(1) Provide hardware and software maintenance of EPA’s “Air Quality Forecast Auxiliary System” and update relevant database in order to facilitate the operation of daily forecast of air quality and UV index; (2) Perform daily simulation of atmospheric dust deflation and transport from East Asia to the Taiwan area, predict the arrival time, duration, concentration and extent of dust incursion so as to provide a 48-hr advance notice. An overall evaluation of the forecast results was conducted. It was found that the TAQM/kosa model performed slightly better than the dust forecast model of US Naval Research Laboratory for the spring of 2004. However, the performance of TAQM/kosa in 2004 degraded somewhat from 2003, owing to a complete miss of a case in January and an overestimation of a case in April. For the former, we found the causes to be errors in the relative humidity and temperature fields in TAQM/kosa. (3) Develop a dynamic air quality model to provide simulations of local and foreign air pollutants including those from biomass burning in East Asia, in order to provide 24 to 48 hr forecasts to assist routine operation of air quality forecasts in the future. Five highly polluted cases have been tested with the dynamic air quality model. The model has been installed on EPA’s operation computer, and pre-operation tests of automatic daily forecasts have been performed. (4) Analyze the data quality EPA’s boundary layer temperature profiler, and analyze the applicability of these data in improving air quality forecasts. The data quality looks fine according to our preliminary analysis. We also found significant positive correlation between the inversion height that derived from the data and the concentrations of major air pollutants. However, further application of these data in the forecasts of air pollution requires in-depth understanding of the causes of the correlation.

dust, biomass burning,stability,air pollution ,UV