Rectification methods comparison for the ASTER GDEM V2 data using the ICESat/GLA14 data in the Lvliang mountains, China

Abstract

ASTER GDEM data have high spatial resolution and wide coverage, which are significant in topographic analysis and geographic study. ICESat/GLA14 data can improve the vertical accuracy of ASTER GDEM data based on its abundant points with high vertical accuracy. It is essential to seek a proper rectification method for ASTER GDEM data. In this research, taking the Lvliang Mountains of China as the study area, proper rectification method is explored for the latest version of the ASTER GDEM data (ASTER GDEM V2 data). The ICESat/GLA14 data are regarded as the real data, so the difference between the two datasets is the vertical error data. The key to rectify the ASTER GDEM V2 data is to seek an appropriate interpolation method to simulate a proper surface for the vertical error data. Through analysis and comparison, four interpolation methods including inverse distance weighted (IDW) method, ordinary kriging (OK) method, simple kriging (SK) method and universal kriging (UK) method are first chosen from numerous interpolation methods. Through comparing the vertical error histograms, the samples and the characteristics of the four interpolation methods, the ICESat/GLA14 data in 2005 are selected as the benchmark to rectify the ASTER GDEM V2 data. Then, the vertical error data are calculated using the difference value between the ASTER GDEM V2 data and the ICESat/GLA14 data. Under optimal configuration the vertical error, the chosen four interpolation methods are used to simulate the surface of the vertical error data. Finally, the rectified results for ASTER GDEM V2 data are achieved by adding the interpolated results to the ASTER GDEM V2 data, and the accuracy of the rectified results is appraised using the ground control points on the topographic maps. The research results show that (1) Accuracy of the rectified results is different by using various interpolation methods. The RMSE value range for all the rectified results is 19.5-26.7 m. (2) Among the four methods, SK is the best; the next are OK and UK methods, which have approximately similar performance, but the UK method is a little worse; on the whole, the IDW method has the worst performance. (3) This research not only provides the optimal parameters for different interpolation methods, but also obtains the appropriate rectification method for ASTER GDEM V2 data using ICESat/GLA14 data, which is beneficial for broader range of applications using ASTER GDEM V2 data in topographic and geographic research.