摘要 Abstract：

Understanding the fate and reactivity of dissolved organic matter (DOM) in river networks is crucial for unraveling carbon cycling in inland water systems. Although physio-geographic factors and water residence time (WRT) are recognized as the primary drivers of DOM dynamics, the interaction and influence of these factors on DOM variability along the river continuum remains unresolved, especially in subtropical small rivers (SSRs). This study investigated the spatiotemporal variation of DOM concentration and composition in two SSRs in Taiwan with different landscapes and anthropogenic impacts. The mean water residence time for the Keelung and Lanyang rivers on the sampling days is around 34 and 23 hours, respectively. We investigate DOM quantity and quality along the river continuum by using dissolved organic carbon (DOC) concentration measurements and optical analyses (absorbance and fluorescence). The results showed that, along the SSRs continuum, the DOC concentrations and optical indexes exhibited slight changes, with significant increases observed only at downstream sites influenced by human activities. Meanwhile, the higher biological index (BIX) and lower humification index (HIX) indicated an increase in autochthonous sources and a decrease in the degree of humic character in the lower reaches. In addition, we observed a positive correlation between WRT and DOC concentration variability, but not significant for DOM composition. When comparing the two rivers, the one with steeper topography and less human influence shows lower levels of DOC concentration and degree of humification. On a global scale, the SSRs seem to have lower DOC concentrations (0.26 - 1.65 mg-C L-1), lower HIX (0.28 - 0.76), and slightly higher BIX (0.8 - 1.9), which could be attributed to Taiwan's steep landscape and shorter water residence time, which limit soil organic carbon (SOC) production and rates of in-stream processes. Through our investigation, DOM concentration and composition in river networks will be better understood and potentially improve the assessment of the global carbon cycle.

關鍵字 Keywords：Dissolved organic matter (DOM), fluorescence and absorbance, river continuum, subtropical small rivers (SSRs), Taiwan

摘要 Abstract：

Human-driven changes in land use have significant impacts on both societies and ecosystem services. For instance, the shift from forests to conventional farms utilizing fertilizers can lead to increased soil nitrogen levels, groundwater degradation, and disruption of downstream ecosystems. This research delves into the complex dynamics of human-induced land-use change, specifically focusing on the transition from forests to tea farm-dominated areas in Taiwan. Here, conventional farming methods with fertilizers affect soil quality, groundwater, and downstream ecosystems. Through the use of the Soil and Water Assessment Tool (SWAT) for nutrient export analysis, our findings demonstrate that when agricultural land use surpasses 2%, there is a substantial increase in the export of nitrate, phosphate, and potassium, ranging from 25% to 150%. Notably, agricultural activities have a greater impact on nitrate levels, exceeding those by 120% and 233% during the dry and wet seasons respectively. Tea farms, comprising a significant portion of the landscape, show an almost tenfold increase in NO3-N yield compared to forests. Implementing a revised fertilization strategy, which includes application during light rainfall events, enhances nitrogen absorption and tea tree yield while reducing nitrogen input by 10%. This study provides actionable recommendations for sustainable agroforestry practices by integrating river and rainwater data with SWAT modeling. Such an approach advances our understanding of hydrological and biogeochemical processes in subtropical tea farm-dominated areas, offering valuable insights into hydrology and biogeochemistry.

關鍵字 Key words：Riverine nutrient export, SWAT, Nonpoint source pollution, Fertilization strategy

摘要 Abstract：

「臺灣氣候變遷科學報告 2023 版」彙整東亞地區及臺灣氣候的重要變遷特徵，包含溫度、降雨、風速、周遭海域海溫及海平面高度、極端天氣、空氣汙染等。未來推估以「第六期耦合模式比對計畫 (CMIP6)」模擬資料為主，考量多種未來情境，進行統計或動力降尺度後分析臺灣氣候的未來變化。氣候變遷對臺灣可能造成的衝擊則包含水資源、坡地、海岸、糧食、生態、健康、城鄉空間等，最後提供風險評估、調適策略及科研缺口。此報告將作為政府各單位進行氣候變遷風險評估，研擬及推動調適政策之科學依據。

