Tuesday, 11 February 2003: 11:00 AM
Use of numerical ozone prediction models in operational air quality forecasting
Advances in chemistry and meteorology models and computational efficiency have allowed operational use of coupled chemistry-transport models by operational air quality forecasters. The adoption of numerical guidance by operational forecasters, the number of whom have increased rapidly in the past several years, depends on the reliability of model guidance in critical high ozone cases. In turn, routine use of numerical models by operational forecasters can result in a positive feedback of information to model developers to further improve model performance. The MAQSIP-RT model, operated as part of the South East Center for Mesoscale Environmental Prediction (SECMEP), has provided a variety of locations with operational numerical forecasts for the past several years. This paper presents an analysis of meteorological conditions and model performance for teh August 1-10, 2001 regional ozone episode in the Washington-Boston Corridor. This episode was of the type commonly observed in the summer season. The standard measure of performance for local air quality forecasts, which drive metropolitan-wide health warnings and public-private pollution control strategies ("Ozone Action Days"), is metropolitan-wide peak ozone. While this measure is not wholly appropriate for evaluation of numerical models, even under this stricter standard the MAQSIP-RT results for Philadelphia improved on standard regression-based approaches on the order of 0-10%. Similar results, with some differences due to bay-land interface considerations, apply for the Baltimore-Washington region. These results are consistent with preliminary results from 2002 including a complex regional case from July 15-19.