline of Benin (see Figure 23). The distribution of canopy heights in Benin is not even, highly clustered on the low value side (see Figure 24). Because of the lowest mean canopy height (4.47 m), Benin has the lowest mean biomass and carbon of mangrove forests among 9 countries: 83.21 Mg/ha and 41.6 Mg/ha for aboveground (see Table 7). The total aboveground biomass and carbon of mangroves in Benin are 159,422 Mg and 79,711 Mg, which are slightly larger than those in Togo (see Table 9). The largest area of mangrove forests in West Africa is in Nigeria with a total area of 9,109.29 km 2, which is 65.54 percent of the entire area of mangroves in our study region (see Table 4). Nigeria provides suitable habitats for mangroves in river delta and island area in the middle of the coastline of Nigeria (see Figure 26). The distribution of canopy height in Nigeria is close to that of the entire region (see Figure 27). Among five canopy height classes, class 2 (5-7 m) takes the largest proportion, and proportions of other four classes are close to each other (see Figure 28). The averaged canopy height in Nigeria is 9.88 m with a standard deviation of 8.69 m (see Table 4). Based on the estimated height of mangrove forests, we calculated the mean aboveground biomass and carbon of mangroves as 138.36 Mg/ha and 69.18 Mg/ha, slightly lower than those in Cote d'ivoire (see Table 7). Nigeria has the highest total aboveground biomass and carbon of mangroves in West Africa, with values of 126,039,670 Mg and 63,019,835 Mg (see Table 9). Similar to Nigeria, Cameroon has a suitable environment for mangrove forests in river delta and island area in northwestern and southeast parts. The total area of mangrove is 1,578.88 km 2, occupying 11.36% of the entire mangrove coverage of West Africa (see Table 4). With respect to canopy height, mangroves in Cameroon are nested in the high value side (see Figure 31), different with other countries in our study region. In other words, most mangroves in Cameroon fall within height class 5 (14-32 m), with the largest mean value of 16.12 m among 9 countries (see Table 4). Therefore, the mean aboveground biomass and carbon of mangroves in Cameroon, 204.47 Mg/ha and 102.24 Mg/ha, are higher than those in other countries in our study region (see Table 7). Furthermore, Cameroon has the second highest total biomass and carbon of mangroves (32,283,823 Mg and 16,141,912 Mg for aboveground), only lower than Nigeria among 9 countries. Center for Applied Geographic Information Science, UNC Charlotte Page 14

6. Conclusion In this study, we conducted GIS-based spatial analysis to estimate area, mean biomass and carbon, and total biomass and carbon of mangrove forests in West Africa, including 9 countries (from Guinea to Cameroon). Based on canopy height data from NASA (spatial resolution: 90 m by 90 m), we estimated the area of mangrove forests at both regional and country levels. Besides, we compared our estimate of mangrove area with those reported in the literature (Giri et al, 2011; Fatoyinbo and Simard 2013). Canopy heights of mangroves forests for each country were analyzed and summarized. Based on the empirical distribution of canopy heights in the entire study region, canopy heights of mangrove forests (ranging from 1 to 32 m) were categorized into five classes among which the number of cells (i.e., area) is balanced. Based on this classification, maps of five canopy height classes in each country were produced to support the analysis of spatial distributions of mangrove forests in our study region. Biomass estimation is very important for us to understand carbon storage and cycling (carbon can be estimated from biomass as 50% of biomass estimate). We estimated the mean biomass and carbon of mangroves in our study area at two levels: regional and country. A global allometric equation (obtained from linear regression; see Fatoyinbo and Simard 2013) was used in this study to calculate mean biomass based on canopy heights of mangroves. With help from the allometric equation, we estimated the mean biomass within each canopy height class by country, weighted by the area of each canopy height class within each country. The total biomass and carbon of each canopy height class within each country was assessed based on the estimated mean biomass and carbon. We compared and discussed the estimated biomass (mean and total) and carbon (mean and total) of mangrove forests in our study region. We further compared biomass and carbon results estimated from canopy height data at coarse and fine resolutions (1 km 1 km and 90 m 90 m), and investigated possible anthropogenic impacts on mangrove forests by deriving a cost distance surface that shows shortest distance between mangroves and cities/towns in the study region. Major Findings from this study are summarized as follow: The total area of mangrove forests in our study area is estimated as 13,897.78 km 2. Mangrove area estimated from the USGS dataset (see Giri et al. 2011) is 11,710.05 km 2, 15.7 percent smaller than our estimate in this study based on NASA dataset. The largest discrepancy between these two estimations is in Nigeria, up to 2,929 km 2. Mangrove area reported in Fatoyinbo and Simard (2013) is close to our estimation, only 4.9% of difference for the entire study area. The occurrence of discrepancy between USGS and NASA datasets can be attributed to the use of different classification algorithms (hybrid supervised and unsupervised classification used in the former; unsupervised classification used by the latter). The estimation of mangrove coverage from USGS and NASA datasets is based on Landsat images (about one thousand scenes used in the former and 117 scenes used by the latter). Center for Applied Geographic Information Science, UNC Charlotte Page 15

