ارزيابي مدل‌هاي استوکستيکي در پيش‌بيني سطح آب زيرزميني دشت همدان ـ بهار
محورهای موضوعی : مدیریت بهینه منابع آب و خاکحامد نوذری 1 * , نادیا صدق نژاد 2 , سجاد پویان فر 3
1 - دانشيار، گروه علوم و مهندسي آب، دانشکده کشاورزي، دانشگاه بوعلي سينا، همدان.
2 - دانشجوي دکتري، گروه علوم و مهندسي آب، دانشکده کشاورزي، دانشگاه بوعلي سينا، همدان.
3 - کارشناسي ارشد، گروه علوم و مهندسي آب، دانشکده کشاورزي، دانشگاه بوعلي سينا، همدان.
کلید واژه: هوش مصنوعي, مدل ماشين بردار پشتيبان (SVM), الگوريتم تبريد شبيه‌سازي (SA), ARIMA, SARIMA,
چکیده مقاله :
زمينه و هدف: ذخاير آب زيرزميني از مهم‌ترين منابع قابل دسترس آب شيرين در جهان محسوب مي‌شوند. امروزه با توجه به تغييرات آب و هوا، تغيير اقليم، افزايش جمعيت و برداشت‌هاي بي‌رويه از آب‌هاي زيرزميني، اين منابع با کاهش قابل توجهي مواجه شده است. با توجه به اينکه ايران در منطقه خشک و نيمه‌خشک واقع شده است، در بسياري از مناطق سطح آب زيرزميني دستخوش تغييرات بسياري قرار گرفته است. سطح آب زيرزميني نيز در منطقه دشت همدان-بهار با کاهش قابل توجهي مواجه شده است. از اين رو پيش‌بيني مقادير سطح آب زيرزميني در دشت همدان-بهار و مديريت عوامل موثر در کاهش آن از اهداف اساسي اين پژوهش مي‌باشد.
روش پژوهش: در گام نخست، در اين پژوهش سعي شد به کمک مدل تلفيقي ماشين بردار پشتيبان با الگوريتم تبريد شبيه‌سازي (SVM-SA) سطح آب زيرزميني با استفاده از مقادير بارش 4 ايستگاه سينوپتيک آق‌کهريز، سد اکباتان، کوشک آباد و مريانج برآورد گرديد و عدم قطعيت مدل SVM-SA نيز مورد تجزيه و تحليل قرار گرفت. در گام بعد، مقادير بارش 4 ايستگاه سينوپتيک مذکور به مدت 5 سال بصورت ماهانه و سالانه به کمک مدل‌هاي ميانگين متحرک يکپارچه خودهمبسته فصلي (SARIMA) و ميانگين متحرک يکپارچه خودهمبسته (ARIMA) مورد پيش‌بيني قرار گرفت و در نهايت با استفاده از مقادير بارش پيش‌بيني شده، سطح آب زيرزميني با استفاده از مدل SVM-SA به مدت 5 سال به‌صورت ماهانه و سالانه مورد پيش‌بيني قرار گرفت.
يافته‌ها: مقادير برآورد سطح آب زيرميني به کمک مدل SVM-SA با استفاده از شاخص‌هاي ضريب تبيين (R2)، جذر ميانگين مربعات خطا (RMSE) و ضريب نش ساتکليف (NSE) مورد بررسي قرار گرفت. نتايج نشان داد که عملکرد مدل در پيش‌بيني سطح آب زيرزميني در دوره‌هاي سالانه و ماهانه مناسب بوده و اختلاف معني داري در نتايج پيش‌بيني ماهانه و سالانه وجود ندارد. اما مدل SVM-SA با ضريب نش ساتکليف 0.993، جذر ميانگين خطاي استاندارد 0.417 و ضريب تبيين 0.993 در دوره واسنجي در برآورد ماهانه سطح آب زيرزميني کمي دقت بالاتري داشته است. در گام بعد جهت دستيابي به بهترين مرتبه مدل SARIMA و ARIMA جهت پيش‌بيني مقادير بارش ماهانه و سالانه از شاخص‌هاي آماري ضريب تبيين (R2)، جذر ميانگين مربعات خطا (RMSE)، ميانگين خطاي استاندارد (SE) و سنجش نيکويي برازش (AIC) استفاده شد. در نهايت با استفاده از مرتبه‌هاي مدل‌هاي منتخب با توجه به شاخص‌هاي ارزيابي براي دوره‌هاي ماهانه و سالانه براي استگاه‌هاي آق‌کهريز به‌ترتيب SARIMA(3,0,1)*(1,0,1) و ARIMA(3,0,2)، ايستگاه سد اکباتان به‌ترتيب SARIMA(1,0,1)*(1,1,2) و ARIMA(3,1,3)، براي ايستگاه کوشک آباد به‌ترتيب SARIMA(1,1,3)*(1,1,1) و ARIMA(2,0,3) و براي ايستگاه مريانج به‌ترتيب SARIMA(1,0,1)*(1,1,2) و ARIMA(3,0,2)، مقادير بارش براي 5 سال بصورت ماهانه و سالانه مورد پيش‌بيني قرار گرفت. در نهايت با استفاده از مقادير بارش پيش‌بيني شده به کمک مدل‌هاي SARIMA و ARIMA، سطح اب زيرزميني با استفاده از مدل SVM-SA بصورت ماهانه و سالانه براي 5 سال آينده مورد پيش‌بيني قرار گرفت.
