With an increase in the amount of research into vegetation-atmosphere interactions, the parameterizations that the current atmospheric models use have gotten increasingly sophisticated in order to treat these complexities with full efficacy. The ability to make remote measurements of vegetation has also improved, giving more data to these parameterizations. With these consistent advances in characterizing the land surface, it is not surprising that many numerical models have improved as well to incorporate current vegetation data into their own models to serve as a basis for model calculations with the land surface.

The Regional Atmospheric Modeling System (RAMS) uses the LEAF-2 submodel (Walko, 2000) to compute the surface fluxes for many different surface and vegetation layers. LEAF-2 is a prognostic model for the temperature and water content of snow cover, soil, vegetation, as well as canopy air, and computes the turbulent and radiative exchange between these components and the atmosphere. With context to the RAMS model, the LEAF-2 model provides a very important component by modeling the interactions between the differing parts of the surface roughness layer and then feeding this data into RAMS. The LEAF-2 model uses a look-up table based on a USGS vegetation classification to decide the various vegetation characteristics of the land surface for a certain grid point. These factors include the leaf area index, vegetation albedo, fractional coverage, root profile, and roughness height. Since the data that is being provided is limited by the 30 differing vegetation classes programmed into LEAF-2, there is not as much heterogeneity as would be using real field data for the model simulation.

This study uses scientific data sets from the MODIS-Terra satellite in order to provide a real land surface input to the model and compares model simulations using the USGS vegetation dataset as well as simulations that utilize MODIS-derived leaf area index and albedo grids. Surface fluxes are compared and the model output is compared to reality in an attempt to improve the model output.