關鍵字 Keywords：臺灣氣候變遷、未來推估、科學報告

摘要 Abstract：

臺灣位於季風氣候區，受中緯度與熱帶天氣系統影響，降水系統多樣而且複雜，春夏有西南季風，夏季有颱風、午後熱對流，冬季有東北季風，在不同的季節有不一樣的降水特徵，但由於特殊的地形環境，雖然降雨量充沛，但降雨時空分佈不均，為了探討此現象，本研究使用氣象局局屬站小時資料，探討臺灣地區降水變化特徵，分析各地區各1月至12月每小時的降水變化。

分析結果顯示，臺灣在不同的季節各地有不同的降水特徵 ，具有明顯的日夜變化，因地域的不同特徵的強弱也有所差距，以臺灣東西半部進行區分，西半部地區降水主要集中在5至8月，東半部地區則集中在8至12月。

在6月至8月全臺皆有降水發生，受到西南季風、颱風以及午後熱對流影響，其中7月的降水在午後有最大值，相較其他月份在晚間的降水相對較弱，主要是受午後熱對流影響，10月開始轉為東北季風主導，東半部有明顯的降水特徵，具有更顯著的日夜變化。

進一步分析2010-2023年間氣象局局屬站雨量資料，將雨量資料進行標準化計算，並透過階層式分群法（Hierarchical Clustering）對站點進行分類，可分類出五大特徵:臺灣西部、台灣東部、春季特徵、冬季特徵以及午後特徵。

未來將詳細分析各地的降水特徵，透過更為密集的觀測資料進行統計分析，瞭解各地受到地理以及地形環境影響，所造就的特殊降水特徵。

關鍵字 Keywords：區域降水特徵、氣象觀測、階層式分群法（Hierarchical Clustering）

摘要 Abstract：

本研究利用能夠明確模擬熱帶氣旋的高分辨率模式的結果，教驗兩個TC熱帶氣旋生成指數在暖化情境下西北太平洋熱帶氣旋的變化。Emanuel (2010)  所提出的熱帶氣旋生成指數（χGPI）和我們提出的修正指數（χMqGPI）都很好反映熱帶氣旋生成頻率(TCGF )的季節變化及其在歷史模擬和觀測數據中的空間分佈。在暖化情境下更溫暖的海洋表面 χGPI 預測 TCGF 將顯著增加，這與模式預測的顯著減少形成對比。這種不一致性可歸因於海溫對於χGPI指數的控制，西北太平洋的高海溫會導致大氣更加穩定，因此熱帶氣旋發生較少。相比之下，考慮總水氣凝結的 χMqGPI 正確預測了 TCGF 的減少，儘管減少的程度小於直接從模式預測熱帶氣旋。 整體來說，χMqGPI比χGPI更準確地反映了暖化情境模擬中TCGF的下降趨勢。 我們的研究顯示，使用 χMqGPI 來預測 WNP 中 TCGF 的未來變化是合理的。

The study utilized the results of high-resolution models, which can explicitly simulate tropical cyclone (TC), and treated the simulated TC genesis as the “truth” to evaluate two TC genesis potential indices in reflecting the projected TC changes in the western North Pacific (WNP) under warming scenario. Both a widely-used TC genesis potential index (χGPI) and a revised index (χMqGPI) well represented the seasonal variation of TC genesis frequency (TCGF) and its spatial distribution in both historical simulation and observation data. The χGPI projected a significant increase in TCGF in response to the future warmer ocean surface, in contrast to the significant reduction in model projection. By contrast, the χMqGPI that considers gross moisture condensation properly projected a TCGF decrease, although the degree of reduction was smaller than that derived directly from TC detection scheme. Overall, the χMqGPI more accurately reflected the decreasing trend of TCGF in the warming simulations than the χGPI.