Inconsistencies between remotely sensed (thermal infrared), in situ, and modeled values of soil water content are examined. First, an important hydraulic parameter in a soil water profile model is varied by one standard deviation to simulate a reasonable degree of spatial variability within a given soil texture class. This results in a large range of drying rates at the soil surface and the formation of a sharp vertical soil water gradient at the surface. The formation of this gradient is dependent upon the soil properties. Thermal infrared remote sensing, which detects soil moisture at the soil surface, is then discussed in this context. Because of the “decoupling” of the soil water profile, we conclude that soil moisture derived from surface radiant temperatures is probably not useful in knowing the column-average soil water content but may provide some insight into the spatial variations in soil texture and hydraulic properties at the surface.