Microwave remote sensing such as soil moisture active passive (SMAP) can provide soil moisture data for agricultural and hydrological studies. However, the scales between station-measured and satellite-measured products are quite different, as stations measure on a point scale while satellites have a much larger footprint (e.g., 9 km). Consequently, the validation for soil moisture products, especially inter-comparison between these two types of observations, is quite a challenge. Spatial autocorrelation among the stations could be a contribution of bias, which impacts the dense soil moisture networks when compared with satellite soil moisture products. To examine the effects of spatial autocorrelation to soil moisture upscaling models, this study proposes a spatial analysis approach for soil moisture ground observation upscaling and Thiessen polygon-based block kriging (TBP kriging) and compares the results with three other methods typically used in the current literature: arithmetic average, Thiessen polygon, and Gaussian-weighted average. Using the Texas Soil Observation Network (TxSON) as ground observation, this methodology detects spatial autocorrelation in the distribution of the stations that exist in dense soil moisture networks and improved the spatial modeling accuracy when carrying out upscaling tasks. The study concluded that through TBP kriging the minimum root-mean-square deviation (RMSD) is given where spatial autocorrelation takes place in the soil moisture stations. Through TBP kriging, the station-measured and satellite-measured soil moisture products are more comparable.