Spatial soil moisture mapping through multi-temporal analysis of ERS-SAR pri data

The scattering of microwaves from soil depends on several surface characteristics, such as the roughness, vegetation and the moisture content of the top layer. Knowledge of the temporal and spatial distribution of this last parameter is of major importance to hydrologic, meteorologic and climatologic modelling. However accurate measurements of the spatial distribution of soil moisture with classical methods have always been a difficult task. Owing to its dependency on soil moisture and its spatial character, radar remote sensing holds much promise. Several empirical and physically based scattering models have been proposed to retrieve soil moisture values from SAR data, but problems occur with the identification of the roughness and vegetation parameters. This can be partly overcome through the use of multi-frequency and/or multi-polarization radar, but this option is often not available on spaceborne platforms. However, single frequency and single polarization data allows one to map saturation-prone areas using a multi-temporal analysis. The use of multi-temporal data makes it possible to retrieve spatial soil moisture patterns within the studied catchment by applying statistical methods to the time series of images. Two methods for the analysis of a winter time series of ERS-1 and ERS-2 images, for which constant roughness and vegetation conditions can be assumed, are suggested. The first method is based on the temporal coefficient of variation. Since the variability of soil moisture is expected to be smaller near a stream then further upslope from the stream, a smaller temporal coefficient of variation of the returned signal is observed near streams. The second method makes use of principal component analysis of the winter time series of images. Both methods lead to a representation of the spatial distribution of the soil moisture at the catchment scale. However, principal component transformation performs better since it can separate the soil moisture component in the backscattered signal from other influencing factors such as topography and land use.