Backscattering model validity across large footprints for global scatterometer applications

Medium to low resolution (1-50 km) active microwave sensors such as spaceborne scatterometers and wide-swath mode SAR have great potential as tools for long term monitoring over land and ice. Their large area coverage and low cost make them ideal for measuring mesoscale land surface changes. One approach to the interpretation of such data is to employ a modeling approach whereby the backscatter model can be inverted to estimate some representative surface parameters. However, to optimize this method it is necessary to use scattering models that are applicable to large (> 1 km) footprints that inevitably contain a range of surface characteristics. This paper investigates the validity of a number of surface scattering models for such a task. The investigation is carried out both from a theoretical modeling approach as well as through comparison to real data from the ERS scatterometer over non-vegetated areas. Modified models that incorporate surface heterogeneity through probability distributions are also introduced. It will be shown than the ERS scatterometer data are better represented using these modified models than the standard theoretical models. Good results are also obtained using the semi-empirical model by Oh et al. that seems to incorporate for the target heterogeneity.