A method for estimating spatially continuous soil moisture from the synergistic use of geostationary and polar-orbit satellite data

• Published in: Journal of hydrology (Amsterdam) 2022-05, Vol.608, p.127590, Article 127590
• Main Authors: Yan, Qiu-Yu, Leng, Pei, Li, Zhao-Liang, Liao, Qian-Yu, Zhou, Fang-Cheng, Han, Xiao-Jing, Ma, Jianwei, Sun, Ya-Yong, Zhang, Xia, Shang, Guo-Fei
• Format: Article
• Language: English
• Subjects: Feng-Yun satellite data; Soil moisture; Spatially continuous; Trapezoid feature space
• ISSN: 0022-1694; 1879-2707
• DOI: 10.1016/j.jhydrol.2022.127590

•Synergistic geostationary and polar-orbit satellites were used to estimate spatially continuous soil moisture.•Soil moisture retrievals with different temporal composing periods were investigated.•Estimated soil moisture were assessed with both reanalysis and microwave-based datasets. Soil moisture (SM) is a key variable in the surface energy balance and water cycle, and its spatiotemporal dynamics are of great significance to climate, agriculture and other fields. Optical remote sensing has been widely used to estimate SM with relatively fine spatial resolution. However, optical observations are easily contaminated by clouds, making it difficult to obtain spatially continuous SM over large regions. In the present study, a semimonthly SM dataset over the study area of the entire Inner Mongolia region with nearly full spatial coverage was derived from the synergistic use of China’s Feng-Yun (FY) geostationary (FY-4A) and polar-orbit (FY-3D) observations, following a previously developed trapezoid feature space in a pixel-to-pixel manner. A preliminary assessment was conducted to evaluate the performances of the proposed method over two main dominant land cover types (grassland and cropland) in the study region, where the China Meteorological Administration Land Data Assimilation System (CLDAS) and the Soil Moisture Active Passive (SMAP) SM products were provided as references. The results indicated that the estimated SM was well correlated to the referenced SM datasets, with a significant correlation coefficient varying from 0.5 to 0.8. Furthermore, for the grassland (cropland), unbiased root mean square errors of approximately 0.062 (0.097) m3/m3 and 0.055 (0.069) m3/m3 can be found when comparing the estimated SM with the CLDAS and the SMAP product, respectively.