以塔塔加地區臺灣雲杉樹輪最大密度重建過往 300 年溫度

Abstract

工業革命後，人類所排放的溫室氣體增加，儀器觀測溫度在近數十年中明顯上升。然因可靠的儀器紀錄僅始於1950年代，是以溫度上升是否全因人類造成迄今仍未能釐清。樹芯等氣候替代資訊(climate proxy)，已被證實能有效重建古氣候，藉以審視與比較今氣候變化與古氣候變化的關係。 本研究利用臺灣中部塔塔加地區的臺灣雲杉(Picea morrisonicola)樹輪密度為材料重建氣候，經叢集經驗模態分解(Ensemble Empirical Mode Decomposition)移除生長趨勢，建立輪寬及最大密度(Maximum Density)、平均密度(Ring Density)、最小密度(Minimum Density)等樹輪年表。以阿里山氣象站氣候資料(1934–2020年)為校準資料，比較氣象站觀測紀錄與各樹輪年表間相關性，作為氣候重建的依據。 結果顯示，樹輪密度年表與溫度呈現顯著正相關，最小密度及最大密度年表顯示與標準化降雨蒸發散指數(Standardized Precipitation Evapotranspiration Index)的相關性微弱。最大密度年表與二到十二月均溫存在高度相關，決定係數達到0.67，因此利用最大密度年表重建1723至2020年的二到十二月均溫。依據所得的重建，本研究觀察到兩個相對暖期：1723至1755、1989至2020年，以及兩個相對冷期：1836至1845，1885至1893；而二十世紀末的增溫幅度大於其他暖期。 重建溫度與大氣環流指數相關場域分析顯示，在1901–2020年期間，重建溫度與東亞、赤道等地區地面溫度正相關；在1870–2020年期間，與西太平洋、印度洋、北大西洋海溫存在正相關。推測重建溫度年際變化可能受到火山影響。十九世紀低溫期可能與小冰期末期或道爾頓極小期(Dalton minimum)等有關，亦觀察到領先大西洋多年代際振盪 (Atlantic Multidecadal Oscillation)的現象。

After the Industrial Revolution, the emission of greenhouse gases by humans increased, and the observed temperatures have significantly increased in the past few decades. Due to the limitations of instrumental records, it remains inconclusive whether the warming is entirely made by people. Climate proxies, such as tree rings, have been demonstrated as effective for reconstructing temperatures to examine temperature changes nowadays. This study aimed to reconstruct high-altitude temperatures by using the tree-ring density of Taiwan spruces (Picea morrisonicola) in the Tatachia area of central Taiwan. The study used ensemble empirical mode decomposition to remove growth trends and an X-ray densitometry machine to measure wood density. Tree-ring chronologies of ring width, maximum density (MXD), ring density, and minimum density were developed. Correlations between climate data from the Alishan weather station (1934-2020) and the chronologies were examined to find the best chronology for reconstruction. The density chronologies exhibited significantly positive correlations with temperature, whereas the minimum density and MXD chronologies show weak correlations with the Standardized Precipitation Evapotranspiration Index. The correlation between MXD and the average monthly mean temperatures from February to December was the highest (R2 = 0.67). Therefore, this study used MXD chronology to reconstruct the February–December mean temperatures from 1723 to 2020 CE. Reconstruction identified two relatively warm periods: 1723–1755 and 1989–2020, and two relatively cold periods: 1836–1845 and 1885–1893. The warming trend at the end of the 20th century exceeded that of other warm periods. Spatial field correlation analysis showed that the temperature reconstruction was positively correlated with East Asia, the equatorial region land temperature (1901-2020), and the sea surface temperatures in the western Pacific, Indian Ocean, and North Atlantic (1870-2020). Volcanic activities might influence the reconstruction's interannual temperature variations. The anomalously cold periods in the 19th century may be related to the end of the Little Ice Age or the Dalton Minimum. This study also revealed that both the observed and the reconstructed temperatures were leading the development of the Atlantic Multidecadal Oscillation.