Watershed Delineation Using The National Elevation Dataset and Semiautomated Techniques


Jay R. Kost, Raytheon ITSS*
Glenn G. Kelly, USGS

Abstract

Federal, State, and local agencies have realized that currently available hydrologic units are not of sufficient scale for many applications. An interagency effort is under way to subdivide hydrologic units into smaller units called watersheds and subwatersheds. The National Elevation Dataset contains the best available elevation data merged into a seamless database for the entire United States. These data can be readily used to delineate watershed and subwatershed basins. Recently developed ArcView tools facilitate the semiautomatic delineation of watersheds and subwatersheds.


*Raytheon ITSS, Work performed under U.S. Geological Survey contract 1434-CR-97-CN-40274.

Any use of Trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.




Introduction

In October 2000 a significant government policy was announced related to how Federal, State, tribal, and local governments would manage natural resources in the future. Titled the "Unified Federal Policy for a Watershed Approach to Federal Land and Resource Management", this policy takes a cooperative approach to managing natural resources on a watershed basis. The U.S. Departments of Agriculture, Interior, Commerce, Defense, and Energy; the Environmental Protection Agency; Tennessee Valley Authority; and the Army Corps of Engineers are some of the agencies involved in carrying out the policy.

In part because of this policy, there has been an increasing need for watershed and subwatershed delineation information and data. In some areas of the country delineations do not exist and in many cases those that do exist are not at a large enough scale to be useful for many applications. Consistency in delineation methods and format has also varied in the past; sometimes making comparison of delineated areas difficult because of variations in size or other delineation criteria.

A multi-agency effort is under way to develop the Watershed Boundary Dataset (WBD). This dataset, being managed in part by the Advisory Committee on Water Information and the Federal Geographic Data Committee, will be a seamless, nationwide dataset consisting of watershed and subwatershed delineations in a consistent, spatially referenced, digital format.

For purposes of consistency, the U.S. Geological Survey (USGS), Natural Resource Conservation Service (NRCS), and many other groups and agencies have combined distinct, but often similar, guidelines related to watershed delineation. This common set of guidelines, "Federal Standards for Delineation of Hydrologic Unit Boundaries", has become the primary document used in developing watershed boundaries. The NRCS is the lead agency and is responsible for watershed and subwatershed certification of the WBD. A status graphic on the certifying office's web site shows the current status of delineations for the conterminous United States.

Several projects are using the guidelines to delineate watershed and subwatershed boundaries. In some cases these efforts have been under way for several years and significant progress has been made. In other areas little progress has been made and there is a growing desire to find a relatively quick, cost-effective, and accurate way to delineate watersheds and subwatersheds.

Work toward developing the EDNA (Elevation Derivatives for National Applications) database has included the development of tools for semiautomated delineation of watershed and subwatershed boundaries. Delineations created with these tools use hydrologic derivatives created from the National Elevation Dataset (NED).


National Elevation Dataset (NED)

The National Elevation Dataset (NED) is a seamless, digital elevation model (DEM) with coverage of the entire United States, Puerto Rico, U.S. Virgin Islands, and many U.S. territories. The NED was developed from individual 7.5-minute DEM's by the USGS with the goals of improving data access and minimizing the need for pre-application processing. Figure 1 shows a shaded-relief rendering of the NED for the conterminous U.S.

The NED is updated bimonthly to provide the best available data to the user community. It is constructed with a consistent datum, elevation unit, and projection. NED data are also edge matched and filtered (Oimoen, 2000) to ensure seamless continuity and minimize artifacts. Figure 2 illustrates the flow patterns across a DEM surface before and after the filtering process.

Currently, the conterminous U.S., Hawaii, Puerto Rico, and Virgin Islands are distributed with a one-arc-second cell size (approximately 30 meters) while Alaska is distributed with a two-arc-second cell size. Future plans incorporate the inclusion of new data sources that will further increase data quality and resolution. The NED data source index is a useful tool for monitoring NED updates, including date of update, resolution, and production method.


Elevation Derivatives For National Applications (EDNA)

The Elevation Derivatives for National Applications (EDNA) project, formerly called NED-H, is a multi-agency effort that takes advantage of the seamless and filtered characteristics of the NED. The goals of the project are to create a hydrologically conditioned NED dataset, systematically create hydrologic derivatives, and vertically integrate these data with other spatial datasets such as the National Hydrography Dataset (NHD).

EDNA data development occurs in three stages, the first of which uses semi-automated techniques to create preliminary derivative data from the unconditioned NED elevation data. The second stage uses the stage 1 data, within a set of ArcView tools, to create preliminary watersheds and subwatersheds and to identify and flag discrepancies between the derivative data and existing data sets that portray true hydrologic features. Data and information created in stage 2 are used in stage 3 to develop a hydrologically conditioned NED dataset. These data are then used to create a more hydrologically correct derivative database. This final step in stage 3 processing results in vertical integration of the final EDNA data with the NHD and other datasets.

