Comparison of Meteorological Simulations created using Nudging, 3DVar and 4DVar with Respect to Key Meteorological Inputs for Air Quality and Dispersion Models

It is well known that meteorological conditions are an important factor in air quality. Meteorological quantities such as temperature, insolation, winds, and relative humidity strongly influence the evolution of chemical species within the atmosphere. Therefore, air quality and dispersion models require the best estimate of the state of the atmosphere to provide accurate predictions of chemical concentrations of species and the dispersion of those species throughout the atmosphere. Numerical meteorological models provide that best spatial and temporal resolutions of meteorological variables that are necessary to drive air quality and dispersion models, but uncertainties remain. Typically, Newtonian relaxation, or nudging, is used to constrain the meteorological model so that errors are kept to a minimum, producing a better meteorological analysis field to drive air quality and dispersion models. However, with continuing advances in meteorological data assimilation techniques and available computation power, it has become of interest to determine the continued effectiveness of nudging when compared to these more sophisticated assimilation techniques. This study assesses the Weather Research and Forecasting (WRF) meteorological model performance when a three-dimensional (3DVar) or four-dimensional (4DVar) variational assimilation technique is used to assimilate observations into the model. The benefit of these variational assimilation techniques is the ability to assimilate non-traditional observations, such as satellite radiances. These simulations are then compared to a traditional WRF model simulation, which employs analysis nudging to improve model performance. All of these model simulations were assessed with respect to insolation, clouds, temperature, wind speed and direction, and mixing ratio, which are key meteorological inputs for air quality studies. It was found that while using 3DVar and 4DVar assimilation with WRF provides slightly better results with respect to temperature and mixing ratio, a WRF model simulation which uses analysis nudging provides the best performance with respect to clouds, wind speed and wind direction which are vitally important to air quality and dispersion models. Further analysis of these three WRF simulations will be presented.