The Water-Energy-Food Nexus and the AI Imperative

This article is the part of “Policy Pathways for Food and Water Security in the MENA Region”

Weather is no longer what it used to be. Floods now submerge cities once thought safe from such calamities, heatwaves scorch regions that historically rarely broke a sweat, and droughts stretch far beyond their usual seasons. The familiar patterns that governed our climate have fractured. What we once called “extreme” is fast becoming the baseline.^[1] Yet, this is not just a meteorological story; it is a civilisational one. Climate change has moved from an abstract threat to lived experience, and the rise in weather extremes is perhaps its most tangible symptom.

Amidst this turbulence, a new technological frontier is emerging: Artificial Intelligence (AI). Properly harnessed, AI could help us anticipate, adapt to, and even mitigate some of the most severe impacts of a rapidly warming planet. If misused, however, it could only amplify inequities and false confidence.

The task before us is not simply to adopt AI, but to integrate it in intelligent ways, as a tool of foresight, not a substitute for scientific understanding. In arid regions, AI must also manage its own resource footprint, especially its water use and energy demand.

The WEF Nexus

Weather shocks are inseparable from the water–energy–food (WEF) nexus especially in vulnerable regions such as the Middle East and North Africa (MENA)^[2] where climate extremes drive water volatility, which then cascades into agriculture and power systems.^[3] Agriculture already accounts for the dominant share of consumptive water in the region, while most cereals are imported—this makes local hydrology and external supply chains, together, impactful on food security.^[4],^[5]

In April 2024, the United Arab Emirates (UAE) experienced the heaviest 24-hour rainfall on record, flooding urban corridors, disrupting aviation, and exposing drainage and planning gaps, a stark local expression of global hydrological volatility.^[6],^[7] Recent history also reminds us that rare tropical cyclones from a warming Arabian Sea can strike Oman and spill impacts into the UAE, as with cyclones Gonu in 2007 and Shaheen in 2021.^[8]

The Age of Extremes

Across the MENA, weather is rewriting records faster than we can update them.^[9] Wildfires ravaged the Mediterranean, droughts dried up major rivers in the Middle East, while in North Africa the balance between deluge and drought has become dangerously erratic. In the Gulf, this now includes record-breaking cloudbursts and rare tropical cyclones from a warming Arabian Sea, alongside long dry spells that stress aquifers and rangelands (e.g., variability in Dhofar’s Khareef).^[10]

This intensification is not a surprise, nor is it unexplainable; it is physics at work. Warmer air holds more moisture, fuelling heavier rain when storms form; hotter land increases evapotranspiration, deepening drought when rains fail; processes now visible in the MENA and known as “weather whiplash” reflecting the abrupt swing between extremes. The result is volatility across scales: heat domes that linger for weeks, tropical storms that intensify overnight, shifting jet streams and ocean currents that redistribute weather patterns in unpredictable ways.

Infrastructure designed for a gentler climate is struggling, from urban drainage overwhelmed in Dubai to aquifer-dependent farms facing salinisation and declining water tables. Crops fail, grids buckle, and urban heat islands turn deadly.

Where Forecasting Struggles

Traditional weather and climate models, based on physical equations of the Earth System, have advanced immensely since the mid-20th century. They underpin everything from cyclone alerts to climate projections. But even the best models falter under the weight of today’s volatility.

Extreme events occupy the statistical tail of probability distributions, meaning there are few examples in the historical record. Models trained or calibrated on past data are, therefore, illequipped to simulate “unprecedented” events. Resolution also matters: many models cannot resolve the small-scale processes, convective storms, coastal interactions, urban heat zones, that trigger local extremes.

The result is uncertainty especially in predicting the timing and intensity of extremes. Uncertainty is most acute for tail risks and locally forced hazards (convective bursts, wadis, urban heat). In the Gulf, a one-hour improvement in nowcast skill can materially reduce flash flood losses. We need tools that learn faster, resolve finer, and are affordable for national meteorological services. Enter AI.

AI Enters the Forecast Room

Artificial intelligence, when coupled with climate science, offers precisely that promise: Speed, adaptability, and scale. AI systems can ingest massive datasets, satellite imagery, radar scans, reanalysis archives, and detect hidden correlations that human-designed algorithms might miss.

Recent breakthroughs such as FourCastNet and GraphCast have shown that learned global models can rival or exceed traditional systems for many variables while enabling large ensembles.^[11] Where physical models take hours on supercomputers, AI can now generate forecasts in minutes. At short lead times, deep generative radar models improve 0–90-minute precipitation nowcasts, the window that matters for MENA flash floods.^[12]

The real transformation, however, lies in hybrid modelling, systems that merge the strengths of physics-based and data-driven approaches. These “physics-informed” neural networks respect the laws of energy and mass conservation while leveraging AI’s ability to learn complex, nonlinear relationships. In essence, they let science set the boundaries while letting AI explore the patterns within them.

