Document Type : Original Article
Authors
1 Associate Professor, Department of Water Science and Engineering, University of Birjand, Birjand - Iran
2 PhD student in Irrigation and Drainage - Department of Water Science and Engineering - University of Birjand - Birjand - Iran
3 Master's degree in Hydraulic Structures - Zabol University - Birjand - Iran
Abstract
In recent years, due to the problem of water scarcity, the issue of optimal use and management of these resources has become particularly important. In order to be aware of the status of these resources and their optimal management, it is necessary to accurately predict groundwater level fluctuations. Today, various models have been presented for predicting groundwater level fluctuations, which can help in the sustainable use of groundwater to meet urban, agricultural, and industrial needs. Therefore, special attention has been paid to intelligent models, which include time series models, wavelet analysis, artificial neural networks, support vector machine models, etc. In this research, a Wavelet-Neuro-Fuzzy hybrid model was used to predict groundwater levels in the Birjand plain. Finally, the results obtained were compared with the results of the wavelet model and the neuro-fuzzy network. The data used in this research includes rainfall, evaporation, maximum temperature, average temperature, minimum humidity, and groundwater level elevation for 16 piezometers over a statistical period of 18 years, which were measured monthly. The results showed that the Wavelet-Neuro-Fuzzy hybrid model, with an Root Mean Square Error of RMSE= 0.19 and a Nash-Sutcliffe efficiency coefficient of NS= 0.95, has higher accuracy in predicting groundwater levels compared to other models. The Wavelet-Neuro-Fuzzy model, due to the combination and integration of the useful features of wavelet transform, neural networks, and fuzzy systems, not only increases the accuracy of prediction but can also provide a more comprehensive view of the data.
Keywords
Main Subjects
)