Performance evaluation of the updated air quality forecasting system for Seoul predicting PM_(10)

摘要

The performance of the updated Air Quality Forecasting System (AQFS) using Weather Research and Forecasting (WRF v.3.1) and the U.S. EPA's Models-3/Community Multiscale Air Quality (CMAQ v4.6) with emphasis on PM_(10) (Particulate Matter with aerodynamic diameters less than 10 urn) forecast is evaluated over the Seoul Metropolitan Area (SMA) for 2010. The simulations of the updated and old forecasting systems are compared with air quality and meteorological measurements in the modeling domain. The results of the analysis show that the updated forecasts of daily PM_(10) can reproduce the magnitude and temporal variation of the observations. The time variations of forecasted PMio are in good agreement with the observations with the range of Index of Agreement (lOA) over 0.7. The forecasted concentrations of daily PM_(10) are underestimated in all forecasting regions with a range of Normalized Mean Bias (NMB) from -10.76% in the Seoul Metropolitan to -21.29% in the North Gyeonggi province. The discrepancy can be attributed to uncertainties in emissions, forecasted meteorology and models. Even with persistent uncertainties in emission data, the PM_(10) forecasts from the updated system with emission inventories of INTEX-B for year 2006 in Asia, as well as CAPSS supplementing fugitive dust and biomass emissions for year 2007 in Korea, perform better than those of the old system, which consists of MM5 (v4.7) and CMAQ (v.4.3) and the emission data from TRACE-P for 2000 in Asia and CAPSS for 2003 in Korea. It is also demonstrated that the forecasting system is effective to detect the onset time of the episode and peak value of PM_(10) in advance which is mainly caused by the long-range transport of aerosols from eastern China to the SMA. The Probability of Detection (POD) for the "C" category of Air Quality Index (AQI), which indicates a health risk for the sensitive group, improves to over 60% by applying the bias-adjustment of hybrid forecast (HF).