-- A wheat grower can spend thousands of pounds on nitrogen fertiliser and still lack a clear answer to a basic question: does the crop need it?
The uncertainty is not for want of experience. Nitrogen decisions are made across fields with different soils, drainage and yield potential, in seasons that can turn from wet to dry in weeks. By the time deficiency is obvious to the eye, the best opportunity to protect yield or grain protein may have passed. Applying too much can be just as costly: it ties up working capital, may not increase output, and leaves nitrogen vulnerable to leaching or loss to the atmosphere.
Messium, a UK-based space agritech company, is trying to close that information gap. It uses hyperspectral satellites to estimate nitrogen concentration in wheat, then combines those readings with models of crop growth, weather, soils and farm management. Its software tells growers when a crop is likely to become deficient, what that could mean for yield, protein and margin, and which fertiliser rate and timing are most likely to pay.
That is more ambitious than a prettier satellite map. Messium is turning orbiting sensors into a live economic model of a field.
Beyond a green map
Most crop-monitoring imagery is multispectral: it collects a small number of unique wavelengths and uses only two, generating data points such as NDVI. Those images show variation in greenness. They are useful, but do not directly reveal the biochemical state of a plant or precisely how much fertiliser it takes up.
Hyperspectral imaging divides reflected light into hundreds of unique wavelengths. Messium works with that richer signal and can detect changes associated with chlorophyll, proteins and other plant constituents that move with nitrogen status.
The challenge is not simply to point a better camera at a crop. Satellites look through an atmosphere affected by water vapour and ozone; clouds interrupt the signal; and different sensors capture different wavelength bands at different resolutions. Messium standardises data from more than 200 multispectral satellites and more than 14 hyperspectral satellites. It combines imagery with atmospheric correction, weather, soil type and moisture, sowing date, row spacing, management history and historical images.
The company was also selected by the European Space Agency to support calibration and validation (Cal/Val) of ESA’s new FLEX hyperspectral mission, proving its technical expertise.
Messium has built its models on a proprietary dataset of 36,000 laboratory-tested wheat samples paired with hyperspectral satellite overpasses. In a blind test organised with Eurofins Agro Testing, Messium predicted nitrogen levels and biomass at geo-referenced sample locations without seeing the laboratory results; with accuracy above 85%.
AI constrained by biology
Messium’s approach brings together two forms of intelligence:
“I think it’s fair to say we’re at the frontier of the fusion of world models and mechanistic/process-based models,” says Rory Geeson, Messium’s Head of Machine Learning.
A world model learns relationships from large volumes of data: how spectral signals, weather, soil conditions and crop outcomes move together. A mechanistic model encodes what agronomists know about crop development, water stress, nitrogen uptake and temperature. Either can fall short alone. A crop model can drift away from the condition of a particular field without fresh measurements; a purely data-driven model can find patterns without knowing whether they remain biologically plausible in another season.
Messium uses frequent nitrogen observations to recalibrate crop-growth models through the season. The result is a digital twin of each field: a model that reflects its current state and forecasts what could happen next.
“Normal satellite imagery gives you a broad view of how a crop looks. Our objective is to understand how it is functioning,” says Vishal Soomaney Vijaykumar, Messium’s co-founder and CTO. “We have to reconcile measurements from a large and changing set of sensors with plant chemistry, atmospheric physics and what we know about how wheat grows.”
The decision is economic as well as agronomic. The system considers variety, yield potential, fertiliser cost and grain contracts. A field may be heading for nitrogen shortage but no longer have the moisture or potential to justify extra spending. Another may still have a credible route to a higher-protein milling premium.
Multidisciplinary Team
Messium’s technical team includes PhDs in atmospheric correction, crop genetics, crop modelling, machine learning and physical chemistry. Its advisers add experience in crop physiology, nitrogen nutrition, precision farming, digital agriculture, crop modelling and remote sensing.
For Messium, the team is part of the product. It combines the ability to learn from incomplete real-world data with the scientific judgement to recognise when an answer does not fit the underlying biology, making this capability possible for the first time.
Proof in farm margins
The environmental and economic case for precision nitrogen is clear: global nitrogen-use efficiency is estimated at 45%, meaning a large share of fertiliser is not taken up by crops and lost to air or water.
The company says three years of UK and New Zealand trials have consistently improved profits against established commercial practice. Its 2025 New Zealand programme reported yield improvements in 75% of trials and a median gross-margin benefit of NZ$115 per hectare - NZ$34,500 across the average farm Messium serves. ADAS independently validated results.
Reducing fertiliser use is not enough if growers feel they are leaving money on the table; equally, boosting yield is no better if the extra inputs cost more than the return. By measuring margin alongside fertiliser savings and yield, Messium meets farmers on the terms by which they judge agronomy products.
Messium has moved from individual trial fields to farm-scale delivery. In 2026, it worked across ~4,300 fields on around 200 UK farms and is working with roughly 20 in New Zealand.
From wheat nitrogen to all of agriculture
Agricultural software has a distribution problem: farmers are fragmented, while trust and purchasing decisions sit with established agronomy, merchant and machinery networks.
Messium’s B2B2F (business-to-business-to-farmer) approach supplies nitrogen intelligence to fertiliser companies, agronomy groups and precision-agriculture platforms, rather than asking growers to adopt another standalone tool.
Partners can embed recommendations in systems, workflows and advisory relationships farmers already use.
The UK Agri-Tech Centre supports Messium with multi-site trials, data collection and farmer-led product development.
Messium also works with Frontier and Hutchinsons to reach more growers through established agronomy channels.
This creates a data flywheel: more partner-delivered fields mean more soil types, varieties, weather patterns and management outcomes, improving validation and recommendations over time.
Messium’s immediate focus remains wheat and nitrogen: deepening its UK presence, activating Australia and servicing commercial demand in France, Germany and Eastern Europe.
The larger proposition is straightforward: repeatedly combining chemical measurements from space with process-based models and farm economics gives Messium a route into other crops, nutrients and uses across the agricultural food system, answering the question every grower faces throughout the season: what does this crop need now, and will acting on it make money?
Contact Info:
Name: gianamrco
Email: Send Email
Organization: xraised
Website: http://xraised.com
Release ID: 89204328
Google
RSS