Fusion Scheme and Implementation Based on SRTM1, ASTER GDEM V3, and AW3D30

Abstract

Multi-source data fusion can help to weaken the original data’s shortcomings while improving data accuracy. The experimental area in this research is Taiyuan City in Shanxi Province, China. Using SRTM1 DEM, ASTER GDEM V3, and AW3D30 DEM, the optimal resolution of the Fused DEM in the research area is determined by analyzing the topographic factor information entropy. Then the optimally weighted fusion coefficient of the DEM with root mean square error (RMSE) as the criterion under different slope classes is determined by traversal exploration and quantitatively evaluates the fusion effect. The results show that the optimal resolution of the Fused DEM is 40 m under the terrain feature constraint of Taiyuan city. The fused DEM decreases by 33.8%, 57.9%, and 11.5% for mean absolute error (MAE), 36.3%, 54.6%, and 1.4% for standard deviation (STD), and 32.8%, 54.2%, and 9.7% for root mean square error (RMSE) compared with SRTM1, ASTER GDEM V3, and AW3D30. The weighted average fusion of multiple intensities increased the accuracy of the original data. The reduced topographic factor errors, such as slope, profile curvature, and TPI, improved the Fused DEM’s topographic representation capacity. Furthermore, the results confirm the high accuracy of Fused DEM in complex mountainous regions.