Enhancing Soil Moisture Forecasting with Ensemble Learning: An Approach towards Sustainable Agriculture

Irrigation plays a pivotal role in agriculture, yet inefficient techniques often result in significant water wastage. An effective approach is needed to meet the increasing demands of global economies through improved resource management. Additionally, precise soil moisture (SM) forecasting is essential for making well-informed agricultural decisions and enabling efficient water use. In the era of evolving sustainable agriculture, digital tools and real-time data analysis can be a solution to this problem. This study addresses the challenge by developing an adaptive SM prediction model, aimed at optimizing irrigation practices. Utilizing ensemble learning algorithms, this research integrates an internet of things (IoT) database for SM prediction. The findings indicate that a Catboost based stacked ensemble regressor demonstrates promising performance in forecasting SM both for small and large datasets. The model is shown to be beneficial for enhancing agricultural output, promoting water conservation and supporting adaptable irrigation systems.