Why Lincolnshire's fields carry unusual stakes
Stand on the edge of a field near Spalding or Holbeach in late summer and the geometry is almost disorienting: flat black earth stretching to a horizon interrupted only by drainage channels, pylons, and the occasional line of leeks or cabbages heading toward their supermarket shelf. Much of this land sits below sea level. The only reason it can be farmed at all is a network of pumping stations, sluices, and drains that have been holding back the North Sea — and the inland rivers — for centuries.
The scale of what grows here is easy to underestimate. Greater Lincolnshire produces roughly one third of England's fresh vegetables, a figure confirmed both by the NFU's Delivering for Britain report and the Greater Lincolnshire Local Enterprise Partnership. The same flat terrain also yields one fifth of England's potatoes and one fifth of its sugar beet. Total agricultural output across the region exceeded £2 billion in 2019, accounting for 12% of England's entire food production — more Grade 1 agricultural land than any other LEP area in the country.
Against that productivity sits a stark warning from the Climate Change Committee: these same low-lying Fenland areas face a projected 16-fold increase in flood risk. The soils themselves — dark, peaty, and among the most fertile in Europe — are also among the most precarious. They are thin, irreplaceable, and dependent on drainage infrastructure that was never designed for the rainfall patterns now emerging.
This combination of national food-security importance and acute climate exposure is the reason environmental forecasting technology carries unusual practical weight in Lincolnshire. The question of whether AI can help farmers here read and respond to water, weather, and soil conditions more accurately is not a speculative technology story. It is a question with direct consequences for what ends up on plates across England.
The environmental forecasting toolkit Lincolnshire farmers can access
Five publicly named tools form the core of what is currently available to growers and councils in the region, though they vary considerably in scope, funding model, and how directly they have been documented for Lincolnshire use.
The Environment Agency's ALERT system draws on LIDAR survey data and satellite imagery to map pollution pathways and flood risk at fine spatial resolution — directly relevant to the low-lying drainage networks that run beneath fen fields. Farmscoper, developed by ADAS, takes a different angle: it models the farm as a pollution source, letting growers test more than 100 mitigation measures — from cover cropping to buffer strips — and quantify their likely impact. Primarily a compliance and grant-application aid, it is national in scope rather than designed specifically for Lincolnshire.
E-Planner, produced by the UK Centre for Ecology and Hydrology, offers land suitability mapping at five-metre resolution, fine enough to distinguish drainage and soil conditions across individual field sections. The Met Office Food and Farming Service goes further into agronomy, providing subscribing growers with bespoke, crop-level climate forecasts for the near term — though access depends on a commercial subscription rather than public funding. Rounding out the suite, the Greater Lincolnshire Nature Partnership's Precision Farming portal integrates geodiversity and soil mapping data into everyday arable planning decisions.
These tools range from freely accessible government services to subscription products, and the evidence on how widely any of them are used by Lincolnshire growers in practice remains limited.
How AI is changing on-farm water decisions
For a vegetable grower managing irrigation across hundreds of acres, the traditional approach has been largely reactive: observe the crop, check the forecast, open the system when conditions look dry. AI-enabled platforms are beginning to change that sequence.
By combining continuous readings from in-field soil moisture sensors with satellite imagery and hyper-local weather modelling, these systems generate dynamic irrigation schedules calibrated to individual crop types and adjusted in near real-time as conditions shift. Water use reductions of up to 50% have been reported by proponents of the technology. That figure should be treated with some caution — independent, large-scale evaluation specific to Lincolnshire growers has not, to date, appeared in the public record — but it reflects a meaningful directional shift in how water is applied.
A related development involves private on-farm reservoirs, which a growing number of growers are expanding and pairing with AI-managed intake and release systems. The logic is dual-purpose: during the East of England's increasingly dry summers, a managed reservoir sustains crops when demand outpaces rainfall. During high-rainfall periods, the same infrastructure can absorb excess water, reducing the pressure on local river and drainage networks downstream. This is a meaningful design shift — from a single-function storage asset to something closer to an adaptive buffer.
The UK Government's Water Management Grant has been one of the main funding routes enabling growers to invest in reservoir capacity and the sensors and control systems that make dynamic management viable — though access is not universal, particularly for smaller holdings.
