臺灣博碩士論文加值系統

台灣水庫淤砂問題日益嚴重，入庫砂量大多來自於汛期的颱風事件，因此若能有效評估並預報颱風事件的輸砂量，權責單位便能提前做出準備與災害因應來減少入庫泥沙量並增加水庫使用年限。水理輸砂數值模式已廣泛應用於水利工程實務上，主要的輸砂量推估是透過經驗公式搭配動量方程式將複雜水砂運動過程進行簡化，然而其需耗費大量的時間在參數的率定與驗證。由於台灣的氣候地形因素，致使河川的變動量非常大，因此上述的數值模式參數會隨著不同的時間而變動，往往一場特定事件的參數可能無法用於其他事件，加上現地量測資料不確定性與短缺之問題，且模式因簡化而導致本身的不確定性，使得最佳參數組合並不易獲得，因此本研究擬透過廣泛運用於氣象預報中的系集演算的原理，用單一模式多重參數進行河道模擬，透過蒙地卡羅法減少模式模擬上的不確定性。
集水區的土砂災害預報的準確性及效率十分重要，近年來，機器學習與深度學習於預測方面的研究已有許多成果，其具有可模擬非線性系統、運算快速、持續學習的優勢，已廣泛應用於水文模擬預測上，像是淹水範圍、淹水深度等皆有取得不錯的成效。
本研究將利用系集演算原理得到的參數組合進行水砂數值模擬，將模擬的成果以機器學習與深度學習進行土砂預報。本研究以石門水庫上游的大漢溪河段為研究區域，首先透過HEC-HMS模式進行前置支流資料的建置，接著用SRH2D模式結合系集演算原理進行輸砂量的模擬，最後以評鑑指標與改善率決定最佳化系集成員數，並以此資料透過MLP、DNN、RNN、LSTM、GRU進行學習並預報，提供未來1至6小時的輸砂量預報。研究結果顯示選用最佳化系集成員數為20組與Parker公式的模擬結果作為預報的資料來源，而深度學習建置預報會來的比機器學習來的穩定且準確，而深度學習中又以DNN表現最好，不論於趨勢或是洪峰輸砂量，皆能準確掌握。

The reservoir sedimentation issue which caused by Typhoon events become serious in Taiwan. If the sedimentation is effectively evaluated and predicted during typhoon events, the authority can make preparations and disaster response in advance to mitigate the quantity of incoming sediment and increase the life of reservoirs. The hydraulic model has been widely applied to predict the sedimentation in engineering practice. The estimation of sediment transport is mainly based on the simplification of the complex process through empirical equations and momentum equations, but it takes a lot of time for the parameters calibration and verification. Considering the climatic and topographic complexities in Taiwan, the variability of rivers is very large, so the parameters of the above numerical model will change with time. Therefore, this study is intended to reduce the uncertainty of model simulation by using the principle of system integration algorithm, which is widely used in meteorological forecasting, to simulate the river channel with multiple parameters of a single model by Monte Carlo method. The accuracy and efficiency of soil and sand hazard prediction in catchment areas are very important. In recent years, there have been many achievements in machine learning and deep learning for prediction, which have the advantages of simulating non-linear systems, fast computation, and continuous learning, and have been widely applied to hydrological simulation and prediction, such as inundation extent and inundation depth.
In this study, we simulate the water and sand values by using the parameter combinations obtained from the coefficient set algorithm, and then use the simulation results for soil and sand prediction by machine learning and deep learning. In this study, the study area is the upper reaches of Shimen Reservoir, and the data of the antecedent tributaries are firstly constructed by HEC-HMS model, and then the sand transport is simulated by SRH-2D model combined with the principle of system integration. The data is learned and predicted by MLP, DNN, RNN, LSTM, and GRU, and the prediction of sand delivery is provided for the next 1 to 6 hours. The results of the study show that the number of optimizing system integrators is 20 and the simulation results of Parker's formula are used as the data source for forecasting.

