A wide range of actors seek to understand the water risks facing agriculture in order to make informed decisions on planning, investment and operations, and a large and growing set of tools supports such assessments. The right tool will depend on the type of user and the nature of the decision (Figure 2). Public authorities need to be equipped with a suite of tools to support decision making and water risk management for agriculture across different timescales and for different purposes. Such purposes include early warning systems to bolster farmers’ preparedness to floods and droughts; water allocation and permitting regimes to ensure sustainable water use across different uses and users; extension and advisory services for farmers for climate-resilient agriculture; or spatial planning decisions to guide the location of agricultural production, among others. In the private sector, farmers looking to optimise yields can use tools to guide on-farm water management decisions; agro-food corporations carry out water risk assessments to understand supply chain resilience and threats to business continuity; and agricultural insurers use sophisticated modelling tools to determine insurance policy coverage and premiums.
Anticipating and monitoring water risks for agriculture
From satellites to sensors, advances in technology are opening up new opportunities to improve the anticipation and monitoring of water risks. Many of the tools being used and developed today are underpinned by increasingly sophisticated data collection, processing and analysis techniques, including remote sensing, Internet-of-Things (IoT) enabled smart devices, modelling approaches, artificial intelligence (AI), machine learning and more.
At the same time, data collected with in situ measurement methods continue to feed into several tools and play an important role in the validation of results. With data at the core of water risk assessment, data quality is critical. Credible and usable tools are those built on or validated with accurate real-world data.
Nevertheless, accurate assessment of water risks is challenging given the inherent complexity of natural systems and human-nature interactions as well as varying degrees of uncertainty surrounding risk drivers. Water risks are interrelated and can impact each other given the nature of water as a hydrologically interconnected resource (OECD, 2013[4]). For example, the occurrence of drought increases the risk of flash floods in case of heavy rains, as drought-affected soils cannot absorb and retain water as much as healthy soils. Risks of shortage, inadequate quality and excess may all increase the risk of undermining the resilience of freshwater systems. These aspects of compounding or cascading risks introduce complexity that is difficult to capture in models and risk assessment tools.
Users should recognise the inherent limitations of tools. Transparency about uncertainties in data, assumptions and models is essential to prevent overconfidence and to support appropriate decision making, particularly as uncertainty increases over longer time horizons.
This section showcases examples of tools that can support the anticipation and monitoring of water risks for agriculture organised according to a typology (Figure 3). It includes tools that give information on water resource status and trends as a component or input to risk assessment (e.g. water use monitoring) as well as those that are more explicitly focused on risk assessment (i.e. assessing probabilities and impacts of water-related hazards). Emphasis is placed on physical water risks as opposed to water-related financial, regulatory or reputational risks, although some of the tools reviewed assess these aspects in addition to physical risk. Many are well-established tools that have a proven track record of use by water risk managers. Others are more experimental in nature, often in initial or pilot phases and being developed in academic settings. More detailed descriptions of individual tools can be found in the Annex.
The selected tools have been chosen with the needs of public authorities in mind, such as ministries of agriculture, public irrigation managers, water agencies, spatial planning offices etc. This paper does not review the full range of tools and technologies designed to support farm-level water management decisions, although it notes the potential benefits of data sharing between farmers and public authorities.
Tools to monitor and assess water-related risks in agriculture vary by decision-making horizon, spanning real-time operational use to medium-term risk management and long-term planning and preparedness. For tools designed to support short-term, often operational decision making (days, weeks), advances in remote sensing, IoT and machine learning are transforming the ability to monitor water resources and water-related hazards in real or near real time. Other tools lend themselves to monitoring and assessing medium-term risks (seasonal, months) or to supporting decisions for longer-term planning and preparedness (years, decades). Some types of tools can be applied for several time horizons. For example, impact assessment tools may be used for short-term decision making, e.g. assessing damage for insurance claims post-disaster, or for longer-term planning and preparedness purposes, e.g. estimating the potential impacts of different drought scenarios to design agricultural support packages.
Hazard monitoring and short-term forecast tools
Real-time monitoring and short-term forecasts provide vital information for several uses, including early warning systems to minimise damage and loss in the occurrence of hazards.