نتيجه‌گيري: از نتايج مهم اين مطالعه مي‌توان به اين نکته اشاره کرد که رابطه معني‌داري بين کاهش بارش و افت شديد آب‌هاي زيرزميني در دشت همدان-بهار وجود ندارد. در واقع نتايج اين پژوهش حاکي از آن است که افت شديد سطح آب زيرزميني در سال‌هاي اخير ناشي از برداشت بي‌رويه از اين منابع ارزشمند مي‌باشد.
Background and Aim: Groundwater sources are considered to be one of the most important available sources of fresh water in the world. Today, due to the changes in weather, climate change, population increase and excessive withdrawals of underground water, these resources have faced a significant decrease. Considering that Iran is located in a dry and semi-arid region, the underground water level has undergone many changes in many areas. The level of underground water in Hamadan-Bahar plain area has also faced a significant decrease. Therefore, the prediction of underground water levels in Hamadan-Bahar plain and the management of effective factors in its reduction are the main goals of this research.
Method: In the first step, in this research, it was tried to predict the underground water level with the help of support vector machine integrated model with Simulated annealing algorithm (SVM-SA) using the rainfall values of 4 synoptic stations of Aghkahriz, Ekbatan Dam, Kooshkabad and Marianaj. The uncertainties of this model are also analyzed. In the next step, the precipitation values of the mentioned 4 synoptic stations were predicted for 5 years monthly and annually with the help of seasonal autocorrelated moving average (SARIMA) and autocorrelated moving average (ARIMA) models, and finally, using the predicted rainfall values, the underground water level was predicted monthly and annually using the SVM-SA model for 5 years.
Result: The estimated values of underground water level were analyzed with the help of SVM-SA model using the indices of explanation coefficient (R2), root mean square error (RMSE) and Nash Sutcliffe coefficient (NSE). The results indicate that there is no significant difference between the performance of the model in predicting the underground water level in annual and monthly periods. But the SVM-SA model with Nash Sutcliffe coefficient of 0.993, root mean standard error of 0.417 and explanatory coefficient of 0.993 in the calibration period has been more accurate in monthly estimation of underground water level. In the next step, in order to achieve the best SARIMA and ARIMA models for predicting monthly and annual rainfall values, statistical indicators of coefficient of explanation (R2), root mean square error (RMSE), mean standard error (SE) and goodness of fit (AIC) are used. Finally, by using the ranks of the selected models according to the evaluation indices for monthly and annual periods for Aghkahriz station, respectively SARIMA(3,0,1)*(1,0,1) and ARIMA(3,0,2), for Ekbatan dam station according to SARIMA(1,0,1)*(1,1,2) and ARIMA(3,1,3), for Kooshkabad station according to SARIMA(1,1,3)*(1,1,1) and ARIMA(2,0,3) and for Marianaj station by SARIMA(1,0,1)*(1,1,2) and ARIMA(3,0,2) respectively, rainfall values for 5 years in monthly and annually forecast it placed. Finally, using the forecasted rainfall values with the help of SARIMA and ARIMA models, the groundwater level was forecasted monthly and annually for the next 5 years using the SVM-SA model.
Conclusion: One of the important results of this study is the absence of a significant relationship between the decrease in rainfall and the sharp drop in groundwater in the Hamedan-Bahar plain. In fact, the results of this research indicate that the sharp drop in the underground water level is caused by the excessive extraction of these valuable resources.
