Sensitivity of water vapor distribution to the land surface parameterization schemes in the Advanced Weather Research and Forecasting model

Land surface parameterization plays a significant role in the accuracy of the meso-local scale water vapor exchange in the numerical models. The goal of this study was to evaluate two (Noah and RUC) land surface models (LSMs) in the Advanced Research and Weather Forecasting model (ARW) for varying topography, landuse and soil characteristics in relation to the moisture transport in the boundary layer. Water vapor exchange depends on several other processes, such as the boundary layer dynamics, microphysical, and convective and radiative processes. To isolate the influence of land surface schemes, we used similar initial and boundary conditions and other physical parameterizations in two ARW model runs. The main objective of the study was to optimize high resolution ARW model with proper physical parameterizations for frost assessment and warning and for other agricultural decision making applications in Georgia. Downscaled surface and boundary layer characteristics during a 30-day period corresponding to different weather events over Georgia are presented, including frost occurrences which are important for agricultural decision making. Model results were initially verified with observations from the Automated Environmental Monitoring Network (AEMN) across Georgia. A comparison of surface temperatures and mixing ratios between the models showed a strong dependence on the terrain, landuse, and soil characteristics and on the large-scale moisture transport. The PBL profiles indicated that Noah land surface model produced drier and colder boundary layers (PBL) compared to RUC land surface model in the presence of low winds and during high pressure situations with more stable boundary layers. While mixing ratios using RUC land surface model were closer to observations at lower elevations, Noah results showed a good comparison over high terrains. Differences between the two LSMs were significant over cropland and grassland areas compared to forested areas, which needs further investigation. These results also have implications to the coupling of ARW with the biosphere models.