Flood monitoring systems are typically based on real or near real-time meteorological information and remote sensing data, combined with hydrological models to provide short-term forecasts. The choice of the most suitable flood forecasting model may depend on the type of watershed and climatic characteristics of the area (Jam-Jalloh et al., 2024[76]). The GloFAS Global Flood Monitoring, for example, integrates real-time precipitation information from multiple satellites into a quasi-global hydrological runoff and routing model to provide near-real time monitoring of global flood risk (Annex A). Forecast rating curves are an alternative tool that can provide early warning of downstream flood events and their probable severity (see Annex C).
National meteorological services have traditionally provided near-term forecasts for extreme weather events, and some have developed apps to deliver information on severe weather directly to users. For example, the NOAA Hi-Def Radar Pro app developed by the US National Weather Service provides alerts of tornadoes, storms, and flood watches and warnings. Numerous commercial weather services offer high-resolution forecasts tailored for farmers at resolutions as low as 1km Many of these commercial services rely on data, infrastructure and technologies developed and maintained by public institutions. For example, one commercial provider employs NASA's digital terrain model to downscale weather information to provide hyperlocal forecasts globally.
Drought monitoring systems map drought conditions based on drought indices (Annex C) calculated from observed meteorological and hydrological data from weather stations and Earth observation data. Drought monitoring systems are often complemented with a forecasting and early warning function that predict the probability of a drought occurring in a region based on historical weather patterns, real-time meteorological monitoring, weather forecasts and modelling (e.g. European Drought Observatory, US Drought Monitor, and the Canadian Drought Monitor; Annex A). An interesting recent development in the field of drought indicators is a shift from meteorological to impact-based indicators, as shown by the Australian Agricultural Drought Indicators (see Impact assessment tools section below and Annex A for more details).
Recent advances are pushing inland water quality monitoring from isolated case studies often requiring the physical presence of technical personnel toward real-time and spatially resolved monitoring. Remote sensing can provide insights into salinity, algal blooms and nutrient fluxes. Machine-learning models can process satellite imagery and sensor data to forecast eutrophication, detect harmful algal blooms and predict tipping points in agricultural systems. However, remote sensing cannot currently capture all aspects of water quality; including total suspended matter and nitrates (Arias-Rodriguez et al., 2023[77]).
Bridging that gap requires fine-scale, in-situ sensing (Annex C). When high-frequency data from in situ sensors are combined with Earth observation and meteorological inputs, hybrid machine learning can outperform traditional empirical indices for irrigation water quality (El Bilali and Taleb, 2024[78]). Combining data from multi-parameter and other in situ measurement techniques with machine learning approaches, the Digital Water City prototype system was developed in Peschiera Borromeo, Italy to forecast potential contamination from the reuse of wastewater for agricultural purposes (DigitalWater.City, 2022[79]). Other innovative monitoring solutions include ‘fish tracking’, a method that monitors fish behaviour to detect changes in water quality over time (see the example of FISHTRAC integrated into the digital twin developed by IWMI/CGIAR Annex A).
Irrigation advisory tools
Remote sensing has become a cornerstone of precision irrigation and soil moisture monitoring. High-resolution soil moisture maps derived from remote sensing can provide actionable data for farmers and authorities, supporting targeted irrigation scheduling and the identification of water-stressed areas (Parra-López et al., 2025[80]; Dari et al., 2021[81]) (Annex C). Italy’s IRRINET is a prime example of remote sensing technologies applied to optimise irrigation scheduling (Annex A). This online irrigation advisory service provides free daily irrigation scheduling via web, SMS or mobile apps. In Spain, the Agroclimatic Information System for Irrigation (Sistema de Información Agroclimática para el Regadío, SiAR), provides daily information on crop irrigation needs through its website, mobile application and web GIS viewer (Annex A).
At the farm level, smart irrigation frameworks that combine remote-sensing, IoT-enabled soil moisture sensors, and AI-based forecasting of weather and water demand are being deployed to optimise water productivity and crop yields (Al Mashhadany et al., 2024[82]). These digital systems blur the line between monitoring and decision support. AI-based predictive models are increasingly embedded in the systems to link monitoring with automated responses. By linking forecasting directly to control such as pump activation and valve adjustment, these systems can autonomously trigger irrigation events or alert managers to pollution risks, thereby supporting a shift from passive monitoring to proactive water management. Drone-mounted multispectral and thermal sensors further enhance this capability. For instance, thermal imagery used in the Crop Water Stress Index (CWSI) enables within-field detection of plant water deficits, allowing highly localised irrigation interventions (Cheng et al., 2023[83]). Drone imagery combined with machine learning has been applied to optimise irrigation across diverse crops, from kiwifruit orchards (Zhu et al., 2023[84]) to vineyards (Romero et al., 2018[85]).
