氣候變遷是整個地球社會面臨的迫切問題。嘗試減少氣候變遷推估的不確定性讓整個社會對於所面對的問題可以有更高得共識是氣候學者應該擔負的責任，本研究計畫的主要目的是想利用耦合模式比較專案第五期(Coupled Model Intercomparison Project, Phase 5; CMIP5)所提供的數量龐大的多重模式氣候變遷情境模擬資料來發展出可以有效的減少系集模擬資料的氣候變遷推估不確定範圍的方法。過去兩年，我們結合個人發展的型態穩定度分析，排序修正法，和完美模式篩選法針對CMIP5 RCP情境的系集模擬資料的全球熱帶海溫和全球平均地表溫度進行氣候變遷推估5-95不確定範圍的校正。計畫執行結果發現，先經過完美模式篩選後再進行排序修正的方法確實可以非常有效的減少系集模擬資料所推估的5-95不確定範圍。在未來一年本計畫的主要研究重點有三：一是探討如何選擇適當的篩選條件；二是結合這種篩選排序修正法和第一年計畫所發展的參考空間型態來嘗試減少全球，區域，個別國家，乃至個別網格尺度的氣候變遷推估的不確定性；三是探討篩選排序修正法在建立emergent constraint的潛力和可能應用。我們預期經由此計畫的執行可以有助於提升我們對於氣候變遷推估能力。 ;Climate change is a pressing issue that the global society is facing. Accurate decadal predictions and projection are the corner stone for mitigation planning of climate change problems. The main purpose of this study is to find a better way to use the Coupled Model Intercomparison Project, Phase 5(CMIP5) climate change scenario runs to reduce uncertainties in climate change projection than just use the mean and spread of Multi-Model Ensembles (MME). In the past two years, we applies pattern stability analysis, rank histogram calibration, and perfect model approach to global tropical sea surface temperature and global mean surface temperature from both observed and model runs to calibrated the 5-95 uncertainty ranges of CMIP5 RCP scenarios MME climate change projections. The results showed that the use of rank histogram calibration to constrained MME (i.e., filtered MME using perfect model approach) indeed could effectively reduce the 5-95 uncertainty ranges of MME climate change projections. In the following year, our study will focus on three aspects. The first is to find a proper way to choose filtering conditions with perfect model approach. The second is to extend the calibration process from individual reference spatial pattern to individual grid scale. The third is to explore the relation between perfect model filtering process and emergent constraints. Through this study, we hope to improve our capability in climate change projections.
