Search

Summary Soil moisture is a key factor in energy and water cycles. Many satellite missions have been planned and implemented for retrieving soil moisture globally. Because the spatial representativeness of a point-scale soil moisture station is rather limited, a station network needs setting up for scale-matching validation of satellite-based soil moisture products. Even so, an upscaling procedure is needed to upscale these station soil moisture values into area-wide one. However, such a procedure itself introduces uncertainties into the upscaled soil moisture. In this study, four upscaling methods (simple average, block kriging, model-based, and apparent-thermal-inertia-based) are inter-compared according to their performance stability for evaluation of soil moisture estimated by assimilating microwave signals into a land surface model. It is found that the performance of the model-based upscaling approach is the most unstable because model simulations are full of uncertainties for representing spatial variability of soil moisture. The block kriging upscaling method performs not worse than the simple averaging approach; the former may generate more representative soil moisture if the range of the soil moisture semivariogram used in the block kriging is comparable to the extent of a satellite footprint. The apparent-thermal-inertia-based upscaling is the most stable one, which has been developed with the aid of high-resolution satellite thermal data. All analyses indicate that choosing a suitable upscaling approach is important for the effective evaluation of satellite-based soil moisture. Otherwise, uncertainties hiding in the upscaling method will be incorrectly attributed to errors in satellite products, undermining our confidence in implementing them into practice. [ABSTRACT FROM AUTHOR]