The total aboveground biomass and carbon of mangrove forests for the entire study area are 197,509,024 Mg and 98,754,512 Mg. The total belowground biomass and carbon of mangroves in the study area are 75,053,429 Mg and 37,526,715 Mg. At the country level, the largest area of mangrove forests is observed in Nigeria, taking the proportion of 65.54 percent of mangrove forests in the entire study area, while Togo has the smallest area (2.29 km 2 ) among 9 countries. Correspondingly, Nigeria has the highest total aboveground biomass and carbon of mangroves (126,039,670 Mg and 63,019,835 Mg) and Togo has the lowest total biomass and carbon of mangroves. The mean canopy height of mangroves in Cameroon is the largest (16.12 m), leading to the highest values of mean biomass and carbon (204.47 Mg/ha and 102.24 Mg/ha) in West Africa. Further, we recommend an approach that combines the use of mangrove datasets available online (from NASA and USGS; spatial resolution: 1 km 1 km) for assessing biomass and carbon of regional- or continental-level mangroves. This approach makes it feasible and convenient for mangrove study at regional or higher level. The comparison of estimation results between coarse (1 km 1 km) and fine (90 m 90 m) resolution data of canopy heights suggests that coarseresolution data tend to overestimate biomass and carbon in our study area with respect to results from fine-resolution data. The estimation of mangrove biomass in this study may serves as the basis for mangrove carbon inventory in West Africa. This biomass estimation work is based on data of mangrove coverage and canopy heights (from NASA and USGS) derived from remote sensing and GIS technologies. The mapping of the spatial distribution of biomass across the entire study area may be of help for a better understanding of spatial characteristics of mangrove carbon stock in West Africa. Further, the coarse-resolution canopy height data, while tending to overestimate mangrove biomass, can still be a useful dataset for planning large-scale assessments (e.g., country or regional) in the world (instead in West Africa). Of course, a generalized Web-based spatial decision support system (SDSS) tool, based on results from this biomass estimation work and cutting-edge Web GIS computing technology, is necessary to guide and automate the design of mangrove carbon inventory in an intelligent and convenient manner. Center for Applied Geographic Information Science, UNC Charlotte Page 16

7. Acknowledgement Funding was provided by USAID West Africa Mission through the International Programs and Center for Forested Wetlands Research, U.S. Forest Service. We thanks help from Dr. Marc Simard and Dr. Lola Fatyinbo from NASA, Mr. Gray G. Tappan and Mr. John Hutchinson from USGS for kindly sharing their datasets to support this study. Center for Applied Geographic Information Science, UNC Charlotte Page 17

8. References Adams, J. B., B. M. Colloty, and G. C. Bate. 2004. The distribution and state of mangroves along the coast of Transkei, Eastern Cape Province, South Africa. Wetlands Ecology and Management 12 (5):531-541. Bolstad, P. 2005. GIS Fundamentals: A First Text on Geographic Information Systems. Eider Press. White Bear Lake, MN, USA. Dahdouh-Guebas, F., R. De Bondt, P. D. Abeysinghe, J. G. Kairo, S. Cannicci, L. Triest, and N. Koedam. 2004. Comparative study of the disjunct zonation pattern of the grey mangrove Avicennia marina (Forsk.) Vierh. in Gazi Bay (Kenya). Bulletin of Marine Science 74 (2):237-252. Dahdouh-Guebas, F., A. Verheyden, W. De Genst, S. Hettiarachchi, and N. Koedam. 2000. Four decade vegetation dynamics in Sri Lankan mangroves as detected from sequential aerial photography: a case study in Galle. Bulletin of Marine Science 67 (2):741-759. Dahdouh-Guebas, F., T. Zetterström, P. Rönnbäck, M. Troell, A. Wickramasinghe, and N. Koedam. 2002. Recent changes in land-use in the Pambala Chilaw lagoon complex (Sri Lanka) investigated using remote sensing and GIS: Conservation of mangroves vs. development of shrimp farming. Environment, Development and Sustainability 4 (2):185-200. Day Jr, J. W., W. H. Conner, F. Ley-Lou, R. H. Day, and A. M. Navarro. 1987. The productivity and composition of mangrove forests, Laguna de Terminos, Mexico. Aquatic Botany 27 (3):267-284. De Boer, W. 2002. The rise and fall of the mangrove forests in Maputo Bay, Mozambique. Wetlands Ecology and Management 10 (4):313-322. FAO. 2007. Mangroves of Africa 1980-2005: Country Reports. Forest Resources Assessment Working Paper NO. 135. Rome: FAO. Fatoyinbo, T. E., and M. Simard. 2013. Height and biomass of mangroves in Africa from ICESat/GLAS and SRTM. International Journal of Remote Sensing 34 (2):668-681. Fatoyinbo, T. E., M. Simard, R. A. Washington-Allen, and H. H. Shugart. 2008. Landscape-scale extent, height, biomass, and carbon estimation of Mozambique's mangrove forests with Landsat ETM+ and Shuttle Radar Topography Mission elevation data. Journal of Geophysical Research: Biogeosciences 113 (G2):G02S06. Fromard, F., H. Puig, E. Mougin, G. Marty, J. Betoulle, and L. Cadamuro. 1998. Structure, above-ground biomass and dynamics of mangrove ecosystems: New data from French Guiana. Oecologia 115 (1-2):39-53. Giri, C., E. Ochieng, L. L. Tieszen, Z. Zhu, A. Singh, T. Loveland, J. Masek, and N. Duke. 2011. Status and distribution of mangrove forests of the world using earth observation satellite data. Global Ecology and Biogeography 20 (1):154-159. Jennerjahn, T. C., and V. Ittekkot. 2002. Relevance of mangroves for the production and deposition of organic matter along tropical continental margins. Naturwissenschaften 89 (1):23-30. Komiyama, A. J.E. Ong, S. Poungparn. 2008. Allometry, biomass and productivity of mangrove forests: A review. Aquatic Botany 89:128-137. Kovacs, J. M., F. F. de Santiago, J. Bastien, and P. Lafrance. 2010. An assessment of mangroves in Guinea, West Africa, using a field and remote sensing based approach. Wetlands 30 (4):773-782. Center for Applied Geographic Information Science, UNC Charlotte Page 18

