There have been a number of studies demonstrating the sensitivity of near-surface atmospheric variables, specifically air temperature, moisture and wind fields, to land surface and sub-surface parameters. Skin temperature, soil moisture, and soil temperature are known to influence near-surface atmospheric variables, including 2 m (screen) temperature and 10 m wind fields. In earlier work, two of the authors (LML, DJS) demonstrated the influence of vegetation on these same variables, as well as on precipitation and moist convective initiation.

In this study, we will report results from the assimilation of Oklahoma Mesonet land surface data and soil parameters into a high-resolution numerical model (HIRES) developed by one of the authors (LML). Using the adjoint of the land surface model, sensitivity studies were performed on a set of equations that have only two independent variables: the vertical coordinate and time. The system of equations is linear, and the assimilation system allows for a very economical and detailed study of the topology of the penalty functional. It was found that the most important parameter in the several days explored in this initial study was the soil temperature. Moreover, the procedure was shown to be efficient even for initial guess (background) errors larger than measurement uncertainties.

The next step is to demonstrate the dependence of near-surface parameters and deep soil moisture on vegetation characteristics, included in the model as leaf area index (LAI) and fractional vegetation coverage (FVEG). Using LAI and FVEG values from the biweekly maximum normalized difference vegetation index (NDVI) composites at 1 km resolution obtained from daily observations from the Advanced Very High Resolution Radiometer (AVHRR), short range predictions of temperature and moisture will be carried out via HIRES over the southern Great Plains and will be verified against observations from the Oklahoma Mesonet.