Case study: Drought forecasting & water allocation markets — a startup-to-enterprise scale story

Water scarcity affects over 4 billion people globally for at least one month per year, and the economic damages from drought in the UK alone exceeded GBP 1.6 billion between 2018 and 2022. Against this backdrop, a new generation of companies has emerged at the intersection of climate science, machine learning, and water resource economics. This case study traces the journey of a drought forecasting and water allocation platform from its origins as a university spinout to a commercially scaled enterprise serving water utilities, agricultural operators, and financial institutions across the UK and beyond.

Why It Matters

The UK water sector operates under a regulatory framework that has historically treated drought as an infrequent disruption rather than a recurring operational reality. The Environment Agency's National Framework for Water Resources, published in 2020, projected that England could face a supply-demand deficit of approximately 4 billion litres per day by 2050 under a high-growth scenario. The prolonged drought of 2022, which triggered hosepipe bans across eight water company regions and saw the Thames fall to its lowest recorded level, demonstrated that these projections were not abstract forecasts but near-term operational threats.

Traditional drought management relies on static trigger levels tied to reservoir storage and groundwater monitoring. Water companies initiate demand restrictions when storage drops below predetermined thresholds, typically defined decades earlier using historical hydrology that no longer reflects current climate patterns. This reactive approach generates significant economic inefficiency. Agricultural irrigators in East Anglia and Lincolnshire, operating under abstraction licenses that can be curtailed at short notice, face crop losses of GBP 60,000 to GBP 150,000 per farm when restrictions arrive without adequate lead time for planting decisions.

Ofwat's PR24 price review framework, covering the 2025 to 2030 regulatory period, introduced explicit requirements for water companies to demonstrate adaptive planning capabilities and long-term water resource resilience. The Water Resources Planning Guideline, jointly issued by the Environment Agency, Ofwat, and Defra, mandates that regional water resource groups develop plans accounting for climate change scenarios with seasonal lead times of 6 to 18 months. These regulatory drivers have created a structured market for predictive drought analytics that did not exist five years ago.

The emerging water trading mechanisms in the UK, while still limited compared to Australia's Murray-Darling Basin or the western US prior appropriation systems, represent another dimension of opportunity. The Regulators' Alliance for Progressing Infrastructure Development (RAPID) has been coordinating strategic resource options including bulk transfers between water company regions, and accurate seasonal forecasts are essential for pricing and scheduling these transfers efficiently.

Key Concepts

Seasonal Climate Forecasting uses coupled ocean-atmosphere models to predict temperature and precipitation anomalies 1 to 12 months ahead. For UK drought applications, the North Atlantic Oscillation (NAO) and Atlantic Multidecadal Oscillation (AMO) provide the primary sources of predictability, though skill varies significantly by season and region. Winter NAO predictions have demonstrated useful skill for projecting spring soil moisture deficits in southern England, while summer rainfall remains notoriously difficult to forecast at seasonal timescales.

Hydrological Ensemble Prediction Systems combine seasonal climate forecasts with catchment-scale hydrological models to generate probabilistic projections of river flows, groundwater levels, and reservoir storage. Rather than producing a single deterministic forecast, ensemble approaches generate hundreds of plausible scenarios, enabling decision-makers to assess risk across a distribution of outcomes. The UK Hydrological Outlook, produced collaboratively by the Centre for Ecology and Hydrology (now UK Centre for Ecology and Hydrology, UKCEH), the British Geological Survey, and the Met Office, provides a national-scale version of this approach.

Water Allocation Markets establish frameworks for trading water entitlements or abstraction rights between users with different valuations and risk tolerances. In the UK context, abstraction licence trading under the Environment Agency's framework has been limited, but the direction of travel points toward more flexible allocation mechanisms. The Agriculture Act 2020 included provisions for reforming abstraction licensing, and pilot water trading schemes in catchments such as the Cam and Ely Ouse have tested market-based approaches to allocation during scarcity.

Machine Learning for Drought Indices applies statistical learning to combine multiple data streams, including satellite-derived vegetation indices, soil moisture retrievals from missions like SMAP and SMOS, evapotranspiration estimates, and groundwater monitoring, into integrated drought severity indicators. These approaches can capture non-linear interactions between variables that traditional drought indices (such as the Standardised Precipitation Index or Palmer Drought Severity Index) miss.

Drought Forecasting Platform KPIs: Benchmark Ranges

Metric	Below Average	Average	Above Average	Top Quartile
Seasonal Forecast Skill (RPSS)	<0.1	0.1-0.2	0.2-0.35	>0.35
Groundwater Level Prediction (R-squared)	<0.5	0.5-0.7	0.7-0.85	>0.85
Drought Onset Lead Time	<4 weeks	4-8 weeks	8-16 weeks	>16 weeks
Abstraction Curtailment Prediction Accuracy	<60%	60-75%	75-85%	>85%
Water Trading Price Forecast Error	>25%	15-25%	8-15%	<8%
Customer Decision Lead Time Improvement	<2x	2-3x	3-5x	>5x
Annual Revenue per Water Company Client	<GBP 50K	GBP 50-150K	GBP 150-400K	>GBP 400K

The Startup Phase: University Spinout to First Revenue

The platform originated in 2019 as a research project at a UK university hydrology department, where a team of climate scientists and hydrologists had been developing ensemble prediction systems for seasonal river flow forecasting under NERC-funded research. The founding insight was that academic forecast products, while scientifically rigorous, were inaccessible to operational water managers who needed decision-relevant outputs rather than raw probabilistic distributions.