Kovacs, J. M., J. Wang, and M. Blanco-Correa. 2001. Mapping disturbances in a mangrove forest using multi-date Landsat TM imagery. Environmental Management 27 (5):763-776. Lucas, R. M., A. L. Mitchell, A. Rosenqvist, C. Proisy, A. Melius, and C. Ticehurst. 2007. The potential of L-band SAR for quantifying mangrove characteristics and change: Case studies from the tropics. Aquatic Conservation: Marine and Freshwater Ecosystems 17 (3):245-264. Saenger, P., and M. Bellan. 1995. The mangrove vegetation of the Atlantic coast of Africa: A review. Universite de Toulouse, Toulouse, France. Saenger, P., and S. C. Snedaker. 1993. Pantropical trends in mangrove above-ground biomass and annual litterfall. Oecologia 96 (3):293-299. Simard, M., V. H. Rivera-Monroy, J. E. Mancera-Pineda, E. Castañeda-Moya, and R. R. Twilley. 2008. A systematic method for 3D mapping of mangrove forests based on Shuttle Radar Topography Mission elevation data, ICEsat/GLAS waveforms and field data: Application to Ciénaga Grande de Santa Marta, Colombia. Remote Sensing of Environment 112 (5):2131-2144. Simard, M., K. Zhang, V. H. Rivera-Monroy, M. S. Ross, P. L. Ruiz, E. Castañeda-Moya, R. R. Twilley, and E. Rodriguez. 2006. Mapping height and biomass of mangrove forests in Everglades National Park with SRTM elevation data. Photogrammetric Engineering & Remote Sensing 72 (3):299-311. Smith III, T. J., and K. R. Whelan. 2006. Development of allometric relations for three mangrove species in South Florida for use in the Greater Everglades Ecosystem restoration. Wetlands Ecology and Management 14 (5):409-419. Wilkie, M. L., and S. Fortuna. 2003. Status and trends in mangrove area extent worldwide. Forest Resources Assessment Programme. Working Paper (FAO). Center for Applied Geographic Information Science, UNC Charlotte Page 19

9. Tables Table 1. Area of mangroves reported in the literature Literature Region Area (km 2 ) Year Approach Saenger and Bellan Atlantic Coast 45,787 Before 1995 Literature review (1995) of Africa De Boer (2002) Maputo Bay, 97.76 1991 Aerial photographs Mozambique Adams et al. (2004) Transkei, South 2.74 1999 Topographic maps and Africa aerial photographs Fatoyinbo et al. Mozambique 2,909 1999-2001 Maximum likelihood (2008) classification Kovacs et al. (2010) Mabala and 104.42 2008 Unsupervised ISODATA Yelitono mangrove classification islands, Guinea Fatoyinbo Simard (2013) and Africa 25,960 1999-2002 Unsupervised ISODATA classification Center for Applied Geographic Information Science, UNC Charlotte Page 20

Table 2. Summary of original datasets used for biomass estimation in West Africa Theme Format Year Scale Source Canopy height (90m 90m) Canopy height (1,000m 1,000m) Mangrove coverage Mangrove coverage Raster TIF 2000 Country Raster TIF 2005 Global Shapefile 2000 Global SRTM dataset from NASA (http://www2.jpl.nasa.gov/srtm/) NASA (see http://lidarradar.jpl.nasa.gov/) USGS data released in 2011 for the year of 2000 (see Giri et al. 2011) Shapefile 2014* Africa Unpublished USGS data by G. Tappin Country boundary Shapefile 2012 Country *unpublished data 2014 Global Administrative Area (GADM) database (see http://www.gadm.org) Center for Applied Geographic Information Science, UNC Charlotte Page 21

