Using AMDAR to Assess the Urban Boundary Layer in WRF

Continued urbanization and a warming climate will exacerbate the urban heat island (UHI) and aggravate health risks of urban populations. In addition to satellite and surface observations, high resolution numerical simulations are often used to study the UHI. Less attention is typically given to the urban boundary layer (UBL) mainly because upper-air observations are sparse or limited to short field campaigns, which also makes verification of model profiles challenging. The continued growth of observations from the Aircraft Meteorological Data Relay (AMDAR) provides essential measurements of the lower atmosphere, which are collected by many commercial airplanes during landing and takeoff. AMDAR data from airports reasonably close to urban centers yields many profiles of the UBL per day that can be used to evaluate model performance and better understand the lower atmosphere over urban areas. The Weather Research and Forecasting (WRF) modelling system is used to simulate the urban environment of the Dallas-Fort Worth metropolitan area during recent heat waves. Model output is evaluated using AMDAR soundings from two airports (DFW and DAL) to not only assess overall performance but also to help identify specific model deficiencies and highlight the important role of the lower atmosphere in influencing surface temperatures and the magnitude of the urban heat island. Focus is placed on the interaction of urban canopy model and planetary boundary layer schemes. Major differences are found in the lower atmosphere primarily in the morning and evening hours with the most similar profiles during the daytime. Warm biases are often present at the surface in the model and are related to deeper mixing in the model than observed. Morning observations often indicate a shallow mixed layer near the surface capped by a stable layer associated with a temperature inversion or an isothermal layer. Some cases reveal situations where the layer of stability is not captured by the model and the temperatures at the top of the layer are often greater than simulations. This likely represents warmer temperatures that remained confined to the residual layer overnight instead of being mixed out as the model suggests. However, certain simulations have been able to resolve the features of the nocturnal boundary layer in some cases, albeit the inversion is usually at a lower level or not as well defined as the observation. The relative accuracies of these simulations are usually dependent on which closure scheme they employ. Further work will attempt to refine the simulations through analysis of model accuracy through adjustment of the parameterizations of both the canopy models and the closure schemes.