Groundwater Level Prediction and Correlative Study with Groundwater Contamination Under Conditional Scenarios: Insights from Multivariate Deep LSTM Neural Network Modeling

摘要

Groundwater is the primary source for drinking water and irrigation in India, and from last few years, due to population burst across the nation, there is a sharp decline in the groundwater level (availability). There is a constant pressure balance among the groundwater and seawater level so that contaminated water cannot seep into and due to lowering level there is an alarming situation for water contamination across India. In this paper, we aim to find the liaison between groundwater level and ground contamination condition through LSTM predictive modeling. The proposed algorithm for groundwater prediction is based on conditional approach through deep LSTM modeling and the ground contamination is calculated using an aggregated scoring approach modeled using Euclidian distance concept. Lastly, a correlative study is being provided to analyze the liaison in between the said variables. There is a high negative correlation among the said variables indicating loss of groundwater level is increasing the contamination level across the taken zones. The experiment has been carried out on the data across the three eastern Indian states, viz. West Bengal, Odhisa, and Bihar for a time span from 2004 to 2017.