Irrigation advisory tools help optimise production by ensuring crops receive the right amount of water – minimising crop water stress or avoiding overwatering. However, such tools do not assess whether water use is sustainable at the catchment level in terms of staying within renewable resource limits or ensuring enough water is left within the system for environmental flows. Public authorities mandated with promoting environmental sustainability will require different tools to assess these water risks, including accurate estimates of agricultural water use.
Water use and consumption monitoring
Accurate information on actual and projected water use and consumption can help gauge the risk of water shortages for agriculture, yet water resource management is often hindered by a lack of reliable data.
While often estimated at the country level, more granular estimates of total water withdrawn from surface or groundwater sources for irrigation usually requires on-the-ground measurement systems. To overcome in situ monitoring gaps, remote sensing-based modelling approaches combine satellite observations and agro-hydrological models1 (see, for example, Zaussinger (2019[86])), although such methods are subject to uncertainty when validated with in situ irrigation data at field and regional scales given discrepancies observed (Ji et al., 2025[87]). In situ monitoring requires on-farm water flow meters or water level sensors (for measuring groundwater levels), which are commercialised for farm-level decision making and regulatory reporting.
Tools exist to estimate crop water consumption, although their coverage tends to be regional rather than global. A few recently created platforms provide real-time or near real-time data on how much water crops consume, mainly through satellite-based estimates of evapotranspiration (ET). A leading example is FAO’s WaPOR, which offers near real-time data on ET and biomass production across Africa and the Near East (Annex A). This enables estimates of water productivity (how much crop is produced per unit of water consumed) and to identify areas where water is being overused or where efficiency gains are possible. Similarly, the OpenET platform makes satellite-based estimates of ET at field to basin scales in the western United States, while the EEFlux app produces field-scale maps of water consumption using satellite imagery allowing users to check water-use maps in near real-time on any mobile device. Nevertheless, these tools may be subject to degrees of error. For example, one study found that the EEFlux tool led to large ET overpredictions and error of up to 181% (Kadam et al., 2021[88]).
Digital twins
A digital twin is a virtual representation of a physical system, such as a river basin or a farm system. Acting as a platform, it combines real-time data streams (from sensors, satellites, IoT) with multiple models (process-based and/or machine-learning models). Users can explore “what-if” scenarios, test interventions and see the simulated consequences in near real time. Furthermore, digital twins can evolve as the physical system changes. Whereas models like SWAT or WEAP are simulation tools that are usually run periodically, a digital twin is live and continuously updating. IWMI has developed a digital twin of the Limpopo River Basin that allows monitoring of reservoir levels, environmental flows and water quality among many other features (Annex A).
Crop monitoring and forecasting tools
Several tools monitor crop conditions globally, providing early warning on crop failures and supply chain disruptions. Timely and reliable information on crop yields supports decision making around food security planning and market access. Crop yield estimates were traditionally generated through field surveys or farmer interviews. Increasingly, countries have developed crop monitoring and yield forecasting systems that combine satellite, meteorological and ground-sourced data (Annex A). For example, the Canadian Crop Metrics application produces reports and data that provide market information and analysis on the situation and outlook for Canadian principal field crops, including grains, oilseeds and some pulse and special crops. NASA Harvest provides a suite of tools including the GEOGLAM Crop Monitor that produces monthly updates about crop supply in major crop producer countries to the Agricultural Market Information System. The FAO Global Information and Early Warning System on Food and Agriculture (GIEWS) monitors the condition of major food crops across the globe to assess production prospects and provides alerts on emerging food shortages. It integrates the Agricultural Stress Index System, based on earth observation data on water stress. CropWatch, developed by the Chinese Academy of Sciences, provides an outlook to the global production situation of major grains (CropWatch, 2025[89]).
Seasonal weather forecasts
Seasonal weather forecasts provide information about the likely average weather conditions over a period of several months, typically three to six months, rather than specific daily weather events. They estimate how variables such as temperature and precipitation for an upcoming season are expected to compare with long-term historical averages, often expressed in terms of probabilities: for example, the likelihood of a warmer-than-normal or drier-than-normal season. Seasonal forecasts help anticipate the influence of climate patterns such as El Niño, La Niña or the North Atlantic Oscillation. Seasonal forecasts do not aim to predict exact weather on particular days but instead provide probabilistic guidance on general trends that can inform planning and risk management. For example, the EU’s Copernicus Climate Change Service (C3S) provides forecasts in the form of charts and other graphical products for the three coming months, updated monthly.