Table 3. Summary of area of mangroves in each country for USGS and NASA datasets Mangrove Area (km 2 ) Difference (km 2 ) Country USGS 1 USGS 2 NASA 1 NASA 2 USGS NASA Benin 39.3993 0.2414 18 19.1602-39.1579 1.1602 Cameroon 2,124.8092 NA 1,483 1,578.8786 NA 95.8786 Cote d'ivoire 42.2715 NA 32 35.3334 NA 3.3334 Ghana 112.9929 211.6785 76 80.6399 98.6856 4.6399 Guinea 2,343.9667 2,191.0755 1,889 1,804.5959-152.8911-84.4041 Liberia 99.8280 399.8551 189 202.6031 300.0271 13.6031 Nigeria 6,179.9272 NA 8,573 9,109.2916 NA 536.2916 Sierra Leone 1,401.0377 1,899.5842 955 1,024.5953 498.5465 69.5953 Togo 6.4419 26.6278 2 2.2903 20.1859 0.2903 Total 11,710.0537 NA 13,217 13,897.7783 NA 680.7783 USGS 1: USGS data released in 2011 for the year of 2000 (see Giri et al. 2011); USGS 2: G. Tappin unpublished data 2014; NASA 1: Fatoyinbo and Simard (2013); NASA 2: estimates from this study Center for Applied Geographic Information Science, UNC Charlotte Page 22

Table 4. Statistics of canopy heights of mangroves for each country (data source: NASA) Canopy Height Country Area (km 2 ) Average (m) Standard Deviation (m) Benin 19.1602 4.4662 2.3331 Cameroon 1,578.8786 16.1200 8.1616 Cote d'ivoire 35.3334 10.7095 8.5204 Ghana 80.6399 7.4192 6.4361 Guinea 1,804.5959 7.8472 4.2402 Liberia 202.6031 8.5870 6.2441 Nigeria 9,109.2916 9.8796 8.6859 Sierra Leone 1,024.5953 8.3876 5.1534 Togo 2.2903 4.6916 2.9052 Total 13,897.7783 10.1785 8.2248 Center for Applied Geographic Information Science, UNC Charlotte Page 23

Table 5. Distribution of canopy heights in the study region Height (m) #Cells Percentage Accumulate Percentage Class ID 1 3 0.0000 0.0000 1 2 30,674 0.0191 0.0191 1 3 77,029 0.0480 0.0671 1 4 147,425 0.0918 0.1589 1 5 197,215 0.1228 0.2817 2 6 10 0.0000 0.2817 2 7 201,052 0.1252 0.4069 2 8 172,761 0.1076 0.5145 3 9 138,082 0.0860 0.6005 3 10 107,539 0.0670 0.6674 4 11 83,254 0.0518 0.7193 4 12 65,915 0.0410 0.7603 4 13 53,725 0.0335 0.7938 4 14 45,394 0.0283 0.8221 5 15 38,210 0.0238 0.8458 5 16 32,783 0.0204 0.8663 5 17 28,471 0.0177 0.8840 5 18 25,079 0.0156 0.8996 5 19 22,173 0.0138 0.9134 5 20 20,021 0.0125 0.9259 5 21 532 0.0003 0.9262 5 22 18,270 0.0114 0.9376 5 23 16,303 0.0102 0.9477 5 24 14,673 0.0091 0.9569 5 25 13,359 0.0083 0.9652 5 26 11,930 0.0074 0.9726 5 27 10,687 0.0067 0.9793 5 28 9,493 0.0059 0.9852 5 29 8,261 0.0051 0.9903 5 30 7,152 0.0045 0.9948 5 31 4,865 0.0030 0.9978 5 32 3,492 0.0022 1.0000 5 Center for Applied Geographic Information Science, UNC Charlotte Page 24

Table 6. Above- and below-ground biomass and carbon for each canopy height class in the study region Above-ground Below-ground Biomass Carbon Biomass Carbon Class Mean Mean Mean Mean Height Density Total (Mg) Density Total (Mg) Density Total (Mg) Density (m) (Mg/ha) (Mg/ha) (Mg/ha) (Mg/ha) Total (Mg) 1 [1,4] 72.342 15,973,451 36.1710 7,986,726 27.4900 6,069,911 13.7450 3,034,956 2 [5,7] 99.904 34,436,070 49.9520 17,218,035 37.9635 13,085,707 18.9818 6,542,853 3 [8,9] 126.1976 33,949,806 63.0988 16,974,903 47.9551 12,900,926 23.9775 6,450,463 4 [10,13] 156.0901 41,936,144 78.0451 20,968,072 59.3142 15,935,735 29.6571 7,967,867 5 [14,32] 248.4822 71,213,553 124.2411 35,606,777 94.4232 27,061,150 47.2116 13,530,575 Total [1,32] 142.1155 197,509,024 71.0578 98,754,512 54.0039 75,053,429 27.0019 37,526,715 Center for Applied Geographic Information Science, UNC Charlotte Page 25

