A quantitative model to simulate the vertical errors of SRTM3 DEM V4 data at the pixel level in the Shanbei Plateau of China. Issue 14 (17th July 2020)

• Record Type: Journal Article
• Authors: Zhao, Shangmin; Zhao, Hengyang; Li, Rongping; Cheng, Weiming; Zhou, Chenghu
• Is Part Of: International journal of remote sensing. Volume 41:Issue 14(2020)
• Page Start: 5257
• Page End: 5276
• Publication Date: 2020-07-17
• Subjects: Remote sensing -- Periodicals; Télédétection -- Périodiques; 621.3678
• DOI: 10.1080/01431161.2020.1734250
• Languages: English
• ISSNs: 0143-1161

Abstract

ABSTRACT: The Shanbei Plateau is located in the central Loess Plateau of China, which exhibits typical loess geomorphology and suffers serious soil and water erosion. Numerous studies are conducted here using digital topographic analysis methods based on digital elevation model (DEM) datasets. Most of these topographic analyses are conducted at the pixel level; thus, it is meaningful to determine the vertical error distributions of the DEM datasets at that scale. Taking the 14th product of the Geoscience Laser Altimeter System on the Ice, Cloud, and land Elevation Satellite (ICESat/GLA14) as the benchmark, the aim of this research is to construct a quantitative model to simulate the vertical error distributions of SRTM3 (Shuttle Radar Topography Mission with a spatial resolution of 3") DEM V4 data for the Shanbei Plateau. Through comparison, the multiple linear regression (MLR) model is chosen and the vertical error surface is determined based on the MLR model and the land surface factor data. Moreover, the rectified SRTM3 DEM V4 data are developed based on the original data version and the vertical error surface. Through accuracy assessment, the root mean squared error (RMSE) value for the rectified version is found to be 9.6 m. Compared to 20.6 m for the original version, this is a clear improvement. Meanwhile, the MLR model has an R 2 (the coefficient of determination) value of 0.484, which shows that it is able to represent the relationships between the vertical error distributions and the land surface factors. This research is not only meaningful for digital topographic analysis and its relevant fields of application but also provides a reference for accuracy assessments of other DEM datasets in different regions.