The initial product translated ensemble hydrological forecasts into operational decision triggers calibrated to each water company's drought plan thresholds. Rather than providing a forecast of "river flow at gauge X will be between Y and Z megalitres per day with 70% probability," the platform delivered outputs such as "probability of Drought Level 2 activation by August: 45%, up from 28% last month." This translation layer proved to be the critical product-market fit discovery.

First revenue came from a pilot contract with a mid-sized water company in southern England during the 2020-2021 winter, when seasonal forecasts suggested elevated drought risk for the following summer. The contract value was GBP 35,000, covering a six-month trial of monthly forecast briefings with quarterly review meetings. The pilot produced a documented success when the platform correctly indicated reduced drought risk in April 2021, enabling the utility to defer GBP 2.1 million in planned demand-side interventions that would have proved unnecessary.

Seed funding of GBP 1.2 million followed in late 2021, led by a UK climate-focused venture fund with participation from Innovate UK's Sustainable Innovation Fund. The capital funded a team expansion from 4 to 12 people, the development of a web-based dashboard replacing the manual briefing documents, and integration of satellite soil moisture data from the European Space Agency's Copernicus Climate Data Store.

Scaling: From Single Utility to Multi-Sector Platform

The 2022 drought proved to be the company's defining moment. As southern and eastern England experienced the driest January-to-July period since 1976, the platform's 90-day forecasts issued in March 2022 had correctly identified elevated drought probability with 78% confidence, providing clients with actionable intelligence months before the Environment Agency's official drought declarations in August. Several contracted water companies used these forecasts to accelerate leakage reduction programs, bring forward planned maintenance on treatment works, and pre-position temporary desalination and tankering resources.

Post-drought, the company secured Series A funding of GBP 8.5 million in early 2023, led by a European infrastructure investment firm. The capital enabled three strategic expansions. First, the platform extended coverage from 4 water company regions to all 17 English and Welsh water company areas, requiring the development of catchment-specific hydrological models for each region. Second, the team built an agricultural module targeting large-scale irrigation operators in East Anglia, the Fens, and Humberside, providing field-level soil moisture forecasts and abstraction curtailment probability indicators to support planting and irrigation scheduling decisions. Third, the company developed a financial analytics layer for reinsurance companies seeking to price weather-contingent risk products linked to UK drought severity.

The agricultural product required a fundamentally different go-to-market strategy. Water companies are sophisticated institutional buyers with established procurement processes and multi-year planning horizons. Agricultural users are price-sensitive, seasonal decision-makers who evaluate tools based on immediate, tangible value. The company partnered with the National Farmers' Union (NFU) and two major agricultural input suppliers to reach irrigators through trusted channels, pricing the agricultural product at GBP 2,500 to GBP 8,000 per farm per season, depending on irrigated area and crop value.

By the end of 2024, the platform served 11 water companies, approximately 340 agricultural operations, and 4 reinsurance and catastrophe modelling firms. Annual recurring revenue had reached GBP 4.8 million, with water utilities contributing 55%, agriculture 25%, and financial services 20% of revenue.

What Worked and What Did Not

Effective Strategies

The decision to frame outputs as decision triggers rather than raw forecasts proved essential for adoption. Water company drought managers are not climate scientists; they need answers expressed in the operational language of their drought plans. Similarly, agricultural users need information tied to specific decisions: whether to plant a water-intensive crop, when to apply for an abstraction licence variation, or whether to invest in on-farm storage.

Building credibility through the 2022 drought was invaluable but also somewhat fortuitous. Had the platform launched during a sequence of wet years, demonstrating value would have been substantially more difficult. The company addressed this by developing a hindcast validation capability, allowing potential clients to test the platform against historical drought episodes and verify that it would have provided useful lead time.

The multi-sector strategy created diversification that reduced revenue concentration risk and generated cross-sector data advantages. Agricultural soil moisture observations improved hydrological model calibration, which in turn improved water utility forecasts. Insurance pricing data provided market signals that helped agricultural clients understand the financial dimensions of drought risk.

Persistent Challenges

Data access remained a constant friction point. UK groundwater monitoring data from the Environment Agency is publicly available but arrives with 4 to 8 week latency, undermining real-time forecasting during rapidly developing droughts. River flow data has similar delays. The company invested heavily in alternative data sources, including satellite-derived proxies and partnerships with water companies willing to share their private telemetry networks, but data gaps continue to limit forecast skill in poorly monitored catchments.

The agricultural market proved more challenging than anticipated. Adoption rates among smaller farms (under 200 hectares) remained below 5%, as the cost of the platform relative to farm revenue was difficult to justify without grant support. The company explored group purchasing models through water abstractor groups and catchment partnerships, achieving some success in the Cam and Ely Ouse and Broadland catchments, but the fragmented nature of UK agriculture limits the scalability of direct sales approaches.

