Comparison of Digital Elevation Models for Aquatic Data Development

Thirty-meter digital elevation models (DEMs) produced by the U.S. Geological Survey (USGS) are widely available and commonly used in analyzing aquatic systems. However, these DEMs are of relatively coarse resolution, were inconsistently produced (i.e., Level 1 versus Level 2 DEMs), and lack drainage enforcement. Such issues may hamper efforts to accurately model streams, delineate hydrologic units (HUs), and classify slope. Thus, the Coastal Landscape Analysis and Modeling Study (CLAMS) compared streams, HUs, and slope classes generated from sample 10-meter drainage-enforced (DE) DEMs and 30-meter DEMs. We found that (1) drainage enforcement improved the spatial accuracy of streams and HU boundaries more than did increasing resolution from 30 meters to 10 meters, particularly in flatter terrain; (2) streams and HU boundaries were generally more accurate when delineated with Level 2 than with Level 1 30-meter DEMs; and (3) the 10-meter DE-DEMs better represented both higher and lower slope classes. These findings prompted us to have 10-meter DE-DEMs produced for the Coast Range Province of Oregon, increased confidence in CLAMS outputs from the 10-meter DE-DEMs, and should benefit others interested in using DEMs for aquatic analyses. Introduction Decision makers and scientists are increasingly planning for and studying aquatic ecosystems over broad spatial extents. The Coastal Landscape Analysis and Modeling Study (CLAMS) (Spies et al., 2002), similar to other scientific assessments that support landscape planning (e.g., Forest Ecosystem Management Assessment Team (FEMAT), 1993; Sierra Nevada Ecosystem Project (SNEP), 1996; and Umpqua Land Exchange Project (ULEP), 2001), is developing broad-scale spatial databases. For the Coast Range Province of Oregon (FEMAT, 1993), we are using these databases in the aquatic component of CLAMS to assemble and apply watershed condition indices and to model instream habitat structure from upslope and streamside attributes (CLAMS, 2002). The value of these indices and models relies heavily on the quality and consistency of the foundation data layers used in their construction. Some of the foundational data can be derived from digital elevation models (DEMs), i.e., slope, elevation, aspect, channel gradient, hydrologic unit (HU) boundaries (FGDC, 2002), and stream traces (Jenson and Domingue, 1988; Jenson, 1991; Moore et al., 1991; Quinn et al., 1991; Tarboton et al., 1991; Band, 1993; Wang and Yin, 1998). These DEM-generated products have many benefits for broad-scale aquatic analyses. For example, streams created from DEMs are precisely registered to Comparison of Digital Elevation Models for Aquatic Data Development Sharon Clarke and Kelly Burnett the DEMs (Jenson and Domingue, 1988; Jenson, 1991), which improves the quality of stream-associated topographic information, such as channel gradient and valley floor width. Furthermore, streams created from DEMs always appear as single lines, rather than as braided channels or double-bank streams, and a single line represents water bodies such as lakes and reservoirs. Thus, it is easier to calculate stream order and route stream networks; the latter is necessary prior to georeferencing stream-associated data. We evaluated the usefulness of DEM data for our specific applications throughout the range of landscape characteristics found in our study area, following the recommendation of many authors that DEM evaluations need to be contextual (Walsh et al., 1987; Weih and Smith, 1990; Shearer, 1991; Carter, 1992, Robinson, 1994; Zhang and Montgomery, 1994). Given that DEM resolution has been demonstrated to affect the accuracy of landform characterization and drainage networks (e.g., Elsheikh and Guercio, 1997; Thieken et al., 1999; Zhang et al., 1999; McMaster, 2002), we wanted to evaluate advantages for aquatic analyses of employing 10-meter drainageenforced DEMs (DE-DEMs) (Osborn et al., 2001) rather than the more widely available 30-meter DEMs. Specifically, results from 10-meter DE-DEMs and 30-meter DEMs (Level 1 or 2) were compared for deriving (1) a consistent density, positionally accurate, single-line stream layer that corresponded to the topography; (2) hydrologic units delineated at approximately the 6-field hydrologic unit (HU) level; and (3) a representation of topography to calculate slope. The 10-meter DE-DEMs we used have better horizontal and vertical resolution and were produced with a more consistent process (i.e., by the same contractor and specifications; Averstar Geospatial Services, now Titan Corp.) than were the 30-meter DEMs. DEM Data Concerns about the vertical and horizontal resolution and inconsistent quality of the 30-meter DEMs prompted preliminary assessment of these data for aquatic analyses. Although the U.S. Geological Survey (USGS) has improved its methods to generate DEMs, some of the 30-meter DEMs in the study area were created with earlier methods, which yielded two classes of quality, Levels 1 and 2. Level 1 DEMs were created by autocorrelation or manual profiling directly from aerial photography. Level 2 DEMs were created from digital line graph (DLG) contours or equivalent (USGS, 1998). The vertical accuracy P H OTO G R A M M E T R I C E N G I N E E R I N G & R E M OT E S E N S I N G December 2003 1367 S. Clarke is with the Forest Science Department, Oregon State University, Corvallis, OR 97331 (Sharon.Clarke@oregonstate. edu). K. Burnett is with the PNW Research Station, USDA Forest Service, Corvallis, OR 97331 ([email protected]). Photogrammetric Engineering & Remote Sensing Vol. 69, No. 12, December 2003, pp. 1367–1375. 0099-1112/03/6912–1367$3.00/0 © 2003 American Society for Photogrammetry and Remote Sensing The use of trade or firm names in this publication is for reader information and does not imply endorsement by the U.S. Department of Agriculture of any products or services. 02-062.qxd 10/30/03 11:49 AM Page 1367