AI at catchment scale: the Internal Drainage Boards and beyond
Internal Drainage Boards are not a recent innovation. These statutory bodies — Lincolnshire has several, covering its low-lying fen areas — have managed the water table across the county's flatlands for centuries, operating the pumps, sluices, and channels that make arable farming possible below sea level. What is changing is the information layer they work with.
The Smart Catchments Project, led by the Black Sluice Internal Drainage Board in partnership with the South Lincolnshire Water Partnership, uses remote sensors and AI models to automate floodwater management across its catchment. Its central challenge mirrors one familiar to every fen farmer: the same drainage network that must empty quickly after heavy rain also needs careful management during dry periods to retain water in the system. No detailed independent evaluation of the scheme has yet been published, but it is reported as active and developing.
Project Groundwater operates a complementary network: telemetry devices installed in rivers and streams send water-level readings every 15 minutes to AI-supported dashboards, which automatically alert local councils and community leaders when levels approach critical thresholds. The granularity matters — 15-minute updates can compress the effective warning window from hours to minutes for communities that have historically had little advance notice.
These schemes sit within a broader institutional architecture. Lincolnshire County Council acts as Lead Local Flood Authority, coordinating alongside Anglian Water and the Environment Agency. AI tools slot into this structure rather than replacing it. They cannot resolve the underlying governance question at its heart: whether water should be moved away from fields faster or held back longer. That remains a negotiation between farming interests, ecological requirements, and downstream communities — one that better data may inform but will not settle.
The University of Lincoln's role in applied agri-tech
Sitting within the broader agri-tech ecosystem is the University of Lincoln, whose Lincoln Agri-Robotics (LAR) centre has become the region's most concentrated source of applied AI for farming since its establishment in 2019. Positioned as the world's first global centre of excellence in agri-robotics, LAR has drawn over £100 million in delivery funding across 149 projects with 250 partners. Its research spans three practical areas: autonomous harvesting using AI computer vision, precision crop-protection drones that can detect disease or pest damage before symptoms are visible to the naked eye, and automated crop phenotyping — analysing plant traits at scale to support the breeding of more resilient varieties.
The clearest evidence of a commercially deployed tool from this ecosystem is FruitCast, a spin-out launched in autumn 2021. Using camera systems that scan millions of berries daily and combining those counts with external weather data, FruitCast delivers yield forecasts at 83–90% accuracy up to six weeks ahead. It currently covers more than 20% of the UK's strawberry crop.
The honest caveat is that Lincolnshire's dominant crops — brassicas, root vegetables, onions, and cereal arable — are not what FruitCast currently serves. Documented AI yield-forecasting equivalents for those sectors have not been publicly established at comparable scale.
Separately, University of Lincoln researchers are using AI modelling to assess groundwater recharge rates and how fen soils may adapt to increasing salinity driven by sea-level rise — work that addresses a slow-moving but serious long-term threat to the county's vegetable-growing capacity.
What the tools don't yet answer for most growers
Much of the evidence reviewed here describes tools that are genuinely in use — but it says relatively little about who, precisely, is using them, and at what scale. No publicly available data documents AI adoption rates among Lincolnshire vegetable growers as a whole, and no systematic cost evaluation for smaller farms has been published. The research base captures what is technically possible and what early adopters have reported; it does not yet tell us how far uptake extends beyond that group.
For a grower managing, say, 80 acres of brassicas or carrot ground near Holbeach, the practical question is not whether AI irrigation scheduling exists in principle — it is whether the hardware costs, subscription models, and connectivity requirements are workable at their scale. That question remains largely unanswered in the available evidence.
The navigation burden is also real. Accessing relevant support currently means engaging with several overlapping systems: the Water Management Grant, the IDB schemes, Environment Agency tools, and Met Office services each sit under different application processes and governance bodies. Knowing which tool applies to which problem, and which funding route covers which hardware, requires time and technical confidence that varies considerably across the farming population.
The technology, in most cases, exists. What the evidence does not yet confirm is whether the policy frameworks, extension services, and cost structures are in place to make it reachable for most growers rather than a well-resourced few.