Automated DEM Extraction from ASTER Images and its Accuracy Assessment in Eyvanekey, Semnan Province

Authors

Abstract

Digital Elevation Model (DEM) is increasingly used these days as an effective tool to extract elevation data in quantitative environmental and land management studies. The improvement of sensors and land resource satellites has made it possible to generate DEM from remotely sensed data. DEMs which are produced from these data are applied not only in the two dimensional (2D) but also in the three dimensional (3D) interpretations of the earth surface. The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER), which is designed to capture along-track stereo images, was selected in this study for DEM generation, and its accuracy assessed, to be applied in land management studies. PCI Geomatica software and Toutin rigorous model were applied for DEM generation. Using accurate and well-distributed ground control points, DEM with a RMSE (Root Mean Square Error) of about 20m was extracted from the data. Image profiles, spatial correlation and statistical parameters were used to assess both the planimetric and vertical accuracy of the produced DEM. Interpretation of the image profiles showed that the generated DEM is very accurate in nearly flat regions and on smooth slopes with either no or sparse vegetation. Errors were generally less than 10m in such areas. Errors increase in dense vegetation, steep slopes and deep valleys because of the shadows. An evaluation of the results showed that the extracted DEM can be successfully employed in land management studies especially in soil-landscape relationship and land evaluation for mapping such elevation parameters as slope, aspect, hypsometry and physiographic as well as catchments discrimination at medium scales of: 1:50000 and 1:100000. It would also be suitable for orthorectifying the satellite images and aerial photos.

Keywords