What is Sentinel?
Sentinel are a groups of satellites launched by the European Space Agency to collect a range of climate and environmental data which can be used by research and commercial organisations.
Performance Live
Sustainable, scalable, cost-effective farm benchmarking
Using satellite data to benchmark farm performance
Integrating satellite data with farm records can reduce the need for manual data collection making farm benchmarking simpler and more accessible.
Farm benchmarking aims to improve understanding of crop performance and close performance gaps in arable farming. You can benchmark within one farm business but in most cases benchmarking is about comparing and learning from your peers.
Initiatives like the Yield Enhancement Network and Farm Bench have demonstrated that benchmarking works. However, they struggled to find a sustainable operating model. Performance Live is an ADOPT-funded project that aims to create a benchmarking system that is easier to run, sustain and is scalable. Using Sentinel satellite data may simplify the data collection process and reduce the need for manual collection. However, significant challenges remain to integrate a wide variety of data sources into comparable farm metrics.
The project brings together two lead farmer groups of around five farms each, based in the South and Midlands. Satellite data specialists Assimila are providing technical input alongside leading farm business consultancy Ceres Rural with their extensive exeperience of farm operations and performance.
Farm Benchmarking
Analysing differences within and between farms helps to improve productivity in underperforming areas.
Operating Model
Previous schemes have succeeded in benchmarking performance but not built a sustainable operating model.
Satellite Data
Field level data from Sentinel satellites can reduce requirement for on the ground data collection.
Integrations
Numerous systems are generating farm data but sharing and combining information from different systems is a technical and administrative challenge.
Farmer Centred
Benchmarking is about sharing facts and experience. Metrics and data are the starting point for a discussion not the end in itself.
Who is involved?
Lead farm – AH Oliver & Son
Farmer and agronomist Will Oliver is leading the Midlands benchmarking group. The family farm is located in Leicestershire.
Lead farm – Passmore Brothers
Farmer David Passmore is leading the Southern benchmarking group. He is based in Oxfordshire.
Evaluator
Benchmarking, dashboards, COM-B evaluation
Data Provider
Satellite crop metrics – ACROPALIS
Managing Farm Data
There is no shortage of data amongst the project participants (satellite time series, soil datasets, 10 years of YEN farm records, farm management software records, weather date). The problem is almost none of it can be shared or combined without significant friction.
Much farm data is compliance-grade, not analysis-grade. Farm management software is used primarily for regulatory compliance, not management.
The data that flows through API connections - where it flows at all - typically has around 10% error rate and needs significant cleaning before it is usable for benchmarking. Only 30-50% of farmers connect via API even where connections are available; the rest revert to manual entry or simply don't participate.
Interoperability between farm data systems requires either common standards or translation layers. Thehe industry has been working on this problem for a decade and has made real progress, but seamless interoperability remains elusive.
Hestia, an open agri-environmental database developed at the Oxford Martin School is used by Defra as a data standard across multiple funded projects. The FIG project (Farm Data Infrastructure and Governance, is working on this at industry level. AHDB's Farm Data Exchange – targeting procurement by end of 2026 – is a permissions and data routing infrastructure (not a database) designed to give farmers control over how their data flows between organisations.
In many cases, farmer data collected under one project cannot be used beyond its original scope without going back to each individual farmer. Even where organisations hold valuable datasets built up over years, they cannot open them without a permissions exercise that is itself a major undertaking. This applies to research datasets, farm management software records, and benchmarking databases alike.
Rather than simple indices like NDVI, Assimila work with biophysically meaningful parameters – Leaf Area Index, canopy cover, chlorophyll content, fraction of photosynthetically active radiation. These are tracked as time series and converted into summary metric such as, overwintering canopy, maximum canopy, area under the curve, timing of green-up and senescence.
Radar data (unaffected by cloud cover) adds biomass and structural information. Weather data is similarly converted into meaningful metrics like growing degree days, vernalisation, timing of wet and dry periods. Yield modelling and in-season forecasting are also possible, with uncertainty estimates.
Farmers will share data where they trust the recipient and can see a direct benefit to themselves. The red line is commercial use – specifically, the fear that their data will be used to benefit a compant not their own farm business, or used against them in pricing. Policy documents and data principles don't close this gap on their own. The organisations with the highest farmer trust are those perceived as non-commercial – levy bodies and farmer-led networks have a structural advantage here that commercial platforms cannot easily replicate.
The Social Element
Focus on the human side of benchmarking not just technology
In any project looking to test new technology, there is a danger that the technical aspects take precedence over how the users implement it. The benchmarking groups with the lead farmers are as important as the data infrastructure.
The benchmarking groups with the lead farmers are as important as to the project's success as the data infrastructure. Peer-to-peer farmer learning works best in person. Data and analysis are inputs to that conversation, not substitutes for it.
The outputs of any benchmarking system need to be robust enough to provoke meaninfgul conversations. But not seek to replace discussion with an entirely data driven version of benchmarking which ends up focused on the things that are easy to measure and can miss some of the qualitative assessments farmers make during the season.
Recording and analytical software are making it easier to incorporate qualitative evaluations into datasets but it is still no replacement for human contact, especially in a job which is seeing more and tasks replaced with technical solutions.
Key Questions for Performance Live
Which metrics matter?
Hundreds of satellite-derived, weather and soil metrics are potentially available. The project aims to identify which ones are most useful for predicting performance
How to encourage information sharing?
Even farmers who engage with benchmarking may resist data sharing when they don't understand exactly how it will be used. Clarity about data use and ownership is crucial.
Within farm or cross farm benchmarking?
There is typically 50% yield variation within a single farm. Understanding the variation on one farm can avoid issues around data sharing and trust but does it provide enough useful insights?
How to make a benchmarking system that is sustainable in the long-term?
YEN and Farm Bench both built genuine value and both struggled to sustain it commercially. Performance Live has ADOPT funding for 15 months and then needs a continuation plan.
How to combine different data sources?
Farms are generating more data than ever but quality and standardisation between different systems is a challenge alongside
Who owns the IP of datasets?
When multiple organisations contribute data to a shared analysis, it's important to define ownership and usage of the combined dataset.