Table 7. Mean biomass and carbon density for each country in the study region Above-ground Below-ground Mean Biomass Density (Mg/ha) Mean Carbon Density (Mg/ha) Mean Biomass Density (Mg/ha) Mean Carbon Density (Mg/ha) Country NASA 1 NASA 2 NASA 1 NASA 2 NASA 1 NASA 2 NASA 1 NASA 2 Benin 76 83.2052 38.0000 41.6026 28.8800 31.6180 14.4400 15.8090 Cameroon 171 204.4731 85.5000 102.2366 64.9800 77.6998 32.4900 38.8499 Cote d'ivoire 124 143.1636 62.0000 71.5818 47.1200 54.4022 23.5600 27.2011 Ghana 97 113.1378 48.5000 56.5689 36.8600 42.9924 18.4300 21.4962 Guinea 108 119.2244 54.0000 59.6122 41.0400 45.3053 20.5200 22.6526 Liberia 113 125.7998 56.5000 62.8999 42.9400 47.8039 21.4700 23.9020 Nigeria 111 138.3639 55.5000 69.1820 42.1800 52.5783 21.0900 26.2891 Sierra Leone 112 125.4823 56.0000 62.7412 42.5600 47.6833 21.2800 23.8416 Togo 78 85.8854 39.0000 42.9427 29.6400 32.6365 14.8200 16.3182 NASA 1: Fatoyinbo and Simard (2013); NASA 2: estimates from this study Center for Applied Geographic Information Science, UNC Charlotte Page 26

Table 8. Mean biomass and carbon density of each canopy height class for each country in the study region (Total: mean biomass/carbon density for all canopy heights) Country Benin Cameroon Cote Sierra Ghana Guinea Liberia Nigeria d'ivoire Leone Togo Class 1 67.2504 74.2885 67.1741 68.8664 72.6595 70.8347 72.5882 71.5928 68.6343 Mean Class 2 97.5484 102.312 99.9234 97.1993 100.219 98.3586 99.8377 99.0688 96.6 Biomass Class 3 124.645 126.972 126.763 126.038 126.13 126.137 126.128 126.2036 123.425 Density Class 4 150.357 158.361 157.018 158.079 153.853 157.075 156.003 156.0535 155.24 (Mg/ha) Class 5 204.2 264.427 238.735 230.057 220.227 235.287 244.312 231.2303 200.086 Total 83.2052 204.473 143.164 113.138 119.224 125.8 138.364 125.4823 85.8854 Above-ground Below-ground Mean Carbon Density (Mg/ha) Mean Biomass Density (Mg/ha) Mean Carbon Density (Mg/ha) Class 1 33.6252 37.1443 33.5871 34.4332 36.3298 35.4174 36.2941 35.7964 34.3172 Class 2 48.7742 51.1562 49.9617 48.5997 50.1093 49.1793 49.9189 49.5344 48.3000 Class 3 62.3227 63.4858 63.3817 63.0190 63.0648 63.0685 63.0638 63.1018 61.7125 Class 4 75.1783 79.1803 78.5090 79.0394 76.9264 78.5377 78.0014 78.0268 77.6200 Class 5 102.1000 132.2136 119.3674 115.0285 110.1133 117.6433 122.1559 115.6152 100.0429 Total 41.6026 102.2366 71.5818 56.5689 59.6122 62.8999 69.1820 62.7412 42.9427 Class 1 25.5552 28.2296 25.5262 26.1692 27.6106 26.9172 27.5835 27.2053 26.0810 Class 2 37.0684 38.8787 37.9709 36.9357 38.0831 37.3763 37.9383 37.6461 36.7080 Class 3 47.3652 48.2492 48.1701 47.8944 47.9292 47.9320 47.9285 47.9574 46.9015 Class 4 57.1355 60.1770 59.6668 60.0699 58.4640 59.6887 59.2811 59.3003 58.9912 Class 5 77.5960 100.4823 90.7192 87.4216 83.6861 89.4089 92.8384 87.8675 76.0326 Total 31.6180 77.6998 54.4022 42.9924 45.3053 47.8039 52.5783 47.6833 32.6365 Class 1 12.7776 14.1148 12.7631 13.0846 13.8053 13.4586 13.7918 13.6026 13.0405 Class 2 18.5342 19.4394 18.9854 18.4679 19.0415 18.6881 18.9692 18.8231 18.3540 Class 3 23.6826 24.1246 24.0850 23.9472 23.9646 23.9660 23.9642 23.9787 23.4508 Class 4 28.5677 30.0885 29.8334 30.0350 29.2320 29.8443 29.6405 29.6502 29.4956 Class 5 38.7980 50.2411 45.3596 43.7108 41.8430 44.7044 46.4192 43.9338 38.0163 Total 15.8090 38.8499 27.2011 21.4962 22.6526 23.9020 26.2891 23.8416 16.3182 Center for Applied Geographic Information Science, UNC Charlotte Page 27

