You are here: Skip Navigation LinksHomeEcosystemsCoral ReefsSummit to Sea Characterization of Coastal WatershedsPuerto Rico and U.S. Virgin Islands

Back to Summit to Sea Project page

Summit to Sea Characterization of Coastal Watersheds
(Puerto Rico and U.S. Virgin Islands)

Produced in collaboration with the World Resources Institute

Summary

Puerto Rico and U.S. Virgin Islands mapAlteration of the natural landscape for development, road construction, or agriculture can have adverse impacts on coral reefs through increased delivery of sediment and pollution to coastal waters. The threat associated with land clearing is higher in areas of steep relief, intense precipitation, and where soils are more erosive in nature. This threat is often evaluated through application of the Revised Universal Soil Loss Equation (RUSLE) developed by the US Department of Agriculture. RUSLE is useful for examining erosion in many agricultural areas, but is not well suited to the very steep and rutted environments of the US Virgin Islands where road construction accounts for most erosion. A simplified version of RUSLE (using slope, land-cover, precipitation and soil characteristics) was used to assess and compare watersheds on each island, based on slope, precipitation, land cover, soils. This study uses several spatial and statistical techniques to characterize watersheds across the US Virgin Islands and Puerto Rico with regard to relative erosion rates and the threat of land-based sources of sediment and pollutant delivery to coastal waters. The spatial analysis of land-based sources of threat to coral reefs has several components:

  1. Delineation of watersheds (basins) based on a hydrologically-corrected digital elevation model (DEM.) These basins reflect all land areas discharging to a single coastal location (outflow point).
  2. Analysis of relative vulnerability of land to erosion based on slope, precipitation, and the soil characteristics.
  3. Analysis of the relative erosivity of land given current land cover.
  4. Estimation of relative sediment delivery at each outflow point.
  5. Estimation of relative threat to benthic habitats, based on distance to, and sediment delivery of the nearest outflow points.
Cartographic Model
This cartographic model depicts the process to map land-based threats (by watershed) to coral reefs from mainland Puerto Rico. The original inputs are elevation, landcover, soils and precipitation.

Watershed (basin) delineation

Hydrologic modeling allows one to develop a series of data sets from a single elevation data set (derived data sets include slope, flow direction, flow accumulation, basins, and outflow points (river mouth). All modeling was performed using ESRI ArcGIS v. 9.1, at 30m resolution in UTM projection zone 20N, WGS84.

  1. Hydrologic modeling for both the USVI and PR are based on the 30m resolution National Elevation Dataset (NED) from US Geological Survey (USGS.)
  2. For the USVI, the digital elevation data (DEM) was hydrologically corrected: the location of known rivers and streams and guts were subtracted from the elevation data. The vector data set reflecting guts from DPNR was converted to a 30 m grid to match the DEM, and coded as -15. This value is used to adjust the original value in the DEM, forcing the hydrologic flow into these "depressed areas." This improves watershed delineation particularly in flat coastal areas and through wetlands. For Puerto Rico, the raw (uncorrected NED) was the basis for the flow modeling.
  3. FLOWDIRECTION is run on the DEM. This results in a data set showing which of eight directions every given 30m cell flows. (coded as integer 1,2,4,8,16,32,64, and 128, with 1 being due East.)
  4. FLOWACCUMULATION is run on the FlowDIR grid. This gives an integer grid reflecting the number of cells flowing into any given cell. This layer can be used to identify rivers (gut) locations. Any cells above a given value (such as 200) represent a potential gut location.
  5. The BASIN command is run on the FlowDirection grid. This produces a dataset of all grid cells discharging to a single outflow point. There is a unique identified (BASIN_ID) within each file to identify the basin.
  6. A dataset reflecting outflow points (river or gut discharge) is created from the point of maximum flow within each basin. These are identified by locating the cell in the basin that matches the maximum FlowAccumulation grid value. These cell are assigned the value of the BASIN_ID to link each outflow point to its basin.
  7. The resulting set of basins and pour points is valuable for summarizing the spatial threat indicators developed in subsequent steps of this analysis.

Vulnerability to Erosion

Physical factors, such as the slope of the land, the texture of the soil, and the precipitation regime influence erosion in an area. Parts of Puerto Rico and many parts of the USVI in particular, are very steep and erosion-prone. In addition, the nature of the soil and intense rainfall events promote severe erosion in these areas. Erosion can be extreme in exposed areas (cleared for a road or residential construction, or where soil is exposed due to cropping patterns or agricultural cycle.)

Asimple indicator of the erosivity of the land is used for this analysis, based on physical factors of the location (slope, precipitation, and a soil characteristic called K-factor, which reflects the erodibility of the given soil type.) This indictor does not consider the current land cover or land use. Rather, it provides an overall indicator of erosion-prone areas, and areas where development / land conversion / road construction should be avoided.

