Evaluation of DEM Accuracy Improvement Methods Based on Multi-Source Data Fusion in Typical Gully Areas of Loess Plateau.

Publisher: MDPI Country of Publication: Switzerland NLM ID: 101204366 Publication Model: Electronic Cited Medium: Internet ISSN: 1424-8220 (Electronic) Linking ISSN: 14248220 NLM ISO Abbreviation: Sensors (Basel) Subsets: PubMed not MEDLINE; MEDLINE

Original Publication: Basel, Switzerland : MDPI, c2000-

Improving the accuracy of DEMs is a critical goal in digital terrain analysis. The combination of multi-source data can be used to increase DEM accuracy. Five typical geomorphic study areas in the Loess Plateau in Shaanxi were selected for a case study and a 5 m DEM unit was used as the basic data input. Data from three open-source databases of DEM images, the ALOS, SRTM and ASTER, were obtained and processed uniformly through a previously geographical registration process. Three methods, Gram-Schmidt pan sharpening (GS), weighted fusion and feature-point-embedding fusion, were used for mutual enhancement of the three kinds of data. We combined the effect of these three fusion methods in the five sample areas and compared the eigenvalues taken before and after the fusion. The main conclusions are as follows: (1) The GS fusion method is convenient and simple, and the three combined fusion methods can be improved. Generally speaking, the fusion of ALOS and SRTM data led to the best performance, but was greatly affected by the original data. (2) By embedding feature points into three publicly available types of DEM data, the errors and extreme error value of the data obtained through fusion were significantly improved. Overall, ALOS fusion resulted in the best performance because it had the best raw data quality. The original eigenvalues of the ASTER were all inferior and the improvement in the error and the error extreme value after fusion was evident. (3) By dividing the sample area into different areas and fusing them separately according to the weights of each area, the accuracy of the data obtained was significantly improved. In comparing the improvement in accuracy in each region, it was observed that the fusion of ALOS and SRTM data relies on a gentle area. A high accuracy of these two data will lead to a better fusion. Merging ALOS and ASTER data led to the greatest increase in accuracy, especially in the areas with a steep slope. Additionally, when SRTM and ASTER data were merged, the observed improvement was relatively stable with little difference.

41871324 National Natural Science Foundation of China; 41807075 National Natural Science Foundation of China; 41930102 National Natural Science Foundation of China; 2018JY0464 Science & Technology Department of Sichuan Province Key project of Applied Basic Research; Kj-2022-13 Sichuan Natural Resources Technology Project; 416627 China West Normal University General Cultivation Project; CXTD2018-10 China West Normal University Research and Innovation Team Project

Keywords: DEM; Gram–Schmidt pan sharpening; elevation RMSE; feature points embedding; weight fusion

Date Created: 20230428 Date Completed: 20230430 Latest Revision: 20230430

20250114

PMC10143287

10.3390/s23083878

37112219