首页|Detection of Eroded Sites of Mild Terrains by Adjusting SRTM DEM Errors
Detection of Eroded Sites of Mild Terrains by Adjusting SRTM DEM Errors
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
The Shuttle Radar Topography Mission (SRTM) digital elevation model (DEM) is the first highresolution seamless elevation surface of global coverage and the only one of its epoch. However, the reported high root mean square error (RMSE) limits its use in detecting elevation changes ofabout 5 m. SRTM DEM accuracy was influenced by terrain parameters such as roughness or smoothness and openness or forestation. The aim here is to present a procedure for adjusting the SRTM 3 arc seconds (3″) DEM in terrains of less than 6% slopes and using the adjusted dataset to detecterosions of about 5 m depth. Adjustment values were determined at 69 ground control points (GCP) across the project area by subtracting SRTM 3″ DEM cell values from elevations of the GCP. An adjustment raster of matching grid with the SRTM 3″ DEM was created for the project area.Using geostatistical interpolations based on the adjustment values at the GCP, adjustment values were assigned to each cell of the adjustment raster, which was used to adjust the published SRTM 3″ DEM. Testing the adjusted SRTM 3″ DEM dataset on a new set of 77 GCP reported thatthe average error value and RMSE were –0.17 m and ±2.42 m, respectively, from the initial –4.34 m and ±4.90 m, respectively. The published and the adjusted SRTM 3″ DEM surfaces were subtracted from a reference 1977 cartometric DEM of the project area to detecterosions. Eroded sites were detected as areas with positive value outliers at the 95% confidence level. The published dataset hardly detected any eroded sites, whereas the corrected dataset detected the two eroded sites in the project area which were confirmed by ground truthing.
DIGITAL ELEVATION MODEL (DEM)SRTMOUTLIER STATISTICSEROSION DETECTIONADJUSTMENTROOT MEAN SQUARE ERROR (RMSE)
NETL
NSTL
