A0526
Title: Hierarchical Bayesian model for unmixing remote sensing data
Authors: Joon Jin Song - Baylor University (United States) [presenting]
Jonathan Hobbs - Jet Propulsion Laboratory (United States)
Colin Lewis-Beck - Iowa State University (United States)
Anirban Mondal - Case Western Reserve University (United States)
Xin Wang - San Diego State University (United States)
Zhengyuan Zhu - Iowa State University (United States)
Abstract: Remote sensing satellites often have coarse spatial resolution and provide aggregate estimates of surface characteristics over wide footprints with heterogeneous ground cover. Disaggregating or un-mixing the observed signal is useful for identifying individual vegetation types and their unique signatures. We propose a hierarchical Bayesian model to deal with the un-mixing problem, which integrates information from multiple sources. Combining additional ground-based data sources with the satellite data allows us to identify the separate crop signals that make-up the original satellite measurements. The methodology is illustrated with data from the SMOS (Soil Moisture and Ocean Salinity) satellite mission.
