Wednesday, 9 January 2013: 9:45 AM
Room 16A (Austin Convention Center)
William F. Ryan, Pennsylvania State University, University Park, PA; and N. Wiles
Coupled chemistry-transport models (CTM) have been used in support of operational air quality forecasting in the United States for nearly a decade and have become a standard guidance tool for short range ozone (O3) forecasting. CTM forecasts of O3 concentrations are sensitive to both meteorological and chemical variables. The most sensitive variations include temperature, wind velocity, boundary layer depth, as well as the emissions and chemical reaction pathways of O3 precursors and the transport of both O3 and its precursors. In addition, the standard predictand for metropolitan scale forecasts is peak domain-wide 8-hour O3 concentrations. These sensitivities, and the extreme value nature of the forecast, suggest that an ensemble approach to numerical air quality forecasting may be useful. The complexities of the CTMs make a standard ensemble approach, one model with a series of perturbations, too computationally expensive to carry out at the current time. However, there are a number of operational and research O3 forecast models currently available. An alternative approach is to develop an ensemble formed from a set of these independent models a so-called poor man's ensemble.
In the summer of 2011, a pilot program was undertaken in the Philadelphia metropolitan area using a set of three O3 forecast models: The NOAA Air Quality Forecast Capability Model (NAQFC), the Barons Meteorological Services Model (MAQSIP) and the SUNY-Albany model (SUNY). The results of that study showed that forecast guidance at the all-important air quality warning threshold Code Orange or 76 ppbv 8-hour average of several ensembles outperformed any individual model. In this study, the 2011 results were used to refine the choice and weighting of ensemble members and extended to two forecast areas in 2012: Philadelphia and the State of Delaware. Preliminary results (as of early August) show improvement in skill scores at the warning threshold using ensemble forecasts. In particular, there is a large reduction in the number of false alarms of poor air quality. As uncertainty exists regarding the future of the NAQFC, this study is able to provide a quantification of its utility in a multi-model setting.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner