空氣污染（大氣氣膠）對全球所造成的影響已成為現今重要的議題，除氣候變遷外，前人研究顯示〖ＰＭ〗_2.5對人類的健康有顯著的影響，空氣污染的監測因此重要。地面〖ＰＭ〗_2.5測站能精準的評估空氣污染，但常受限於測站的空間分佈，因此本研究使用ＡＥＲＯＮＥＴ的氣膠光學厚度（aerosol optical depth, ＡＯＤ）及〖ＰＭ〗_2.5 測站資料，並根據沙塵（ＤＳ）、人為污染（ＡＰ）及生質然燒（ＢＢ）等氣膠種類來建立ＡＯＤ及〖ＰＭ〗_2.5的關係式，應用於衛星監測以克服地面測站之限制。除氣膠種類外，本研究亦進一步考慮相對濕度對ＡＯＤ及〖ＰＭ〗_2.5關係式的影響。結果顯示不同種類ＡＯＤ及〖ＰＭ〗_2.5的相關係數在使用ＮＧＡＩ（Normalized Gradient Aerosol Index）進行氣膠種類分類後，相關係數分別改進為0.815(DS) 、0.693(AP)及0.741(BB)， 加入相對濕度後則可提昇至0.853（DS）、0.707(AP)及0.768(BB)。最後將研究所建立的關係式應用至ＭＯＤＩＳ的ＡＯＤ產品進行〖ＰＭ〗_2.5濃度之估算，在時間與空間的分布上均獲得不錯之結果，顯示本研究所建構之ＡＯＤ及〖ＰＭ〗_2.5濃度在衛星反演〖ＰＭ〗_2.5濃度具相當的實用價值。;Air pollution is a hot issue caused serious problems all around the world. Previous studies have shown that PM2.5 concentrations have a strong influence on human health. PM2.5 ground-based stations are appropriate to evaluate air pollution with high accuracy but typically limited in spatial distribution. This study used AOD from AERONET stations and PM2.5 concentrations within study area to representing 3 types of aerosols: dust (DS), anthropogenic pollutants (AP) and biomass burning (BB) to build relationships between AOD and PM2.5 concentrations. In addition, the study figures the relative humidity effect on the relationship between AOD and PM2.5 concentrations. The results show that the correlation coefficient between AERONET AOD and PM2.5 concentrations were 0.6, 0.652 and 0.729 for DS, AP, and BB source region respectively. After NGAI (Normalized Gradient Aerosol Index) method applied, the correlation coefficient improved to 0.815, 0.693 and 0.741 for DS, AP, and BB, respectively. The correlation coefficient became as high as 0.853, 0.707 and 0.768 for DS, AP, and BB, respectively after considering RH correction. Then, the linear regression method was used to calibrate spectral AODs of MODIS before applying to the estimation of PM2.5 concentration based on spectral AODs of AERONET. The results demonstrate the potential of satellite observation to estimating the concentration and characteristics of PM2.5 concentrations after comparing the ground-based measurements.
