In computer models of the atmosphere, calculations are carried out at horizontal intervals. Thus a calculation is representative of an area in which the fine-grain detail of natural variations is smeared out. Formulas included in the model that attempt to account for important thermal and dynamic effects of such detail, at least in some statistical sense, are referred to as "parameterizations". The land surface, in particular, norm- ally contains significant horizontal variations in topography, vegetative cover, moisture, and thermal characteristics within a model grid square of typical dimensions (ranging from 10 sq km in mesoscale models to 1 million sq km in models of global extent). The two approaches used in current state-of-the art parameter- izations either do not account for subgrid-scale heterogenity of complex biophysical mechanisms, or neglect the effects of known subgrid scale land-atmosphere interactions such as slope-valley winds, lake breezes, and other thermal circulations caused by varying vegetative cover. Before improvements in current model computer algorithms are implemented it is important to assess the impact of such variations on the models' behaviour at differing spatial scales. Avissar proposes to carry out such assesments, by using a combination of a very flexible (in terms of optional resolution) and well-tested computer model (the Colorado State University- Regional Atmospheric Modelling System), and detailed low-level measurements made by flight transects over hilly, bare, agricultural, and forested areas in two past observational programs. The PI's prior research in boundary-layer and agricultural meteorology as well as in developing regional models of the atmosphere will contribute to advances in this problem area, which has application to improvement of short, medium and long- term forecasting, as well as research in climate change on regional and global scales.
