Evaluation of Four Tree Algorithms in Predicting and Investigating the Changes in Aquifer Depth

One of the key elements in improved management and better planning for aquifer maintenance is the ability to predict changes in aquifer depth. In order to forecast changes in aquifer depth in Qazvin plain, four methods, including Classification and Regression Tree (CART), Reduced Error Pruning Trees (RepTree), M5-Pruned (M5P), and M5Rule, were used in this work. The absolute mean error (MAE) and coefficient of determination (R2) data show that the CART algorithm performs better than other algorithms at forecasting changes in aquifer depth. The CART algorithm's prediction findings showed that the aquifer's behavior in the two seasons was entirely different. In the first stage, which began in November and continued through April, there was an annual average depth of 0.045 m. The aquifer depth has been greatly influenced by rising precipitation and falling air temperature. The aquifer experiences an average decline of 0.15 m in the second portion, which runs from May to October. Aquifer depth has significantly decreased as a result of declining natural water supplies and rising agricultural water use. It is advised to utilize a crop scheme with reduced water need when rainfall reduces due to the strong effect of changes in aquifer depth from rainfall with a delay of one to three months ago.