Data Ethics

Our Privacy Policy explains what we do with personal data and what the law requires of us. This statement is different: it sets out the principles we hold ourselves to for all the data we work with — soil surveys, sensor readings, field boundaries, imagery, public datasets and the models we build from them — including data that is not personal but still carries real consequences for the people whose land it describes.

These principles draw on widely recognised frameworks, including the UK Government’s Data Ethics Framework, the OECD Principles on Artificial Intelligence, the FAIR data principles (Findable, Accessible, Interoperable, Reusable) and the EU Code of Conduct on Agricultural Data Sharing by Contractual Agreement.

1. Clear public benefit

Every use of data has a defined purpose that we can state plainly and that serves a legitimate benefit — better decisions for a farm, a stronger evidence base for sustainable land management, or research that is shared openly. We do not collect or retain data “just in case”, and we do not repurpose data for goals the people who provided it would not expect.

2. Data providers keep control

Farmers and land managers own the data about their land. When you work with us you decide what is collected, how it is used, whether it contributes to aggregated or research datasets, and when it is returned or deleted. We make those choices easy to understand and easy to change, and we honour them.

3. Proportionate collection

We gather the minimum data needed for the stated purpose, at the coarsest resolution that still does the job. Precise field locations, yield figures and other commercially or personally sensitive details are treated as confidential, access-controlled, and aggregated or anonymised wherever a task does not require the raw values.

4. Transparency about methods

We are open about where our data comes from, how our models work, what assumptions they make, and how confident we are in a result. Modelled figures are labelled as modelled and reported with their uncertainty. If a number would materially change a decision, we explain how it was produced.

5. Human judgement stays in the loop

Models inform decisions; they do not make them on their own. A qualified person reviews the outputs that carry weight — management recommendations, compliance reports, carbon and biodiversity claims — before they are relied on, and there is always a route to question or appeal a result.

6. Fairness and representativeness

Soil and climate vary enormously between regions, and a model trained mostly on one context can mislead in another. We check where our data and models are well supported and where they are not, we say so, and we work to broaden coverage rather than quietly extrapolating beyond the evidence.

7. Security and long-term stewardship

Data is encrypted in transit and at rest, held only as long as it serves its purpose, and looked after so that it stays accurate and usable over time. We keep provenance records so that any figure can be traced back to its source and method.

8. Sharing value back

Where data can advance soil science without harming the people it came from, we favour publishing methods and aggregated results openly, and returning insight to the landholders and communities who made it possible. Research partnerships are set up so that the benefit is shared, not extracted.

9. Accountability and review

We treat these principles as commitments we can be held to. They are reviewed as our work, our tools and the standards around them evolve, and concerns can be raised with us directly at any time.

10. Contact

To ask about how we handle a particular dataset, or to raise a data-ethics concern, contact ethics@soilwater.com.