A sensitivity analysis of WRF-simulated cold air damming using variable parameterization schema
Simms, Morgan M.
Cold air damming (CAD) is a common phenomenon in Northern New England U.S.A. in which cold air persists to the south and east of the northern Appalachian Mountains during region-wide warm air advection. CAD often allows for wintry precipitation to persist for extended periods and can negatively impact transportation and local commerce in Northern New England, especially during the winter months. Numerical models are known to struggle with the evolution of CAD, particularly the erosion process and the portrayal of the vertical temperature profile and precipitation type. Numerical models often mix-out the cold air too soon, which can lead to the unanticipated continuation of hazardous wintry weather. This study assessed how the WRF-ARW simulated surface cold pool evolution associated with CAD, and the changes to this evolution associated with different planetary boundary layer (PBL) and microphysical parameterization schema. The project repeated this analysis for three different classes of CAD based on the classification scheme of Bailey et al. (2003) to generate respective eight-member ensembles of 2-meter xvi temperature forecasts, as well as time-height analyses of temperature and vertical temperature gradients from the surface to 800 hPa. Analysis of the ensembles showed that more definitive temperature biases existed between PBL schema, while no distinct temperature bias existed between microphysical parameterizations. The local Mellor-Yamada-Janjic (MYJ) scheme simulated an erosion of the surface cold pool more in line with observations than the non-local Yonsei University (YSU) scheme. However, both PBL schemes completed the erosion process quicker than observations and produced warmer temperatures than observed in the three analyzed cases. The sample size of cases is too small to definitively say whether the WRF accurately simulates a particular class of CAD better than another, or whether a clear temperature bias occurs during the entire CAD event rather than just the erosion process.