A Statistical Downscaling Approach of Soil Moisture Estimations by Synergistically using Optical Remote Sensing and Ground Soil Moisture Measurements at the Valencia Anchor Station

A multifractal technique has been used to downscale 1 km optical remote sensing MODIS derived soil moisture index (SMI) to the scale of interpolated soil moisture map produced by ground measurements at the Valencia Anchor Station (VAS) during the SMOS Validation Rehearsal Campaign (2008) with the spatial resolution of 32 meters. Scale invariance assessment shows a constant behavior of soil moisture variability at all scales of aggregation. This result proves the homogeneity of the VAS region from a mathematical point of view and exempts or allows us from using ancillary data such as topography, soil texture and vegetation characteristics in our downscaling model. Our predicted soil moisture values compared to the observed ground data show RMSE ranges from 0.026 to 0.039 for 2008/05/02, indicating accurate predictions for this date. However, there are high RMSE values in the range of 0.761 to 0.784 for 2008/04/24, due to rainfall events (30 mm accumulated) occurring in the region a few days prior to the measurements, which influenced the result of the downscaling model. At the same time, the strong correlation (77%) between the predicted and the observed data is promising and warrants further application of the model to other homogeneous areas with or without rainfall events.