Table 9. Total biomass and carbon of each canopy height class for each country in the study region (Total: total biomass/carbon density for all canopy heights) Cote Sierra Country Benin Cameroon Ghana Guinea Liberia Nigeria Togo d'ivoire Leone Class 1 79,777 239,031 58,206 257,432 2,343,312 409,530 10,650,097 1,900,679 10,302 Class 2 47,738 1,357,900 45,275 150,263 5,692,611 480,997 24,034,928 2,556,648 4,474 Total Class 3 21,877 2,575,596 64,548 88,516 5,919,776 371,172 22,614,218 2,206,931 1,693 Biomass (Mg) Class 4 8,970 5,844,246 132,804 135,811 5,179,836 555,632 27,108,108 2,801,289 1,997 Class 5 1,059 22,267,048 205,010 280,318 2,379,654 731,409 41,632,316 3,391,305 1,201 Total 159,422 32,283,823 505,845 912,341 21,515,191 2,548,743 126,039,670 12,856,853 19,669 Above-ground Below-ground Total Carbon (Mg) Total Biomass (Mg) Total Carbon (Mg) Class 1 39,889 119,516 29,103 128,716 1,171,656 204,765 5,325,049 950,340 5,151 Class 2 23,869 678,950 22,638 75,132 2,846,306 240,499 12,017,464 1,278,324 2,237 Class 3 10,939 1,287,798 32,274 44,258 2,959,888 185,586 11,307,109 1,103,466 847 Class 4 4,485 2,922,123 66,402 67,906 2,589,918 277,816 13,554,054 1,400,645 999 Class 5 530 11,133,524 102,505 140,159 1,189,827 365,705 20,816,158 1,695,653 601 Total 79,711 16,141,912 252,923 456,171 10,757,596 1,274,372 63,019,835 6,428,427 9,835 Class 1 30,315 90,832 22,118 97,824 890,459 155,621 4,047,037 722,258 3,915 Class 2 18,140 516,002 17,205 57,100 2,163,192 182,779 9,133,273 971,526 1,700 Class 3 8,313 978,726 24,528 33,636 2,249,515 141,045 8,593,403 838,634 643 Class 4 3,409 2,220,813 50,466 51,608 1,968,338 211,140 10,301,081 1,064,490 759 Class 5 402 8,461,478 77,904 106,521 904,269 277,935 15,820,280 1,288,696 456 Total 60,580 12,267,853 192,221 346,690 8,175,773 968,522 47,895,075 4,885,604 7,474 Class 1 15,158 45,416 11,059 48,912 445,229 77,811 2,023,518 361,129 1,957 Class 2 9,070 258,001 8,602 28,550 1,081,596 91,389 4,566,636 485,763 850 Class 3 4,157 489,363 12,264 16,818 1,124,757 70,523 4,296,701 419,317 322 Class 4 1,704 1,110,407 25,233 25,804 984,169 105,570 5,150,541 532,245 379 Class 5 201 4,230,739 38,952 53,260 452,134 138,968 7,910,140 644,348 228 Total 30,290 6,133,926 96,111 173,345 4,087,886 484,261 23,947,537 2,442,802 3,737 Center for Applied Geographic Information Science, UNC Charlotte Page 28

