Integrating Vegetation Phenology in Soil Moisture Change Detection from ASAR Wide Swath Images
Van doninck, Jasper; De Baets, Bernard; Verhoest, Niko
Ghent University, BELGIUM

The estimation of soil moisture variations using change detection algorithms and multitemporal SAR data is becoming increasingly popular, and will continue to do so with the upcoming launch of the first Sentinel satellite. Change detection algorithms are based on the assumption that soil moisture changes occur on much shorter time scales than changes in soil roughness and vegetation cover, so that changes in SAR backscatter can be attributed to changes in soil moisture content. In reality, however, the influence of vegetation will change seasonally throughout the year, as will the influence of roughness.

In the multitemporal change detection algorithm for the coarse resolution Scatterometer and ASCAT, developed at TU Wien, seasonal effects are eliminated using the multi-angular capability of these sensors. In this study, an attempt is made to integrate phenology in a change detection algorithm for ASAR Wide Swath data. The study is performed on a time series of 130 images over the region of Calabria, Italy. A soil moisture model was used for validation. Correlation between modelled soil moisture and ASAR WS soil moisture were found to be moderate to strong, depending on land cover.