細懸浮微粒於2007年南加州森林大火之時空變異分析

Abstract

火災為嚴重迫害生態環境與人體健康的主要原因，近年來由於全球暖化的因素，森林大火發生頻率不斷增加，在美國西岸加州又有聖安納焚風的助燃下，導致每年都會有森林大火的發生，本研究目的為使用PM2.5與PM10比值關係及氣象變數與PM2.5迴歸關係推估火災前後時PM2.5濃度變化情形，比較火災造成空氣污染之嚴重性。 本研究採用貝氏最大熵法及地理加權迴歸分析。貝氏最大熵法不僅可考慮確定性資料(實際觀測值)，同時還可以加入不確定資料(比值關係或迴歸關係下產生之資料)增加推估的準確性，而地理加權迴歸分析改善了一般線性迴歸所忽略的空間變化，於模型建置時納入空間的概念，並解決自相關的問題。 火災時的驗證中，可發現同時加入比值關係與氣象關係下產生之不確定性資料驗證誤差為8.35μg/m3，無火災時誤差更降低至4.81μg/m3。於火災下 =0.27，無火災時 提升到0.71。 在PM2.5時空推估圖中，可以發現濃度較高的發生區域為San Diego及Los Angles區域，小時濃度最高超過300μg/m3。於無火災時濃度大部分區域都下降，不過部份區域PM2.5尚未消散，因此濃度還是高於一般標準，約為60~80μg/m3間。 貝氏最大熵法利用加入不確定性資料的概念確實改善了推估的準確性。但由於本研究事件並非常態性，若此概念應用於常態性研究或是任何其他領域的研究，並能使得推估更加完整且精確。

A wildfire is one of issues to seriously damage the ecological environment and human health. Due to the impact of global warming and the foehn at Santa Ana, the frequency of wildfire occurrences is increasing in recent years. This study used PM2.5/PM10 ratios and meteorological variables to estimate the PM2.5 spatiotemporal distribution before and after the fire, in order to compare the quality caused by the wildfire at south California during Oct.21 ~Oct.29,2007. In this study, Bayesian maximum entropy method and geographical weighted regression are used. Bayesian maximum entropy method can account for both certain and uncertain information to improve the accuracy of estimation. The Geographically Weighted Regression model is applied to modify the traditional regression model, which cannot capture spatial variations, and to solve the spatial non-stationary. The results show that the relative error and the r-square during the period of the wildfire are 8.35μg/m3 and 0.27, respectively. The low r-square can result from the extreme events of PM2.5 during the period. The spatial distribution maps show that the higher concentration of PM2.5 occurred in San Diego and Los Angles, which is accord with the smoke shown in the satellite images. The study applied BME method to assimilate the empirical relationship of PM2.5 derived by GWR and uncertain information. More information is required to account for the extreme events caused by the fires.