Table 10. Comparison of biomass and carbon estimated from datasets at coarse and fine spatial resolutions Cote Country Benin Cameroon Ghana Guinea Liberia Nigeria d'ivoire Above-ground Below-ground Area (km 2 ) Mean Biomass (Mg/ha) Total Biomass (Mg) Mean Carbon (Mg/ha) Total Carbon (Mg) Mean Biomass (Mg/ha) Total Biomass (Mg) Mean Carbon (Mg/ha) Total Carbon (Mg) Sierra Leone 1km 1km 20 1,968 22 81 2,217 87 5,320 1,346 3 90m 90m 19.1602 1,578.88 35.3334 80.6399 1,804.60 202.6031 9,109.29 1,024.60 2.2903 1km 1km 133.28 240.4799 218.6 117.8 137.8752 205.8138 186.3563 180.5031 107 90m 90m 83.2052 204.4731 143.1636 113.1378 119.2244 125.7998 138.3639 125.4823 85.8854 1km 1km 266,560 47,326,440 480,920 954,180 30,566,940 1,790,580 99,141,560 24,295,720 32,100 90m 90m 159,422 32,283,823 505,845 912,341 21,515,191 2,548,743 126,039,670 12,856,853 19,669 1km 1km 66.6400 120.2400 109.3000 58.9000 68.9376 102.9069 93.1782 90.2516 53.5000 90m 90m 41.6026 102.2366 71.5818 56.5689 59.6122 62.8999 69.1820 62.7412 42.9427 1km 1km 133,280 23,663,220 240,460 477,090 15,283,470 895,290 49,570,780 12,147,860 16,050 90m 90m 79,711 16,141,912 252,923 456,171 10,757,596 1,274,372 63,019,835 6,428,427 9,835 1km 1km 50.6464 91.3824 83.0680 44.7640 52.3926 78.2092 70.8154 68.5912 40.6600 90m 90m 31.6180 77.6998 54.4022 42.9924 45.3053 47.8039 52.5783 47.6833 32.6365 1km 1km 101,293 17,984,047 182,750 362,588 11,615,437 680,420 37,673,793 9,232,374 12,198 90m 90m 60,580 12,267,853 192,221 346,690 8,175,773 968,522 47,895,075 4,885,604 7,474 1km 1km 25.3232 45.6912 41.5340 22.3820 26.1963 39.1046 35.4077 34.2956 20.3300 90m 90m 15.8090 38.8499 27.2011 21.4962 22.6526 23.9020 26.2891 23.8416 16.3182 1km 1km 50,646 8,992,024 91,375 181,294 5,807,719 340,210 18,836,896 4,616,187 6,099 90m 90m 30,290 6,133,926 96,111 173,345 4,087,886 484,261 23,947,537 2,442,802 3,737 Togo Center for Applied Geographic Information Science, UNC Charlotte Page 29

Table 11. Summary of the shortest distance from mangroves to nearest city for each country in the study area Country Distance (km) Benin 3.5443 Cameroon 0.0000 Cote d'ivoire 0.6741 Ghana 0.4722 Guinea 0.0000 Liberia 0.3818 Nigeria 0.2071 Sierra Leone 4.6531 Togo 22.926 Center for Applied Geographic Information Science, UNC Charlotte Page 30

10. Figures Figure 1. Map of canopy heights in Africa (data source: NASA; spatial resolution: 1 km 1 km; our study area covers countries from Guinea to Cameroon) Center for Applied Geographic Information Science, UNC Charlotte Page 31

Figure 2. Map of canopy height in West Africa (data source: NASA; resolution: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 32

Figure 3. Histogram of canopy heights for the entire study area (cell size: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 33

Figure 4. Histogram of canopy heights for the entire study area (max height: 50 m; cell size: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 34

Figure 5. Map of canopy height classes in Guinea (resolution: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 35

Figure 6. Histogram of canopy heights in Guinea (cell size: 90 m 90 m) Figure 7. Histogram of canopy height classes in Guinea (cell size: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 36

Figure 8. Map of canopy height classes in Sierra Leone (resolution: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 37

Figure 9. Histogram of canopy heights in Sierra Leone (cell size: 90 m 90 m) Figure 10. Histogram of canopy height classes in Sierra Leone (cell size: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 38

Figure 11. Map of canopy height classes in Liberia (90 m 90 m resolution) Center for Applied Geographic Information Science, UNC Charlotte Page 39

Figure 12. Histogram of canopy heights in Liberia (cell size: 90 m 90 m) Figure 13. Histogram of canopy height classes in Liberia (cell size: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 40

Figure 14. Map of canopy height classes in Cote d Ivoire (spatial resolution: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 41

Figure 15. Histogram of canopy heights in Cote d Ivoire (cell size: 90 m 90 m) Figure 16. Histogram of canopy height classes in Cote d Ivoire (cell size: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 42

Figure 17. Map of canopy height classes in Ghana (spatial resolution: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 43

Figure 18. Histogram of canopy heights in Ghana (cell size: 90 m 90 m) Figure 19. Histogram of canopy height classes in Ghana (cell size: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 44

Figure 20. Map of canopy height classes in Togo (spatial resolution: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 45

Figure 21. Histogram of canopy heights in Togo (cell size: 90 m 90 m) Figure 22. Histogram of canopy height classes in Togo (cell size: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 46

Figure 23. Map of canopy height classes in Benin (spatial resolution: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 47

Figure 24. Histogram of canopy heights in Benin (cell size: 90 m 90 m) Figure 25. Histogram of canopy height classes in Benin (cell size: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 48

Figure 26. Map of canopy height classes in Nigeria (spatial resolution: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 49

Figure 27. Histogram of canopy heights in Nigeria (cell size: 90 m 90 m) Figure 28. Histogram of canopy height classes in Nigeria (cell size: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 50

Figure 29. Map of canopy height classes in Cameroon (spatial resolution: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 51

Figure 30. Histogram of canopy heights in Cameroon (cell size: 90 m 90 m) Figure 31. Histogram of canopy height classes in Cameroon (cell size: 90 m 90 m) Center for Applied Geographic Information Science, UNC Charlotte Page 52

Figure 32. Map of the mean biomass of mangroves for the study area in West Africa Center for Applied Geographic Information Science, UNC Charlotte Page 53