摘要 I
Abstract III
目錄 V
圖目錄 VII
表目錄 X
第一章 緒論 1
1.1 研究動機 1
1.2 研究目的 2
1.3研究架構 3
1.4 研究流程 4
第二章 文獻回顧 5
2.1 水理輸砂數值模式 5
2.2 不確定性及系集演算原理之應用 6
2.3 機器學習與深度學習之應用 9
第三章 研究方法 11
3.1 HEC-HMS模式介紹 11
3.1.1模式理論介紹 12
3.2 SRH-2D模式介紹 16
3.2.1 模式介紹 16
3.2.2水理控制方程式 17
3.2.3 輸砂控制方程式 19
3.2.4 模式參數說明 24
3.3系集演算法 25
3.4 機器學習介紹 27
3.4.1 多層感知器(Multi-Layer Perceptron，MLP) 28
3.4.2 深度神經網路(Deep Neural Networks，DNNs) 29
3.4.3 遞迴神經網路(Recurrent Neural Network，RNN) 31
3.4.4長短期記憶網路(Long Short-Term Memory，LSTM) 32
3.4.5閘門循環單元神經網路(Gated Recurrent Unit，GRU) 35
3.5 評鑑指標 37
第四章 基本資料蒐集及模型建置 40
4.1基本資料蒐集 40
4.2 HEC-HMS模式建置 46
4.3 SRH-2D模式建置 50
4.4 砂量預報建置 58
第五章 結果與討論 60
5.1 HEC-HMS模式之率定驗證 60
5.2 SRH-2D模式之率定驗證 63
5.2.1水理模式結果 63
5.2.2輸砂模式結果 66
5.3 建立土砂量預報 78
第六章 結論與建議 91
6.1 結論 91
6.2 建議 92
參考文獻 93

1.Alfieri, L. (2013). GloFAS – global ensemble streamflow forecasting and flood early warning. Hydrology and Earth System Sciences, 17(3), 1161-1175. doi:10.5194/hess-17-1161-2013
2.Benke, K. K., Lowell, K. E., & Hamilton, A. J. (2008). Parameter uncertainty, sensitivity analysis and prediction error in a water-balance hydrological model. Mathematical and Computer Modelling, 47(11-12), 1134-1149. doi:10.1016/j.mcm.2007.05.017
3.Beven, K., & Binley, A. (1992). The future of distributed models: Model calibration and uncertainty prediction. Hydrological Processes, 6(3), 279-8. doi:10.1002/hyp.3360060305
4.Bouzeria, H., Ghenim, A. N., & Khanchoul, K. (2017). Using artificial neural network (ANN) for prediction of sediment loads, application to the Mellah catchment, northeast Algeria. Journal of Water and Land Development, 33(1), 47-55. doi:10.1515/jwld-2017-0018
5.Butts et.al (2004). An evaluation of the impact of model structure on hydrological modelling uncertainty for streamflow simulation. Journal of Hydrology, 298(1-4), 242-266. doi:10.1016/j.jhydrol.2004.03.042
6.Chiang, Y. et.al (2017). Evaluating the contribution of multi-model combination to streamflow hindcasting by empirical and conceptual models. Hydrological Sciences Journal, 62(9), 1456-1468. doi:10.1080/02626667.2017.1330543
7.Freer, J., Beven, K., & Ambroise, B. (1996). Bayesian Estimation of Uncertainty in Runoff Prediction and the Value of Data: An Application of the GLUE Approach. Water Resources Research, 32(7), 2161-2173. doi:10.1029/95wr03723
8.Gneiting, T. (2005). ATMOSPHERIC SCIENCE: Weather Forecasting with Ensemble Methods. Science, 310(5746), 248-249. doi:10.1126/science.1115255
9.Hsiao et.al (2013). Ensemble forecasting of typhoon rainfall and floods over a mountainous watershed in Taiwan. Journal of Hydrology, 506, 55-68. doi:10.1016/j.jhydrol.2013.08.046
10.Hu, C., Wu, Q., Li, H., Jian, S., Li, N., & Lou, Z. (2018). Deep Learning with a Long Short-Term Memory Networks Approach for Rainfall-Runoff Simulation. Water, 10(11), 1543. doi:10.3390/w10111543
