کاربرد روش‌های شبکه‌ی بیزین و حداقل مربعات ماشین بردار پشتیبان در پیش بینی تراز سطح آب دریاچه ارومیه

نوع مقاله: مقاله پژوهشی

نویسندگان

1 دانشگاه ارومیه

2 دانشگاه ارومیه- گروه مهندسی آب

3 دانشیار گروه مهندسی آب، دانشگاه ارومیه

چکیده

سابقه و هدف: دریاچه ارومیه به عنوان یک اکوسیستم آبی مهم در شمال غرب ایران واقع شده است. در 14 سال اخیر میانگین تراز سطح آب دریاچه ارومیه به 2/1272 متر تقلیل پیدا کرده و این به این معنی است که اختلاف تراز سطح اکولوژیک دریاچه و تراز سطح کنونی 2 متر است. خشک شدن دریاچه ارومیه باعث بروز مسائل و بحران‌های جدی برای حوضه، استان‌های مجاور و کشور خواهد شد. در این تحقیق از پارامترهای موثر مستقیم و غیر مستقیم در پیش-بینی تراز سطح دریاچه از جمله تبخیر، رواناب ورودی به دریاچه، بارش، دما، باد، میانگین رطوبت هوا و تراز سطح آب دریاچه در ماه قبل به عنوان ورودی‌های مدل استفاده شده است. مقایسه کارایی مدل‌های شبکه‌ی بیزین که یک مدل احتمالاتی تحت شرایط عدم قطعیت و الگوریتم ماشینی حداقل مربعات ماشین بردار پشتیبان، هدف اصلی تحقیق حاضر است.
مواد و روش‌ها: در تحقیق حاضر از دو روش حداقل مربعات ماشین بردار پشتیبان و شبکه بیزین در فرایند مدل-سازی استفاده گردید. در این مطالعه عوامل موثر در پیش بینی تراز سطح دریاچه در ماه قبل به عنوان ورودی و تراز سطح آب دریاچه در ماه کنونی به عنوان خروجی مدل‌ها مورد بررسی قرار گرفت. جهت تخمین دما، تبخیر، بارش، باد و میانگین رطوبت هوا بر روی سطح دریاچه از داده‌های پنج ایستگاه سینوپتیک مجاور دریاچه با برآورد ضریب تیسن هر ایستگاه و داده‌های 13 ایستگاه هیدرومتری واقع بر رودخانه‌های منتهی به دریاچه جهت تشکیل پارامترهای ورودی دو مدل محاسبه گردید.
یافته‌ها: تحلیل داده‌های ایستگاه‌های هیدرومتری نشان داد که تنها چهار ایستگاه از 13 ایستگاه هیدرومتری از نرمال پیروی می‌کنند. مقایسه و بررسی نتایج دو مدل با بررسی ضرایب R2، RMSE، MBE و ناش ساتکلیف حاکی از برتری عملکرد مدل حداقل مربعات ماشین بردار پشتیبان در مقایسه با شبکه بیزین است. این مقادیر برای مدل برتر به ترتیب برابر با 3/92%، 082/0متر، 012/0-متر و 86/0 بدست آمد.
نتیجه‌گیری: در این تحقیق معیارهای ارزیابی نشان داد که مدل حداقل مربعات ماشین بردار پشتیبان نسبت به مدل شبکه بیزین دارای برتری است. نکته حائز اهمیت در مقایسه دو مدل آن است که ماهیت مدل حداقل مربعات ماشین بردار پشتیبان، ماشینی است ولی ماهیت شبکه بیزین یک مدل احتمالاتی تحت شرایط عدم قطعیت است که در آن از توزیع نرمال جهت آموزش متغیرهای شبکه استفاده شده است. از آنجایی که ماهیت وقوع رخدادهای طبیعی تصادفی است، بنابراین استفاده از مدل شبکه بیزین نسبت به مدل حداقل مربعات ماشین بردار پشتیبان می‌تواند توصیه گردد.

کلیدواژه‌ها


عنوان مقاله [English]

Application of Bayesian Network and LS-SVM Methods in Predicting Water Surface Level of Urmia Lake

نویسندگان [English]

  • Sajjad Karimzadgan 1
  • Javad behmanesh 2
  • Hossein Rezaie 3
1 Urmia University
3 Urmia University
چکیده [English]

Background and Objectives: Urmia lake, as an important water ecosystem, is located in the northwest of Iran. Over the past 14 years, the average of Urmia Lake water surface level has decreased to 1272.2 m and it means that the difference between the ecological level of the lake and the present water level is 2-meter. Drying the Urmia Lake will cause serious problems and crises for the watershed, adjacent provinces and the country. In this research, direct and indirect effective parameters in predicting of the lake level such as evaporation, input runoff to the lake, precipitation, temperature, wind, mean air humidity and water level in the previous month have been employed. The comparison of the efficiency of two models namely the Bayesian network, which is a probabilistic model under conditions of uncertainty and the machine algorithm of Least Square Support Vector Machine (LS-SVM) is the main objective of the present research.