Inputs:

  1. Slope (percentage) - for each 30m grid cell, derived from the (raw) DEM (without hydrologic correction.)
  2. Precipitation during the peak rainfall month (in millimeters) - The long-term average monthly precipitation values for climate stations across the area were downloaded from NOAA's National Climate Data Center (NCDC). The mean monthly precipitation for the peak rainfall month was selected for each point location, and a 30m resolution grid was interpolated using an inverse distance weighting interpolation method. This variable was chosen because it is more indicative of the rainy season and more extreme events during the year.
  3. Soil erodibility factor (K-factor) - the K-factor was obtained from the SSURGO database of the USDA.

Equation 1:

Vulnerability = Slope(%) x precip (mm for peak rainfall month) x soil k-factor

The resulting grid reflects an estimate of relative vulnerability of land to erosion. It is a relative (unitless) value. Summary statistics have been derived from this 30m resolution grid for both hydrologic basins and official watersheds.

Relative Erosion Potential (REP)

Agriculture and other land use activities far inland can have an adverse impact on coral reefs through the increased delivery of sediment and pollution to coastal waters. Particularly in steep areas of the USVI and Puerto Rico, land cover change can increase erosion and ultimately sediment delivery to coastal waters. A watershed-based analysis of land-based sources of pollution (LBS) was implemented to develop a preliminary estimate of this threat.

Analysis Method
Watersheds are an essential unit for analysis, since they link land areas with their point of discharge to the sea. We have implemented a watershed-based analysis of sediment and pollution threat to coral reefs. This analysis incorporates land cover type, slope, soil erodibility factor (k-factor), and precipitation for all land areas, using a simplified version of the Revised Universal Soil Loss Equation (RUSLE) in order to estimate relative erosion rates for each 30m resolution grid cell. These relative erosion estimates are summarized by basin. Since not all erosion makes its way to the river mouth, sediment delivery ratios (based on watershed size) were applied in order to estimate relative sediment delivery at the river mouth. It should be noted that relative erosion rates and sediment delivery are being used as a proxy for both sediment and pollution delivery.

Model Implementation

Step 1) The first step of the analysis involves estimating likely relative erosion rates for each 30 m resolution grid cell using a modified, simplified form of the Revised Universal Soil Loss Equation (RUSLE). Information on slope, land cover type, precipitation, and soil porosity were integrated to develop an indicator of relative erosion potential (REP) for all land areas in Puerto Rico and the USVI.

Inputs: (REP relies upon four input data sets, three of which were described under the analysis of vulnerability to erosion, above.)

  1. Slope (percentage) - for each 30m grid cell, derived from the (raw) DEM (without hydrologic correction.)
  2. Relative erosion rate by land cover type (eros). Relative erosion rates for each land cover type were determined from previously published work and calculated for each land cover data set. For the USVI, data from UVI/CDC reflecting vegetation cover was used as the base. For Puerto Rico, two different datasets were used. A classification of Landsat imagery for 2000 by Jennifer Gebelein of Florida International University was used for a more detailed analysis, while NASA 1990 and 2000 Geocover, developed by MDA Federal was used in order to assess the effects of land cover change on erosion potential between 1990 and 2000. Relative erosion potential is related to vegetative cover, so land cover types such as developed/urban have much higher relative erosion rates than forest or secondary vegetation. Land cover categories were reclassified to relative erosion rates, ranging from 15 for forest to 220 for barren land (shown below for each dataset).
  3. Precipitation during the peak rainfall month (in millimeters) - Long-term average monthly precipitation values for the peak rainfall month of the year is an interpolated grid based on data for climate stations from NOAA's National Climate Data Center (NCDC). This variable was chosen because it is more indicative of the rainy season and more extreme events during the year.
  4. Soil erodibility factor (K-factor) - the K-factor was obtained from the SSURGO database of the USDA.
Table 1. Land Cover and Associated Relative Erosion Rates (USVI/CDC)
Land Cover Type Relative Erosion Rate
Water Body 5
Forest 15
Woodland 15
Shrubland 45
Hedge 50
Mixed Vegetation 50
Mangrove 80
Salt Flat 80
Salt Pond 80
Swamp 80
Pasture 120
Grassland 125
Cropland 200
Developed 210
Rock Pavement 210
Beach 220
Barren Land 220
Table 2. Land Cover and Associated Relative Erosion Rates (NASA GeoCover)
Land Cover Type Relative Erosion Rate
Forest, Deciduous 15
Forest, Evergreen 15
Shrub/Scrub 50
Grassland 125
Barren 220
Urban/Built up 210
Agriculture 200
Wetland/Permanent 80
Wetland/Mangrove 80
Water 5
Table 3. Land Cover and Associated Relative Erosion Rates (J. Gebelein)
Land Cover Type Relative Erosion Rate
Water 5
Urban 210
Crop/Natural Vegetation 120
Cropland 200
Shrubland 50
Pasture Land 125
Grassland 125
Evergreen 15
Mixed Forest 15
Mangrove 80

Equation 2:

Relative Erosion Potential (REP) = Slope(%) x precip (mm for peak rainfall month) x soil k-factor x relative erosion rate for land cover type / 1000

The grid is divided by 1000 and converted to integer for a better data range and easier display.
The analysis was implemented in ESRI's ArcMap.

