Using area-average remotely sensed surface soil moisture in multipatch land data assimilation systems

In coming years, Land Data Assimilation Systems (LDAS) two-dimensional (2-D) arrays of the relevant land-surface model) are likely to become the routine mechanism by which many predictive weather and climate models will be initiated. If this is so, it will be via assimilation into the LDAS that other data relevant to the land surface, such as remotely sensed estimates of soil moisture, will find value. This paper explores the potential for using low-resolution, remotely sensed observations of microwave brightness temperature to infer soil moisture in an LDAS with a "mosaic-patch" representation of land-surface heterogeneity, by coupling the land-surface model in the LDAS to a physically realistic microwave emission model. The past description of soil water movement by the LDAS is proposed as the most appropriate, LDAS-consistent basis for using remotely sensed estimates of surface soil moisture to infer soil moisture at depth, and the plausibility of this proposal is investigated. Three alternative methods are explored for partitioning soil moisture between modeled patches while altering the area-average soil moisture to correspond to the observed, pixel-average microwave brightness temperature, namely, 1) altering the soil moisture by a factor, which is the same for all the patches in the pixel, 2) altering the soil moisture by adding an amount that is the same for all the patches in the pixel, and 3) altering the change in soil moisture since the last assimilation cycle by a factor which is the same for all the patches in the pixel.