Hazard and risk maps
Hazard maps allow decision makers to spatially visualise hazard levels. Complemented with information on exposure and vulnerability, hazard maps can be expanded into risk maps, with risk often calculated through observed or expected impact, e.g. losses due to exposure and vulnerability to the hazard. For example, the European Drought Risk Atlas provides drought risk for five agricultural crops at subnational level in the EU, quantifying average annual losses to a 1-in-50-year drought under current climate conditions (JRC, 2023[90]).
Drought hazard maps have become increasingly advanced, moving beyond historical data to integrate machine learning and climate modelling that offer forward-looking insights for decision making. Early drought hazard maps typically estimated hazard levels by extrapolating observed or historical data. Today, methods such as machine learning allow for the quantification of hazards via indices calculated from hydrometeorological data (Annex C), while climate models can project future hazard levels under different warming scenarios. For example, the European Drought Risk Atlas combines conceptual models of drought risk with data-driven assessment of sectoral drought risk based on machine learning, and provides maps showing projected changes in drought hazard across the EU under different climate scenarios (JRC, 2023[90]) (see Annex A for more details). Given increasing climate variability, such tools may be more reliable than those that rely solely on historical data.
Flood hazard and risk maps are underpinned by different types of flood modelling (empirical, hydrodynamic and simple conceptual models), each with their own advantages and limitations (Annex C, Table C.2.). Advances in remote sensing are improving empirical approaches (see IWMI flood risk mapping for South Asia, Southeast Asia and Nigeria in Annex A). Remote sensing is also essential to calibrate and validate hydrodynamic models. The data and computational demands for hydrodynamic modelling can be significant and simply unavailable in many places. Simple conceptual models overcome some of these challenges. For example, Natural Resources Canada developed a simplified flood model covering the entire country, which only requires topographic data of a watershed and the shape and depth of the river network (see HAND model, Annex C). Ultimately the most suitable model will depend on the purpose, data availability and computational demands, with multi-model, multi-discipline approaches representing a promising area of research (Teng et al., 2017[91]).
In recent years, modelling approaches have complemented observed data to develop indicators that can proxy water quality and the resilience of freshwater ecosystems (e.g. modelling of salinity levels and biological oxygen demand based dynamical surface water quality models). The WWF Water Risk Filters provide hazard maps of several indicators on water quality and freshwater ecosystem resilience (Annex A).
Hydrological infrastructure maps
Irrigation and drainage maps provide information that helps to assess agriculture’s vulnerability to water-related hazards as well as agriculture’s potential impact on water risks. High-resolution satellite data and large-scale machine learning are replacing agricultural survey or census data as a way to produce irrigation maps. Irrigation maps can help identify over-abstraction of water resources and identify illegal irrigation, supporting regulatory enforcement of water allocation regimes. Brazil’s national water agency, ANA, produces an Irrigation Atlas using satellite data and machine learning techniques to identify irrigated areas. The agency uses the information to estimate water use and update water balances (ANA, 2024[92]).
Drainage water from agricultural areas can often contain heavy nitrate, pesticide and other contaminant loads. Drainage mapping is therefore an important tool for assessing water quality risks. Drainage maps can also be used for flood risk assessment, as water travelling through artificial drainage systems may impact flood events. In Latvia, the country’s drainage maps have been digitalised, providing open access to information on drainage systems for agricultural land.2 In France, the Agricultural Drainage Database (“BD Drainage”) is being built as an interactive mapping tool to provide information on the path taken by water from agricultural plots that have been drained, through to its discharge into the natural environment, watercourses or groundwater (Annex A).
Impact assessment tools
Advance knowledge of how water-related hazards might impact farming can guide policymakers in planning measures that strengthen resilience and adaptation. While hazard mapping tools often show the extent of hazard-affected agricultural areas (exposure), relatively fewer tools provide quantitative estimates of losses in terms of yields or income. One such tool is the FAO Drought Impact Assessment Platform (d-iap). This web-based platform integrates advanced crop modelling with economic, climate and soil datasets to estimate drought impacts on yields and income under both present and future climate scenarios. In Australia, the ABARES farmpredict model helps anticipate production and financial outcomes for Australian farms using a machine-learning micro-simulation incorporating weather and climatic conditions, commodity prices and farm-level survey data, with a particular aim to foresee potential drought impacts (Annex A). Australia is also focusing on drought impacts through its recently developed Australian Agricultural Drought Indicators. Unlike most drought indicators (see section above), these indicators aim to anticipate and monitor agricultural and economic impacts of drought for non-irrigated extensive agriculture (Hughes et al., 2025[93]).
