本研究旨在應用支撐向量機及隨機森林兩種機器學習法於中央氣象局QPESUMS (Quantitative Precipitation Estimation and Segregation Using Multiple Sensors)劇烈天氣系統所提供之雷達估計降雨資料，建立雨量預報模式以提供一即時預報資訊。本研究針對六場颱風事件，以翡翠水庫、德基水庫及曾文水庫三個水庫集水區為研究區域，建立各颱風事件前置1至3小時之即時雨量預報模式。其中為改善即時雨量預報模式之表現，測試不同可能輸入變量，包含各水庫集水區網格之雷達估計降雨、網格坐標、網格高程以及颱風因子(颱風中心位置、颱風中心與網格位置之距離)。此外，本研究比較兩種建模方式於預報之表現，分別為(1)建模方式一：前置1至3小時預報模式為各小時使用相同模式與(2)建模方式二：前置1至3小時預報模式為各小時獨立建模，分析結果顯示，建模方式一於即時雨量預報有較好之表現。以相關係數和均方根誤差來看，前置1小時，兩種機器學習法皆有不錯的表現。而前置2至3小時，針對小雨量的部分，隨機森林之預報值容易有低估現象，其相對係數較低，且均方根誤差較高；針對大雨量的部分，支撐向量機則有高估或低估問題，但就整體預報結果來看，支撐向量機之預報雨量有較好的表現。最後，為驗證即時雨量預報模式之準確度及可靠度，將預報結果和地面雨量站觀測值進行分析比較，比較結果顯示，支撐向量機表現較好，而隨機森林則有較嚴重之低估情況。

The purpose of this study is to develop a real-time rainfall forecasting model to predict lead-time 1~3 hours rainfall by using two machine learning techniques, Random Forests (RF) and Support Vector Machine (SVM), with QPESUMS data as input. In this study, the data were collected from six typhoon events for three various reservoir catchments (i.e., Feitsui, Deji, and Zengwen reservoir catchment). A number of variables including QPESUMS data, gird XY position, grid elevation and typhoon information were examined for finding suitable input variables in rainfall forecasting. Besides, two model structures were also tested: (1) single-mode and (2) multiple-mode for deciding a better model structure. Based on two performance indexes (i.e., correlation coefficient and root mean squared error), the results suggest that the single-mode model structure gives a better performance and both machine learning techniques show reasonable performance. However, for lead-time 2~3 hours, RF will underestimate in low rainfall amount and SVM will overestimate or underestimate in high rainfall amount. Moreover, because of the extended of prediction time, time lag and the decrease of accuracy occur. Overall, SVM-based rainfall forecasting model can give better results than RF-based model. In order to verify the ability and reliability of the proposed real-time rainfall forecasting model, the study compared the predictions and rain gauge data. The results indicate that SVM performs better than RF.

王盼(2014).基於隨機森林模型的需水預測模型及其應用.水資源保護.第30卷.第1期.第34-37頁.
丘台光(2002) .劇烈天氣監測系統QPESUMS之服務與應用.中央氣象局.
巫孟璇(2013).地文性淹水即時預報模式之發展與應用.水利及海洋工程學系.台南,國立成功大學.碩士論文.
李志昕(2011).區域系集預報系統研究：物理參數化擾動.中央氣象局氣象資訊中心.
李明軒(2007).支撐向量機與模糊推論於流量預報即時誤差修正之研究.水利及海洋工程學系.台南,國立成功大學.碩士論文
沈哲緯(2015).運用隨機森林演算法進行莫拉克颱風災區土石流發生因子關聯性分析.財團法人中興工程顧問社防災科技研究中心.
林忠義.雷達資料同化於極短期定量降雨預報之研究.台灣颱風洪水研究中心.
洪國展(2012).利用ABLER法發展臺灣地區雷達回波外延估計.多采科技有限公司.
張亦凡(2005)支撐向量機在即時河川水位預報之應用.水利及海洋工程學系.台南,國立成功大學.碩士論文
張保亮(2007). QPESUMS產品應用與客製化介紹.中央氣象局氣象衛星中心.
