DOWNSCALING SMAP SOIL MOISTURE WITH ECOSTRESS PRODUCTS USING A GRAPH-BASED INTERPOLATION METHOD

Technologies such as radiometry, radar or synthetic aperture radar have demonstrated the potential of remote sensing to observe the Earth’s land surface and estimate variables that influence the climate system. Among the remote sensed variables, soil moisture has be- come increasingly important as the impacts of climate change on water resources continue to intensify. Soil moisture products, nor- mally retrieved from microwave remote sensing data, are typically not suitable for regional hydrological and agricultural applications such as irrigation management and flood predictions, due to their coarse spatial resolution. Providing accurate information on soil moisture at an appropriate temporal and spatial scale is challeng- ing for traditional interpolation methods, due to the high variability of soil moisture. Aiming to provide fine resolution soil moisture es- timations, in this paper we evaluate the performance of our proposed graph-based downscaling method to obtain fine resolution soil mois- ture data (9km) from coarse satellite observations (9km) using very fine resolution evapotranspiration data (30m). Our approach for- mulates the data interpolation problem as a signal reconstruction on a graph, where coarse soil moisture observations are signals at the nodes, while high resolution evapotranspiration data is used to com- pute the weights of the edges connecting the nodes.