Evaluation of SMAP/Sentinel 1 High-Resolution Soil Moisture Data to Detect Irrigation Over Agricultural Domain

Irrigation is not well represented in land surface, hydrological, and climate models. One way to account for irrigation is by assimilating satellite soil moisture data that contains irrigation signal with land surface models. In this study, the irrigation detection ability of SMAP enhanced 9km and SMAP-Sentinel1 3km and 1km soil moisture products are evaluated using the first moment (mean) and the second moment (variability) of soil moisture data. The SMAP enhanced 9km soil moisture product lacks irrigation signals in an irrigated plain south of Urmia, whereas SMAP-Sentinel1 products record irrigation signal in soil moisture variability. Despite observing higher variability over irrigated areas, there are only small and inconsistent wet biases observed over irrigated pixels relative to nearby non-irrigated pixels during the irrigation season. This is partly attributable to the climatology vegetation water content used in the SMAP-Sentinel 1 soil moisture retrieval algorithm that is not accounting for crop rotattion and land management . Thus, in the second part of this study, we updated the retrieval algorithm to use dynamic vegetation water content. The update increased vegetation water content up to 1 kg/m2, which corresponds with a 0.05 cm3/cm3 increase in soil moisture during irrigation season. The update does not notably change soil moisture retrievals off season. This study shows that irrigation signals are present in both the first and second moment of soil moisture time series, and employing dynamic vegetation water content in the SMAP-Sentinel 1 algorithm can enhance the irrigation signal over agricultural regions.