Aderemi, B. A., Olwal, T. O., Ndambuki, J. M., and Rwanga, S. S. (2023). Groundwater levels forecasting using machine learning models: A case study of the groundwater region 10 at Karst Belt, South Africa. Systems and Soft Computing, 5, 200049.
Arya Azar, N., Kardan, N., and Ghordoyee Milan, S. (2023). Developing the artificial neural network–evolutionary algorithms hybrid models (ANN–EA) to predict the daily evaporation from dam reservoirs. Engineering with Computers, 39(2), 1375-1393.
Behzad, M., Asghari, K., and Coppola Jr, E. A. (2010). Comparative study of SVMs and ANNs in aquifer water level prediction. Journal of Computing in Civil Engineering, 24(5), 408-413.
Box, GEP., Cox, DR. 1964. An analysis of transformations. Journal of the royal statistical society series b-methodological, 26(2), 211–252.
Brunner, P., and Simmons, C. T. (2012). HydroGeoSphere: a fully integrated, physically based hydrological model. Ground water, 50(2), 170-176.
Cercignani, C., and Cercignani, C. (1988). The boltzmann equation (pp. 40-103). Springer New York.
Fathi, A., Lee, T., and Mohebzadeh, H. (2019). Allocating underground dam sites using remote sensing and GIS case study on the southwestern plain of Tehran Province, Iran. Journal of the Indian Society of Remote Sensing, 47, 989-1002.
Granata, F., Papirio, S., Esposito, G., Gargano, R., and De Marinis, G. (2017). Machine learning algorithms for the forecasting of wastewater quality indicators. Water, 9(2), 105.
Javadi, S., Saatsaz, M., Shahdany, S. M. H., Neshat, A., Milan, S. G., and Akbari, S. (2021). A new hybrid framework of site selection for groundwater recharge. Geoscience Frontiers, 12(4), 101144.
Kardan Moghaddam, H., Ghordoyee Milan, S., Kayhomayoon, Z., Rahimzadeh kivi, Z., and Arya Azar, N. (2021). The prediction of aquifer groundwater level based on spatial clustering approach using machine learning. Environmental Monitoring and Assessment, 193, 1-20.
Kayhomayoon, Z., Azar, N. A., Milan, S. G., Moghaddam, H. K., and Berndtsson, R. (2021). Novel approach for predicting groundwater storage loss using machine learning. Journal of Environmental Management, 296, 113237.
Kayhomayoon, Z., Ghordoyee Milan, S., Arya Azar, N., and Kardan Moghaddam, H. (2021). A new approach for regional groundwater level simulation: clustering, simulation, and optimization. Natural Resources Research, 30, 4165-4185.
Khorasani, M., Ehteshami, M., Ghadimi, H., & Salari, M. (2016). Simulation and analysis of temporal changes of groundwater depth using time series modeling. Modeling Earth Systems and Environment, 2, 1-10.
Liu, D., Mishra, A. K., Yu, Z., Lü, H., and Li, Y. (2021). Support vector machine and data assimilation framework for Groundwater Level Forecasting using GRACE satellite data. Journal of Hydrology, 603, 126929.
Marashi, A., Kouchakzadeh, S., & Yonesi, H. A. (2023). Rotary gate discharge determination for inclusive data from free to submerged flow conditions using ENN, ENN–GA, and SVM–SA. Journal of Hydroinformatics, 25(4), 1312-1328.
Milan, S. G., Roozbahani, A., and Banihabib, M. E. (2018). Fuzzy optimization model and fuzzy inference system for conjunctive use of surface and groundwater resources. Journal of hydrology, 566, 421-434.
Mirarabi, A., Nassery, H. R., Nakhaei, M., Adamowski, J., Akbarzadeh, A. H., and Alijani, F. (2019). Evaluation of data-driven models (SVR and ANN) for groundwater-level prediction in confined and unconfined systems. Environmental Earth Sciences, 78, 1-15.
Mohammadi, G. M., Ebrahimi, K., & Araghinejad, S. (2012). Evaluation impact of drought, extraction and construction of dam on the groundwater drop-case study Saveh aquifer.
Mohanasundaram, S., Suresh Kumar, G., & Narasimhan, B. (2019). A novel deseasonalized time series model with an improved seasonal estimate for groundwater level predictions. H2Open Journal, 2(1), 25-44.