Two indicators indicative of erosion within the watershed were calculated for each basin:
mean REP for the basin (an indicator of average erosion rates for the basin) (REP_MEAN), and total relative erosion within the basin (REP_SUM).

Step 2) An indicator of relative sediment delivery at the river mouth was estimated by multiplying total relative erosion in the basin (REP_SUM) by the sediment delivery ratio (SDR) for the basin, which is a function of watershed size. SDR reflects the percentage of erosion within the basin (REP_SUM) which reaches the river mouth.

Equation 3:

Sediment Delivery Ratio (SDR) = 0.41 x basin area (in sq km)^-0.3.

Equation 4:

Sediment delivery at the river mouth (SED_DELIV) = REP_SUM x SDR

Using the Sediment Delivery at the water mouth, an estimate of threat to benthic habitats from land-based sources of sediment was calculated (benthic_threat). The point density tool was used to create this continuous benthic threat layer, using the sediment delivery at the outflow point as input, and a kernel radius of 5 km for Puerto Rico, 1km for USVI

Summary Indicators

Summary statistics at the basin or watershed level can be developed using the Zonal Statistics Tool in Spatial Analyst Toolbox in ArcGIS. A basin layer is used as the data set reflecting zones, while the indicator (vulnerability or REP) is used as the data set in which the function is performed.

Shapefile attributes:

  1. BASIN_ID - Watershed or basin ID
  2. AREA_M2 - basin area in m2
  3. AREA_KM - basin area in km2
  4. VULN_MEAN - mean vulnerability to erosion for the basin
  5. SDR - sediment delivery ratio (percentage of erosion (REP_SUM) reaching the river mouth
  6. MAX_PREC - mean precipitation (mm) in the basin during the peak rainfall month
  7. MAT_PREC - mean total annual precipitation (mm) in the basin
  8. VULN_MEAN - mean vulnerability to erosion for the basin
  9. REP_MEAN - mean relative erosion potential (REP) for the basin - (Puerto Rico - using J. Gebelein Landcover c1999; USVI - CDC landcover)
  10. REP_SUM - sum of REP for the basin - (using CDC landcover for relative erosion potential)
  11. SED_DEL - relative sediment delivery to the pour point (river mouth) - (Puerto Rico - using J. Gebelein Landcover c1999; USVI - CDC landcover to calculate relative erosion potential)
  12. ROAD_DENS - average basin road density, where density is the kernel density of the road coverage
  13. ROAD_PERC - amount of road area/basin area
  14. REP90_MEAN - mean relative erosion potential (REP) for the basin - (using 1990 Geocover data to determine relative erosion potential)
  15. REP90_SUM - sum of REP for the basin - (using 1990 Geocover data to determine relative erosion potential)
  16. SED_DEL90 - relative sediment delivery to the outflow point - (using 1990 Geocover data to determine relative erosion potential)
  17. SED_DEL00 - relative sediment delivery to the outflow point - (using 2000 Geocover data to determine relative erosion potential)
  18. REP_CHANGE - change in relative erosion potential between 1990 and 2000 (using Geocover datasets)
  19. SED_CHANGE - change in sediment delivery between 1990 and 2000 (using Geocover datasets)

Download the data (and metadata)

The USVI and Puerto Rico zipfiles include FGDC metadata the following data: (All raster data are in geotiff format, UTM Zone 20N, WGS84) Metadata for individual files, as well as a universal metadata file.

