雷達降雨應用於農業災害預警之可行性研究

A Feasibility Study on Early Warning for Agricultural Disaster by Using Radar Rainfall Data

呂玟潔(WEN-JYE LEU)

指導教授 ： 虞國興

共同指導教授 ： 李柏青

淡江大學/工學院/水資源及環境工程學系碩士班/碩士(2018年)

https://doi.org/10.6846/TKU.2018.00396

摘要

臺灣常遭梅雨鋒面及颱風侵襲，高強度降雨事件使得農田積淹、水閘門無法即時操作及設施損壞，目前國內政府機關已有淹水與土石流之預警系統，可於颱風豪雨期間發布警戒資訊，提供防災單位及民眾採取必要應變措施，惟目前尚無針對農業灌區淹水災害建立一套預警系統，另因雷達降雨其空間解析度優於地面雨量站，故本研究擬探討雷達降雨應用於農業災害預警之可行性，期能建立一套雷達降雨應用於農業災害預警之方法及運作機制，提供農業相關單位參考應用。
本研究以石門水利會農田灌區為研究區域，並以2012年0610豪雨、2014年鳳凰及2015年蘇迪勒颱風事件為研究案例，蒐集前述三場颱風豪雨劇烈天氣監測系統（QPESUMS）雷達降雨資料與地面雨量站觀測資料，藉由克利金法計算其區域平均及區域最大雨量值，並與雷達降雨網格分析資料進行相關性分析，以探討雷達降雨替代地面雨量站之可行性；另藉由蒐集研究區域內之歷史淹水事件、訂定雨量警戒值，並利用三場事件之雷達降雨及地面雨量站區域平均及區域最大雨量資料與選取灌區適當雨量站之最大雨量資料，分別就有無超過警戒值及有無淹水進行比對分析，同時建立警戒評估指標，並藉由評估指標據以評估灌區雷達降雨雨量值採計方式，以探討雷達降雨應用於農業災害預警之可行性。
由研究結果顯示，雷達降雨(QPESUMS)與地面雨量站之雨量相關係數均高達0.9以上，且具有較高空間解析度，可知雷達降雨替代地面雨量站之降雨監測具高度可行性；另雷達降雨應用於農業災害預警之實際應用可行性研究結果顯示，其預警方式於實際應用上亦具高度可行性，並可利用雷達降雨之區域最大雨量值作為灌區雨量採計值，以比對灌區雨量警戒值作為預警之雨量監測指標。因國內目前似尚無針對農業災害預警建立一套預警方法及運作機制，本研究實例應用之預警方法及運作機制，可提供農業相關單位應用或相關研究參考，同時提出此預警方法未來可再改善之建議。

關鍵字

雷達降雨 ； 農業災害預警 ； 雨量警戒值

並列摘要

Taiwan is often hit by the plum-rain fronts and typhoons. High-intensity rainfall events have caused farmland flooding, gates not to perform real-time operations, and damage to flood control facilities. For example, 0610 torrential rain of the year 2012 gave rise to a severe flooding in irrigation districts of the Shihmen Irrigation Association. At present, the domestic government agencies have already provided early warning systems for flooding and landslides. During the typhoon and torrential-rain events, they release official warning information to provide relevant disaster prevention units and the public with necessary response measures. So far, no early warning system has been set up for flood-affected agricultural irrigation districts. Therefore, the purpose of this study is to explore the feasibility by applying rainfall radar to the agricultural- disaster early warning. In the meantime, setting a method and operational mechanisms for early warning of agricultural disasters is expected to be able to provide the relevant agencies an essential reference.
In this research, the irrigated farmland of the Shihmen Irrigation Association was used as the study area. Moreover, three severe hydrological events (i.e.0610 torrential-rain in 2012, Fung-wong typhoon in 2014, and Soudelor typhoon in 2015) were implemented as the case study. The rainfall data of the weather monitoring system (QPESUMS) was interpolated and meshed.by the Kriging method. The regional average and the maximum rainfall value were calculated, and the correlation analysis was conducted with the radar rainfall data. The feasibility of radar rainfall (instead of surface rainfall stations) for the study area was explored through collecting the historic flood events, setting rainfall warning values for flooding, and employing the radar rainfall of the three events as well as the regional average of surface rainfall stations. The maximum rainfall data in the study area and the rainfall data from one selected rainfall station in the irrigation area were compared. Furthermore, the study analyzed the warning value and flooding occurrence to establish the warning evaluation index. The results also compared the warning effect of radar rainfall and ground rainfall station using the evaluation index to discuss the feasibility of establishing the agricultural-disaster early warning system with radar rainfall.
The results show that the QPESUMS rainfall is capable of replacing precipitation stations and it is feasible to apply the maximum rainfall value to the agricultural disaster early warning. It can launch early warning action against a sudden torrential rain event in Taiwan and assists the relevant units to operate gates and disaster prevention measures for reducing more flooding losses caused by heavy rainfall. It is expected that the related units will, referring to the research results, provide their staffs/officers with the applications of radar rainfall data and warning values to perform disaster prevention and early warning.

並列關鍵字

Radar Rainfall ； Early Warning for Agricultural Disaster ； Rainfall Warning Value

參考文獻

15.黃鵬豪，2008，「應用QPESUMS高解析降雨資料改良洪水預報模式之研究」，臺灣大學生物環境系統工程學系碩士論文。

18.陳淵翔、施國肱、張麗秋，2013，「雷達定量降水資料結合類神經網路於颱風時期降雨量與流量推估之研究」，淡江大學水資源及環境工程學系碩士論文。

19.陳振宇、劉維則、許家祥，2017，「使用QPESUMS雨量資料建立崩塌災害預警模式」，中華水土保持學報，48(1)，44-55。

21.鍾侑達、郭峻菖、陳昶憲，2009，「台灣區域降雨趨勢分析」，農業工程學報，55(4)，1-18。

25.鄭士仁、李正豐、謝惠紅、李如晃，2010，「應用降雨特性評估雨量估計方法及其應用」，農業工程學報，56(2)，9-20。