Vulnerability to watershed erosion and coastal deposition in the tropics

Over half of the global population is projected to live in the tropics by 2050. Sustainable land development will be challenged by enhanced sediment erosion and deposition, which can negatively impact water quality and ecosystem services in inland and coastal waterways. Existing erosion assessments treat watersheds and coastal zones separately, but we connect them in a two-part vulnerability index to watershed erosion and coastal deposition at 0.0004° (~ 45 m) resolution throughout the tropics. We use open-source datasets and a simple, GIS-based method geared toward tropical, novice end-users. Part 1 of the index reveals a majority of the tropics is vulnerable to erosion. Vulnerability is highest where there are co-occurrences of earthquakes, steep slopes, and high precipitation such as the Caribbean and Southeast Asia. In Part 2, we assess erosion vulnerability at 4 watersheds and include their coastal systems, which can enhance or diminish vulnerability of the entire system to coastal deposition.


Introduction
The following Supplemental Material contains: detailed methods and rationale behind the creation of the EVI and EDVI, step-by-step instructions to generate all EVI and EDVI files in ArcGIS as well as the EVI and EDVI of any watershed in the tropics, a figure describing how to calculate the Coastal Protection variable, a table with detailed information regarding all used variables, and a table demonstrating characteristics of the 4 case studies chosen.

