Continued advances in availability and quality of digital terrain models (DTMs) have made topographically based modelling approaches one of the most popular themes in catchment hydrology literature ( Moore et al. 1991). Automated procedures are commonly used to delineate basin geometry and to derive flow pathways from digital maps of topography. Catchment and subcatchment boundaries and the stream drainage network are calculated by these routines, as are the slope, aspect, flow direction, and upstream contributing area for each grid cell (e.g., Moore et al. 1991; Martz and Garbrecht 1992, 1993; Eash 1994). Over the past four years many GIS software packages have been developed explicitly to facilitate extraction of hydrologically relevant information from DTM data. For example, Arc/Info's GRID module (version 6.1), released in 1992 ( ESRI 1992), implemented versions of the terrain analysis algorithms first put forth by Jenson and Dominque (1988).
Although most of the GIS-based approaches to topographic analysis use gridded data, a vector-based approach in conjunction with contour data is possible. Moore and Grayson (1991) used the vector-based TAPES-C (Topographic Analysis Programs for the Environmental Sciences--Contour) programs to partition the landscape into geographical elements utilizing the ``stream-tubes'' concept. (Stream tubes are portions of a catchment that drain into a segment of the stream.) Dawes and Short (1994) present a method of considering topological relationships in such contour based algorithms; they show that their method is faster computationally and that it results in more accurate delineation of catchment boundaries and drainage networks relative to previously used methods.
Despite the popularity of topographic analysis using DTM data, there are significant problems that arise. Limitations in using automated procedures for extracting drainage networks result from systematic errors in the production of DTMs; Garbrecht et al. (1994) demonstrated these limitations, focusing on the quality of US Geological Survey (USGS) 30-m digital elevation models (DEMs) of topography. Artificial pits or sinks, which are cells that do not have any downslope neighbors, result from inherent DEM resolution restraints and from errors in DEM generation. Fairfield and Leymaire (1991) and Martz and Garbrecht (1992, 1993) present approaches for dealing with problematic pit areas.