張斐章(2012).融合多重雨量資訊於水庫集水區即時雨量推估及入庫流量預報技術之研究(1-2).國立台灣大學.
郭家妏(2014).隨機森林在河川水位即時預報之應用.水利及海洋工程學系.台南,國立成功大學.碩士論文
陳忠煒 (2014).應用遙測及地面雨量資訊於WRF颱風時雨量預報. 水利及海洋工程學系.台南,國立成功大學.碩士論文
陳憲宗(2006).支撐向量機及模糊推理模式應用於洪水水位之即時機率預. 水利及海洋工程學系. 台南,國立成功大學.博士論文.
陳薇伊(2011).克利金法即時修正大甲溪雷達估計降雨.水利及海洋工程學系.台南,國立成功大學.碩士論文.
曾財益(1995).洪水即時預報模式及不確定性分析之研究.水利及海洋工程學系.台南,國立成功大學.碩士論文.
黃子晉. 運用支持向量機與類神經網路於金融時間序列之預測與比較.龍華科技大學.
黃椿喜(2015).氣象局官方與主要數值天氣預報指引之定量降水預報校驗與綜合比較.中央氣象局氣象預報中心.
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. 生物環境系統工程學系.台北,國立台灣大學.碩士論文.
葉世瑄(2015).極短期定量降水預報技術於梅雨季節之校驗結果.中央氣象局氣象預報中心.
劉鑌鋈(2009).利用機器學習修正QPESUMS.水利及海洋工程學系.台南,國立成功大學.碩士論文
歐鐙元(2015).應用隨機森林(Random Forest)演算法於WorldView-2衛星影像大蒜分類判釋之研究.土地管理學系.台中.逢甲大學.碩士論文.
蔡禹明(2012).應用ARMOR方法於台灣強降雨事件定量降水預報之研究. 中央氣象局氣象預報中心.
謝章廷(2007).應用雷達降雨於分布型水文模式.水利及海洋工程學系.台南,國立成功大學.碩士論文.
Teague (2013). “Radar Rainfall Application in Distributed Hydrologic Modeling for Cypress Creeek Watershed, Texas.” Journal of Hydrologic Engineering. Vol. 18, Issue 2.
Berne (2004). “Temporal and spatial resolution of rainfall measurements required for urban hydrology.” Journal of Hydrology. Vol. 299, Pages 166-179.
Breiman (2001). “Random Forests.” Machine Learning, 45(1), 5-32.
Chen (2012). “Statistical Uncertatinty Estimation Using Random Forests and Its Application to Drought Forecast.” Mathematical Problems in Engineering.
Gianluca Bontempi (2013). “Machine Learning Strategies for Time Series Prediction.” Machine Learning Group, Computer Science Department.
Goormans, T. and Willems, P. (2012). “Using Local Weather Radar Data for Sewer System Modeling: Case Study in Flanders, Belgium.” Journal of Hydrologic Engineering. Vol. 18, Issue 2.
Gourley (2002). “An Exploratory Multisensor Technique for Quantitative Estimation of Stratiform Rainfall.” Journal of Hydrology. Vol. 3.
Gustavo E. A. P. A. Batista (2004). “A study of the behavior of several methods for balancing machine learning training data.” ACM SIGKDD Explorations Newsletter. Vol. 6, Pages 20-19, June.
Gwo-Fong Lin (2013). “Development of an Effective Data-Driven Model for Hourly Typhoon Rainfall Forecasting.” Journal of Hydrolohy. Vol. 495, Pages 52-63.
J.A.K. Suykens (2002). “Weighted least squares support vector machines : robustness and sparse approximation.” Neurocomputing. Vol. 48, Pages 85-105, October.
J.T.B. Obeysekera (1987). “On Parameter Estimation of Temporal Rainfall Models.” Water Resoureces Research. Vol. 23, No. 10, Pages 1837-1850, October.
Jin Li (2011). “Application of machine learning methods to spatial interpolation of environmental variables.” Environmental Modelling & Software. Vol.26, Issue 12, Pages 1647-1659.