Materials and Methods: In the present research, two methods including least squares support vector machine and Bayesian were employed in the modelling process. In this study, effective factors in predicting Urmia lake water level in the previous month as inputs for the used models and water level in the present month as outputs were investigated.
The inputs of the models were including
The surface of lake level such as temperature, evaporation, wind, average humidity, precipitation on the lake surface, total input runoff to the lake and the surface of lake water level in the previous month as input of models and surface of level Lake water was studied as the output of the current month. To estimate the temperature, evaporation, wind rainfall and average air humidity on surface lake, five synoptic stations in the adjacent lake data are estimated by evaluating the Thiessen coefficient of each station and the data of 13 hydrometric stations located on the rivers were calculated to lead the lake as input parameters of two model.
Result: The analysis of hydrometric stations data showed that the data of only 4 stations among 13 stations are fitted to the normal distribution. Comparison and investigation of the results of two model by examining the coefficients R2, RMSE, MBE and Nash-Sutcliff indicate superiority of the least squared vector machine model compared to the Bayesian network. These values for the superior model were 92.3%, 0.082, 0.122 and 0.86, respectively.
Conclusion: In this research, the criteria for evaluating and comparing the accuracy of the two model in prediction based on R2, RMSE, MBE and Nash-Sutcliff criteria indicated that the least square vector machine is superior to the Bayesian model. But the important point in comparing the two model is that the nature of the least squares support vector machine is a machine, however, the nature of the Bayesian network is a probabilistic model under uncertainty conditions in which the normal distribution is used to train network variables because the nature of events is random. Therefore, in this research, the use of Bayesian network model is recommended to the least-squares support vector machine model.

کلیدواژه‌ها [English]

