Effect of Spatial Scale on Soil Moisture Retrieval From Passive Microwave Sensors

Near-surface soil moisture (SM) retrieval from Lband (1.4 GHz) passive microwave brightness temperature (TB) measurements has been demonstrated from tower and airborne experiments (Wang 1983, Jackson et al. 1999). Current passive microwave techniques for SM retrieval are based on inversion of radiative transfer models which simulate the microwave emission from the earth surface given a specified soil water content, soil temperature, vegetation water content, surface roughness, and so on. These models assume homogeneous conditions within each pixel, and have been developed from radiometer data of homogenous pixels with resolutions on the order of 10’s of meters. However, the strong nonlinearity that exists in such models with respect to land surface parameters (i.e., vegetation and soil moisture), stems the question whether the same model should be applied to heterogeneous pixels at satellite footprint scales (~40km), where the land surface heterogeneity is expected to be significant. As shown in figure 1, even in the simple case of a smooth, non-vegetated surface, the non-linearity would imply that the field average SM retrieved from a single low resolution footprint is different then that calculated by averaging the SM retrieved from high resolution pixels over the same area.