An evaluation of the land surface-atmosphere interactions over a heterogeneous landscape in numerical mesoscale model

Land surface-atmosphere processes play an important role in the water and energy cycle balance. However, there are still uncertainties of about 10-20 % in their estimates from current methods (Brutsaert, 1998). This is partly due to the irregularity of turbulence and the variability of the land surface that are often met in reality. Additionally, in current numerical models, most schemes are designed for steady flow over uniform surfaces (e.g., Brutsaert, 1998; Arola, 1998; Bunzli, 1998) so that they do not account for surface heteorgeneity. Therefore, a better representation of these physical processes in numerical model is a key for reliable forecast. Land-atmosphere interactions are defined by a number of physical quantities such as the surface fluxes and the mixing-layer height. Although, the theoretical framework for these quantities is understood, their representation and computations in numerical model still pose problems. In this study, we propose to use both the Weather Research and Forecasting (WRF) model and in-situ observations over the Howard University Beltsville research site (39.054°N; 76.877°W) that will be supplemented with flux data from the National Oceanic and Atmospheric Administration (NOAA) DCNet/URBANet network to evaluate the energy exchange and water transport between the surface and the atmosphere over a rural-urban environment. Intensive observation period was conducted during the summer of 2004 and preliminary evaluation of the surface fluxes and the mixing-layer height was made for two contrasting case studies. The model results were contrasted with in-situ observations made at the Howard University Beltsville site. Results showed that the model underestimated the height of the mixed layer during the stable case, and overestimated it during the unstable case. This is probably due to an inconsistency in the mechanism limiting the mixing in the model. Furthermore, for both cases, the model overpredicted the sensible heat flux. This is most likely in response to an inherent problem in the model as in the calculation of the eddy diffusivity coefficients. Sensitivity experiments and model performance assessment will be performed to evaluate the key parameters in the physical processes and to identify the bias of the model.