Urban water security needs better data, governance and AI support

Water security research is expanding as climate change, urbanization, pollution and rising demand place growing pressure on freshwater systems, but artificial intelligence (AI) remains underused in the tools needed to assess and manage the crisis, according to a systematic review by Karunya Baburaj and Aavudai Anandhi of Florida Agricultural and Mechanical University.

The review analyzes 146 scientific articles and finds that water security remains unevenly defined, inconsistently measured and only weakly linked to artificial intelligence applications, despite AI's growing potential for monitoring, prediction, treatment, infrastructure planning and decision-making.

The study, titled "Water Security: A Systematic Review of Definitions, Indicators, and Artificial Intelligence Applications," was published in Water.

Water security expands beyond supply as climate and urban pressures rise

Freshwater systems are facing pressure from population growth, rapid urban expansion, climate change, pollution, rising consumption and overexploitation. The review finds that the concept is no longer limited to whether enough water is physically available. It now includes whether water is safe, affordable, accessible, reliable, equitably distributed and managed through institutions capable of protecting people, economies and ecosystems.

The authors used the PRISMA systematic review framework to screen research from Google Scholar, Web of Science and Scopus. From 289 records initially identified, 146 articles were included in the final synthesis. The review examined how water security is defined, which indicators are used to measure it, what methods are applied in urban water security assessments and how artificial intelligence and machine learning are being used in the field.

The review shows that scientific attention to water security has increased sharply since 2015, with a peak in 2024. The research base is geographically uneven, with the highest number of studies focused on China, followed by the United States, India and the United Kingdom. Other countries contribute smaller clusters of work, showing that water security is a global concern but that research capacity and policy attention remain unevenly distributed.

Water security lacks one widely accepted definition. Among the reviewed studies, 54 articles offered definitions, but they varied widely. Some treated water security as a dynamic and multidimensional concept. Others focused on safe access, sufficient quantity, acceptable quality, governance, sustainability, risk reduction or protection from water-related disasters.

The researchers synthesize these fragmented definitions into a broader structure. They identify access, quality, quantity, governance, sustainability, risk, capacity, needs, protection, beneficiaries, scale, impacts and outcomes as core components of water security. Access includes stable, affordable and equitable supply. Quality refers to water that is suitable for human, domestic, agricultural, industrial and environmental uses. Quantity covers adequate and sufficient water for different sectors and communities.

The review shows that water security depends not only on water resources, but also on planning, management, stakeholder participation, political stability, institutional capacity and public services. Risk is another central component, covering droughts, floods, contamination, water-related hazards and the ability to reduce exposure to future shocks.

The broader framing matters because water insecurity can occur even where water exists in physical terms. A city may have water resources but lack reliable infrastructure. A rural area may have groundwater but face contamination, unaffordable access or poor governance. A basin may have seasonal supply but remain vulnerable to drought, overuse or conflict between sectors. The review shows that water security must be assessed through social, environmental, technical and political lenses at the same time.

The study also shows that water security is examined across multiple scales, from households and communities to cities, basins, regions, countries and global systems. This scale variation makes comparison difficult. Urban assessments often focus on infrastructure, service delivery and population growth, while basin-level studies focus on resource availability, ecosystem health and competing water demands. National and global studies often give more attention to governance, economic development, food systems and climate risk.

Indicators and urban assessment methods remain fragmented

Water security indicators are widely used, but the field remains fragmented. Among the 146 articles, 36 provided dimensions and indicators for assessment. The authors classify these indicators into quantitative, qualitative and combined types, creating a clearer structure for understanding how water security is measured.

Quantitative indicators are the most common. They include freshwater resources per capita, water consumption per capita, piped water coverage, sewage coverage, supply reliability, wastewater treatment capacity, water loss, water productivity, water demand, annual precipitation, annual temperature, population density, urbanization rate, greenhouse gas emissions, water use intensity and efficiency of water supply systems.

These measures are important because they provide clear data points for tracking water availability, supply coverage, infrastructure performance and resource stress. They are used across domains such as water supply, water resources, sanitation and health, environment and ecosystems, water management and socio-economic systems.

Qualitative indicators receive less attention but remain essential. They include hygiene, perceived water quality, contamination risk, waterborne disease, compliance with standards, recreational water safety and local public health conditions. These measures are harder to standardize, but they capture dimensions that physical supply data alone cannot explain.

Combined indicators integrate measurable data with governance, risk and institutional factors. They include flood frequency, drought frequency, disaster preparedness, public health risk, stakeholder engagement, access to data, strategic planning, infrastructure reliability, leakage rates, environmental protection spending, water source protection and management capacity. The review shows that these mixed indicators are especially important because water security depends on both technical systems and human institutions.

