高屏溪為高屏地區各標的用水主要供水來源，其來自於夏秋兩季熱帶氣旋所帶來的豐沛降雨，但因地勢陡峭導致河川坡陡流急，無法有效蓄存水源，容易形成一季水源四季使用之情形。隨著經濟的發展，該地區對穩定水源的依賴日漸加深，然而成本及其他因素的考量使得流量站不易全面性的設置，而所設置的流量站也可能因為損壞或廢站而記錄期限不長，因此，選擇一能有效且準確延伸流量記錄之模式，是水資源規劃與管理重要的工作項目之一。
類神經網路近年來蓬勃發展，已發展出不同的模式相繼被應用在許多領域上，因其線性及非線性的統籌歸納演算能力，透過網路自我訓練調整參數，可解決許多參數給定不易的問題。本研究目的為比較不同類神經網路模式包含倒傳遞類神經(back propagation neural network, BPN)及徑向基底函數網路(radial basis function neural network, RBFN)與傳統面積比例法(area ratio method, ARM)及線性迴歸(linear regression, LR)於日流量資料延伸之適用性。本研究以位於台灣南部高屏溪流域之里嶺大橋測站為研究對象，首先從高屏溪各主支流上選取紀錄期限較長、紀錄較完整之測站作為模式輸入，另考量稽延時間對流量的影響，加入前1至3天的日流量紀錄進行里嶺大橋站日流量紀錄延伸，並以平均絕對誤差(mean absolute error, MAE)、均方根誤差(root mean square error, RMSE)及效率係數(coefficient of efficiency, CE)評估不同模式之延伸成果。
本研究以BPN、RBFN、AMR及LR進行流量資料之延伸比較，目的為尋求一較為適用及準確之資料延伸方式，結果顯示BPN及RBFN之誤差均小於ARM及LR，表示類神經網路之方式具有相當程度之穩定性，可達到較好的延伸成果，對於水資源規劃設計能提供較精確的時序列流量資料。綜合比較研究中所使用的四種模式，其中以BPN-II1之成果最為優異，獲得最小的MAE、RMSE及最高的CE，分別為47.73 cms、102.85 cms及0.956，說明BPN在此案例中適用性高於其他三模式。

Acquisition and analysis of streamflow is an essential step for water-resources engineering planning and design. However, streamflow gauge stations are not established basinwide since high construction and maintenance costs. Insufficient streamflow data are frequently met in hydrologic studies. Thus, streamflow data extension became an important and inevitable task in water-resources planning and management.
The artificial neural network (ANN) is a well-developted tool applied in hydrologic studies. The ANN is an adaptive system that change its structure based on external or internal information for flowing through the network during the learning phase, thus ANN can be used to model complex hydrologic processes with nonlinear relationship. The purpose of this study is to find the best data-extension models, wehich include back-propagation neural network (BPN)、radial basis function neural network (RBFN)、area ratio method (ARM), and linear regression (LR). The proposed approach is carried in the upstream watershed of Koaping creek basin located in southern Taiwan. The inputs of the models include different time-lag streamflow from upstream sgauged station. The mean absolute error (MAE)、root mean square error (RMSE), and coefficient of efficiency (CE) are used as criteria evaluate models .
The ANN models are offer accurate extension in water-researches planning. The research shows that the BPN-II1 is the best model, which results in the minimum MAE, RMSE and the highest CE of 47.73 cms, 102.85 cms and 0.956.

1.Ahmad S., and Simonovic S.P., (2005) An artificial neural network model for generating hydrograph from hydro-meterological parameters. Journal of Hydrologic Engineering, 315:236-251.
2.Besaw L.E., Rizzo D.M., Bierman P.R., and Hackett W.R., (2010) Advances in ungauged streamflow prediction using artificial neural networks. Journal of Hydrology, 386:27-37.
3.Chang L.C. and Chang F.J., (2001) Intelligent control for modeling of real-time reservoir operation. Hydrological Processes 15(9):1621-1634.
4.Chang L.C., Chang F.J., and Chiang Y.M., (2004) A two-step ahead recurrent neural network for streamflow forecasting. Hydrological Processes, 18:81-92.
5.Chang F.J., Chang L.C., Kao H.S., and Wu G.R., (2010) Assessing the effort of meteorological variables for evaporation estimation by self-organizing map neural network. Journal of Hydrology 384(1-2):118-129.
6.Chen S., Cowan C.F.N., and Grant P.M., (1991) Orthogonal least squares learning algorithm for radial basis function networks. IEEE Transactions on Neural Networks, 2(2):302-309.
7.Dastorani M.T., Moghadamnia A., Piri J., and Ramirez M.R., (2010) Application of ANN and ANFIS models for reconstructing missing flow data. Environmental Monitoring and Assessment, 166:421-434.