Such models are already improving forecasts of heavy rainfall, cyclone intensification, and drought onset. They are also enabling faster ensemble generation, where thousands of simulated scenarios can be produced to quantify risk. In the age of extremes, that probabilistic foresight is invaluable for the MENA region.

Beyond Forecasting: Detection, Attribution, and Action

AI is also expanding the horizons of what forecasting means. It now plays a growing role in detecting, attributing, and translating weather extremes.

Detection involves scanning continuous data streams to flag early anomalies; subtle shifts in sea-surface temperature, unusual humidity buildups, or atmospheric “blocking” that might signal a heatwave. AI excels at spotting such weak signals amid noise.

Attribution, once a slow, post-disaster process, is being accelerated by machine learning. Scientists can now estimate, in near real-time, how much human-driven warming has increased the likelihood or severity of a specific event. This not only informs policy but also strengthens accountability in climate diplomacy.

And then there is translation, turning complex forecasts into operational decisions in the WEF nexus: Which neighbourhoods will flood, which crops are at risk, which basins to recharge, which reservoirs to pre-release, and which irrigation districts to throttle. Gulf examples include Oman’s National Multi Hazard Early Warning Centre, both platforms that AI can enhance with smarter triggers and uncertainty information.^[13],^[14],^[15]

Glimpses of Progress

Across MENA, tangible use-cases are emerging. For instance, irrigation optimisation that combines satellite data and machine learning is supporting date-palm and horticultural production in arid settings, including pilots in Egypt.^[16] Groundwater risk mapping now fuses GRACE trends (Gravity Recovery and Climate Experiment), pumping records, and climate reanalysis to highlight hot spots in systems such as the Saqram and the Tigris–Euphrates.^[17] Moreover, AI-augmented inflow prediction is being tested to inform multi-objective reservoir operations on the Nile. Beyond speed and precision, cloud platforms are lowering barriers-toentry by making advanced forecasting tools available without supercomputing infrastructure.

The Cautions We Must Heed

Machine learning models are only as good as the data that train them, and historical climate data are uneven. Regions like the MENA remain under-observed. If AI learns from incomplete or inaccurate data, it may fail to predict extremes precisely where vulnerability is highest. Indeed, the risks of overreliance on AI are real. Physics-free systems may be fast but wrong.

Indeed, the risks of overreliance on AI are real. Physics-free systems may be fast but wrong. Without open hydrometeorological and agriculture data, AI could only widen regional divides. Then comes the problem of interpretability. Decision-makers cannot base public warnings or policy actions on “black box” outputs. They need explainable AI, systems that reveal why a forecast says what it does, and how certain that prediction is.

There is also the question of access. High-performance AI systems demand enormous computational resources and proprietary data. Without deliberate investment in open data and shared architectures, AI could widen the digital divide in climate resilience, making rich nations smarter and poor ones more exposed.

Finally, AI’s own carbon footprint is rising. Data centres in the MENA must rely on renewables for both cooling and operations.^[18]

Charting a Smarter Future

To harness AI effectively across the MENA water–energy–food nexus, stakeholders should co-design meteorology–water–agriculture services that translate forecast probabilities into operational decisions. In this first step, collaboration is key. Meteorologists, data scientists, and AI engineers must co-design models that combine physics, data, and operational realism. Cross-sector partnerships between research institutions, governments, and the private sector are essential to translate innovation into action.

Second, open science must become the norm. Shared datasets, transparent benchmarks, and open-source models can democratise access and foster trust.

Third, capacity-building is critical. Many developing nations in the MENA need support to adopt and adapt AI tools, from computational infrastructure to technical training. The next generation of meteorologists must be as fluent in machine learning as in thermodynamics. Finally, AI should serve not just as an early warning system but as an adaptive planner. From optimising reservoir operations to designing heat-resilient cities, AI can help societies prepare before the next shock hits. It must be embedded not at the periphery of policy but at its core.

Conclusion: From Prediction to Preparedness

AI will not stop the storms or the heatwaves, but in arid regions it can protect people, farms, and grids and conserve water used by the digital systems themselves. The measure of success is anticipatory action and prudent resource use across the WEF nexus. Therefore, AI can help us see the extremes coming sooner, understand them better, and respond more wisely. It offers not the illusion of control, but the possibility of foresight.

As the atmosphere grows hotter and more chaotic, our best defence is intelligence, not just artificial, but collective. A smarter planet begins with smarter choices: to integrate AI with science, to share knowledge openly, and to ensure that those most at risk are the first to benefit. The climate of the future is already here. The question is whether our tools and our will can keep pace with it.

