Document Type : Research Paper
Authors
1 Assistant Professor, Faculty of Agriculture, University of Tabriz, Iran
2 MSc, Faculty of Technology and Engineering, Islamic Azad University of Ahar, Iran
3 Assistant Professor, Faculty of Agriculture, Shahrekord University, Iran
Abstract
Floods are the main natural disasters that produce serious agricultural, environmental, and socioeconomical damages in many parts of the world. Accurate estimation of river flow in streams can have a significant role in water resources management and in protection from possible damages. This study aims to compare the abilities of Support Vector Machine (SVM), M5 model trees and Linear Regression (LR) methods in forecasting hourly discharge flow of Aharchay River. The hourly water level-discharge and 14 flood events data of Aharchay River measured at the Tazekand hydrometric station was used for modeling. The results showed that the SVM method gives more accurate results than the M5 model and LR method in forecasting river flow for next one and two hours with the R2=0.96 and RMSE=0.0472 (m3s-1) and the R2=0.90 and RMSE=0.1596 (m3s-1), respectively. Comparing the performance of SVR and M5 models indicated that, however the SVR approach may present more accurate results than the M5 model tree, but the M5 model provides more understandable, applicable and simple linear relation in forecasting hourly discharge. Thus, the M5 model tree can be used as an alternative method in forecasting hourly discharge.
Keywords
- Adamowski, J. and S.O. Prasher. 2012. Comparison of machine learning methods for runoff forecasting in mountainous watersheds with limited data. Journal of Water and Land Development, 17(7-8): 89–97.
- Alikhanineghad, M., A.J. Kamali and B. Nematollahi. 2013. Ground water forecasting using support vector machine, case study: Kerman watershed. 2nd International Conference on Plant, Water, Soil and Weather Modeling, may 8-9 2013, Kerman, Iran (in Persian).
- Bhattacharya, B. and D.P. Solomatine. 2003. Neural networks and M5 model trees in modeling water level-discharge relationship for an Indian river. In European Symposium on Artificial Neural Networks, Bruges (Belgium), 23–25 April; 407–412.
- Bhattacharya, B. and D.P. Solomatine. 2005. Neural networks and M5 model trees in modeling water level–discharge relationship. Neurocomputing, 63: 381–396.
- Emamifar, S. and A. Rahimi Khob. 2011. Evaluation of M5 model tree and experimental model Angstrom for estimating radiation reaching the Earth's surface. 1st National Conference on Agricultural Meteorology and Water Management, College of Agriculture and Natural Resources, Tehran University, Tehran, Iran (in Persian).
- Hamel, L.H. 2009. Knowledge discovery with support vector machines. Wiley, 262 pages.
- Lin, J.Y., C.T. Cheng and K.W. Chau. 2006. Using support vector machines for long term discharge prediction. Hydrological Sciences Journal, 51(4): 599–612.
- Mahjobi, A. and M. Tajrishi. 2010. Comparison of artificial neural network algorithms and decision trees in predicting changes in river salinity, case study: Karun River. 4th Conference and Exhibition on Environmental Engineering, Tehran Iran (in Persian).
- Pai, P.F. and W.C. Hong. 2007. A recurrent support vector regression model in rainfall forecasting. Hydrological Process, 21: 819-827.
10. Sattari, M.T., M. Pal, H. Apaydin and F. Ozturk. 2013. M5 model tree application in daily river flow forecasting in Sohu stream, Turkey. Water Resources, 40(3): 233–242.
11. Seefi, A., M. Mirlatifi and H. Reahi. 2013. Introduction and application of Least Square Support Vector Machine (LSSVM) for simulation of reference evapotranspiration and uncertainty analysis of results, A case study of the Kerman city. Journal of Irrigation and Water Engineering, 13: 67-79 (in Persian).
12. Solomatine, D.P. and Y. Xue. 2004. M5 model trees compared to neural networks: application to flood forecasting in the upper reach of the Huei River in china. Journal of Hydrologic Engineering, 9(6): 491-501.
13. Stravs, L. and M. Brilly. 2007. Development of a low-flow forecasting model using the M5 machine learning method. Hydrological Sciences Journal, 52(3): 466-477.
14. Vapnik, V.N. 1998. Statistical learning theory. Wiley, 736 pages.
15. Yu, P.S, S.T. Chen and I.F. Chang. 2006. Support vector regression for real-time flood stage forecasting. Journal of Hydrology, 328: 704-716.
16. Zahiri, A.R. and K.H. Ghorbani. 2013. Flow discharge prediction in compound channels by using decision model tree M5. Journal of Water and Soil Conservation, 20(3): 113-132 (in Persian).
)