Paper Title
Machine Learning Based Prediction of Water Level and Volume in Farm Wells for Sustainability
Abstract
Water resources are crucial for agriculture, and wells located near agricultural fields are often used as water
sources. However, there is significant temporal variability in water use patterns, which can make it difficult to predict future
water availability. This study focuses on the analysis of water level and volume in wells located near agricultural fields,
using water level and volume data collected over a one-year period. Daily water level and volume patterns were well
described by autoregressive integrated moving average (ARIMA) models. Model development was supported by
unsupervised clustering analysis that revealed similarities in water use patterns and confirmed the time-series water use
model attributes. The inclusion of ambient temperature as a model attribute improved ARIMA model performance for daily
water level and volume in a well. The study aims to provide insights into the behavior of the water system and enable better
decision-making in agricultural operations that depend on this water source. The ARIMA model can help predict future
water availability, which can be critical for planning and managing agricultural activities such as irrigation and crop
selection. The study can contribute to more efficient and sustainable use of water resources in agriculture.