Stage 1 processing is being completed by the National Weather Service's Severe Storms Laboratory and the USGS. This stage 1 processing uses well tested GIS techniques to create derivative data, including initial synthetic streams and reach catchments (drainage areas corresponding to each stream in the synthetic streamline coverage). Completion of this database is expected by the fall of 2001. A stage 1 status graphic is available from the EDNA web site.

Data processing for stage 2 utilizes the local expertise of many cooperators across the country to help identify and delineate watershed and subwatershed boundaries. Local expertise is also an important component in the discrepancy identification and flagging process. Cooperators include Federal, State, and local agencies, private organizations, and academia. The watershed and subwatershed delineations created in stage 2 are used by many of these cooperators for various applications and will be used in efforts to develop the WBD dataset. A status graphic for stage 2 processing is also available.

Tool development for stage 3 processing is directed at creating algorithms for quickly and efficiently modifying the DEM data so flow derivatives more accurately represent actual flow. Some of the utilities developed thus far use algorithms for modifying DEM flow direction so the resultant synthetic streams more closely match true flow conditions. Other developments in stage 3 tools include utilities for centerline flow enforcement within water bodies and obstruction (e.g. highway overpasses) modification to allow flow.


Semiautomated Techniques For Watershed Delineation
Setting

Watersheds can be delineated by several methods. One used extensively is hand delineation based on the contour information depicted on USGS 7.5-minute topographic maps. Even with the advent of GIS technology, this method is often still used prior to creating a digital watershed dataset. While this manual method can result in accurate delineations, it is a time-consuming and expensive task. The availability of digital topographic maps, in the form of Digital Raster Graphic (DRG) data, has made heads-up digitizing methods possible, but this method can also be slow and costly.

While the use of digital elevation data in watershed delineation is not new, the ability to use the data in an easy and efficient manner has been harder to realize. Because the watershed delineation process is often a subjective one that depends not only on the hydrologic characteristics of a given location (and how it is represented by the data or information used), but also on the requirements of the delineator, a fully automated system is not practical for many purposes.

Tools have been developed for EDNA stage 2 processing that use semi-automated techniques for delineating upstream areas, based on a user-selected outlet, and creating watershed and subwatershed data layers. The tools also facilitate the watershed and subwatershed attributing requirements and expedite and document the quality control phase (i.e. flagging) of stage 2 processing. These tools were developed within ArcView using the Avenue programming language. Since stage 1 processing generated the preliminary derivative products, stage 2 processing does not require the DEM data. Because of this independence from the DEM data, watershed and subwatershed delineation can be done in a very efficient manner with small data processing requirements.

Functionality

The EDNA stage 2 tools were developed with an intuitive functionality. Each button and tool is laid out in the order of processing flow. Each of these buttons and tools also has descriptive information included in the ArcView help system to help explain its functionality and usage. Figure 3 shows the ArcView help topics for the stage 2 tools.

Description of Stage 2 Tools

The underlying functionality of the tools for watershed delineation is built on the reach catchments created in stage 1 processing. In stage 1 processing, reaches based on a 5000-pixel drainage area (about 2 mi2 ) are created from the DEM. A catchment is subsequently produced from the DEM for each one of these reaches. The stage 2 tools enable the user to select an outlet catchment and automatically aggregate the upstream catchments to form watersheds or subwatersheds. This automatic aggregation is implemented by use of a Pfafstetter coding scheme.

The Pfafstetter coding scheme is based on work done by Otto Pfafstetter, an engineer for the Brazilian government (Pfafstetter, 1989). Based on topological information, the system lends itself to use with DEM derived drainage networks (Verdin and Verdin, 1999).

In addition to allowing the user to aggregate catchments into watersheds and subwatersheds, the tools also facilitate and manage the attributing of each watershed or subwatershed. While some attributes are filled in automatically (e.g. hydrologic unit number), others are populated through the use of a simple form. For example, naming is accomplished by automatically accessing NHD reach names, their associated levels, and length information, and filling in the form.

Examination of watershed and subwatershed boundaries and synthetic streams for problem areas is also an important part of stage 2 processing. The tools provide for systematic review, identification, and documentation of any discrepancies between the DEM-derived drainage basin delineations and streamlines and ancillary datasets (e.g. NHD, DRG data). Spatial identification of discrepancies is accomplished through use of flag polygons and lines. These discrepancies are subsequently attributed with descriptive information related to the cause and corrective actions required in stage 3. The flagging process is facilitated in the tools through use of drawing tools and an attribute form.

