Towards Efficient Irrigation: Machine Learning-Based Water Level Forecasting

Abstract

Efficient irrigation is a serious challenge in modern agriculture with growing population, climate change, and resource constraints. This research explores the use of machine learning models to forecast soil water content based on environmental factors acquired from IoT sensors to maximize precision irrigation. Four prediction models were examined: linear regression, random forest, XGBoost, and stacking based on a neural network (MLP) as a meta-model. Models were evaluated using R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> and RMSE metrics. The results show that ensemble techniques, particularly XGBoost and stacking, deliver higher quality predictive outcomes, with stacking achieving the best R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> (0.9742) and lowest RMSE (5.9476). These results demonstrate the feasibility of predictive intelligent irrigation systems and their application in supporting eco-friendly and resource-efficient agriculture.