Introduction and application of Least Square Support Vector Machine (LSSVM) for simlulation of reference evapotranspiration and uncertainty analysis of results, A case study of the Kerman city

Document Type : Original Article

Authors

Abstract

Estimating reference evapotranspiration (ETo) is one of the most important variables at water supply and distribution, irrigation management, irrigation systems design, agriculture and hydrological operations. The need for accurate estimates of ETo, complexity of ETo, unknowing mathematical of phenomenon, lack of reliable meteorological data, the cost of using lysimeters and their absent in most areas magnifies the need for developing new data mining methods. In this paper, the Least Square Support Vector Machine (LSSVM) model with three kernels function of RBF, Linear and Polynomial based on Gamma test used for estimating ETo and their results compared with other methods including Artificial Neural Networks (ANN) and Adaptive Neuro Fuzzy Inference System (ANFIS) models. In this analysis, annual daily meteorological data of Kerman synoptic station and ETo lysimeter data used. In this research, Gamma test was used for selecting the best combinations of input parameters for various models used instead of using trial and error classic methods. The combination including maximum and dew point temperatures, relative humidity, wind speed, and solar radiation selected as the best combination for estimating ETo and modeling was based on this combination. The LSSVM model with RBF kernel performed better than the SVM model with polynomial and linear kernels. Additionally, the distribution of prediction error was calculated that the ANFIS and LSSVM-RBF created less error in train and test steps, respectively. At the end study, Monte-Carlo uncertainty analysis was performed on results of different models that were used in this study. According to the results, predictions of LSSVM models showed less uncertainty than the ANFIS and ANN models were used.

Keywords

References

1-    بختیاری، ب.، ع. لیاقت، ع. خلیلی و م.ج. خانجانی. 1388. ارزیابی دو مدل ترکیبی برآورد تبخیر-تعرق مرجع چمن در بازه زمانی ساعتی (مطالعه موردی اقلیم کرمان). علوم و فنون کشاورزی و منابع طبیعی، علوم آب و خاک، سال سیزدهم، شماره 50.
2-    دلاور، م. 1384. تحلیل و ارائه مدل نوسانات تراز آب دریاچه ارومیه و آنالیز ریسک مناطق ساحلی. پایان نامه کارشناسی ارشد سازه­های آبی، دانشکده کشاورزی، دانشگاه تربیت مدرس. 140 ص.
3-    دلاور، م.، ا. مرید و م. شفیعی­فر. 1387. شبیه­سازی، تحلیل حساسیت و عدم قطعیت تراز آب دریاچه ارومیه نسبت به مولفه­های بیلان آبی آن. مجله هیدرولیک، جلد 3، شماره 1.
4-    رضایی، ع.، ب. بختیاری، ف. هوشیاری پور و م. دهقانی امیری. 1386. ارزیابی روشهای مختلف برآورد تبخیر تعرق گیاه مرجع با استفاده از سنجش های لایسیمتری (مطالعه موردی: شهر کرمان). نهمین سمینار سراسری آبیاری و کاهش تبخیر، کرمان.
5-    ریاحی، ح. و س.ع. ایوب­زاده. 1387. تخمین ضریب پراکندگی طولی آلودگی با استفاده از سیستم استنتاج  فازی- عصبی تطابقی. مجله علمی و پژوهشی آب و فاضلاب، شماره 67، صفحات 46-34.
6-    Allen, R. G.; L. S. Preira; D. Raes; M. Smith. 1998. Crop evapotranspiration guidelines for computing crop water requirement. FAO Irrigation and Drainage Paper, NO.56, Rome, Italy.
7-    Asefa, T.; MW. Kemblowski; M. Mckee; A. Khalil. 2006. Multi-time scale stream flow prediction: The support vector machine approach. Hydrology, 318:7-16.
8-    Asefa, T.; MW. Kemblowski; G. Urroz; M. Mckee; A. Khalil. 2004. Support vector-based groundwater head observation networks design. Water Resource Research, 40, W11509, doi:10.1029/2004WR003304.
9-    Behzad, M.; K. Asghari; M. Eazi; M. Pallhang, 2009. Generalization performance of support vector machines and neural networks in runoff modeling. Expert Systems with Applications, 36: 7624-7629.
10- Eckhardt, K., L. Breuer; H-G. Frede. 2003. Parameter uncertainty and the significance of 12 simulated land use change effects. Journal of Hydrology, 273:164–176.
11- Eslamian, S.S.; J. Abedi-Koupai; MJ. Amiri; SA. Gohari. 2009. Estimation of daily reference evapotranspiration using support vector machines and artifical neural networks in greenhouse. Environmental Sciences, 4: 439-447.
12- Jones, A. J. 2004. New tools in non-linear modeling and prediction. Computational Management Science, DOI: 10.1007/s10287-003-0006-1, 109-149.
13- Khemchandani, R.; Jayadeva; S. Chandra. 2009. Regularized least squares fuzzy support vector machine time series forecasting. Expert System with Application, 36: 132-138.
14- Liong ,SY.; C. Sivapragasam. 2002. Flood stage forecasting with support vector machines. American Water Resource Association, 38:173-186.
15- Marce, R.; M. Comerma; J.C. Garcia; J. Armengol. 2004. A neuro-fuzzy modeling tool to estimate fluvial nutrient loads in watersheds under time-varying human impact. Limnology and Oceanography, 2:342-355.
16- Moghadamnia, A.; M. Ghafari Gousheh; J. Piri; S. Amin; D. Han. 2009. Evaporation estimation using artifical neural network and adaptive neuro-fuzzy inference system technique. Advance in Water Resource, 32: 88-97.
17- Moghadamnia, A.; M. Ghafari; J. Piri; D. Han,. 2008. Evaporation estimation using support vector machines technique. Engineering and Technology, 33:14-22.
18- Pai, PF.; WC. Hong. 2007. A recurrent support vector regression model in rainfall forecasting. Hydrological Process, 21:819-827.
19- Piri, J.; S. Amin; A. Moghadamnia; A. Keshavarz; D. Han; R. Remesan. 2009. Daily pan evaporation modeling in a hot and dry climate. Hydrologic Engineering, 14(8): 803-811.
20- Remesan, R.; M.A. Shamim; D. Han. 2008. Model data selection using gamma test for daily solar radiation estimation. Hydrological Processes, 22:4301-4309.
21- Riahi-Madvar, H.; S. A. Ayyoubzadeh. 2010. Uncertainty analysis of ANN and ANFIS technique for predicting bed load transport. Hydraulic modeling and uncertainty, 2-4 June, Sophia Antipolis.
22- Riahi-Madvar, H.; S. A. Ayyoubzadeh; R. Noori. 2010. Uncertainty analysis of ANN and ANFIS technique in comparison with regime equations for determination of regime channel geometry. Tenth Symposium on Stochastic Hydraulics, Fifth International Conference on Water Resources and Environment Research, 5-7 July,  Quebec City, Canada.
23- Suykens, J.A.K.; J. Vandewalle. 1999. Least Squares Support Vector Machine Classifiers. Kluwer Academic Publishers. Printed in the Netherlands.
24- Valyon, J.; G. Horvath. 2005. A robust LS-SVM regression. World Academy of Science, Engineering and Technology, 7:148-153
25- Vapnik, V.N.1998. Statistical Learning Theory. Wiley, New York.
26- Yu, P.S.; S.T. Chen; I.F. Chang. 2006. Support vector regression for real-time flood stage forecasting. Hydrology, 328: 704-716.
Irrigation and Water Engineering
Volume 4, Issue 1 - Serial Number 13
December 2013
Pages 67-79

Files

Share

How to cite

Statistics