Mapping High Spatiotemporal-Resolution Soil Moisture by Upscaling Sparse Ground-Based Observations Using a Bayesian Linear Regression Method for Comparison with Microwave Remotely Sensed Soil Moisture Products

In recent decades, microwave remote sensing (RS) has been used to measure soil moisture (SM). Long-term and large-scale RS SM datasets derived from various sensors have in environmental fields. Understanding the accuracies of products is essential for their proper applications. However, due mismatched spatial scale between ground-based observations, truth at pixel may not be accurately represented by especially when density situ measurements low. Because observations are often sparsely distributed, temporal upscaling was adopted transform a few into values 1 km introducing temperature vegetation dryness index (TVDI) related SM. The upscaled showed high consistency with could capture rainfall events. considered as reference data evaluate different scales. regard validation results, addition correlation coefficient (R) Soil Moisture Active Passive (SMAP) being slightly lower than that Climate Change Initiative (CCI) SM, SMAP had best performance terms root-mean-square error (RMSE), unbiased RMSE bias, followed CCI. Ocean Salinity (SMOS) were worse agreement inferior R value X-band Advanced Microwave Scanning Radiometer 2 (AMSR2). conclusion, study area, CCI more reliable, although both underestimated 0.060 cm3 cm−3 0.077 cm−3, respectively. If biases corrected, then improved an 0.043 0.039 will hopefully reach application requirement accuracy less 0.040 cm−3.