  • Bayesian network
  • Ecological level
  • Least Square Support Vector Machine
  • prediction
  • Urmia Lake
1.Abbaszadeh, A.A., Khalili, K., and Behmanesh, J. 2015. Application of Combined AR-ARCH
model in Forecasting Urmia Lake Water Level. J. Soil Water Know. 25: 4/2. 175-186.
(In Persian)
2.Ahmadi, F., and Radmanesh, F. 2016. Application of Bayesian Networks and Genetic
Programming for Predicting Daily River Flow (Case study: Barandoozchay River). J. Irrig.
Sci. Engin. 39: 4. 213-223. (In Persian)
3.Ahmadi, F., Radmanesh, F., and Mir Abbasi, N.R. 2016. Comparing the performance of
Support Vector Machines and Bayesian networks in predicting daily river flow (Case study:
Baranduz Chai River). J. Water Soil Cons. 22: 6. 171-186. (In Persian)
4.Alexandersson, H. 1986. A homogeneity test applied to precipitation data. Inter. J. Climatol.
6: 6. 661-675.
5.Alizadeh, A. 2010. Principles of Applied Hydrology. Imam Reza University Press, 911p.
(In Persian)
6.Anbari, M.J., and Tabesh, M. 2016. Failure Event Probability Calculation in Wastewater
Collection Systems Using the Bayesian Network. J. Water Wastewater. 3: 48-61. (In Persian)
7.Bishop, C.M. 2006. Pattern recognition and machine learning. Springer.
8.Download.hugin.com. 2017. Introduction to the Hugin Development Environment / Manual.
[online] Available at: http://download.hugin.com/webdocs/manuals/Htmlhelp/descr_NPC_
algorithm_pane.html [Accessed 31 Aug. 2017].
9.Gelman, A., Carlin, J.B., Stern, H.S., Dunson, D.B., Vehtari, A., and Rubin, D.B. 2014.
Bayesian data analysis (Vol. 2). Boca Raton, FL: CRC press.
10.Ghgajarnia, N., Liyaghat, A., and Danesh Kar Araste, P. 2014. Verifying precipitation data
of TAMAB and meteorology institute in Urmia basin. Water and Soil resources conservation
4: 1. 91-109. (In Persian)
11.Ghorbani, A.M., and Dehghani, R. 2014. Application of Bayesian Neural Networks, Support
Vector Machines and Gene Expression Programming Analysis of Rainfall - Runoff Monthly
(Case study: Kakareza River). J. Irrig. Sci. Engin. 39: 2. 125-138. (In Persian)
12.Hesar, A.S., Tabatabaee, H., and Jalali, M. 2012. Structure learning of Bayesian networks
using heuristic methods. In Proc. of International Conference on Information and Knowledge
Management (ICIKM 2012).
13.Kardan Moghadam, H., and Roozbahani, A. 2015. Evaluation of Bayesian networks model
in monthly groundwater level prediction (Case study: Birjand aquifer). J. Water Irrig.
Manage. 5: 2. 139-151. (In Persian)
14.Kelts, K., and Shahrabi, M. 1986. Holocene sedimentology of hypersaline Lake Urmia,
northwestern Iran. Paleogeography, Paleoclimatology, Paleoecology, 54 (1-4): 105-130.
15.Koop, G. 2010. Bayesian econometrics. Chichester [u.a.]: Wiley.
16.Lin, G.F., Chen, G.R., Wu, M.C., and Chou, Y.C. 2009. Effective forecasting of hourly
typhoon rainfall using support vector machines. Water Resources Research, 45: 8.
17.Madadgar, S., and Moradkhani, H. 2014. Spatio-temporal drought forecasting within
Bayesian networks. J. Hydrol. 512: 134-146.
18.Mahmoudi, Kh.S., and Rezaie, H. 2013. Investigating and evaluating causes of fluctuations
in Lake Urmia water level and providing various solutions. Urmia University, Faculty of
Agriculture. 233p. (In Persian)
19.Mahsafar, H., Maknoun, R., and Saghafian, B. 2011. The Impact of Climate Change on
Urmia Lake Water Level. J. Iran Water Resour. Res. 7: 1. 47-58. (In Persian)
20.Margaritis, D. 2003. Learning Bayesian Network Model Structure. Ph.D. Thesis. School of
Computer Science, Carnegie Mellon University.
21.Mehdizadeh, S., Behmanesh, J., and Saadatnejad Gharahassanlou, H. 2017. Evaluation of
gene expression programming and Bayesian networks methods in predicting daily air
temperature. J. Agric. Meteorol. 4: 2. 1-13. (In Persian)
22.Mellit, A., Massi Pavan, A., and Benghanem, M. 2013. Least squares support vector machine
for short term prediction of meteorological time series. Theor. Appl. Climatol. 111: 297-307.
23.Mohajerani, H., Kholghi, M., Mosaedi, A., Saidodin, A., and Meftah Halaghi, M. 2013.
Quantitative Management of Groundwater Using Bayesian Decision Network. J. Water Soil.
26: 6. 1522-1534. (In Persian)
24.Moravaj, M., Khalili, K., and Behmanesh, J. 2016. Forecasting Lake Urmia water level using
linear time series models. J. Water Soil Cons. 22: 5. 287-296. (In Persian)
25.Noori, R., Karbassi, A.R., Moghaddamnia, A., Han, D., Zokaei-Ashtiani, M.H., Farokhnia,
A., and Gousheh, M.G. 2011. Assessment of input variables determination on the SVM
model performance using PCA, Gamma test and forward selection techniques for monthly
stream flow prediction. J. Hydrol. 401 (3-4): 177-189.
26.Pang, A.P., and Sun, T. 2014. Bayesian networks for environmental flow decision-making
and an application in the Yellow River estuary, China. Hydrology and Earth System
Sciences, 18 (5): 1641.
27.Parkes, B., and Demeritt, D. 2016. Defining the hundred-year flood: A Bayesian
approach for using historic data to reduce uncertainty in flood frequency estimates. J.
Hydrol. 540, 1189-1208.
28.Seifi, A. 2011. Develop an expert system to predict daily reference evapotranspiration using
a backup vector machine (SVM) and compare its results with ANFIS, ANN and empirical
methods. Tarbiat Modarres University, Faculty of Agricalture. 153p. (In Persian)
29.Subramanya, K. 2013. Engineering Hydrology, 4e. Tata McGraw-Hill Education.
30.Vapnik, V. 1998. Statistical Learning Theory. John Wiley&Sons. Inc., New York.
31.Xue, J., Gui, D., Zhao, Y., Lei, J., Zeng, F., Feng, X., and Shareef, M. 2016. A decisionmaking framework to model environmental flow requirements in oasis areas using Bayesian
networks. J. Hydrol. 540: 1209-1222.