Abstract :
Sustainable irrigation water management is increasingly challenged by water scarcity, climate variability, land-use change, and growing agricultural demand, particularly in semi-arid regions. Conventional irrigation planning and operational approaches often lack the spatial, temporal, and predictive capabilities required to support adaptive and data-driven decision-making. In recent years, the integration of Remote Sensing (RS) and Machine Learning (ML) techniques has emerged as a powerful framework for improving the monitoring, assessment, and forecasting of irrigation water demand and system performance. This paper presents a comprehensive review of the state of the art in RS- and ML-based applications for irrigation water management, with a particular focus on semi-arid agricultural systems. The review synthesizes advances in satellite-based land use and land cover mapping, vegetation indices —especially the Normalized Difference Vegetation Index (NDVI)—and their role in estimating cultivated areas and irrigation requirements. It further examines the application of statistical and machine learning models, including time-series forecasting and supervised learning algorithms, for predicting irrigation demand and hydraulic parameters within canal-based irrigation networks. Current integrated RS-ML frameworks are critically assessed in terms of data requirements, model performance, transferability, and practical implementation challenges. Key limitations related to data availability, model interpretability, and institutional adoption are discussed, and emerging research directions are highlighted, including hybrid modelling, climate-informed forecasting, and decision-support systems. The review provides a consolidated knowledge based to support research, practitioners, and water authorities in developing more resilient and sustainable irrigation water management strategies.