8.Dawson C.W., Abrahart R.J., Shamseldin A.Y., and Wilby R.L., (2006) Flood estimation at ungauged sites using artificial neural networks. Journal of Hydrology, 319:391-409.
9.Hebb D.O., (1949) The organization of behavior. A Neuropsyological Theory.
10.Hopfield J.J, (1982) Neural networks and physical system with emergent collective computational abilities. Proceeding of the National Academy of Scientists, 79:2554-2558.
11.Kagoda P.A., Ndiritu J., Ntuli C., and Mwaka B., (2010) Application of radial basis function neural networks to short-term streamflow forecasting. Physics and Chemistry of the Earth, 35:571-581.
12.Kisi O., Nia A.M., Gosheh M.G., Tajabadi M.R.J., and Ahmadi A., (2012) Intermittent streamflow forecasting by using several data driven techniques. Water Resources Management, 26:457-474.
13.Lin G.F., and Wu M.C., (2007) A SOM-based approach to estimating design hyetographs of ungauged sites. Journal of Hydrology, 339:216-226.
14.McClelland T.L., and Rumelhart D.E., (1986) Parallel distributed processing MIT press and the PDP Research group. Cambridge.
15.McCulloch W.S, and Pitts W., (1943) A logical calculus of the ideas immanent in nervous activity. Bulletin of Mathematical Biophysics, 5:115-133.
16.Minsky M.L., and Papert S.A., (1969) Perceptrons. Cambridge.
17.Moustris K.P., Larissi I.K., Nastos P.T., and Paliatsos A.G., (2011) Precipitation forecast using artificial neural networks in specific region of Greece. Water Resources Management, 25:1979-1993.
18.Rosenblatt F., (1958) The precptron: A probabilistic model for information storage and organization in the Brain. Psychological Review, 65:386-408.
19.Rumelhart D.E., and McClelland J.L., (1986) Parallel distributed processing: Explorations in the microstructure of cognition. Cambridge.
20.Sattari M.T., Apaydin H., and Ozturk F., (2012) Flow estimation for the Shou stream using artificial neural networks. Environmental Earth Science, 66:2031-2045.
21.Schulze R.E., Dent M.C., Lynch S.D., Schafer N.W., Kienzle S.W., and Seed A.W., (1994) Rainfall. In:Schulze R.E.(Ed.) Hydrology and agrohydrology: a text to accompany the ACRU 3.00 agrohydrological modeling system. Agricultural Catchments Research Unit, Dept. of Agricultural Engineering , University of Natl, Pietermaritzburg, RSA, Report 41, 34p.
22.Shamseldin A.Y., (1997) Application of a neural network technique to rainfall-runoff modeling. Journal of Hydrology, 199:272-294.
23.Singh P., and Deo M.C., (2007) Suitability of different neural networks in daily flow forecasting. Applied Soft Computing, 7:968-978.
24.Whigham P.A., and Crapper P.F., (2001) Modeling rainfall-runoff using genetic programming. Mathematical and Computer Modeling, 33:707-721.
25.Zadeh M.R., Amin S., Khalili D., and Singh V.P., (2010) Daily outflow prediction by multi-layer perceptron with logistic sigmoid and tangent sigmoid activation function. Water Resources Management, 24:2673-2688.
26.周仲島、陳永明和黃柏誠(2007)，侵臺颱風劇烈降雨之氣候變異分析，水資源管理會刊，19-29。
27.汪中和(2009)，氣候暖化與台灣環境危機，方濟生態論壇3。
28.李品輝(2009)，以類神經網路探討全台蒸發量區域性分類與推估之成效，國立台灣大學生物資源暨農學院生物環境系統工程學系研究所碩士論文。
29.林政華(2012)，以類神經網路探討雲林沿海地區地下水砷濃度與水質特徵，國立台灣大學生物資源暨農學院生物環境系統工程學系研究所碩士論文。
30.徐宏瑋(2004)，降雨量變遷趨勢檢定與分析，國立台灣大學生物環境系統工程學系研究所碩士論文。
31.陳柏蒼和陳昶憲(2002)，未設測站流量推估-利用類神經網路建構模式，第七屆海峽兩岸水利科技交流研討會，第543-550頁。
32.張斐章和張麗秋(2006)，類神經網路，東華書局有限公司。
33.葉怡成(1993)，類神經網路模式應用與實作，儒林圖書有限公司。
34.詹仕堅(2003)，使用類神經網路在洪水推估之研究-以集水區地文特徵為基礎，國立台灣大學地理環境資源學研究所博士論文。
35.鄭伊婷(2009)，以複合型模式分析區域淹水潛勢，淡江大學水資源及環境工程研究所碩士論文。　
36.鍾侑達(2005)，遺傳規劃與類神經網路在河川演算上之比較，逢甲大學土木及水利工程研究所碩士論文。
37.羅華強(2005)，類神經網路-Matlab的應用，高立圖書有限公司。