2024 Volume 80 Issue 16 Article ID: 23-16066
In this research, the authors studied how to improve the versatility of the existing soil moisture estimation algorithm developed on the premise of using ALOS/PALSAR. We first showed that the existing algorithm, though originally set to assume that the incident angle is constant, can use the ALOS2/PALSAR2 multi-polarized wave data by adapting it to process changes in the incident angle. The problem is that multi-polarized wave mode observation is highly infrequent. However, with multiple SAR observations and estimated soil moisture maps created based on them, it is technically possible to estimate the ground surface roughness by performing in-verse estimation using the microwave scattering model. Therefore, with the creation of such datasets in mind, we experimentally estimated soil moisture from the Sentinel-1/C-SAR data using the random forest method, a machine learning approach.
