Modelling groundwater level fluctuations in urban areas using artificial neural network

Rapid increase in urban areas due to rise in population and industrialization has increased the demand for water resources. Further, the changing land use pattern due to urban development is affecting the groundwater recharge. So, forecasting groundwater has become the basic condition for efficient groundwater resource development. The technique of Artificial Neural Networks (ANN) which has ability to model the complex water resources problems is used. ANN is a mathematical model which mimics the working of nerve cells. It is capable of recognizing the relationship between the input and output without understanding the physical connection between them. In this study, an ANN model was used to predict the pre and post monsoon groundwater fluctuation at the Tikri Kalan observation well situated in the West Delhi, India. The model comprises of three layers feed forward network trained with Levenberg-Marquardt (LM) algorithm and activated with log sigmoid function. Four different architectural networks are formulated and the optimum one is identified on the performance of statistical indices. The architecture of 3-15-1 is found to be generating the best results and can be employed in urban areas for forecasting groundwater fluctuation and managing water resources effectively.

Automated summary

The rapid increase in urban areas due to population growth and industrialization has led to higher demand for water resources, making groundwater forecasting essential. The use of Artificial Neural Networks (ANN) for predicting groundwater fluctuations in urban areas has been proven effective, with the 3-15-1 architecture identified as the best performing model for managing water resources effectively.