Craig Johnston U.S Geological Survey (603) 226-7843 cmjohnst@usgs.gov
The accuracy of the Census90 population data is bound by the accuracy of the U.S. Census and Canadian census data, census block group and enumeration area delineations, and by the even method by which the population was evenly distributed within each block group or enumeration area.
Stream density (attribute item Str_densit) was computed by dividing the reach length by the total area of the corresponding reach catchment. The reach length is computed by the sum length of the reach as represented in the 1:100,000-scale medium-resolution NHD. Thus, stream density is also somewhat scale dependent.
Average slope of the land surface is provided for each catchment in the attribute Slope_mean. Average slope was determined for each catchment by use of the 30-meter National Elevation Dataset (NED). For accuracy statements of NED, visit the NED web site at <http://gisdata.usgs.net/ned/>. Slope, calculated in GRID is the first derivative of the NED and is expressed as a percent equal to the "rise over run" (or maximum change in elevation divided by the horizontal distance between the centers of adjacent grid cells).
Nitrogen atmospheric deposition listed in the attribute Ndepo were from an existing spatial model of atmospheric deposition of total nitrogen in the northeastern United States (Ollinger and others, 1993). In the Ollinger model, total nitrogen deposition is a function of latitude, longitude, and total precipitation. Contoured data from Ollinger and others (1993) were extended into Canada using a shaded relief map as a general guide for extrapolating the atmospheric deposition contours for the New England SPARROW model area into Canada.
Municipal and industrial wastewater discharges of nitrogen or phosphorus (Nsewer, Psewer, Ppaper) for the New England SPARROW models were based on a USEPA permitted wastewater-discharge dataset (Steven Rubin, U.S. Environmental Protection Agency, written commun., January 2000). This dataset includes estimates of nutrient loads and other pollutants as average yearly or monthly estimates of total nitrogen or total phosphorus. Loads were estimated by USEPA using a methodology developed by NOAA to characterize wastewater loads to coastal waters and watersheds (National Oceanic and Atmospheric Administration, 1993). These estimates were based on a hierarchy of data sources. The highest priority source was derived from data from the USEPA's National Pollutant Discharge Elimination System (NPDES) program as reported in each facility's discharge monitoring report (DMR). When this information was not available, permitted discharge limits set for the facility were used. If neither monitoring nor permit pollutant data were available, engineering values, associated with either the facility's industrial activity or level of wastewater treatment, were used for the estimate (National Ocean and Atmospheric Administration, 1993). Spot checking of the estimates, with a more recent nitrogen dataset for Connecticut and a phosphorus dataset for western Vermont, showed some large discrepancies. However, the wastewater discharge estimates were used in the models because they were the best available information covering the entire New England model area. References:
National Oceanic and Atmospheric Administration, 1993, National Coastal Pollutant Discharge Inventory (NCDPI) point source methods document, 251 p.
Ollinger, S.V., Aber, J.D., Lovett, G.M., Millham, S.E., Lathrop, R.G., and Ells, J.M., 1993, A spatial model of atmospheric deposition for the northeastern United States: Ecological Applications, v. 3, no. 3, p. 459-472.
For most reaches in NE_NHD less than 30-meters in length, a corresponding catchment was not produced due to the 30-meter resolution of the catchment grid. In these cases where reach lengths were too short to generate catchments, the SPARROW model was set up to pass the nutrients to the next downstream NHD reach.
Modifications were made to the National Elevation Dataset (NED) prior to catchment delineation. These modifications were considered necessary because the drainage path defined by the NED surface rarely matches the 1:100,000-scale NHD. In many cases, the NHD streams and NED-derived streams run parallel with each other at some offset distance. If this offset distance is greater than one grid cell width, then part of the catchment may not be identified as being uphill of the NHD stream reach. This part of the catchment area is apt to be erroneously excluded from the delineated catchment.
Pre-processing stream network
The process required the preprocessing of the source stream-network data before the DEM hydrologic-conditioning process. The stream-network coverage was first projected into the same coordinate system as the NED data. The NHD Route.Drain route system was converted to a line feature class coverage. Any reaches without assigned flow relationships were deleted from the working stream network. Several modifications and corrections were applied to the NHD for use as the stream network in the models. For example, NHD headwater streams crossing the NRCS 12-digit subbasin divides were trimmed back. For further information on the New England SPARROW version of the NHD, see the Metadata documentation for the related coverage NENHD.
