| 摘要: | 近年來，休閒活動在大眾生活中的重要性不斷提升，而溫泉區始終是人們旅遊的熱門目的地。隨著永續觀念的興起，觀光發展需同時考慮溫泉的永續利用。本研究透過整合異質場地質模型和地下水流數值模式，評估礁溪溫泉適當的溫泉使用量。本研究首先蒐集不同來源的地質鑽探資料，以及2020年間觀測井水位、水溫資料和抽水量等水文觀測數據。接著，運用地下水模擬系統(Groundwater Modeling System, GMS)軟體，通過馬可夫鏈法(Markov chain)產製數十個實現場的異質性沉積層地質模型，並將各個實現場與底部均質性基盤整合為三維地質模型。因實務上無法針對每個地質模型進行數值模式率定，因此本研究提出一套篩選流程，選定一代表地質模型。隨後，再將其轉換為MODFLOW地下水流數值模型，並結合MT3DMS套件進行水-熱模擬，以獲取地下水流場和推測熱水來源。最後通過模擬在不同抽水量下的水位及溫度場變化分布，導入管理水位，協助礁溪溫泉永續發展。
研究結果顯示，水文地質材料在東北-西南方向上有最大連續性，與沉積物粒徑分布一致。當三維地質模型建置數達30個時，整體地下水位的變異數及平均值趨於穩定。因此，本研究再針對此30個地質模型模擬的地下水位，以決定係數(R-squared)最高及平均絕對誤差(Mean absolute error)最小的統計特性，挑選模型21為代表地質模型。本研究整合了異質性沉積層和均質性基盤，並轉換為地下水流與溫度傳輸數值模型。地下水流模擬顯示，沉積層模擬水位與實際觀測水位相當一致(R^2=0.92)，顯示地下水流模式具有一定的代表性。熱傳輸模擬結果顯示，溫度場的空間分布與目前溫度觀測相似，表示本研究模擬出主要的熱源可能來自靠山區的兩處。將模擬的地下水位結果結合管理水位，顯示多數的監測井水位未低於嚴重下限水位，表示在目前的抽水量下，礁溪地區的水資源尚可持續使用，但仍需持續觀測與注意。當建立之模式在抽水量增加為現在的2.5倍時，監測井地下水位平均約下降1.2公尺且造成冷熱水交界的退縮，此狀態已有溫泉水資源枯竭的潛在風險。本研究建立之研究流程與架構，可以作為不同地區水資源管理的依據，實現對溫泉資源的永續發展目標。;In recent years, leisure activities have become increasingly important and popular for travel. With rising sustainability awareness, tourism must address the sustainable use of hot springs. This study aims to evaluate the appropriate usage of the Jiaoxi hot spring by integrating the hydro-thermal numerical simulation in a heterogeneous hydrogeological model. First, we collected the geological drilling data from various sources, as well as hydrological observation data including groundwater level (GWL) and temperatures from monitoring wells in 2020, along with pumping rates. Next, using the Groundwater Modeling System (GMS) software and the Markov chain method, we generated dozens of realizations of a heterogeneous sedimentary layer and integrated them with the homogeneous basement to construct a 3D geological model. Since calibrating each model individually is impractical, we proposed a process to select a representative model. Subsequently, the model is converted into a numerical groundwater flow model and combined with the MT3DMS package for hydro-thermal simulations to obtain groundwater flow fields and infer the heat sources. Finally, we simulated GWL and temperature changes under different pumping rates and incorporated GWL management to support the sustainable development of Jiaoxi hot springs.
The results revealed that the hydrogeological materials have the highest continuity in the northeast-southwest direction, consistent with sediment grain size distribution. When the number of 3D geological model reached 30, the standard deviation and mean value of the overall GWL stabilized. Consequently, model 21 was selected as the representative geological model due to its highest R-squared and lowest mean absolute error in GWL simulations among the 30 models. The study integrated heterogeneous sedimentary layers and a homogeneous basement into a numerical model of groundwater flow and temperature transport. The groundwater flow simulation demonstrates a remarkable consistency (R²=0.92) between simulated and observed data, indicating the model′s representativeness. Furthermore, the heat transport simulation results revealed a spatial distribution of the temperature field similar to temperature observations, indicating that the main heat sources likely originate from two areas near the mountainous region. Combining the simulated GWL results with management levels shows that most monitoring well water levels did not fall below critical low levels. This indicates that, under the current pumping rates, the groundwater resources can still be sustainably used. However, continuous observation and attention are still necessary. However, if the pumping rate were to increase to 2.5 times the current rate, the average GWL in the monitoring wells drops by approximately 1.2 meters, causing the interface between hot and cold water to retreat. This situation poses a potential risk of thermal water resource depletion. The research framework established in this study provides a valuable reference for groundwater resource management in other regions, contributing to the realization of sustainable development goals for hot spring resources. |
