Investigation internal parameters of neural network model for Flood Forecasting at Upper river Ping, Chiang Mai
- 作者:Tawee Chaipimonplin
- 关键词:neural network ; flood forecasting ; bayesian regularization ; levenberg ; marquardt ; internal parameters ; upper river ping ; chiang mai
- 刊名:KSCE Journal of Civil Engineering
- 出版年:2016
- 出版时间:January 2016
- 年:2016
- 卷:20
- 期:1
- 页码:478-484
- 全文大小:1,532 KB
- 参考文献:Anctil, F. and Lauzon, N. (2004). “Generalisation for neural networks through data sampling and training procedures, with applications to streamflow predictions.” Hydrology and Earth System Sciences, Vol. 8, pp. 940–958.CrossRef
Anctil, F., Lauzon, N., Andreassian, V., Oudin, L., and Perrin, C. (2006). “Generalisation for neural networks through data sampling and training procedures, with applications to streamflow predictions.” Journal of Hydrology, Vol. 328, Nos. 3–4, pp. 717–725.CrossRef
Anctil, F., Michel, F., Perrin, C., and Andreassian, V. (2004). “A soil moisture index as an auxiliary ANN input for stream flow forecasting,” Journal of Hydrology, Vol. 286, Nos. 1–4, pp. 155–167.CrossRef
Anctil, F., Perrin, C., and Andreassian, V. (2004). “Impact of the length of observed records on the performance of ANN and of conceptual parsimonious rainfall-runoff forecasting models.” Environmental Modeling & Software, Vol. 19, No. 4, pp. 357–368.CrossRef
ASCE (2000a). “Artificial neural networks in hydrology, II: Hydrologic applications.” Journal of Hydrologic Engineering, Vol. 5, No. 2, pp. 124–137.CrossRef
ASCE (2000b). “Artificial neural networks in hydrology, I: Hydrologic applications.” Journal of Hydrologic Engineering, Vol. 5, No. 2, pp. 115–123.CrossRef
Beale, M. H., Hagan, M. T., and Demuth, H. B. (2011). Neural Network Toolbox TM 7: User's guide, http://www.mathworks.com .
Chaipimonplin, T., See, L. M., and Kneale, P. E. (2010). “Using radar data to extend the lead time of neural network forecasting on the River Ping.” Disaster Advances, Vol. 3, No. 3, pp. 35–43.
Chaipimonplin, T., See, L. M., and Kneale, P. E. (2011a). “Improving neural network for flood forecasting using radar data on the Upper Ping river.” Proceeding 19th International Congress on Modelling and Simulation, Modelling and Simulation Society of Australia and New Zealand, pp. 1070–1076, (MODSIM2011).
Chaipimonplin, T., See, L. M., and Kneale, P. E. (2011b). “Comparison of neural network learning algorithms: BR and LM, for flood forecasting, Upper Ping catchment.” Proceeding 10th International Symposium on New Technologies for Urban Safety of Mega Cities in ASIA, Chiang Mai, Thailand, (USMCA 2011) (poster).
Chen, C. S., Chen, B. P. T., Chou, F. N. F., and Yang, C. C. (2010). “Development and application of a decision group back-propagation neural network for flood forecasting.” Journal of Hydrology, Vol. 385, Nos. 1–4, pp. 173–182.CrossRef
Dawson, C. W. and Wilby, R. (1999). “A comparison of artificial neural networks used for river flow forecasting.” Hydrology and Earth Sciences, Vol. 3, No. 4, pp. 529–540.CrossRef
Dawson, C. W., Abrahart, R. J., and See, L. M. (2007). “Hydro test: A web-based toolbox of evaluation metrics for the standardized assessment of hydrological forecasts.” Environmental Modelling & Software, Vol. 22, No. 7, pp. 1034–1052.CrossRef
Department of Disaster Prevention and Mitigation (2011). Flood Situation report, http://disaster.go.th/dpm/flood/flood.html .
Department of Water Resources (2007). Ping catchment, http://mekhala.dwr.go.th/mekhala/Basin_North.asp.
Derakhshan, H. and Talebbeydokhti, N. (2011). “Rainfall disasggregation in non-recording gauge stations using space-time information system.” Scientia Iranica, Vol. 18, No. 5, pp. 995–1001.CrossRef
Foresee, F. D. and Hagan, M. T. (1997). “Gauss-Newton approximation to Bayesian learning.” Proceeding the 1997 International Joint Conference on Neural Network, pp. 1930–1935.
Fun, M. H. and Hagan, M. T. (1996). “Levenberg-Marquardt training for modular networks.” Proceeding the 1996 International Conference on Neural Network, Vol. 1, pp. 468–473.
Hornik, K. (1993). “Some new results on neural network approximation.” Neural Networks, Vol. 6, No. 8, pp. 1069–1072.CrossRef
Hydrology and Water Management Centre for Upper Northern Region, (2007). Upper northern region
flood warning brochures, http://www.hydro-1.net/.
Hydrology and Water Management Centre for Upper Northern Region, (2012). http://www.hydro-1.net/08HYDRO/PORTAL/IMAGES/100119-BOAD-P.1aaaa.jpg.
Kisi, O. (2007). “Streamflow
forecasting using different artificial neural network algorithms.” Journal of Hydrologic Engineering, Vol. 12, No. 5, pp. 532–539.CrossRef
Lorrai, M. and Sechi, G. M. (1995). “Neural net for modeling rainfallrunoff transformations.” Water Resources Management, Vol. 9, pp. 299–313.CrossRef
Mackay, D. J. C. (1992). “Bayesian interpolation.” Neural Computation, Vol. 4, No. 3, pp. 415–447.CrossRef
Maier, R. H. and Dandy, G. C. (1996). “The use of artificial neural networks for the prediction of water quality parameters.” Water Resources Research, Vol. 32, No. 4, pp. 1013–1022.CrossRef
Maier, R. H. and Dandy, G. C. (1998). “The effect of internal parameters and geometry on the performance of back-propagation neural network: An empirical study.” Environmental Modelling & Software, Vol. 13, No. 2, pp. 193–209.CrossRef
MathWorks (2012a). Documentation-Neural Network Toolbox, http://www.mathworks.com/help/toolbox/nnet/ref/trainlm.html.
文摘
The flood issue for forecasters at Chiang Mai derives from the monsoon rainfall, which leads to serious out-of-bank flooding two to four times a year. Data for stage and rainfall per hour in Upper Ping catchment is limited as the historical flood record is limited in length. Neural Network forecasting models are potentially very powerful forecasters where the data are limited. However, insufficient data for Neural Network training reduces the model performance. All data for Neural Network is divided into three datasets: training, validation and testing. In addition most of learning algorithms require validation data unlike Bayesian Regularization (BR) algorithm with no validation data. The power of BR to forecast effectively where data set are limited. Therefore, this algorithm is worth exploring for the Upper Ping catchment, also comparison performance with the Levenberg-Marquardt algorithm (LM) that is the fastest training. In addition, for the best model performance hidden nodes are set as 50%, 75% and 2n+1 of input nodes. The Neural Network model is used to predict water stage at P.1 and P.67 station at lead times of 6 and 12 hours with two different learning algorithms. The results have found that Neural Network performance training with LM algorithm is better than BR algorithms by improving the peak stage. The overall performance of the model that has the hidden nodes less than input nodes of 50% and 75% has the best performance.