Abstract:
Remote sensing advancements in recent years have enabled the monitoring of the Earth’s land surface withunprecedented scale and frequency. In the past two decades, for example, remote sensing observations of the land surface (e.g. soil moisture, snow cover, terrestrial water storage, land surface temperature, vegetation, among others) have become available from a number of satellite instruments and platforms. At NASA, a comprehensive land data assimilation environment called Land Information System (LIS; lis.gsfc.nasa.gov) has been developed to enable the effective synthesis of these remote sensing observations with modeled estimates. LIS has been utilized for the assimilation of these remote sensing retrievals both serially and concurrently, over continental and global domains. These assimilation studies have demonstrated the beneficial impact of remote sensing measurements both for improving the representation of water, energy and carbon processes as well as downstream applications (e.g. weather forecasting, drought/flood monitoring etc.). Despite these advancements, there are significant challenges related to assimilation strategies, limitations in model formulations, and observational data processes, that limit the potential utility of the remote sensing measurements. These issues are particularly challenging over the land surface, where the impacts of natural heterogeneity and human management are complex and difficult to characterize accurately. We will present results from recent studies that describe the limitations of the data assimilation strategies when unmodeled processes dominate the observed signals. Many of these limitations can be attributed to the legacy of the land surface models, which have essentially limited the observability of the modeled outputs (e.g. soil moisture). Along with improved data assimilation strategies, use of advanced data fusion techniques and fundamental changes to the model representations are necessary for the realizing the full information content of remote sensing data through data assimilation.