The Effect of Boundary-Layer Scheme on WRF model simulations of the Joint Urban 2003 Field Campaign

Atmospheric plume models require wind direction, wind speed, atmospheric stability, boundary-layer depth and/or turbulence input parameters in order to accurately predict the transport and dispersion of contaminants released within the atmospheric boundary layer. It can be difficult to extract all of the required information from existing meteorological monitoring networks due to temporal and spatial resolution of the measurements and the types of measurements that are typically included in the network. This makes the use of mesoscale numerical weather prediction codes, such as the Weather Research and Forecasting (WRF) model, an appealing source of meteorological data for transport and dispersion modeling. We used the Joint Urban 2003 dataset to evaluate the ability of the WRF model to provide the necessary meteorological data required by transport and dispersion simulations in the place of in situ measurements. Three different boundary-layer schemes were used in the simulations including the Mellor-Yamada-Janjic (MYJ), Yonsei-University (YSU), and Mellor-Yamada-Nakanishi-Niino (MYNN) schemes. The amount of error in the predicted WRF model fields was found to be dependent on the boundary-layer scheme used in the simulation. One of the most notable differences between the simulations using different boundary-layer schemes was that there were multiple time periods over the course of the Joint Urban 2003 field campaign when the MYJ and YSU schemes both predicted large deviations from the prevailing wind direction patterns that were not found in the observations. The simulation using the MYNN scheme did not produce the same spurious predictions contributing to a superior mean absolute error in predicted wind direction over the course of the field campaign.