Key points are not available for this paper at this time.
1994 examine the effect of digital elevation model (DEM) grid size on landscape characterization and hydrologic simulation of catchment response. They conclude "that a grid size of 10 m would suffice for many DEM-based applications of geomorphic and hydrologic modeling" (p. 1027). In this comment the findings of the study are reviewed in terms of the information content of the DEMs. We are concerned that the resolution of the initial topographic data from which the DEMs of various resolutions were generated may have significantly contributed to the conclusions of the study. We suggest that general guidelines for DEM grid size for hydrologic applications should consider landscape characteristics, hydrologic model type, and model parameters. In Zhang and Montgomery's 1994 study, the various DEMs generated for the Tennessee Valley catchment are interpolated from spot elevation data which were "obtained from low-altitude aerial photographs using a stereo digitizer at a density about every 10 m" (p. 1020). These spot elevation data are therefore only capable of representing actual topographic features of the landscape that can be resolved at an observational spacing of 10 m or more. The 10-, 30-, and 90-m DEMs interpolated from this spot elevation data can be expected to provide a reasonable representation of actual topographic features resolvable at the nominal resolution. However, topographic features smaller than the original t0-m observation spacing will not be represented in the interpolated 2-and 4-m DEMs. The information content of a DEM cannot exceed that of the spot elevation data from which it is interpolated. Since the original spot elevation data have an implicit resolution of about 10 m, the topographic information content of the 2-, 4-, and 10-m DEM is essentially the same. While the information content of the original elevation data can be degraded by interpolation to a coarse resolution (i.e., 30 and 90 m), it cannot be enhanced by interpolation to a finer resolution (i.e., 2 and 4 m). Similar considerations apply to the Mettman Ridge catchment. Even though detailed specifications of the base map are not reported in the paper, the scanning and vectorizing of the base map are stated to reproduce identical contours to those of the original topographic map and assumed to provide an accurate portrayal of the land surface. The contour map for the Mettman Ridge catchment is shown in Zhang and Montgomery's 1994 Figure 2a. An examination of the map scale and the spacing of the contours reveals that on the average, the reso~
Garbrecht et al. (Wed,) studied this question.