氣候變遷衝擊評估研究中用以推估未來氣候變化之工具為環流模式,目前IPCC網站上提供多個環流模式之未來氣候推估結果,然而,不同的環流模式其假設及考慮的邊界條件皆不相同,因此,應用大尺度氣候推估於小區域研究隱含著不確定性。本研究建立模式推估資料之評估方法,分析環流模式與研究區域歷史雨量資料統計特性,剔除對研究區域無法提供合理推估資料之環流模式,以強化簡易降尺度應用之合理性與衝擊評估之不確定性。本研究參考國內外相關文獻制定之排名法與缺點扣分法,以年相關係數、豐水期均方根誤差以及枯水期均方根誤差為評分項目,分析比較環流模式基期年(1961-1990年)之推估值與研究區域基期觀測之月平均降雨量資料。研究中將評分方法應用於新店溪、大漢溪與頭前溪流域,研究結果顯示於3個不同流域間均具備良好的適用性之環流模式排序前3名依序為MPEH5、NCCCSM、GFCM2.1、INCM3及MRCGCM。
The general circulation models (GCMs) are often used to project future climate and further applied to assess the impact of climate change. On the IPCC website, it provides many GCMs' projections considering different greenhouse gases emission scenarios. However, different GCMs may simulate climate based on different assumptions and boundaries. Thus, it has significant uncertainty when large scale climate projections are applied to a local scale impact assessment study. The purpose of this study is to establish methods to evaluate the projection abilities of GCMs. According to the comparsions between GCMs' projections and statistics of historical rainfall data, those GCMs, which cannot provide reasonable projections, are not considered for further impact assessment. It is expected to strengthen reasonably application of simple downscale and to reduce uncertainty of impact assessments. The ranking method and score reduction method are used to evaluate the projection ability of GCMs. The results show the MPEH5, NCCCSM, GFCM2.1, INCM3 and MRCGCM models can provide the most reasonable projections for all the Xindian River Basin, Dahan River Basin and Touchien River Basin.