Given the technological limitations of a spaceborne C band sensor, the measurement resolution is about 60 km. At this scale, there is significant spatial heterogeneity in vegetation type and density, surface roughness, and near-surface soil moisture. The effect of such spatial variability on the soil moisture retrieval is poorly understood. Because of nonlinearities in the radiative transfer processes, retrieved quantities may not reflect area-averages for footprints containing mixtures of different surface types. Due to the myriad difficulties in validating large-scale remotely-sensed soil moisture estimates from surface observations, we believe that measurements alone are inadequate for this purpose, and that a hybrid measurement-modeling approach is better. For this reason, we are utilizing a coupled hydrologic/radiobrightness model to evaluate AMSR-derived brightness temperature and soil moisture products. The primary advantages of using such a modeling system are to minimize the problems of spatial continuity, vertical profile effects, and asynchronous satellite and in situ sampling. In our approach, surface and aircraft measurements are assimilated into the model when available to adjust the model soil moisture and provide an optimal validation data set.
The Soil Moisture Experiments in 2002 (SMEX '02) were conducted in Iowa from June 24 - July 12, primarily for the purpose of validating AMSR soil moisture retrieval algorithms and products. In situ data collected during the experiment include near-surface and profile soil moisture, surface and soil temperatures, vegetation height and biomass, and surface water, energy and CO2 fluxes. Several aircraft were deployed with radiometers for measuring visible, near-infrared and infrared radiative fluxes, microwave radiometers and radars across a range of frequencies, and energy flux instrumentation. Utilizing these data as well as topographic and hydrographic data, we have performed model simulations over a 100 x 60 km region to characterize the temporal, spatial and vertical distributions of soil moisture and temperature. This measurement-modeling framework provides estimates of soil moisture and brightness temperatures at 'field scale' (< 1 km) which are in turn aggregated to the larger AMSR footprint scale to estimate the soil moisture retrieval error and the space-time error variability. We also provide uncertainty estimates by considering the uncertainties inherent in soil and vegetation properties as well as errors associated with physics of the hydrology/radiobrightness model. This allows an evaluation of the AMSR products with respect to their anticipated retrieval accuracies.