Watershed Delineation

Upon initialization of the tools, the necessary stage 1 data layers are loaded into an editing window, which is used exclusively for all subsequent processing. Figure 4 shows an example of an editing window with stage 1 data loaded. The tools automatically load all data shown in the editing window. The "P_catchments" represent the Pfafstetter coded reach catchments derived in stage 1 processing and are used to create watershed and subwatershed delineations. Also, notice that the NHD is brought in as an ancillary data set.

The flow diagram in Figure 5 shows the general process used to delineate watersheds and subwatersheds from the Pfafstetter-coded catchments. The entire delineation process simply involves deciding where the outlet should be and then pointing and clicking. The tools use the Pfafstetter connectivity to automatically select catchments upstream from the selected outlet. Figure 6 illustrates how the upstream catchments are aggregated automatically after a user has selected the outlet.

Once the analyst has decided if the aggregated catchments represent the desired watershed area, each selected catchment is coded with a specified 2-digit number, which is subsequently attached to the 8-digit number of the subbasin. The user need only enter the 2-digit number once for each watershed. This process is repeated, from upstream to downstream, until all catchments have been coded for the entire subbasin. Once the catchments have been coded, the next step involves the merging of the catchments into polygons representing each defined watershed. This is done automatically by the tool scripts, which use the field previously populated with unique watershed numbers. Figure 7 shows an example of merged catchments and the resultant watershed shapefile.

After watershed data have been created, the tools can be used to complete the attributes included within the watershed shapefile table. These attributes are based on those specified in the Federal guidelines. Attributing is completed by activating the attribute tool and then sequentially clicking on the watersheds. As each watershed is selected, a form is displayed. The form contains attribute information that has been filled in automatically, as well as those fields that require further input. Figure 8 shows the form as it appears after selection of a watershed, and Figure 9 shows an example of watersheds with their names and 10-digit numbers. The same procedure can be used to delineate subwatersheds. In this case the analyst enters a 4-digit number representing each subwatershed, which is subsequently attached to the 8-digit number (resulting in a 12-digit subwatershed number) and put into the catchment table for use in merging.

While the previous paragraphs provide only a brief overview of the EDNA stage 2 tool functionality, the discussion and examples illustrate the utility of the tools for creating preliminary watershed and subwatershed delineations. The semiautomated functionality of the tools is an efficient and, thus, cost-effective means of developing preliminary boundaries. The delineations produced from the DEM derivatives can be used as preliminary boundaries readily edited to meet map accuracy requirements or as a comparison dataset for checking existing digital delineations.

Although the described tools are not yet publicly available, they are available for use by USGS cooperators. This is primarily because of the research nature of the tools and the need for specially processed data not yet available to the public. Cooperators receive the data from the USGS for use with the tools and then supply delineation and flagging data back to the USGS. Cooperators can then use the preliminary delineations as a starting point for creating watershed and subwatershed boundaries or for comparison with existing delineations.


Summary

With the advent of new policies such as the Unified Federal Policy for a Watershed Approach to Federal Land and Resource Management, an increasing number of Federal, State, local, and tribal agencies and organizations have found a need for more detailed watershed delineation data and information than currently exist. While many agencies and organizations are working on new watershed and subwatershed delineations, many of their methods are time consuming and costly.

The NED contains the best available elevation data merged into a seamless database for the entire United States. The EDNA project, a multi-agency effort, takes advantage of the seamless and filtered characteristics of the NED to develop hydrologic derivatives. These digital elevation data can be readily used to delineate preliminary watersheds and subwatersheds. Recently developed ArcView tools were designed to facilitate the semiautomatic delineation of watersheds and subwatersheds.


References

Oimoen, M.J., 2000, "An Effective Filter For Removal Of Production Artifacts In U.S. Geological Survey 7.5-Minute Digital Elevation Models", Proceedings of the Fourteenth International Conference on Applied Geologic Remote Sensing, 6-8 November, Las Vegas, NV.

Pfafstetter, O., 1989, "Classification Of  Hydrographic Basins: Coding Methodology", unpublished manuscript, DNOS, August 18, 1989, Rio de Janeiro; translated by J.P.Verdin, U.S. Bureau of Reclamation, Brasilia, Brazil, September 5, 1991.

Verdin, K.L. & J.P. Verdin, 1999, "A Topological System for Delineation and Codification of the Earth's River Basins", Journal of Hydrology, vol. 218, nos. 1-2, pp. 1-12.

 



Jay R. Kost, Scientist (jkost@usgs.gov)
Raytheon ITSS*
U.S. Geological Survey
EROS Data Center
Sioux Falls, SD 57198
(605) 594-6931
(605) 594-6529 (fax)

Glenn Kelly, Cartographer (kelly@usgs.gov)
U.S. Geological Survey
EROS Data Center
Sioux Falls, SD 57198
(605) 594-6016
(605) 594-6529 (fax)