Skip to main content
ARS Home » Research » Publications at this Location » Publication #189586
Title: DIGITAL ELEVATION ACCURACY AND GRID CELL SIZE: EFFECTS ON ESTIMATED TERRAIN ATTRIBUTES

Author
item Erskine, Robert - Rob
item Green, Timothy
item RAMIREZ, JORGE - COLORADO STATE UNIVERSITY
item MACDONALD, LEE - COLORADO STATE UNIVERSITY

Submitted to: Soil Science Society of America Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/6/2007
Publication Date: 7/1/2007
Citation: Erskine, R.H., Green, T.R., Ramirez, J.A., Macdonald, L.H. 2007. Digital elevation accuracy and grid cell size: effects on estimated terrain attributes. Soil Science Society of America Journal. Vol 71: No4, July-August 2007, pp1371-1380; doi:10.2136/ssaj2005.0142.

Interpretive Summary: Terrain attributes are commonly used to explain the spatial variability of various processes. The attributes studied here (elevation, slope, aspect, and curvature) are computed readily from digital elevation models (DEMs), but DEM accuracy depends upon the data source, sampling method, and interpolation scheme. The primary objective of this study is to quantify relative differences in computed attributes due to changes in DEM accuracy and grid cell size. Three agricultural fields (approximately 65 ha each) in northeastern Colorado were surveyed using global positioning systems (GPS) and GPS data were interpolated by kriging onto 5, 10, 20, and 30-m grid DEMs. Accuracies of three data sources were compared: 1) real-time kinematic GPS (RTKGPS); 2) satellite-differentially corrected GPS (DGPS); and 3) U.S. Geological Survey (USGS) 30-m DEMs. RTKGPS provided '0.01 m horizontal and '0.02m vertical accuracy. DEMs derived from DGPS produced root mean squared errors (RMSEs) in elevation ranging from 0.89 m to 1.27 m relative to RTKGPS-derived DEMs. The USGS DEMs yielded similar elevation errors (0.58 m to 1.49 m). Calculated slopes, aspects, and curvatures were sensitive to the magnitude and spatial distribution of elevation errors. Small elevation errors (0.03 m RMSE) produced differences in curvature larger than the standard deviation for curvature. The sensitivity of these attribute values to elevation errors decreased as grid cell size increased. Impacts of DEM errors are illustrated using correlations of grain yield versus computed terrain attributes. These results provide new guidance for identifying the DEM accuracy and grid cell size needed for spatial terrain analyses.

Technical Abstract: Terrain attributes are commonly used to explain the spatial variability of various processes. The attributes studied here (elevation, slope, aspect, and curvature) are computed readily from digital elevation models (DEMs), but DEM accuracy depends upon the data source, sampling method, and interpolation scheme. The primary objective of this study is to quantify relative differences in computed attributes due to changes in DEM accuracy and grid cell size. Three agricultural fields (approximately 65 ha each) in northeastern Colorado were surveyed using global positioning systems (GPS) and GPS data were interpolated by kriging onto 5, 10, 20, and 30-m grid DEMs. Accuracies of three data sources were compared: 1) real-time kinematic GPS (RTKGPS); 2) satellite-differentially corrected GPS (DGPS); and 3) U.S. Geological Survey (USGS) 30-m DEMs. RTKGPS provided '0.01 m horizontal and '0.02m vertical accuracy. DEMs derived from DGPS produced root mean squared errors (RMSEs) in elevation ranging from 0.89 m to 1.27 m relative to RTKGPS-derived DEMs. The USGS DEMs yielded similar elevation errors (0.58 m to 1.49 m). Calculated slopes, aspects, and curvatures were sensitive to the magnitude and spatial distribution of elevation errors. Small elevation errors (0.03 m RMSE) produced differences in curvature larger than the standard deviation for curvature. The sensitivity of these attribute values to elevation errors decreased as grid cell size increased. Impacts of DEM errors are illustrated using correlations of grain yield versus computed terrain attributes. These results provide new guidance for identifying the DEM accuracy and grid cell size needed for spatial terrain analyses.