Supplemental Methods
Several erosion triggers/enhancers either did not exist at a global resolution (e.g. tidal range, watershed connectivity, and various human development activities) or are modeled products with questionable accuracy (e.g. rainfall erosivity, presence of dams or other engineering structures, and soil type) and were not included in the EVI. Accurate tidal range at this scale was not possible and is a lesser driver of erosion in most cases when compared to watershed erosion. To account for human infrastructure we used a dataset called Global Urban Footprint (Esch et al., 2018) developed by the German Aerospace Center (DLR) at their Earth Observation Center. Global Urban Footprint identifies 'settlements', not roads or other similar transportation infrastructure.
Of the 10 EVI variables (ultimately 7 Risk Factors), 8 global raster datasets were available and the remaining 2 were in vector format (point or polygon data) (Supplemental Table 1). All datasets were downloaded from open access servers and are free for public use (Supplemental Table 1). The point-based vector dataset (mining) was rasterized at a resolution of 0.01° (Supplemental Table 1). The polygon-based vector dataset (bedrock lithology) was rasterized at the same resolution downloaded (0.67° x 0.67°) (Supplemental Table 1). All rasters were clipped to the tropical band (23.5° N to 23.5° S) and converted to a geographic projection system (GCS WGS84) before analysis.
The values for each variable were reclassified to Risk Factors to demonstrate erosional risk with values ranging from 1 to 5 (Table 1). Risk Factor rasters were resampled using the nearest neighbor method (no new cells generated) to 0.004° x 0.004° grid cell resolution. Land-based datasets (land cover type, AGMD, and bedrock lithology) lacked data over the ocean and had a coarser resolution along the coastline as a result. These Risk Factors' null values were reclassified to Very High Risk Factor thus changing all cells over the ocean, and some on the coastline, to Very High Risk Factor. The coastline generally lacks vegetation and is higher energy, making it prone to erosion, hence the Very High Risk Factor assignment. This does not create much false data because of the high resolution of the three datasets (Supplemental Table 1). The combined AGMD Risk Factor raster was created by taking the maximum value at a given grid cell from each of the 4 contributing rasters. This was done to limit the individual contributions of each variable.
Step by Step Instructions to generate the EVI or EDVI Step by Step Instructions to generate a watershed EDVI from EVI values. 2. Open ArcCatalog and create a new shapefile by right clicking then select "New" and "Shapefile" a. Name the file the name of your watershed (e.g. exwtshd.shp), select "polygon" for format. b. Click the "edit" button and choose a geographic WGS84 projection for the coordinate system. Then click "ok", and close ArcCatalog.
3. In ArcDesktop add the newly created shapefile in this example 'exwtshd.shp' 4. Using the editor toolbar click the "editor" button and click "start editing" a. A dialog box will appear, select 'exwtshd.shp' and click "ok". c. Copy and Paste your watershed, it will prompt as to which layer to paste into and ensure that it is your new shapefile 'exwtshd.shp'. d. Click "Editor" again and select "stop editing" and click "yes" to save your edits.
c. For "Output Extent" select 'exwtshd.shp'. d. Check the box for "Use Input Features for Clipping Geometry". e. Use "Output Raster Dataset" to specify the location and name for the new file (e.g. exwtshdevi).
f. Check the box for "Maintain Clipping Extent" and click "ok".
6. Turn the new raster into integer. Use the 'Int' tool a. In "Arc Toolbox" it is under "Spatial Analyst"  "Math"  "Int" b. A dialog box will appear, for "Input Raster" select 'exwtshdevi'. c. Use "Output Raster" to specify the location and name for the new file (e.g. int_exwtshdevi) and Click "ok". ii. Use "Output Excel File" to specify the location and name for the new file (e.g. exwtshdevitbl.xls) and Click "ok". 11. In ArcDesktop use the "Measure" button to draw a line perpendicular from your watershed outlet to 10 kilometers from your previously determined coastline. a. Tip: Using the draw toolbar draw this line and then adjust it to 10 km. b. Using the 3D Analyst Toolbar select the Mean Coastal Slope raster ('g19slope') and click the "Interpolate Line" button (green square with a zig-zag line above it).
i. Draw a line with two points just beside your line.
ii. Click the "Profile Graph" button (button with a graph on it) iii. This will generate a profile of your slope. Right click the profile and click "export" iv. Select the "Data" tab and click "preview". Copy the data and paste it into a new sheet in your excel table.
c. In Excel write step copy the following step and paste it two cells below the last cell with data in the B Column in your excel sheet i. =average(B2:B25) ii. NOTE: B25 is where our data ends in the example given adjust to the end of your real data.
iii. Write the value generated in cell D21 in your main sheet.
12. The EDVI value is now calculated properly in G18 and H18.
i. G18 represent the Vulnerability Class Value and H18 represents the Vulnerability Class; 1 = Very Low, 2 = Low, 3 = Medium, 4 = High, 5 = Very High Step by Step Instructions to generate a global or tropical or watershed EVI.
For each individual EVI variable follow these steps.
1. Use the Project tool to reproject the raster to a constant geographic projection. a. In "Arc Toolbox" "Data Management"  "Projections and Transformations"  "Project". b. For "Input Dataset or Feature Class" select the target raster c. For "Output Coordinate System" specify the same geographic projection for all files for our study we used 'GCS_WGS84'. d. Select the file path and execute.
2. Use the Resample Tool to resample the data to at least 0.004° x 0.004°. a. In "Arc Toolbox" "Data Management"  "Raster"  "Raster Processing"  "Resample". b. For "Input Raster" select the projected raster.
c. For "Output Cell Size" 'X' type in 0.004 and 'Y' type in 0.004 d. For "Resampling Technique" select "Nearest" e. Select the file path and execute.
3. Use the "Reclassify" Tool to reclassify the values to Risk Factor values in Table   1.
Cleanup, Tips, and Subtracting Lakes 1. Subtract lakes from EVI raster a. Note: Depending on the resolution of the EVI raster it may need to be resampled in order to obtain higher resolution coastline data. b. In "Arc Toolbox" "Spatial Analyst"  "Extraction"  "Extract by Mask" c. For "Input Raster" select the created EVI raster file. d. For "Input Raster or Feature Layer" select the shapefile 'tropics_lknull.shp' (this was created using (ESRI, 2018)). e. Select the file path and execute.
2. Turn the new raster into integer. Use the 'Int' tool a. In "Arc Toolbox" it is under "Spatial Analyst"  "Math"  "Int" b. A dialog box will appear, for "Input Raster" select 'exwtshdevi'. c. Use "Output Raster" to specify the location and name for the new file (e.g. int_exwtshdevi) and Click "ok". d. The generated raster can be exported and develop a histogram as in Step 7 of the above Step by Step Instructions to generate a watershed EDVI from EVI values.
3. Tips a. We found that resolutions above 0.004° x 0.004° would often cause errors with "Raster Calculator" or other tools while generating the EVI or even some Risk Factors. More work will be done on this in the future.

If using this Step By
Step Instruction Guide please use the citation for this publication.
Supplemental Figure 1. Risk Factor distribution for the Coastal Protection variable. Coastlines are idealized to be compared to actual coastlines in order to categorize risk. Figure S1 was created in Adobe Illustrator www.adobe.com/products/illustrator.html