Flood modelling or mapping, as described above, can be combined with flood damage models to predict or evaluate flood impacts on agriculture. Examples of tools include the Agricultural Flood Damage Analysis Model, the floodam-agri, developed in France to estimate flood damage at the plot level, and AGRIDE-c, a conceptual model for the estimation of flood damage to crops designed to be applicable to different territories (Annex C). Flood damage can be modelled in many ways, varying by model type, categories of damage assessed; and flood parameters (water height, flooding duration etc.) (Modjeska and Brémond, 2022[94]). Nevertheless, a lack of observed data on flood damage to agriculture limits the development of empirical models or the validation of synthetic ones (Molinari et al., 2019[95]).
At a broader scale, macroeconomic models can be used to assess economy-wide and agriculture-sector economic impacts of water-related phenomena. The World Bank has developed a model that estimates the macroeconomic implications of climate change impacts and adaptation options, many of which are water-related, highlighting the sources and magnitude of countries’ vulnerability (World Bank, 2025[96]).
Water resource monitoring and assessment tools
In addition to hazard risk assessment, policymakers need information on freshwater availability to assess agricultural water resource security. Many tools in this category rely on modelled estimates rather than observed data. In many instances, the risk assessment information derived from these tools will need to be communicated effectively to farmers to support better on-farm water management decisions.
A starting point can be integrated water availability assessments, which examine the spatial and temporal distribution of water quantity and quality in surface water and groundwater in a defined geographical area. Such assessments typically examine water supply (through climatic inputs), water quality, and water use, and may analyse water availability under different climate scenarios. The United States Geographical Survey (USGS) has developed a National Water Availability Assessment Data Companion to provide regularly updated, model-based estimates of water availability and use, derived from USGS scientific models that underlie the US National Water Availability Assessment.
Water supply monitoring and projection tools allow the agricultural sector to respond more flexibly to water shortages. Monitoring and forecasting of available water supplies is often used by administrators to make decisions about how much water to release and to whom (e.g. across senior and junior water rights holders), as well as by irrigation organisations to determine whether and how to implement water restrictions if current or projected water supplies are insufficient to meet demand. The California Water Watch monitors current water availability and provides short-term forecasts using daily updated weather data, groundwater, reservoir and snowpack levels, streamflow, as well as soil moisture and vegetation conditions (Annex A).
Model-based tools have been developed to better plan water resources. The Soil and Water Assessment Tool (SWAT) is one of the most widely used tools to simulate the quality and quantity of surface and groundwater at the watershed scale (Annex A). Another commonly used model is Water Evaluation and Planning (WEAP) that integrates water supply and demand (including agriculture), allocation rules and scenarios to help optimise allocation and can be used to test how policy changes, drought or new demands affect water availability for irrigated agriculture (Annex C). Hydrologic models such as HydroGeoSphere can simulate the movement of water between surface, soil, and groundwater systems and can be used to evaluate the impact and risk associated with climate for water resources. Canada1Water is a continental-scale model of Canada’s complete hydrologic system based on the HydroGeoSphere platform (Annex A).
At the macro scale, global water models are widely used to understand changes in water resources over time, but differences between them mean their results should be interpreted with caution. Numerous models have been developed over the last 30 years with more recent versions improved by incorporating the interactions of human activities with the water cycle (e.g. irrigation, reservoir construction). Nevertheless, disagreements between simulated variables make model-based inferences uncertain (Gnann, Reinecke and Stein, 2023[97]).
Satellite technology is improving empirical data on freshwater resources and provides complementary techniques to assessing water resources and calibrating water models. For example, the Soil Moisture Active Passive (SMAP) and the Soil Moisture and Ocean Salinity (SMOS) missions have demonstrated an ability to quantify moisture extremes impacting agricultural production. The GRACE satellite missions provide information on changes in “terrestrial water storage” – the total amount of freshwater stored on land that includes surface water, soil moisture and water in underground aquifers. This technology is being applied to fields such as groundwater monitoring, which has typically been hampered by a lack of data (NASA, 2022[98]). Remaining constraints include the relatively coarse spatial and temporal resolution of satellite data as well as the need to improve techniques to fuse data from multiple sources (Ibrahim et al., 2024[99]).
