目前降雨量的資料主要由地面雨量站取得，而雨量站僅提供點的降雨資料，對於防洪設計或是水資源管理等，所需要的多是空間降雨總量，因此需給予雨量站一個權重，或以空間內插推求才能得到可利用資訊。目前空間降雨的推估方法，較常用的有徐昇氏法、算術平均數及等雨量線法等，但在各有其基本假設條件下，所推估之空間降雨不一定是最合適的。 本研究選定蘭陽溪流域，選用其中雨量資料較完整之雨量站，以禁忌演算法搜尋雨量站權重組合進而推估其空間降雨量，再將推估之空間降雨量代入GWLF之水平衡模式推估河川流量，最後以歷史河川流量檢定所求得之雨量站權重組合是否較佳。以禁忌演算法搜尋雨量站權重組合時，根據其中移步、禁忌名單、破禁原則、候選名單及停止原則等進行搜尋，再以模擬流量之RMSE (Root Mean Square Error)判斷所搜尋之權重組合的優劣。 由禁忌演算法所搜尋之雨量站權重結果，經過驗證後發現其模擬河川流量與實測河川流量之RMSE值小於以徐昇氏法及算術平均數所推估結果，因此以禁忌演算法所搜尋之雨量站權重，的確優於徐昇氏權重及算術平均數權重之雨量站權重。根據長時間序列及豐水期分析結果，雨量站權重較大者均集中於集水區上游山區之雨量站，而枯水期時平地雨量站權重幾乎趨近於0。 Raingauges are the major source of rainfall data for engineering applications and science studies in hydrology. They only provide observed information at scattered points in space. For the design of flood control measures and the management of water resources, the amount of rainfall over a region is often required. Therefore, the estimation of areal rainfall is obtained by interpolation or weighted-average of raingage data. Most approaches can be achieved by assigning a weighting factor for each raingage in space, such as Arithmetic Average and Thiessen Method. Some use spatial interpolation of raingage data and/or location, such as Isohyetal Method and Kriging Method. However the amount of areal rainfall estimated by different methods might be quite distinctive due to different assumptions embedded in each method. It is difficult and may not be suitable to use such values for hydrological applications. Lan-Yang river basin was the region of interest in this study. Raingages with complete historical rainfall data were selected to incorporate the Tabu search for finding the optimized combination of weighting factors. The river module of GWLF model was applied to compute river discharges. The RMSE (root mean square error) between computed and observed river discharges was the objective function for finding the best combination of weighting factors in the Tabu search. Move, Tabu List, Aspiration Level, Candidate List and Stopping Criterion, and RMSE check are included as the major searching procedures. Another set of raingage data was applied for validation to ensure the results of Tabu search can be suitable for future applications. The RMSE computed by using the raingage weighting found by the Tabu search is the smallest one compared to those computed by using either Arithmetic Average or Thiessen Method. The weighting of upstream raingages have larger values than those of downstream raingages in both cases of using long-term and wet season rainfall data. There is almost no contribution from downstream raingages in the case of using dry season rainfall data.
