頁籤選單縮合
題 名 | 桃園地區每日最大臭氧濃度之多重迴歸預測的可行性探討=Prediction of the Daily Maximum Ozone Concentration at Tao-Yuan Using Multiple Linear Regression Analysis |
---|---|
作 者 | 胡婷堯; 林能暉; | 書刊名 | 環境保護 |
卷 期 | 26:2 2003.12[民92.12] |
頁 次 | 頁253-279 |
分類號 | 412.33 |
關鍵詞 | 臭氧; 線性迴歸; 綜觀氣候; Ozone; Linear regression; Synoptic weather; |
語 文 | 中文(Chinese) |
中文摘要 | 本文主要的目的乃是嘗試結合逐步複迴歸法與天氣類型分析方法,期能建立一個短期的臭氧統計預報模式,以預測桃園地區每日之最大臭氧濃度。利用9/1/93至8/31/94期間的環保署臭氧監測資料及中正機場的氣象資料所建立之逐步複迴歸模型中,吾人可解析出對桃園地區每日最大臭氧濃度有高度相關的氣象因子包括:早上九點的風速,當日最大混合層高度,早上八點的臭氧通量,早上九點的通風指數,當日最高溫與最低溫之差,以及前一日的最大臭氧濃度。此模式之適合度(R2值)為0.6,臭氧預測值與實際值間誤差的均方根為13.4 ppb,但在高值的部份(大於80 ppb),預測值普遍低估。為了改進模式「高值低估」的現象,吾人利用天氣類型的分類及較小尺度的氣象參數,訂定出修正的條件與方法來修正預測值。結果顯示修正後的預測與實際值間誤差的均方根與原先相差不大,但對高濃度時的低估現象已有改善。基於本文的結果,吾人認為僅以考慮氣象因素所建立之多重迴歸臭氧統計預報模式尚無法對桃園地區每日之最大臭氧濃度作出與國外研究相當之結果(R2>0.8),即使納入天氣類型分析方法修改,改變亦有限。究有原因,可能是複雜的地形與其他非線性之化學因素的干擾所致。 |
英文摘要 | In this paper we construct s statistical model for predicting the daily maximum ozone concentration at Tao-Yuan using the multiple linear regression analysis and synoptic climatological method. Based on the meteorological data observed at the CKS airport and EPA sites, we found the meteorological parameters which are highly correlated with ozone concentration include the wind speed at 9:00, maximum mixing height, ozone flux at 8:00, ventilation index at 9:00, different between the maximum and minimum temperatures, and the maximum ozone concentration of the previous day. The R2 of the mode is around 0.6. The rms of observed and predicted ozone concentrations is 13.4 ppb. The model generally underpredicts the maximum ozone of >80 ppb. In order to improve above situation, we modify the model by considering the relationship between ozone concentration and synoptic weather patterns as well as small-scale meteorological parameters. Consequently, above underprediction is improved. However, the R2 and rsm remain about the same. Our results are not comparable with some others in USA since our metetorological conditions are more complicated due to imhomogeneity of the topography. In addition, inclusion of chemical parameters in our model may improve its performance as well. |
本系統中英文摘要資訊取自各篇刊載內容。