Stream burning
The method used to alleviate the displacement problem between NHD and NED drainage was to modify the NED data in a process that integrates the NHD vector drainage into the raster NED data, often referred as "stream burning" (Saunders, 2000). In creating catchments for the New England SPARROW model, a technique developed by Hellweger (1997) called Digital-Elevation-Model (DEM) surface reconditioning was used and modified. A series of computer algorithms combined in one Arc Macro Language (AML) program called AGREE, developed by Hellweger (1997), was used. AGREE "burns" into the DEM a "canyon", by using a specified vertical distance, which is subtracted from the DEM elevation grid cells beneath the NHD vector stream lines. The vertical exaggeration of the burnt in canyon is controlled by the AGREE "Sharp Drop / Raise Distance". Any sharp drop distance can be specified with AGREE. For the New England SPARROW model, a negative sharp drop distance of 10,000 meters was applied to insure that the burned in NHD stream flow path would remain after the filling process in subsequent processing steps for catchment delineation. Any very large negative sharp drop distance would produce the desired results.
Elevation smoothing
AGREE also "smooths" the elevation cells in the DEM around the corresponding NHD stream cell locations in the DEM within a specific buffer distance. The buffer distance is chosen by the operator using the AGREE program. Typically, the buffer distance is related to a common horizontal displacement error between NHD and NED-derived streams, which is seldom exceeded. In New England, the typical displacement error was found to be 80 meters or less. For the New England SPARROW model, 160 meters was specified as the buffer distance in AGREE. The 160-meter buffer distance creates an 80-meter buffer around the center of the NHD streamline. The smoothing process changes the DEM grid cell elevations within the buffered area to create a downward sloping gradient on both sides of the NHD stream towards the modeled canyon beneath the NHD streams. The steepness of the gradient slope within the buffer distance is controlled by the AGREE "Smooth Drop / Raise Distance" option. Any smooth drop distance can be specified with AGREE. For the New England SPARROW model, a smooth drop distance of 500 meters was specified with acceptable results.
Exaggerating walls
The hydrologic-conditioning of the NED data also includes a series of computer algorithms to force the DEM to recognize existing digital basin-divide data. For the New England SPARROW model, preliminary subbasin watersheds manually delineated from 1:24,000-scale topographic maps at the 12-digit hydrologic unit level were obtained from Natural Resources Conservation Service (NRCS). The NRCS 12-digit watershed data for New England will be incorporated into the planned nationwide Watershed Boundary Dataset (WBD). The process of conditioning DEM data to watershed boundaries is called "walling". The process uses an AML, written by Moore and Johnston, with a series of ARC/INFO Workstation GRID commands that vertically exaggerate DEM elevations by a specific constant at cells that correspond to the location of WBD ridgelines. Breaks in the walls were created at locations where the stream network crosses the WBD.
Creating catchments
The SPARROW stream-network data was rasterized to serve as the source grid for the ARC/INFO, GRID module command, WATERSHED (ESRI, 1999); where the stream cells represent the sinks above which, the catchments were determined with use of a flow direction grid. The flow direction grid was derived from a filled version of the hydrologic-conditioned DEM.
The areas of the catchments in square meters, square miles, and square kilometers were added to the catchment GRID attribute table.
References:
Environmental Systems Research Institute, Inc., (ESRI), 1999, Using ARC GRID with ARC/INFO: ESRI, Redlands, Calif., v. 2, 436 p.
Hellweger, Ferdi and Maidment, David. (1997). "AGREE-DEM Surface reconditioning system." University of Texas. <http://www.crwr.utexas.edu/gis/gishyd98/quality/agree/agree.htm>
Saunders, W., 2000, Preparation of DEMs for Use in Environmental Modeling Analysis, in Maidment, D., Djokic, D. [eds.], 2000, Hydrologic and Hydraulic Modeling Support: Redlands, Calif. p. 29-51 .
The atmospheric depostion grid was resampled to a 30-meter resolution to match the resolution of the catchment grid. Atmospheric depostion was overlaid with the catchment grid to sum for each catchment, the total nitrogen deposited from the atmosphere.
Reference: Alexander, R.B., Elliott, A.H., Shankar, Ude, and McBride, G.B., 2002, Estimating the sources and transport of nutrients in the Waikato River Basin, New Zealand: Water Resources Research, v. 38, no. 12, p. 1268-1290.