A data assimilation-based approach for estimating land surface variables

Abstract

Earth system models and many other applications require high-quality land surface variables retrieved from remote-sensing data. The current land-surface parameter products are generally retrieved from remote-sensing data acquired only at a specific time, which results in a lack of spatial and temporal continuity as well as inaccuracy for some vegetation types. Data assimilation methods can incorporate prior knowledge into the inversion process in an objective manner and consider errors in both satellite observations and dynamic models to produce spatially and temporally continuous land surface variables with relatively higher quality, as well as allowing for uncertainty in the retrieved land surface variables. This article introduces the data assimilation framework and two case studies are presented to illustrate the estimation of land surface variables from remote sensing data using data assimilation techniques.