Shuttle radar topography mission accuracy assessment and evaluation for hydrologic modeling

Pablo Alberto Mercuri, Purdue University


Digital Elevation Models (DEMs) are increasingly used even in low relief landscapes for multiple mapping applications and modeling approaches such as surface hydrology, flood risk mapping, agricultural suitability, and generation of topographic attributes. The National Aeronautics and Space Administration (NASA) has produced a nearly global database of highly accurate elevation data, the Shuttle Radar Topography Mission (SRTM) DEM. The main goals of this thesis were to investigate quality issues of SRTM, provide measures of vertical accuracy with emphasis on low relief areas, and to analyze the performance for the generation of physical boundaries and streams for watershed modeling and characterization. The absolute and relative accuracy of the two SRTM resolutions, at 1 and 3 arc-seconds, were investigated to generate information that can be used as a reference in areas with similar characteristics in other regions of the world. The absolute accuracy was obtained from accurate point estimates using the best available federal geodetic network in Indiana. The SRTM root mean square error for this area of the Midwest US surpassed data specifications. It was on the order of 2 meters for the 1 arc-second resolution in flat areas of the Midwest US. Estimates of error were smaller for the global coverage 3 arc-second data with very similar results obtained in the flat plains in Argentina. In addition to calculating the vertical accuracy, the impacts of physiography and terrain attributes, like slope, on the error magnitude were studied. The assessment also included analysis of the effects of land cover on vertical accuracy. Measures of local variability were described to identify the adjacency effects produced by surface features in the SRTM DEM, like forests and manmade features near the geodetic point. Spatial relationships among the bare-earth National Elevation Data and SRTM were also analyzed to assess the relative accuracy that was 2.33 meters in terms of the total standard deviation for flat areas of central Indiana with a positive bias for SRTM relative to the national elevation dataset. Physical watershed boundaries, streams vector files and topographic attributes have been produced to investigate the SRTM global coverage performance in watershed modeling applications. The verification of the precision of hydrologic delineations in Indiana with the availability of Hydrologic Units datasets at watershed and subwatershed levels showed very small differences in drainage areas calculated, even at subwatershed levels. The hydrologic boundaries and drainage network extraction for the Arrecifes basin in the Argentine Pampas was checked for quality assurance with Landsat data and georeferenced digital topographic quadrangles, demonstrating the feasibility of using the SRTM global coverage for water resources mapping. SRTM data presents unique challenges, but the results of this research have shown that high quality results can be obtained. However, there are numerous issues related to preprocessing of SRTM data for certain surface hydrology applications to be addressed regarding SRTM in such as watershed characterization, land use planning, hazards assessment, energy resources assessment and many other uses around the world.




Engel, Purdue University.

Subject Area

Agricultural engineering|Remote sensing

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server