This study covers the analysis of atmospheric PM2.5 chemical composition, measurement of PM2.5 mass concentration in flue gas, monitoring of PM2.5 concentration near bus stops, as well as the forecast of air quality in Taichung. Whenever the air quality tends to deteriorate with Air Quality Index (AQI) higher than 150 within the next 48 hours, the Environmental Protection Bureau (EPB) of Taichung City Government will be notified in advance to take emergency measures. The chemical composition of PM2.5 at Houli, Dali, and THU sites is analyzed. The ranges of relative amount for each component in PM2.5 are organic carbon (19.2-24.9%), sulfate (20.0-23.3%), ammonium (4.9-7.9%), elemental carbon (6.2-8.9%), nitrate (6.2-8.9%), and metallic elements (2.8-9.9%). The result of measuring PM2.5 in flue gas from Taichung Power Plant in 2019 shows that the mass concentrations of filterable particulate matters (FPM) and condensable particulate matters (CPM) fall in the ranges of 0.7-1.85 mg/Nm3 and 11.9-24.4 mg/Nm3, respectively. The concentrations of PM2.5 are monitored at bus stops in the morning, at noon, and in the evening. The result indicates that the concentration of PM2.5 is highest during the commuting rush hours in the morning while that of PM2.5 is lowest during the heavy traffic in the evening, implying that the diurnal variation of PM2.5 cannot reflect the change in the volume of traffic flow. However, the trend of PM2.5 at bus stops is similar to that at ambient monitoring sites, which means that the influence of background atmosphere on the PM2.5 concentration at bus stops is more significant than that of the volume of vehicles. The accuracy of air quality forecast is 55% in April and 73% in May. The relatively lower accuracy in April was due to the capricious weather systems in spring along with the higher uncertainty of the weather forecast. The Community Multiscale Air Quality Modeling System (CMAQ) is integrated with the air quality forecasting model of the European Centre for Medium-Range Weather Forecasts (ECMWF) and adopts neural network to assist with the forecasting and correct the predicted numerical values since September. Afterwards, the accuracy of air quality forecast steadily increases and reaches 77% in October.

Composition of PM2.5、Filterable Particulate Matter (FPM)、Condensable Particulate Matter (CPM)、Air Quality Forecast