Many water scarcity or water stress metrics have been developed. Octavianti and Staddon (2021[100]) identify 67 distinct assessment tools that use resource-based metrics. Some are designed specifically for agriculture. Calculation of indices rest on estimates of processes of hydrology, crops and human activities which can introduce uncertainties. The spatial resolution can be relatively large meaning that results can be inaccurate at local level.
Water footprint and virtual water trade analysis
Information on the water footprint of agriculture can inform assessments of water resource sustainability. Hoekstra and Chapagain (2008[101]) were among the first to apply the concept of water footprint to agriculture, aiming to provide a comprehensive measurement standard to define the amount of water required to produce a specific crop (Wang et al., 2023[102]). While early studies of water footprint only considered the volume of blue water consumed, more recent studies tend to incorporate the green water footprint, sometimes grey water footprint (the volume of freshwater required to dilute pollutants to meet water quality standards), 3 and increasingly the sustainability of water footprints (Annex C).
Water footprint analysis can be combined with trade flow data to enable virtual water trade analysis. This can inform assessments of imported water risks by looking at the proportion of trade flows that depend on unsustainable water use.
Projection and scenario tools for long-term trends
Projection and scenario tools can help identify potential long-term trends and emerging risks. These can be useful inputs into broader cost-benefit, risk-based and options-based planning approaches that incorporate uncertainty, local data and economic feasibility to guide governments and producers in their choice of interventions.
Climate scenarios
Informed by climate models, climate scenario-based tools have been developed to support the anticipation of long-term water-related risks for agriculture. CANARI-Europe shows projections based on moderate and high-emissions scenarios for 2020‑2050 and 2050‑2100 for agro-climatic indicators and risk for specific agricultural sectors (e.g. water deficit for winter cereals) (Solagro and Makina Corpus, 2025[103]). Similarly, the AgriAdapt Webtool for Adaptation (Annex A) provides simulations of the next 30 years that can be compared with the past three decades to show agro-climatic trends, with information tailored for specific agricultural production areas (Agri-Adapt, n.d.[104]). The AquaPlan App enables users to identify how different climate scenarios will change exposure of farms to drought and water scarcity, and what adaptation measures can be taken to mitigate these risks (Annex A).
In Australia, the My Climate View platform helps farmers plan for long-term climatic shifts, providing an on overview of how seasonal climatic variables will change for over twenty-two agricultural commodities at specific farms. The interactive user interface allows farmers to compare projections for 2030s, 2050s and 2070s under medium and high-emissions scenarios to past climatic data. In addition, the tool also provides seasonal (1‑3 months) forecasts (MyClimateView, n.d.[105]).
Climate modelling is also used to assess long-term trends in water quality, covering nutrients, salinity and climate–agriculture interactions by combining hydrological and water-quality models with climate and socio-economic scenarios.
Water demand forecasts
Water demand forecasts are used to anticipate potential water shortages for agriculture, especially in the context of rising demand from competing needs in other sectors. Perez et al (2024[106]) estimate a 17% increase in irrigation water demand between 2020 and 2050, while WRI estimate that global water demand is projected to increase by 20‑25% by 2050 (World Resources Institute, 2023[107]). Nevertheless, forecasting future water demand is highly challenging. The factors that affect demand are numerous and interconnected and include population, socio-economic development, increasing climatic variability, industrial structure, technological innovation, water prices, consumer behaviour, regulations, etc. In addition, water demand is highly variable from season to season. Water demand forecasting has been notoriously inaccurate as methods have struggled to cope with this complexity. Traditional methods rely on mathematical and statistical techniques that at their simplest involve multiplying current per capita water consumption by expected future population, which tends to overestimate future demand (Fang et al., 2024[108]). Statistical methods struggle with nonlinear and irregular changes in water demand data, and simply extrapolating past trends is unlikely to accurately reflect a changing future (Maußner, Oberascher and Autengr, 2025[109]). Artificial intelligence is being applied with promising results (Fang et al., 2024[108]).
