Temporal and spatial modeling of underground water level using Kriging models and Artificial Neural Networks (case study: Minab Plain)
Subject Areas : Analysis, design and construction of water structuresVahid Sohrabi 1 , Mohammad Ibrahim Afifi 2
1 - Department of Geography, Faculty of Humanities, Islamic Azad University, Larestan Branch, Fars, Iran.
2 - Department of Geography, Faculty of Humanities, Islamic Azad University, Larestan Branch, Fars, Iran.
Keywords: Temporal modeling, Water level, Kriging models, Neural Networks, Minab,
Abstract :
Most of the world's water sources are underground water sources. Rapid population growth, agricultural development, and unresponsive surface water levels have led to an increase in water pumping, resulting in a drop in groundwater levels and depletion of aquifers. Life in arid and semi-arid regions is completely de-pendent on groundwater resources due to low rainfall, so proper management of groundwater in such regions is very critical. This research is aimed at modeling the spatial and temporal level of underground water in the Minab Plain. In order to know and evaluate the process of illegal withdrawal from the underground water table of the Minab Plain, the available information related to water resources (observation wells) and hydrogeological of Minab Plain in a period of 17 years (1376-1392) was used as the basis of the work. In addition to the geostatistical Kriging method, the Radial Basis Functions (RBF) method, which is based on a neural network, has also been used for zoning and interpolation of the underground water level of Minab Plain. In this context, the interpolation models of implementation and changes were investigated temporally and spatially in Minab Plain. The level of accuracy for each of the models was examined, and Kendall's time series method was used to examine the changes, and finally, spatio-temporal modeling was done during the years under review. The results of this research showed that the average balance had a completely decreasing trend. Among the kriging methods, the simplified kriging method with a determination coefficient of 0.89 showed the most accuracy among other models. Among the RBF methods, the fully regular spline method with a coefficient of determination of 0.67 was the most accurate compared to other models. After choosing the kriging method as the optimal method, spatial trending was done using annual level maps and it was found that the amount of level drop has a geographically uneven distribution and in some areas, the level drop was much more intense. All spatial-temporal trend zoning maps show that the decreasing trend of Minab Plain's underground water level is different from each other in terms of spatial distribution, and in different parts of the water level has changed with different intensity, which is necessary to choose the best method. Finally, different kriging and neural network models were used from the results of RMSE statistics and it was found that the kriging method has better modeled the water level changes in the Minab Plain compared to the RBF method.
Affandi, A. & Watanabe, K. (2007). Daily groundwater level fluctuation forecasting using soft computing estimation of reference evapotranspiration from pan evaporation in a semi-arid environment. Irrigation Science.
Ahmadizadeh, M., & Maroufi, P. (2017). Application of Bayesian analysis and particle filter in rainfall-runoff models and uncertainty analysis, Journal of Water and Soil Protection Research, 24(1), 251. https://doi.org/10.22069/jwfst.2017.12108.2663. (In Persian)
Aziz, O. I. A., & Burn, D. H. (2006). Trends and variability in the hydrological regime of the Mackenzie River Basin. Journal of hydrology, 319(1-4), 282-294.
Bierkens, M. (1998). Modeling water table fluctuations by means of a stochastic differential equation. Water Resources Reserch,34(10), 2485-2499.
Boroujerdi, A., & Feridouni, M. (2015). Simulation of underground water level using support vector models and its comparison with fuzzy neural models and neural wavelet (case study of Shiraz Plain), Water Engineering Conference and Exhibition. https://civilica.com/doc/407797/. (In Persian)
Connor KM, Davidson JR, Churchill LE, Sherweed A, & Foa EB. (2012). Psychometric properties of the Social Phobia Inventory (SPIN). The British Journal of Psychiatry, 176, 379-386.
Fakher, A. (2000). Application of physical models in teaching engineering courses. Iranian Engineering Educational Journal, 2(6), 39-72. https://doi.org/10.22047/ijee.2000.2014. (In Persian)
Feridouni, M., & Khojand, M. (2014). Application of geostatistical model (Kriging) and its integration with neural fuzzy inference system in underground water level zoning (case study: Shiraz plain), 2nd National Conference on Water, Man and Earth. https://cpro.ir/product/586034/. (In Persian)
Ghorbani, M.A., Khatibi, R., Hasanpour Kashani M., & Kisi, O. (2019). Comparison of three artificial intelli-gence techniques for discharge routing. Journal of Hydrology, 4. https://doi.org /10.22034/hydro.2024.60223.1310. (In Persian)
Hosseini Soumae, M., Roshani, A., & Zebah, A. (2020). Modeling of groundwater level changes based on methods based on artificial intelligence (Case study of Zaveh Torbat Heydariyeh plain). Water-shed Management, 11(21), 223-235. https://doi.org/10.52547/jwmr.11.21.223. (In Persian)
Kazemi, T., Fadeli, M., & Farzin, M. (2021) Assessment and analysis of peak flow with different return periods for Bashar river using HEC-HMS model, Iranian Water Resources Research, 17(4), 50-69. https://doi.org/20.1001.1.17352347.1400.17.4.3.0. (In Persian)
Mashaikhi, S., Ebrahimi, K., Modaresi, F., & Iraqhinejad, S. (2022). Integrated management of Minab basin water resources using modified sustainability index, Iranian Journal of Irrigation and Water Engineering, 12(47). (In Persian)
Mir Arabi, A., & Nakhai, M. (2008). Prediction of underground water level fluctuations in Birjand Plain using artificial neural network, 12th conference of Geological Society of Iran. (In Persian)
Moghadisi, M., Mardian, M., & Parsa, M. (2021). Comparison and evaluation of smart models and geostatistics in order to analyze the spatial changes of groundwater quality (Dasht Kamijan), Watershed Management Research Journal, 12(24), 54 -64. https://doi.org/10.52547/jwmr.12.24.54. (In Persian)
Mubaraki, M. & Fereidoni, M. (2015). Simulation of underground water level using the combination of wavelet, neural fuzzy models and its comparison with the neural fuzzy model, International Conference on New Research Achievements in Civil Engineering and Urban Architecture. (In Persian)
Nayak, P., Satyaji Rao, Y.R. & Sudheer, K.P. (2001). Groundwater level forecasting in a shallow aquifer using artificial neural network approach. Water Resources Management, 20,77-90.
Rajaei, T. & Zainivand, A. (2014). Modeling the underground water level using the hybrid wavelet-artificial neural network model (case study: Sharifabad plain), Reist Civil and Environmental Engineering Journal, 44.4(77), 51-63. (In Persian)
Ramezani Charmehineh, A. & Zounematkermani, M. (2017). Investigating the efficiency of support vector regression, multilayer perceptron neural network and multivariable linear regression methods to predict the level of underground water (case study: Shahrekord Plain). Watershed Management Research Journal, 8(15), 1-12 . https://doi.org/10.29252/jwmr.8.15.1. (In Persian)
Sarzabi, H. & Esmaili, K. (2013). Investigation and quantitative modeling of groundwater (Case study: Neishabur Plain), Irrigation Engineering Sciences (Agriculture Scientific Journal), 36(4). https://doi.org/20.1001.1.25885952.1392.36.4.8.5. (In Persian)
Shirzaei, S. (2019). A review of variable selection methods in regression models with time series errors, Master's Thesis, Shahrood University of Technology. (In Persian)
Yazidi, A., Davari, K., Alizadeh, A., Kahraman, B., & Haqati Moghadam, S.A. (2007). Prediction of water level using artificial neural networks (case study: Neishabur Plain). Iranian Journal of Irrigation and Drainage, 1(2), 59-71. (In Persian)