Determining Water Body Characteristics of Doana Shallow Marshes Through Remote Sensing

We analyze the potential of Landsat TM and ETM images to discriminate inundation, depth and turbidity patterns in the very heterogeneous shallow marshes of Donana National Park. According to the results we will reconstruct historical changes in such variables with a long time series of Landsat images (MSS, TM and ETM+). For this purpose we sampled 334 ground-truth points simultaneously to 6 Landsat scenes during springtime of 2004, 2005 and 2006. Then we applied statistical models to field data and we predict inundation level, depth and turbidity at every sample unit with reflectivity data. Results show that SWIR band is the best predictor of inundation level at any point (especially in sediment-charged waters and medium-high plant cover). Therefore we propose two statistical models explaining 31 % of deviance for water turbidity and 70 % of deviance depth in inundated areas.