Jyothis Joseph (2013). “Rainfall Prediction using Data Mining Techniques". International Journal of Computer Applications.” 83.8.
38. Konstantine P. Georgakakos (2010). “Quantitative Precipitation Forecast Techniques for Use in Hydrologic Forecasting.” American Meteorological Society.
Lijuan Cao (2002). “Dynamic support vector machines for non-stationary time series forecasting.” Intelligent Data Analysis. Vol.6, Issue 1, Pages 67-83.
Lijuan Cao (2003). “Support vector machines experts for time series forecasting.” Neurocomputing. 51, Pages 321-339.
Lily Ingsrisawang (2008). “Machine Learning Techniques for Short-Term Rain Forecasting System in the Northeastern Part of Thailand.” World Academy of Science, Engineering and Technology. Vol. 2.
Ling-Feng Hsiao (2013). “Ensemble forecasting of typhoon rainfall and floods over a mountainous watershed in Taiwan.” Journal of Hydrology.
M.A. Mohandes (2003). “Support vector machines for wind speed prediction.” Science Direct. Vol. 29, Issue 6, Pages 939-947.
Marcus Paulat (2008). “A gridded dataset of hourly precipitation in Germany : Its construction, climatology and application.” Meteorologische Zeitschrift. Vol. 17, No. 6, 719-732, December.
Mark N. French (1992). “Rainfall forecasting in space and time using a neural network.” Journal of Hydrology. Vol. 137, Issues 1-4, Pages 1-31.
Mark N. French (1994). “A model for real-time quantitative rainfall forecasting usingre remotes sensing.” Water Resources Research. Vol. 30, Issue 4, Pages 1085-1097.
Palmer Ds (2007). “Random forest models to predict aqueous solubility. Journal of Chemical Information and Modeling.” Journal of Chemical Information and Modeling. 47(1), Pages 150-158.
Paolo Burlando (1993). “Forecasting of short-term rainfall using ARMA models.” Journal of Hydrology. Vol. 144, Issues 1-4, Pages 193-211.
Pathak, C. (2013). “Special Issue on Radar Rainfall Data Analyses and Applications.” Journal of Hydrologic Engineering. Vol. 18, Issue 2.
Anantha M. Prasad (2006). “Newer Classification and Regression Tree Technique-Bagging and Random Forests for Eclolgical Prediction.” Ecosystems, 9(2), 181-199.
Rui Huang (2013). “Using Random Forest to Intergate Lidar Data and Hyperspectral Magery for Land Cover Classification.” School of Communication and Information Engineetring Shanghai University.
Sumi S. Monira (2010). “Comparison of Artificially Intelligent Methods in Short Term Rainfall Forecast.” Computer and Information Technology. Pages 23-25.
Urs Germann (2012). “Can Lagrangian Extrapolation of Radar Fields Be Used for Precipitation Nowcasting over Complex Alpine Orography.” American Metorologist Society.
Vladimir Vapnik & Corinna Cortes (1995). “Marchine Learning.” 20,273-297
Wei-Chiang Hong (2008). “Rainfall forecasting by technological machine learning models.” ScienceDirect. Vol. 200, Issue 1, Pages 41-57.
X. Suna (2000). “Flood estimation using radar and raingauge data.” Journal of Hydrology. Vol. 239, Pages 4-18, December.
Xiaoyu Wu (2015). “A Two-Stage Random Forest Method for Short-term Load Forecasting.” School of Electrical Enginerring. Beijing Jiaotong University.
Yonas B. Dibike (2001) “Model Induction with Support Vector Machines: Introduction and Applications.” Journal of Computing in Civil Engineering. Vol. 15, Issue 3.
Yu Pao-Shan (2006). “Support vector regression for real-time flood stage forecasting.” Journal of Hydrology. Vol. 328, Issues 3-4. Pages 704-716.
Yu-Chi Wang (2007). “Application of QPESUMS System with Distributed Rainfall-Runoff Model.” 2nd International Conference on Urban Disaster Reduction.