Recent concerns over pfiesteria in Eastern Shore rivers has focused attention on P runoff from cropland. You have been hired to analyze a proposed riparian vegetative buffer strategy for the upper portion of the Nanticoke watershed in southern Delaware. The "nanticoke" and "PERMANENT" mapsets within the GRASS "de_utm83" location contains the following raster maps:

• nanticoke.dem: digital elevation data (30-meter resolution)
• soils: (30-meter resolution)
• susx_lu92: 1992 land-use/land cover (30-meter resolution)
• spot94.r, spot94.g, spot94.b: registered SPOT HRV band files from August 30, 1994 (20-meter resolution)
• spot94.pan: registered SPOT panchromatic image from August 30, 1994 (10-meter resolution)

• delmar.hyd, ellendale.hyd, georgetown.hyd, greenwood.hyd 1:24,000-scale USGS hydrography
• de_susx.wtr: 1:100,000-scale TIGER water features
• de_susx.rd2, de_susx.rd3 and de_susx.rd4: primary, secondary and neighborhood roads
• de_susx.rr: railroad lines
• de_susx.bndy: county boundary

1. Set your region to match the DEM, but with a cell resolution of at least 60 meters. Create a hillshaded DEM of the study area. Create a nice 3D image of it with sufficient vertical exaggeration to show terrain variation. Okay, so much for the fun part.

2. Run r.watershed on the DEM to obtain a map of basins and drainage in the study area. (Note: in some cases r.watershed works better on DEM's smoothed with an r.neighbors neighbor-average filter to eliminate small basin anomalies. You may want multiply your DEM by some factor and then smooth it. Alternately, you might run r.fill.dir to create a new version of the DEM with local depressions filled in, although this module sometimes crashes.) It's easiest to run r.watershed interactively. You don't have a "depression map" so skip that. Provide names for the basin and accumulation maps to be created. Use an exterior (edge) basin threshold size of 1 square mile. Skip creating the other maps, and skip the "lumped parameter hydrologic/soil erosion model" query. Request "both" tabulations, basin-only and basin-and-upstream.

   Once the r.watershed module is done, check the basin and accumulation maps. Identify and delete the basins in the basin map which don't drain into the Nanticoke. Group the remaining basins by major tributary, reclassing them into 3 - 6 major basins (you will notice the sequence of basin ID's pretty much follows major drainages).

3. If you check these basin boundaries against your vector water map, you will see some streams crossing basin boundaries. Use r.digit to define areas of basins to be recategorized so stream features don't cross any basin boundaries. You're basically shooting the gaps between stream heads. r.digit is pretty easy to figure out: it creates a new raster map which r.mapcalc can use to correct the original basin map.

4. The vector TIGER water map de_susx.wtr is incompletely labeled and needs some cleanup. Create your own copy of it and run v.digit on it to label all unlabeled lines as category 1. Choose (none) for digitizing with the mouse. If you explore around v.digit's menus, you will get the hang of it pretty quickly. The bulk-label utility doesn't always work, so you will probably have to label unlabeled lines individually. Be patient: there are a number of small unlabeled segments you will have to zoom to find. Next, edit in any missing stream segments by breaking lines as necessary to create new nodes and snapping lines to nodes. After quitting v.digit, run v.support to build topology for your streams map. d.frame -e will clear your monitor window. When you are done, your steam map should have no breaks or inconsistencies.

5. Calculate the total acreage of each major basin in the upper Nanticoke watershed. Use the susx_lu92 map to calculate the total acreage in all developed uses, agriculture, forest and wetlands in each major basin.

6. Rasterize your water vector map at 20-meter resolution. Create 40- and 100-meter buffer zones around water features. Create a riparian zone map indicating all areas within 40 meters of any water plus areas with slopes greater than 5 percent within 100 meters of any water.

7. Calculate a Normalized Difference Vegetation Index (NDVI) map of the study area where NDVI is the normalized ratio 255 * ( spot94.r - spot94.g ) / (spot94.r + spot94.g ). Truncate all negative values to zero. Calculate mean NDVI for four land-use categories: all developed land, cropland, forest and wetland. Compare a histogram of NDVI values in the entire study area against a histogram of NDVI values in the just the riparian zone. Is there any evidence of extra vegetation as indexed by NDVI in riparian zones which would control P runoff into streams?

8. Calculate how much cropland is located within the riparian zones. Suppose the government idles all cropland within riparian zones, paying farmers $50 per acre annually to establish and maintain vegetative buffers in these areas. What would be the annual cost of this program in the study area basins draining into the Nanticoke?

Write up a consultant's report in HTML summarizing your procedures and findings, and including relevant maps. Send me an e-mail to tell me the URL.