Nadiri, A. A., Naderi, K., Khatibi, R., and Gharekhani, M. (2019). Modelling groundwater level variations by learning from multiple models using fuzzy logic. Hydrological sciences journal, 64(2), 210-226.
Nhu, V. H., Shirzadi, A., Shahabi, H., Singh, S. K., Al-Ansari, N., Clague, J. J., ... and Ahmad, B. B. (2020). Shallow landslide susceptibility mapping: A comparison between logistic model tree, logistic regression, naïve bayes tree, artificial neural network, and support vector machine algorithms. International journal of environmental research and public health, 17(8), 2749.
Nozari, H., & Zali, A. (2014). Investigating Groundwater Extraction from the Hamedan-Bahar Plain, s Aquifer. Water and Soil Science, 23(4), 277-289. (In Persian)
Nozari, H., and Tavakoli, F. (2020). Forecasting hydrologic parameters using linear and nonlinear stochastic models. Journal of Water and Climate Change, 11(4), 1284-1301.
Nozari, H., Azadi, S., Sedghnejad, N., & Pouyanfar, S. (2023). Predicting monthly evaporation using linear and nonlinear time series models Case study: meteorological station of Ekbatan Dam. Journal of Agricultural Meteorology, (In Persian).
Pham, B. T., Jaafari, A., Prakash, I., and Bui, D. T. (2019). A novel hybrid intelligent model of support vector machines and the MultiBoost ensemble for landslide susceptibility modeling. Bulletin of Engineering Geology and the Environment, 78, 2865-2886.
Poormohammadi, S., Dastorani, M. T., Cheraghi, S. A. M., Mokhtari, M. H., & Rahimian, M. H. (2011). Evaluation and Estimation of Water Balance Components in Arid Zone Catchments Using RS and GIS Case Study: Manshad Catchment, Yazd Province. Journal of Water and Wastewater; Ab va Fazilab (in persian), 22(3), 99-108.
Pouyanfar, S., Nozari, H., and Khodamorad Pour, M. (2023). Comparison of the performances of the gene expression programming model and the RegCM model in predicting monthly runoff. Journal of Water and Climate Change, 14(10), 3810-3829.
Radhika, Y., and Shashi, M. (2009). Atmospheric temperature prediction using support vector machines. International journal of computer theory and engineering, 1(1), 55.
Sarma, R., & Singh, S. K. (2022). A comparative study of data-driven models for groundwater level forecasting. Water Resources Management, 36(8), 2741-2756.
Satish Kumar, K., & Venkata Rathnam, E. (2019). Analysis and prediction of groundwater level trends using four variations of Mann Kendall tests and ARIMA modelling. Journal of the Geological Society of India, 94, 281-289.
Sedghnejad, N., Nozari, H., & Marofi, S. (2024). Comparative analysis of classification techniques and input-output patterns for monthly rainfall prediction. Water Science, 38(1), 192-208.
Takafuji, E. H. D. M., Rocha, M. M. D., & Manzione, R. L. (2019). Groundwater level prediction/forecasting and assessment of uncertainty using SGS and ARIMA models: a case study in the Bauru Aquifer System (Brazil). Natural Resources Research, 28(2), 487-503.
Tapak, L., Rahmani, A. R., and Moghimbeigi, A. (2013). Prediction the groundwater level of Hamadan-Bahar plain, west of Iran using support vector machines. Journal of research in health sciences, 14(1), 82-87.
Tran, N. H., and Tran, K. (2007). Combination of fuzzy ranking and simulated annealing to improve discrete fracture inversion. Mathematical and Computer Modelling, 45(7-8), 1010-1020.
Wang, W. C., Xu, D. M., Chau, K. W., Chen, S. (2013). Improved annual rainfall-runoff forecasting using PSO–SVM model based on EEMD. J. Hydroinformatics, 15(4), 1377-1390.
Xu, Z., Huang, X., Lin, L., Wang, Q., Liu, J., Yu, K., and Chen, C. (2020). BP neural networks and random forest models to detect damage by Dendrolimus punctatus Walker. Journal of forestry research, 31, 107-121.
Yuan, Y. (2013). Forecasting the movement direction of exchange rate with polynomial smooth support vector machine. Mathematical and Computer Modelling, 57(3-4), 932-944.
Zhang, X., Chen, X., and He, Z. (2010). An ACO-based algorithm for parameter optimization of support vector machines. Expert systems with applications, 37(9), 6618-6628.