Evaluation of six satellite- and model-based surface soil temperature datasets using global ground-based observations

The comprehensive evaluation of model- and satellite-based surface soil temperature (ST) products is a prerequisite for applications these datasets in hydrology, ecology, climate change, as well passive microwave moisture retrieval algorithms. Distinguished from existing regional validations, this study used ground observations approximately 800 stations 5 sparse 15 dense networks worldwide to fully assess six ST April 2015 December 2017. consist five model-based the Modern-Era Retrospective Analysis Research Applications, version 2 (MERRA-2), Goddard Earth Observing System Model Forward Processing (GEOS-5 FP), Global Land Data Assimilation (GLDAS) Noah, ERA-Interim newly released ERA5 produced by European Centre Medium Range Weather Forecasts (ECMWF), one retrieved Advanced Microwave Scanning Radiometer (AMSR2) using Parameter Retrieval (LPRM). accuracy was comprehensively assessed per availability networks, interval, land cover, zone, elevation. results show that GEOS-5 exhibits smallest averaged unbiased root mean square difference (ubRMSD) 1.84 K among all products. All high skill capturing temporal trend with an correlation coefficient larger than 0.97. obtains visible improvements compared its predecessor showing smaller ubRMSD absolute bias values. generally lower values ST, while their tends be warmer increase except highest conditions. Moreover, unstable performance most shrubland grassland, tropical, arid cold climate, elevation regions also demonstrated LPRM 3.04 K, more attentions should paid impacts elevations underlying surfaces on improving product. These new findings will valuable future refinement datasets, algorithms estimate satellite data, various disciplines.