Modelling and prediction of groundwater level using wavelet transform and machine learning methods: A case study for the Sahneh Plain, Iran

Due to the complexity of numerical models and the need for much information and data in these models, one of the important solutions is to use artificial intelligence models with a small number of required inputs and a simpler structure. In this study, using wavelet transform analysis (cross wavelet transform and wavelet transform coherence), a model was developed to extract the relationship between climatic data (i.e., temperature and precipitation) and groundwater level changes. The developed model provides the best relationship between the main effective climatic parameters and groundwater level changes in the Sahneh Plain. The results showed that the climatic variables of average temperature and precipitation are the parameters with the highest correlation coefficient with groundwater level changes in the main wells of the study area. The results of verification and validation of the numerical prediction model of groundwater level based on the climatic parameters indicate that the average error is between 3% and 10% in different areas of the plain, which indicates the high accuracy of the developed model. In areas with a lack of data and information required for numerical simulation, the proposed model used in this research can be used as a comprehensive model for assessing and predicting groundwater levels based on climatic parameters.