Data Listing - USVI

  1. Usvi_basins: Watershed boundaries (shapefiles), or hydrological units and their physical and erosion related characteristics;
  2. Usvi_bathy: Estimated depth represents the relative bathymetry of the U.S. Virgin Islands shallow waters based on Landsat imagery;
  3. Usvi_bathy: Estimated depth represents the relative bathymetry of the U.S. Virgin Islands shallow waters based on Landsat imagery;
  4. Usvi_benthic_threat: Potential threat of sediment delivery and land-based sources of pollution to coral ecosystems derived from land cover (LC) database (Created by spectral analysis of consistently orthorectified Landsat Thematic Mapper (TM) imagery), USGS National Elevation Dataset, NRCS Soils Database and NOAA Monthly Surface Data;
  5. Usvi_elev: Digital Elevation Model (DEM) created from the existing U.S. Geological Survey, National Elevation Dataset (NED) by filtering artifacts, converting to the NAD83 datum, edge-matching, and sliver filling of missing data;
  6. Usvi_slope: topographic slope for Puerto Rico, derived from the U.S. Geological Survey's National Elevation Dataset, calculated by the Slope function in ArcGIS;
  7. Usvi_eros00: Relative erosion rate by land cover type 2000 were determined from a table;
  8. Usvi_eros90: Relative erosion rate by land cover type 1990 were determined from a table;
  9. Usvi_kffact: Erodibility of soils (Kffact) based upon soil composition derived from the SSURGO soils data base;
  10. Usvi_matprecip: Mean annual precipitation (mm), derived from climate data collected at weather stations from 1990-2000 (NOAA Monthly Surface Data, NCDC);
  11. Usvi_maxprecip: Maximum monthly precipitation, derived from monthly precipitation for all available weather stations in U. S. V. I.;
  12. Usvi_outflow: Watershed outflow points, (areas of maximum flow accumulation within hydrological basins) calculated from FLOWDIRECTION and FLOWACCUMULATION which were derived from the Digital Elevation Model (DEM);
  13. Usvi_rep00: Relative Erosion Potential - 2000 (an indicator of sediment and pollution runoff) derived from USGS National Elevation Dataset, NRCS Soils Database and NOAA Monthly Surface Data;
  14. USVI_rep90: Relative Erosion Potential - 1990 derived from USGS National Elevation Dataset, NRCS Soils Database and NOAA Monthly Surface Data;
  15. Usvi_vuln: Relative vulnerability to erosion - 2000 derived from land cover (LC) database (Created by spectral analysis of consistently ortho-rectified Landsat Thematic Mapper (TM) imagery), USGS National Elevation Dataset, NRCS Soils Database.

Data listing - Puerto Rico

  1. Pr_basins: Watershed (basin) boundaries (shapefiles), the hydrological units and their physical and erosion related characteristics;
  2. Pr_bathy: Estimated bathymetry of the shelf derived from Landsat images, using a semi-automated method (Smith and Shapiro, 2003);
  3. Pr_benthic_threat: Calculated potential threat of sediment delivery and land-based sources of pollution to benthic habitats;
  4. Pr_elev: A filtered, Digital Elevation Model (DEM), subset from the U.S. Geological Survey's National Elevation Dataset;
  5. Pr_eros00: Relative erosion rate by land cover type 2000 were determined from a table 1;
  6. Pr_eros90:Relative erosion rate by land cover type 1990 were determined from table 2;
  7. Pr_kffact: Erodibility of soils (Kffact) based upon soil composition derived from the SSURGO soils data base;
  8. Pr_matprecip: Mean annual precipitation (mm), derived from climate data collected at weather stations from 1990-2000 (NOAA Monthly Surface Data, NCDC);
  9. Pr_maxprecip: Maximum monthly precipitation, derived from monthly precipitation for all available weather stations in Puerto Rico;
  10. Pr_outflow: Watershed outflow points, (areas of maximum flow accumulation within hydrological basins) calculated from FLOWDIRECTION and FLOWACCUMULATION which were derived from the 11. Digital Elevation Model (DEM);
  1. Pr_slope: topographic slope for Puerto Rico, derived from the U.S. Geological Survey's National Elevation Dataset;
  2. pr_rep00: Relative Erosion Potential - 2000 (an indicator of sediment and pollution runoff) derived from USGS National Elevation Dataset, NRCS Soils Database and NOAA Monthly Surface Data;
  3. pr_rep90: Relative Erosion Potential - 1990 derived from USGS National Elevation Dataset, NRCS Soils Database and NOAA Monthly Surface Data;
  4. pr_benthic_threat: Potential threat of sediment delivery and land-based sources of pollution to coral ecosystems derived from land cover (LC) database (Created by spectral analysis of consistently 16. orthorectified Landsat Thematic Mapper (TM) imagery), USGS National Elevation Dataset, NRCS Soils Database and NOAA Monthly Surface Data;
  5. pr_vuln: Relative vulnerability to erosion - 2000 derived from J. Gebelein Landcover (Created by spectral analysis of consistently orthorectified Landsat Thematic Mapper (TM) imagery), USGS National Elevation Dataset, NRCS Soils Database.

Additional data available from the World Resources Institute's Electronic Atlas: "Land-based sources of threats to coral reefs in the U.S. Virgin Islands"

Project Partners

Back to Summit to Sea Project page