Figure 33. Map of the mean biomass of mangroves in Guinea Center for Applied Geographic Information Science, UNC Charlotte Page 54

Figure 34. Map of the mean biomass of mangroves in Sierra Leone Center for Applied Geographic Information Science, UNC Charlotte Page 55

Figure 35. Map of the mean biomass of mangroves in Liberia Center for Applied Geographic Information Science, UNC Charlotte Page 56

Figure 36. Map of the mean biomass of mangroves in Cote d'ivoire Center for Applied Geographic Information Science, UNC Charlotte Page 57

Figure 37. Map of the mean biomass of mangroves in Ghana Center for Applied Geographic Information Science, UNC Charlotte Page 58

Figure 38. Map of the mean biomass of mangroves in Togo Center for Applied Geographic Information Science, UNC Charlotte Page 59

Figure 39. Map of the mean biomass of mangroves in Benin Center for Applied Geographic Information Science, UNC Charlotte Page 60

Figure 40. Map of the mean biomass of mangroves in Nigeria Center for Applied Geographic Information Science, UNC Charlotte Page 61

Figure 41. Map of the mean biomass of mangroves in Cameroon Center for Applied Geographic Information Science, UNC Charlotte Page 62

Figure 42. Map of canopy height classes in Guinea (spatial resolution: 1 km 1 km) Center for Applied Geographic Information Science, UNC Charlotte Page 63

Figure 43. Map of canopy height classes in Sierra Leone (spatial resolution: 1 km 1 km) Center for Applied Geographic Information Science, UNC Charlotte Page 64

Figure 44. Map of canopy height classes in Liberia (spatial resolution: 1 km 1 km) Center for Applied Geographic Information Science, UNC Charlotte Page 65

Figure 45. Map of canopy height classes in Cote d Ivoire (spatial resolution: 1 km 1 km) Center for Applied Geographic Information Science, UNC Charlotte Page 66

Figure 46. Map of canopy height classes in Ghana (spatial resolution: 1 km 1 km) Center for Applied Geographic Information Science, UNC Charlotte Page 67

Figure 47. Map of canopy height classes in Togo (spatial resolution: 1 km 1 km) Center for Applied Geographic Information Science, UNC Charlotte Page 68

Figure 48. Map of canopy height classes in Benin (spatial resolution: 1 km 1 km) Center for Applied Geographic Information Science, UNC Charlotte Page 69

Figure 49. Map of canopy height classes in Nigeria (spatial resolution: 1 km 1 km) Center for Applied Geographic Information Science, UNC Charlotte Page 70

Figure 50. Map of canopy height classes in Cameroon (spatial resolution: 1 km 1 km) Center for Applied Geographic Information Science, UNC Charlotte Page 71

Figure 51. Map of distance from mangroves to nearest city Center for Applied Geographic Information Science, UNC Charlotte Page 72

Figure 52. Map of differences of mangrove area in Guinea for different data sources (USGS1: USGS data released in 2011 for the year of 2000; NASA2: estimates from this study) Center for Applied Geographic Information Science, UNC Charlotte Page 73

Figure 53. Map of differences of mangrove area in Sierra Leone for different data sources (USGS1: USGS data released in 2011 for the year of 2000; NASA2: estimates from this study) Center for Applied Geographic Information Science, UNC Charlotte Page 74

Figure 54. Map of differences of mangrove area in Liberia for different data sources (USGS1: USGS data released in 2011 for the year of 2000; NASA2: estimates from this study) Center for Applied Geographic Information Science, UNC Charlotte Page 75

Figure 55. Map of differences of mangrove area in Cote d Ivoire for different data sources (USGS1: USGS data released in 2011 for the year of 2000; NASA2: estimates from this study) Center for Applied Geographic Information Science, UNC Charlotte Page 76

Figure 56. Map of differences of mangrove area in Ghana for different data sources (USGS1: USGS data released in 2011 for the year of 2000; NASA2: estimates from this study) Center for Applied Geographic Information Science, UNC Charlotte Page 77

Figure 57. Map of differences of mangrove area in Togo for different data sources (USGS1: USGS data released in 2011 for the year of 2000; NASA2: estimates from this study) Center for Applied Geographic Information Science, UNC Charlotte Page 78

Figure 58. Map of differences of mangrove area in Benin for different data sources (USGS1: USGS data released in 2011 for the year of 2000; NASA2: estimates from this study) Center for Applied Geographic Information Science, UNC Charlotte Page 79

Figure 59. Map of differences of mangrove area in Nigeria for different data sources (USGS1: USGS data released in 2011 for the year of 2000; NASA2: estimates from this study) Center for Applied Geographic Information Science, UNC Charlotte Page 80

Figure 60. Map of differences of mangrove area in Cameroon for different data sources (USGS1: USGS data released in 2011 for the year of 2000; NASA2: estimates from this study) Center for Applied Geographic Information Science, UNC Charlotte Page 81