The scope of this project includes stack monitoring of PM2.5, investigation of particle size and extinction coefficient characteristics, measurement of visibility using digital image processing, and air quality forecasting.
In the year 2021, PM2.5 stack monitoring were performed on the municipal waste incinerator in Houli , as well as the solvent waste incinerator of Shine-yo Resource ＆ Recycling Co., Ltd. and the electric arc furnace of Feng Hsin Steel Co., Ltd. The results indicate that FPM（Filterable Particulate Matter） were 0.60 mg/Nm3, 2.06 mg/Nm3 and 0.060 mg/Nm3. Also, CPM （Condensable Particulate Matter）were 1.50 mg/Nm3, 18.1 mg/Nm3 and 0.37 mg/Nm3.
We monitored the characteristics of particulate matter and light extinction in the coastal Longjing District of the City of Taichung. The extinction coefficient was between 3 Mm-1 and 115 Mm-1. Low visibility was mainly attributed to scattering. On average, the contribution of scattering was 69%, while the contribution of absorption was 31%.
The multiple linear regression formula was based on the actual measured scattering coefficient and the chemical composition of PM1. Among the chemical species, the highest scattering efficiency was nitrate, at 15.8 m2/g, and followed by sulfate and organic carbon, at 4.6 m2/g and 4.2 m2/g. For Taichung City, nitrate has the greatest impact on visibility. Most nitrates are secondary pollutants, and the precursor is nitrogen oxides (NOX). Therefore, the control strategies to improve the visibility effectively and quickly is that should focus on reducing the emissions of nitrogen oxides in industrial and traffic.
We used digital image processing to measure visibility in the Shalu, Beitun, and Wuri Districts of the City. The methodology used was the following:
1. Capture an image with a camera.
2. Transform the image into grayscale values with the aid of software.
3. Transform differences in values into values of visibility with a formula.
Analysis data from March through November, value differences of imagery brightness exhibited a linear relationship with visibility at the Beitun station and the Hungkuang station, while the relationship was exponential at the Wuri station. We compiled the daily visibility above 20 km statistics at noon. At the Beitun station, the visibility with the lowest probability was at 13.6%. And followed by the Hungkuang station with 32.8%. The visibility above 20 km with the highest probability was at the Wuri station, at 56.6%.
We conducted air quality forecasting for the upcoming 48 hours, using data from March through November. Regarding three-hourly AQI （Air Quality Index）predictions, the accuracy for AQI=101~150 was between 82% and 94%. For AQI>150, the forecasting accuracy was between 88% and 96%.