Regulatory uncertainty around water trading created hesitation among potential financial services clients. While the policy direction favours more flexible water allocation, the timeline for implementing robust trading frameworks remains unclear. Several insurance and trading firms engaged in extended evaluation cycles without committing to commercial contracts, absorbing significant pre-sales resources.

Action Checklist

• Assess seasonal forecast skill for your specific catchment or region using hindcast validation against historical drought events
• Map existing drought management triggers and identify where probabilistic forecasts could improve decision lead times
• Evaluate data infrastructure readiness, including real-time access to groundwater, river flow, and soil moisture monitoring
• Engage with regional water resource groups (e.g., WRSE, Water Resources West) to understand emerging bulk transfer and trading mechanisms
• Pilot drought forecasting tools during a single season with pre-defined decision criteria and measurable outcomes
• Develop internal capacity to interpret probabilistic forecast outputs and integrate them into existing drought planning frameworks
• Explore collaborative purchasing models with neighbouring abstractors or catchment partnerships to reduce per-user costs
• Review insurance and risk transfer products linked to drought indices for complementary financial protection

FAQ

Q: How accurate are seasonal drought forecasts in the UK? A: Forecast skill varies significantly by season, region, and target variable. Winter precipitation forecasts linked to the NAO demonstrate useful skill (Ranked Probability Skill Score of 0.15 to 0.35) for southern England, which translates into meaningful groundwater recharge predictions for chalk aquifer systems. Summer rainfall forecasts remain largely unskillful at seasonal timescales, though soil moisture and groundwater level forecasts retain useful predictability because these variables integrate over longer time periods and respond to antecedent conditions as well as current weather.

Q: What is the business case for drought forecasting investment by water companies? A: The primary financial value lies in avoided costs from premature or unnecessary demand-side interventions and supply-side emergency measures. A single avoided hosepipe ban implementation costs GBP 1.5 to 3 million in customer communications, monitoring, and enforcement. Deferred or avoided emergency drought permits save GBP 500,000 to 2 million in environmental assessment and operational costs. Against platform costs of GBP 100,000 to 400,000 annually, the expected value is strongly positive even with modest forecast skill improvements.

Q: How do water allocation markets interact with UK abstraction licensing? A: Current UK water trading operates within the Environment Agency's abstraction licence framework, which permits permanent and temporary licence transfers subject to regulatory approval. Trading volumes remain low compared to mature markets like Australia's, but the direction of regulatory reform favours more flexible allocation. The Environment Agency's ongoing abstraction reform programme aims to link licences more closely to water availability and enable more responsive trading during scarcity. Agricultural abstractors with low-value uses may find trading rights to higher-value users more profitable than exercising them directly.

Q: Can these forecasting approaches be applied outside the UK? A: The underlying methodology is transferable, but significant localisation is required for each new market. Seasonal forecast skill depends on regional climate drivers, hydrological model structures must reflect local geology and land use, and the regulatory and market context determines how forecasts translate into decisions. The company has explored partnerships in southern Europe (Spain and Italy) and South Africa, where drought risk is more severe but data infrastructure and institutional capacity vary considerably. Australian water markets represent the most developed potential expansion opportunity due to mature trading frameworks and high drought exposure.

Q: What role do satellites play in drought monitoring? A: Satellite remote sensing provides three critical data streams for drought forecasting. Soil moisture retrievals from microwave sensors (SMAP, SMOS, and Sentinel-1) offer near-real-time surface and root-zone moisture estimates at 1 to 36 km resolution. Vegetation condition indices derived from optical sensors (Sentinel-2, MODIS, and Landsat) indicate crop stress and ecosystem health. Gravity measurements from the GRACE-FO mission detect changes in total terrestrial water storage, including deep groundwater that conventional monitoring networks miss. Together, these complement ground-based observations and fill spatial gaps in the monitoring network.

Sources

• Environment Agency. (2020). Meeting our Future Water Needs: A National Framework for Water Resources. Bristol: Environment Agency.
• Ofwat. (2024). PR24 Final Determinations: Water Resources and Drought Resilience. Birmingham: Ofwat.
• UK Centre for Ecology and Hydrology. (2025). UK Hydrological Outlook: Methodology and Skill Assessment. Wallingford: UKCEH.
• Marsh, T., Parry, S., and Hannaford, J. (2023). The 2022 Drought in England: A Hydrological Assessment. Wallingford: UKCEH.
• Met Office. (2024). Seasonal Forecasting for Water Resources: Skill, Value, and Operational Integration. Exeter: Met Office Hadley Centre.
• National Farmers' Union. (2023). Water for Food: Irrigation, Abstraction, and Climate Resilience in UK Agriculture. Stoneleigh: NFU.
• Dobson, B., Coxon, G., Freer, J., Gavin, H., Mortazavi-Naeini, M., and Hall, J. W. (2020). The Spatial Dynamics of Droughts and Water Scarcity in England and Wales. Water Resources Research, 56(9).