11.Jiang, S., Ren, L., & Yang, X. (2014). Multi-model ensemble hydrologic prediction and uncertainties analysis. Proceedings of the International Association of Hydrological Sciences, 364, 249-254. doi:10.5194/piahs-364-249-2014
12.Lai, Y. G., & Wu, K. (2014). Combined Vertical And Lateral Channel Evolution Numerical Modeling.Water
13.Lopes, V. L. (1996). On the effect of uncertainty in spatial distribution of rainfall on catchment modelling. Catena, 28(1-2), 107-119. doi:10.1016/s0341-8162(96)00030-6
14.Melo, G. A. (2019). A new approach to river flow forecasting: LSTM and GRU multivariate models. IEEE Latin America Transactions, 17(12), 1978-1986. doi:10.1109/tla.2019.9011542
15.Napolitano, G., Serinaldi, F., & See, L. (2011). Impact of EMD decomposition and random initialisation of weights in ANN hindcasting of daily stream flow series: An empirical examination. Journal of Hydrology, 406(3-4), 199-214. doi:10.1016/j.jhydrol.2011.06.015
16.Pappenberger, F., & Beven, K. J. (2004). Functional classification and evaluation of hydrographs based on Multicomponent Mapping (Mx). International Journal of River Basin Management, 2(2), 89-100. doi:10.1080/15715124.2004.9635224
17.Perrin, C., Michel, C., & Andréassian, V. (2001). Does a large number of parameters enhance model performance? Comparative assessment of common catchment model structures on 429 catchments. Journal of Hydrology, 242(3-4), 275-301. doi:10.1016/s0022-1694(00)00393-0
18.Reichle, R. H., McLaughlin, D. B., & Entekhabi, D. (2002). Hydrologic Data Assimilation with the Ensemble Kalman Filter. American Meteorological Society, 130(1), 103-114.doi:10.1175/1520-0493(2002)130<0103:HDAWTE>2.0.CO;2
19.Song, X., Zhan, C., Kong, F., & Xia, J. (2011). Advances in the study of uncertainty quantification of large-scale hydrological modeling system. Journal of Geographical Sciences, 21(5), 801-819. doi:10.1007/s11442-011-0881-2
20.Supharatid, S. (2003). Tidal-Level Forecasting and Filtering by Neural Network Model. Coastal Engineering Journal, 45(1), 119-137. doi:10.1142/s0578563403000695
21.Tiwari, M. K., & Chatterjee, C. (2010). Uncertainty assessment and ensemble flood forecasting using bootstrap based artificial neural networks (BANNs). Journal of Hydrology, 382(1-4), 20-33. doi:10.1016/j.jhydrol.2009.12.013
22.Wu, Yeh, & Lai. (2019). A Combined Field and Numerical Modeling Study to Assess the Longitudinal Channel Slope Evolution in a Mixed Alluvial and Soft Bedrock Stream. Water, 11(4), 735. doi:10.3390/w11040735
23.Xu, W. et.al.(2020). Using long short-term memory networks for river flow prediction. Hydrology Research, 51(6), 1358-1376. doi:10.2166/nh.2020.026
24.尹雄銳等(2006) 水文模擬與預測中的不確定性研究現狀與展望
25.王瀅婷(2010) 巴陵壩潰壩後對上游河床變遷影響之研究
26.李文獻(2012) 二維水理輸砂模式於壩體移除分析應用之探討
27.江宙君、陳嬿竹、吳德榮(2012) 定量降雨系集演算加值分析–以2012年6月梅雨鋒面為例，天氣分析與演算研討會論文彙編:171-175
28.陳憲宗、游保杉(2007) 洪水位之即時機率預報－結合支撐向量機與模糊推理
29.堂榮華、周復雄(2017) 基於Riverflow2D的來賓市城區洪水淹沒分析
30.經濟部水利署北區水資源局(2009) 石門水庫集水區泥沙推估與處置綜合評析計畫
31.黃麗文(2014) 序率水文模擬及二維水理輸砂模式於壩移除後河川形貌變遷分析應用