The authors also examine urban water security in detail. Urban water systems are under particular strain as cities expand, climate risks increase and infrastructure ages. The review identifies 25 studies focused on urban water security assessment and groups their methods into eight categories: index-based, model-based, framework-based, spatial and geospatial, data-driven, governance and qualitative, climate and scenario-based, and risk-based methods.

Index-based approaches use indicators to produce a composite score. Model-based approaches rely on hydrological, system or simulation models. Framework-based approaches organize variables using structured concepts. Spatial and geospatial methods assess geographic variation and inequality. Data-driven methods use statistics or machine learning. Governance and qualitative methods bring in institutional and stakeholder perspectives. Climate and scenario-based methods examine possible future conditions. Risk-based methods assess hazards, vulnerabilities and impacts under uncertainty.

Previous studies often examined methods separately, but the researchers bring them into one organized structure, resulting in a clearer map of how cities can evaluate water security and where assessment gaps remain.

The review also makes clear that urban water security cannot be measured through infrastructure alone. A city may have treatment plants and distribution networks but still face insecurity due to unequal access, poor maintenance, weak governance, flood exposure, drought risk, rising costs or dependence on distant water sources. Reliable assessment must account for the interaction between physical systems, social equity, environmental conditions and political decisions.

A narrow focus on water supply can miss deeper risks. Cities need assessment systems that can show where insecurity comes from, whether from scarcity, contamination, poor infrastructure, climate exposure, affordability problems or governance weaknesses. Without that broader view, investments may fail to address the roots of water insecurity.

Artificial intelligence offers promise but remains underused

The review finds a clear gap between the potential of AI and its current use in water security research. Out of 146 articles, only 25 addressed AI in relation to water security indicators or applications. This is a small share given the scale of the water challenges facing cities, basins and rural communities.

The AI-related studies used both machine learning and deep learning. The techniques included artificial neural networks, fuzzy logic, knowledge-based systems, genetic algorithms, adaptive agents, extreme learning machines, support vector machines, adaptive neuro-fuzzy inference systems, random forests, gradient boosting, XGBoost, recurrent neural networks, long short-term memory models, gated recurrent units, k-nearest neighbors, clustering and other predictive methods.

The review groups AI applications into six themes: general water system operations, monitoring and detection, water system management and decision-making, water quality and treatment, prediction and forecasting, and data analysis.

• In water system operations, AI has been applied to measurement platforms, preventive maintenance, fault detection, system diagnosis, target tracking, routing, design analysis and infrastructure management. These uses can help utilities detect problems earlier, reduce failures and improve the reliability of water delivery.
• In monitoring and detection, AI can identify anomalies in water flow, detect malfunctioning water meters, locate polluted areas, monitor water systems across different spatial and temporal scales, detect wastage or pilferage during transmission and identify areas at high risk of groundwater depletion. These applications can support faster intervention and better oversight of water systems.
• In water system management and decision-making, AI can help analyze water distribution networks, optimize extraction schedules, support irrigation planning, control aquifer-related variables and guide strategy selection. These uses are particularly relevant for regions where groundwater depletion, agricultural demand and urban consumption are competing pressures.
• In water quality and treatment, AI has been used to model pollution removal, evaluate adsorption processes, support heavy metal remediation, estimate membrane performance, improve treatment systems and predict water quality indicators such as ammonia nitrogen, total phosphorus and chemical oxygen demand. These applications can help improve treatment efficiency and reduce risks to public health.
• Prediction and forecasting represent one of AI's strongest potential contributions. The review identifies applications in lake water-level prediction, rainfall-runoff simulation, groundwater forecasting, nitrate contamination prediction, cloud forecasting, pipe-break prediction, leak detection, water flow prediction, nutrient concentration modeling and future water-energy-food demand assessment. Such tools could help water managers prepare for droughts, floods, pollution events and infrastructure failures before they escalate.
• Data analysis is another major area. AI can process large hydrological, meteorological, environmental and socio-economic datasets, uncover hidden patterns, fill missing data, improve spatial interpolation, support basin delineation and improve the reliability of water security assessments.

Despite these uses, the review finds that AI is still not fully integrated into water security frameworks. Many studies focus on technical model performance without linking AI outputs to broader water security goals such as equity, governance, sustainability, resilience, affordability and risk reduction. This limits the value of AI for policy and management.

The authors argue that AI will be more useful when combined with hydrological models, governance frameworks, climate scenarios, risk assessments and urban planning tools. Integrated systems could support early warning, resource allocation, infrastructure investment, water quality protection and long-term resilience.

The review also reveals limitations in the current research base. The study used keyword-based searches, meaning some relevant work may not have been captured. Screening and data extraction were conducted by a single reviewer, which may introduce selection bias. The small number of AI-focused studies also means that the full potential of AI in water security cannot yet be fully assessed.