Application and Evaluation of MM5 for North Carolina with a 4-km Horizontal Grid Spacing

Accurate meteorological predictions are essential for the accurate simulations of ambient air quality. In this study, the Fifth-Generation NCAR/ Penn State Mesoscale Model (MM5) with a 4-km horizontal grid spacing is applied for two 1-month episodes (August and December ) in year 2002 over the state of North Carolina (NC). The objective is to generate meteorological inputs for the U.S. EPA Community Multiscale Air Quality (CMAQ) modeling system to simulate the fate of emissions from animal and agricultural facilities on a regional scale. The MM5 simulations are conducted with and without nudging. A 3-D analysis nudging is performed for temperature and moisture aloft, and both 3-D and surface analysis nudging are conducted for wind fields. The MM5 results are being analyzed and evaluated against observational data at over 90 sites in NC using various graphical and statistical methods, domain-wide and at specific sites. A preliminary evaluation shows that MM5, when nudged, predicts the wind speed very well; overestimating the observed values by roughly 7% (9% without nudging). These statistics are calculated by ignoring the observed values less than the minimum threshold of the instrument measuring the windspeed (3 knots). The temperature is underestimated by 14% when nudged compared with 13% without nudging. The relative humidity is slightly better predicted without nudging (under prediction by 1.2% vs. 0.8% with nudging). The hourly time series plots also show a similar trend to statistical measures. Possible reasons leading to these discrepancies are being examined. The MM5 predictions with the 4- and 12-km horizontal grid spacings are being compared to access the model's capability in reproducing observations at different grid resolutions. The inaccuracies and uncertainties in predicted meteorological fields will be analyzed; their implications for simulating the fate of air pollutants, especially ammonia, with CMAQ will be discussed. Further study will be conducted to evaluate the effect of using a smaller time resolution (e.g., 10 minutes rather than 1 hour) for MM5 meteorological fields on air quality predictions.