圖書典藏及影音數位平台

經濟部水利署南區水資源局（以下簡稱本局）所轄管之曾文水庫、阿公店水庫、牡丹水庫及台灣自來水股份有限公司第六區管理處轄管之南化水庫係供應南部地區水源之水庫，鑑於氣候變遷之影響，致使水庫汛期間防洪操作需嚴密搭配氣象水情之推估及預警方能有效達到水庫防汛減災及蓄水利用之效果。 因此由本計畫進行入流基本資料之蒐集及更新藉以分析庫區上游降雨時間與空間分佈特性及庫區水位歷線等關係產出之成果來作為防汛應變之用。廠商須提供氣象及水情資訊簡報，依本局需求進行水情資料蒐集、預估及分析等，提供本局於緊急應變操作參考。 近年因氣象預報技術亦有新的推展，可藉由蒐集與研判分析提升水情預估準確度。本局為確保緊急應變期間水情掌握及預警推估之準確性，爰擬辦理本計畫進行資料蒐集、分析與觀測系統整合、氣象與水情資訊提供等工作。 另因應本局防淤隧道完工啟用在即，未來除防洪功能外亦兼負排砂操作任務。為精確研判水庫異重流到達之條件與時機，使曾文防淤隧道利用達最大化，進行排砂操作模擬，以利本局未來排砂操作時之重要參考依據。

This annual project provides rainfall forecasts and decision support for reservoir operation during floods in 2019, for the watersheds of reservoirs and weirs of the Southern Region Water Resources Office (SRWRO) of Water Resources Agency (WRA). The aim is to achieve the multiple operating goals of reservoirs, including flood mitigation, dam safety, water conservation and reservoir desilting. During the flood season of 2019, 57 rainfall forecasts for multiple flood events were provided and archived as power point file. The provided information primarily includes 24-hours-lead-time rainfall forecasts from the model ‘QPESUMS+QPF’ of the Central Weather Bureau, and 72-hours-lead-time rainfall forecasts from the model ‘QPESUMS+WRF’. Comparing the actual and forecasted rainfall depths suggests that the QPESUMS+WRF model is more accurate for typhoon events, while the QPESUMS+QPF model is more robust for rainfall induced by lager scale weather systems, such as the plum rain season front or the southwesterly flow. This annual project also provided consultations for flood control operation of Twengwen Reservoir during 9 events, which are the torrential rainfall events of 18-May, 28-May, 10-June, 13-June, 23-June, 02-July, 11-July, 19-July, 15-August, and Typhoons Danas, Lekima and Bailu. The most significant event during which the spillway and desilting tunnel were both operated is the torrential rainfall following Typhoon Lekima from 12-August to 20-August. Before Typhoon Lekima, the reservoir water surface level (WSL) was 226 El.m. It was expected that the typhoon-induced inflow volume could not elevate the WSL to attain El. 230 m after the flood. Thus, pre-release operation was not recommended. The reservoir release was kept at lower discharge until the noon of 11-August, when rainfall was significantly increased due to the following southwesterly flow. The reservoir release was increased progressively, to achieve a maximum discharge as 1050 m3/s. The actual and forecasted discharges of reservoir peak inflow during this periods were 1167 and 1362 m3/s respectively, and the WSL of reservoir was maintained below 229 El.m when the 1st peak occurred. Afterwards the release was gradually reduced to store the recession inflow to achieve the desired end-of-operation WSL. Nonetheless, the impact of the southwesterly flow was unexpectedly extended to almost one week, lead to the 2nd and 3rd flood peaks with discharge as 2,581 and 2,564 m3/s respectively. The reservoir release was then re-elevated to a maximum of 1950 m3/s to slow down the rise of WSL. It was kept below El. 227.5 m during the 2nd and 3rd peaks. Generally, the operation satisfies multiple objectives including ensuring dam safety, flood mitigation, flushing downstream deposition from previous dredging and securing future water supply, and the suggested operating strategies conform well to the actual operation results. During the flood events, this project also estimated the possible occurrence of turbidity current and its arrival time at the dam. The aim is to precisely estimate the arrival time, so that the desilting tunnel can be used to effectively vent the highly concentrated current. Nonetheless, the scale of floods of 2019 were all too small to meet the conditions of occurring of a turbidity current as well as its arriving at the dam. This evaluation is valid with the field measurements, which show very low sediment concentration from the release of desilting tunnel during flood events. In order to enhance the efficiency of forecasting and decision support analysis, all the available measurements including rainfall, reservoir WSL, inflow and release, downstream river stages and the sediment concentration at different locations within the reservoir impoundments were all integrated into a web-based platform. This platform and its database were properly maintained, with some extensions of functions and data during this project. It is suggested all the necessary analyzing tools should be continuously integrated into this platform, thus to allow the automation of decision support analysis in the future.