Soil Moisture Estimation of Grassland Using Public Copernicus Satellite Radar Sentinel-1

作者：Chang-Feng Yang, Stephan Mäder, T. Rudolph, Peter Goerke-Mallet, Andreas Müterthies · 年份：2022 · DOI：10.3997/2214-4609.202221031 · 研究领域：Soil Moisture and Remote Sensing、Precipitation Measurement and Analysis、Meteorological Phenomena and Simulations

Summary Soil moisture directly affects yields and benefits of grassland. Therefore, monitoring soil moisture is essential to understand environmental footprint of grass. The relevant work helps decision-makers, for example, to handle the escalation of global food crisis. Remote sensing techniques are widely applied to the long-term monitoring task of an extensive area. Among them, satellite radars provide a global visibility in cloud-covered areas day and night. Our study aims to predict the soil moisture of a specific grassland based on water cloud model (WCM) and satellite radar data. Compared with previous works, our modelling demands only moisture records and dual-polarization radar data; the extra field work for model calibration is spared thanks to rigorous data collection and processing. We tested our approach using Sentinel-1 data in two cases of grassland around 50 km northwest of Berlin. In validation, the correlations are 0.87 and 0.93; the mean absolute errors are 3.92 % and 2.29 %. We also checked the model transferability between the two grasslands. The correlation and mean absolute error are 0.92 and 4.47 %; 0.85 and 5.32 % vice versa. Our result shows a prospect of training a model at low cost for a global use.