以臺灣黃杉早材重建過去200年來七家灣溪溪流量

臺灣在水文議題上，多以觀測流量進行評估與研究。然而，短時間的紀錄僅能得知頻率較高的乾濕事件變異，無法瞭解低頻乾濕趨勢變異，因此本研究將針對樹輪與水文的關係進行研究。本研究延伸前人對於七家灣溪流量的研究，於更靠近七家灣溪的樣區取樣，加上更長流量紀錄的基礎下，以臺灣黃杉(Pseudotsuga wilsoniana Hayata)樹輪輪寬重建溪流量(Streamflow, SF)與標準化溪流量指數(Standardized Streamflow Index, SSFI)。本研究以COFECHA軟體檢驗校正定年，篩選出72根樹芯，並以集成經驗模態分解法(Ensemble Empirical Mode Decomposition)分別建立早材、晚材與全輪年表。分析年表與SF及SSFI間的相關性，最後以早材重建1785–2009與1785–2013年前一年九到當年二月的七家灣溪流量(SFSF)與SSFI (SSFISF)，和1785–2009年前一年九到當年三月的七家灣溪流量(SFSM)。本研究也將觀測及重建的七家灣SFSF、SFSM及SSFISF，與荷蘭皇家氣象署Climate Explorer氣象資料庫的海平面溫度及海平面氣壓做場域分析。結果顯示，研究地區臺灣黃杉早材生長受濕季末期(九月)與乾季(十至三月)的土壤水分影響大。重建的七家灣SFSF、SFSM及SSFISF與觀測值及前人重建的七家灣SSFI波動相似，因此，研究結果支持臺灣黃杉早材重建流量因子的可行性。重建七家灣SF有較高的解釋率，但重建SSFI對於判別乾濕事件發生更有優勢。藉由年表頻譜分析加上重建七家灣SSFISF，可觀察到大約每10–20年有乾旱事件。海面溫度場域分析(Field analysis)顯示太平洋十年際振盪及聖嬰–南方振盪現象的跡象；海平面氣壓的場域分析則顯示亞澳季風系統與北極振盪(Arctic Oscillation)對此地區冬季雨量及流量的影響。尤其九到十月雨量及流量除了受颱風影響外，也受北極振盪暖相位直接與間接的影響。

關鍵字

樹木水文學；臺灣黃杉；溪流量重建；標準化流量指數；遙相關

並列摘要

Taiwan often uses flow observation data to study and evaluate hydrologic issues. However, the short-term instrumental records could only reflect high frequency variability of dry/wet events. Therefore, to understand possible low frequency variability, this research focused on the relationship between the tree-ring and streamflow. This study reconstructed the past variability of Chichiawan (CCW) streamflow. This study used a set of sampled Pseudotsuga wilsoniana Hayata (Taiwan Douglass-fir, TDF) trees that were closer to the CCW than those used in the previous research. It also used gaged flow data with a longer observation period. This study then reconstructed the CCW streamflow (SF) and its standardized streamflow index (SSFI) based on the tree-ring width variations of TDF. This study used COFECHA to statistically validate the cross-dating, and 72 cores were selected to construct earlywood, latewood, and total ring width chronologies developed using Ensemble Empirical Mode Decomposition. After analyzing the correlations between chronologies and SF/SSFI, this study selected earlywood ring width chronology to reconstruct CCW SF/SSFI from September to February for 1785–2009 and for 1785–2013 (SFSF / SSFISF), and CCW SF from September to March (SFSM) for 1785–2009. After the reconstructions, this study correlated the observed and reconstructed CCW SFSF / SFSM / SSFISF with the sea surface temperature and sea level pressure fileds. The results showed that TDF earlywood growth in the study area was affected by the soil moisture conditions from the end of wet season (September) and through out the dry season (October to March). The fluctuations of the observed and reconstructed CCW SFSF、SFSM or SSFISF were similar. Also, they were similar to the October to February SSFI reconstruction of a previous study. Thus, the results supported the possibility of using TDF earlywood to reconstruct CCW dry season SF and SSFI. Although the reconstructed SF had a higher explanation power, the reconstructed SSFI was better for determining the occurrences of dry/wet events. The reconstructed SSFISF showed that dry events could happen around every 10 to 20 years. The sea surface temperature field analysis showed traces of the cold phase of the Pacific Decadal Oscillation and El Niño-Southern Oscillation, whereas the sea level pressure field analysis showed that the precipitation and the streamflow in the studied area were affected by the Asian-Australian monsoon system and the Arctic Oscillation. The precipitation and the streamflow in September to October were affected not only by typhoons, but also directly and indirectly by the warm phase of the Arctic Oscillation.