Diana Francis is Professor of Atmospheric and Climate Science, Khalifa University, UAE.

Endnotes

^[1] Georgios Zittis et al., “Climate Change and Weather Extremes in the Eastern Mediterranean and Middle East,” Reviews of Geophysics 59, 2021: e2021RG000762, https://doi.org/10.1029/2021RG000762.

^[2] United Nations Economic and Social Commission for Western Asia, The Water, Energy and Food Security Nexus in the Arab Region, Beirut, ESCWA, 2015, https://www.unescwa.org/publications/escwa-water-development-report-6-water-energy-and-food-security-nexus-arab-region.

^[3] International Energy Agency, “Installed Desalination Capacities by Technology in the Middle East and North Africa, 2000–2035,” 2025, https://www.iea.org/data-and-statistics/charts/installed-desalination-capacities-by-technology-in-the-middle-east-and-north-africa-2000-2035.

^[4] Manzoor Qadir et al., “The State of Desalination and Brine Production: A Global Outlook,” Science of the Total Environment 657, 2019: 1343–1356, https://doi.org/10.1016/j.scitotenv.2018.12.076.

^[5] World Bank, Beyond Scarcity: Water Security in the Middle East and North Africa, Washington DC: World Bank, 2017, https://www.worldbank.org/en/topic/water/publication/beyond-scarcity-water-security-in-the-middle-east-and-north-africa.

^[6] Diana Francis et al., “From Cause to Consequence: Examining the Historic April 2024 Rainstorm in the United Arab Emirates through the Lens of Climate Change,” Climate and Atmospheric Science 8 (2025): 183, https://doi.org/10.1038/s41612-025-01073-1.

^[7] World Weather Attribution, “Heavy Precipitation Hitting Vulnerable Communities in the UAE and Oman Becoming an Increasing Threat as the Climate Warms,” April 29, 2024, https://www.worldweatherattribution. org/heavy-precipitation-hitting-vulnerable-communities-in-the-uae-and-oman-becoming-an-increasingthreat-as-the-climate-warms/.

^[8] Diana Francis, Ricardo Fonseca, and N. R. Nelli, “Key Factors Modulating the Threat of the Arabian Sea’s Tropical Cyclones to the Gulf Countries,” Journal of Geophysical Research: Atmospheres 127, 2022: e2022JD036528, https://doi.org/10.1029/2022JD036528.

^[9] Diana Francis and Ricardo Fonseca, “Recent and Projected Changes in Climate Patterns in the Middle East and North Africa (MENA) Region,” Scientific Reports 14, 2024: 10279, https://doi.org/10.1038/s41598-024- 60976-w.

^[10] N. R. Nelli et al., “Drivers and Trends of Summertime Convection over the Southeastern Arabian Peninsula,” Geophysical Research Letters 52, 2025: e2025GL118960, https://doi.org/10.1029/2025GL118960.

^[11] Remi Lam et al., “Learning Skillful Medium-Range Global Weather Forecasting,” Science 382, no. 6669 (2023): eadi2336, https://doi.org/10.1126/science.adi2336.

^[12] Suman Ravuri et al., “Skilful Precipitation Nowcasting Using Deep Generative Models of Radar,” Nature 597, 2021: 672–677, https://doi.org/10.1038/s41586-021-03854-z.

^[13] World Meteorological Organization, “Early Warnings for All (EW4All) Initiative Overview,” 2025, https://wmo.int/activities/early-warnings-all.

^[14] World Meteorological Organization, “Overview of the Early Warnings for All: Executive Action Plan 2023–2027,” 2025, https://wmo.int/media/magazine-article/overview-of-early-warnings-all-executive-action-plan-2023-2027.

^[15] World Meteorological Organization, “Black Sea and Middle East Flash Flood Guidance System,” 2025, https://wmo.int/governance/black-sea-middle-east-flash-flood-guidance-system.

^[16] Food and Agriculture Organization of the United Nations, “WaPOR: Remote Sensing for Water Productivity— Data Portal,” https://www.fao.org/in-action/remote-sensing-for-water-productivity/wapor-data/en.

^[17] K. A. Voss et al., “Groundwater Depletion in the Middle East from GRACE with Implications for Transboundary Water Management in the Tigris–Euphrates–Western Iran Region,” Water Resources Research 49, no. 2 (2013): 904–914, https://doi.org/10.1002/wrcr.20078.

^[18] International Organization for Standardization and International Electrotechnical Commission, ISO/IEC 30134-9: Information Technology—Data Centres Key Performance Indicators—Part 9: Water Usage Effectiveness (WUE), Geneva: ISO/IEC, 2022, https://www.iso.org/standard/77692.html.