Soil Moisture Retrieval From Sentinel-1 Time-Series Data Over Croplands of Northeastern Thailand

In this letter, we propose a dual-temporal dual-channel (DTDC) algorithm for soil moisture retrieval by using time-series observations from the Sentinel-1 C-band synthetic aperture radar. This algorithm utilizes the ancillary information of vegetation water content derived from optical images and assumes no variation on the surface roughness during the two consecutive radar measurements. Therefore, with the DTDC backscatter observations, four equations could be established using forward models, while three unknowns (the two consecutive soil moisture values and one roughness parameter) could be solved simultaneously by minimizing a cost function. The algorithm was tested with a series of Sentinel-1 dual-channel (VV + VH) data over croplands (sugarcane and cassava) of Northeast Thailand with an upscaling resolution of 1 km. Results show that the proposed algorithm could well capture the temporal change of soil moisture with root-mean-square errors within 0.06 m(3)/m(3) when ignoring days with precipitation, and could achieve a similar spatial pattern of soil moisture as detected from the Soil Moisture Active Passive mission, indicating the Sentinel-1 might be a proper tool for agricultural water management.

Automated summary

In this study, a dual-temporal dual-channel (DTDC) algorithm was proposed to retrieve soil moisture using Sentinel-1 SAR data. By utilizing ancillary information from optical images and assuming constant surface roughness, the algorithm could solve for two consecutive soil moisture values and one roughness parameter simultaneously. The algorithm was tested on croplands in Northeast Thailand and demonstrated good performance in capturing temporal soil moisture changes and achieving similar patterns as a reference mission. This suggests that Sentinel-1 can be a suitable tool for agricultural water management.