Reconstruction of the SMAP-based 12-hourly soil moisture product over the CONUS through water balance budgeting

Passive microwave remote sensing soil moisture retrievals from the Soil Moisture Active Passive (SMAP) mission have been widely used in a variety of hydrologic and climatological studies because of their exceptional performance. However, the application of such a data set relying on a space-borne radiometer is often restricted by its revisit frequency, likely resulting in underestimations of critical soil water responses to suddenly intense rainfall events. Rather than simply interpolating the missing observations, we produced a continuous 7-year (2015 - 2021), 12-hourly soil moisture product over the Contiguous United States (CONUS) using a simplified water balance scheme based on three hydrologic-cycle components: precipitation input, hydrologic loss, and soil moisture. Using a half-hourly precipitation data set from the Global Precipitation Mission (GPM) and the loss estimated exclusively by the SMAP soil moisture product during dry-down periods, each posterior soil moisture was forecast by the soil moisture at the prior time step and two site-specific optimized parameters obtained by minimizing the errors with the coincident SMAP data. This new data set performs closer to the SMAP observations during warm seasons (May - October), with a median Pearson Correlation (R) of 0.66 and a median unbiased root-mean-square error (ubRMSE) of 0.06 m3/m3 compared to its whole-season metric medians of 0.57 (R) and 0.07 m3/m3 (ubRMSE). Although the simulated soil moisture data reflect precipitation features well, their magnitudes are somewhat limited by inappropriate runoff estimations. The amount of dry-down soil moisture used to build the loss function is critical in ensuring good performance of the reconstructed soil moisture, while the above-mentioned parameters calibrated over various periods do not appear to alter the quality of simulated soil moisture. The validation results from a case study using in-situ measurements of the Texas Soil Observation Network (TxSON) concur that this observation-driven approach incorporated with a simplified water balance equation can produce a temporally gapless soil moisture data set that performs comparably to the SMAP product.