Strategic foresight
Strategic foresight is a structured and systematic approach of exploring plausible futures to anticipate and better prepare for change. Rather than making predictions based on linear extrapolation of past and current trends, foresight cultivates the capacity to anticipate alternative futures and an ability to imagine multiple and non-linear pathways. Strategic foresight helps policymakers improve the effectiveness of governments by identifying opportunities, challenges, risks and disruptions that may arise over the coming years. It draws on multiple methodologies including horizon scanning, megatrends analysis, scenario planning and visioning and back-casting (OECD, 2025[110]). The European Parliament carried out a foresight study on future of water availability and use in the EU that included a literature review, qualitative and quantitative data analysis, scenario development and stakeholder consultation, to develop policy options (Duin, Van Lanen and Zehnder, 2025[111]). In France, the Centre for Studies and Strategic Foresight (CEP) is the internal think-tank of the French Ministry of Agriculture created in 2008 to provide the ministry with strategic monitoring, strategic foresight work and policy evaluation. CEP has examined how the agricultural sector has been taken into account in foresight studies (see (CEP, 2014[112]).
Monitoring and assessment tools still provide limited coverage of some critical water-related risks, particularly those linked to resilience of freshwater ecosystems, water quality, groundwater monitoring and terrestrial moisture recycling (green water) (Table 1). The risks related to loss of freshwater ecosystem services to agriculture are difficult to capture as many ecosystem services are hard to quantify or value. When this risk is integrated into assessment tools, the focus is often narrow, for example, calculating environmental flow requirements. While useful, this focus overlooks key dimensions such as water quality and broader ecosystem health, including nutrient loads and pollutants. Many current indicators also rely on simplistic proxies – for example, measuring kilometres of unobstructed rivers – which do not adequately reflect ecological condition or function. More comprehensive approaches are needed to assess the quality, health and resilience of freshwater ecosystems and the risks their degradation poses to agriculture.
Few established tools have so far been developed and deployed for monitoring and assessing water quality risks for agriculture, compared to other risks such as water shortage and excess. Nevertheless, several databases that track water quality are available. Annex B provides examples of the databases that can facilitate monitoring water quality related risks.
Groundwater monitoring also remains a persistent challenge. While satellite data on groundwater storage and ground motion has advanced, its relatively coarse spatial and temporal resolution still limits practical application. This is particularly concerning because aquifer depletion is often non-linear: water levels may appear stable even as underlying conditions deteriorate until a tipping point is reached and depletion accelerates rapidly.
Techniques to assess the dependence of agriculture on moisture recycled through terrestrial ecosystems are still at an early stage of development. Although tracking of moisture recycling networks is improving, significant uncertainties remain. Atmospheric water-tracking models are constrained by limited validation data, and complex meso-climatic interactions make regional vulnerability difficult to predict. Emerging academic work is advancing the field – for example, mapping atmospheric water connectivity in agricultural supply chains to show how green water evaporation from Brazilian soy contributes to rainfall in Argentina, Bolivia and Paraguay (Cigna et al., forthcoming[113]) – but operational tools for decision makers are still lacking. More robust, validated approaches are needed to reliably integrate concepts such as precipitationsheds and evaporationsheds into risk assessment. Future research avenues could explore the sensitivity of rainfall distribution to vegetation management, and how shifts in land use and land cover may propagate hydrological impacts across scales (Dupont et al., 2025[114]).
More broadly, while hazard monitoring is relatively well developed, translating this information into meaningful risk assessments for agriculture remains challenging. Risk analysis requires integrating hazard data with agricultural data, such as planted areas, crop types and yields, as well as socio-economic data. These datasets are often fragmented, outdated or incompatible, limiting the ability to assess vulnerability and potential impacts. More integrated data systems are needed to support comprehensive risk assessments.
Finally, a persistent challenge is the mismatch between the scale of available data and the scale at which decisions are made. Spatially, farmers often require field-level information, whereas public authorities typically require basin or regional level tools. Temporally, seasonal data is often key yet missing in water risk assessment for agriculture. For example, some tools rely on static or outdated crop layers that do not reflect actual planting patterns, underscoring the need for timely seasonal crop masks and yield estimates. In addition, risks often compound and cascade over time, yet most tools still assess hazards in isolation. Bridging these spatial and temporal gaps is essential for effective risk management.
← 1. Hydrological-based, crop demand-based and integrated methods (Brookfield et al., 2024[191]).
← 2. The drainage map is available here: https://www.melioracija.lv/?lang=EN&loc=503959;271613;4.
← 3. In other contexts, the term “grey water” can refer to the wastewater generated from household activities such as bathing, washing hands and laundry.