Detecting illegal water abstractions using Earth Observation and artificial intelligence: challenges, research gaps, and future directions

Elena Kakoulli , Anastasios Antoniades, Stelios Neophytides 

Water scarcity and unsustainable withdrawals are major environmental challenges, particularly in semi-arid and Mediterranean regions where agriculture places strong pressure on limited hydrological resources. Illegal or unreported water abstractions intensify these pressures by undermining ecosystem stability, policy compliance, and long-term water security. Although Earth Observation (EO) and artificial intelligence (AI) have substantially advanced irrigation monitoring, evapotranspiration estimation, soil moisture retrieval, and agricultural analysis, the detection of potentially unauthorized water use remains insufficiently addressed. Most existing studies focus on describing water use patterns or estimating irrigation-related variables rather than identifying suspicious abstraction behaviour in governance settings. This paper presents a structured literature review of recent research relevant to the detection of illegal water abstractions using EO and AI. The review synthesizes evidence across irrigation monitoring, biophysical estimation, anomaly detection, and operational water governance studies, with particular attention to methodological limitations affecting real-world deployment. The analysis shows that the field is constrained less by the absence of useful technical components than by the lack of integrated, validated, interpretable, and scalable monitoring frameworks. Key challenges include heterogeneous data integration, limited ground-truth validation, weak linkage to administrative and legal context, and the predominance of descriptive rather than predictive approaches. The review identifies a critical research gap: the absence of governance-ready systems capable of proactively screening abnormal water-use behaviour across heterogeneous agricultural landscapes. It therefore outlines a research agenda centred on EO data fusion, integrated spatiotemporal modelling, and predictive anomaly detection for risk-informed monitoring and decision support.