From Soil to Satellite: A Practical Farm Data Guide

Published by Faminga. Prepared with AI-assisted research and drafting; sources checked on 5 September 2026. Product descriptions are attributed to their providers. Examples and evaluation checklists are editorial guidance, not results of a comparative field trial.

Farm data becomes useful when it helps someone make a decision and later check the result. A soil reading, satellite image or crop photograph is a starting point. To support a smallholder farmer, it must be connected to the right field, date, crop and action.

This guide provides a practical record structure for farmers, advisers and cooperatives. It is an editorial framework, not a report of measured outcomes from a Faminga farmer cohort. The purpose is to make observations comparable and decisions easier to review, whether records begin on paper or in an app.

Start with one question the farmer wants answered

A broad instruction to “collect more data” creates work without defining its value. Choose a question instead: which irrigation visits were necessary, how quickly did an adviser respond to a crop problem, or which plot's harvest record is incomplete?

That question determines what to record. If the aim is to examine water use, soil moisture alone is insufficient; the record also needs information about water delivered and the relevant area. If the aim is faster crop support, capture the time the issue was reported and the time a useful response reached the farmer.

Agree on the decision before selecting the technology. This prevents a dashboard from becoming a collection of measurements that nobody is responsible for using.

A minimum field record

Begin with a field identifier that stays the same throughout the season and a name the farmer recognises. Associate it with the crop, planting date and area, including how that area was measured. Keep the person responsible for the field distinct from the person who entered the record.

For each observation, keep these elements together:

• Where: field and, where relevant, the location within it.
• When: observation time, plus entry time if different.
• What: value, image or description, with units.
• How: instrument, method or source.
• Quality: known limitations, missing information or uncertainty.
• Follow-up: who should inspect or respond.

A record with an honest missing value is more useful than a fabricated estimate that looks measured. If an estimate is necessary, label it and retain the method used to produce it.

Keep observations, interpretations and actions separate

An observation states what was recorded. An interpretation explains what it may mean. An action records what somebody actually did. Combining all three into a single “recommendation completed” field makes later review difficult.

Consider an illustrative case: a satellite layer flags a change in part of a plot. The adviser requests a visit. The farmer finds a blocked irrigation line and records a repair. Those are three separate events. The satellite did not directly diagnose the blockage; it helped identify a place worth checking.

Preserve the original observation even if the interpretation changes. A correction should clarify the record, not erase the path by which the team reached its decision.

Match timestamps before combining soil and satellite data

A sensor reading from this morning and an image from two weeks ago are not simultaneous evidence. A dashboard should display both dates rather than present everything beneath a single “live” label.

Satellite vegetation measures also differ from soil observations. USGS explains NDVI as a vegetation greenness indicator. It is not interchangeable with a moisture reading taken below ground. Our satellite guide explains the index and spatial-resolution issues.

To connect these sources, record the image footprint and the sensor position. Ask whether they describe the same management area. If not, keep that limitation visible in the interpretation.

Make activity records specific enough to review

For irrigation, record the field, date and measured volume if available. If only pumping time is known, record time and label the delivery rate assumption separately. For input applications, capture the product identity, quantity, unit and treated area, alongside the responsible person's instructions.

For crop problems, retain the original photographs, symptom description and relevant history. Store the app response separately from an agronomist's assessment. Our disease detection guide explains why these distinctions matter.

For harvesting, agree how weights are measured and whether they refer to the entire plot, one picking or only the marketable portion. Consistent definitions make seasonal comparisons meaningful.

Measure whether the information was used

A device reporting data is one operational result. A farmer understanding an alert is another. An action taken and a subsequent outcome are further steps. Keep those stages distinct when reviewing a service.

A cooperative can begin with a simple monthly review: how many reported issues received a response, how many responses were understood, and how many actions were recorded? Examine a few incomplete cases to find the cause. The obstacle may be unavailable labour, missing inputs or an unclear message rather than a software fault.

Do not assume that registration, training attendance and active use are the same measure. Define each one and its reporting period. A person can attend a demonstration without adopting the service.

Compare outcomes without overstating causation

A change between seasons can be real without being caused entirely by a new technology. Weather, crop variety, planting date, area and other management decisions may also differ. Record those factors before attributing a result.

For an illustrative water comparison, a reduction from 10,000 to 8,000 measured litres is a 20% reduction in that recorded volume. It does not, by itself, show equal crop performance, equal area or the reason for the reduction. Keep the arithmetic separate from the causal claim.

Before publishing an impact figure, define the population, period, baseline, measurement method and exclusions. Explain whether it is a trial result, an estimate or a target. A smaller number supported by a clear method is more useful than an impressive figure nobody can reproduce.

Design for people and limited connectivity

Use names and units that farmers recognise, with a clear way to correct mistakes. Where phones are shared, establish who can see which records. Ask the provider how data can be exported and how access ends when a staff member leaves a cooperative.

For offline work, show which entries are saved on the phone and which have synchronised. Check that a delayed upload preserves the original observation date. Read our offline farming guide for an acceptance test.

Faminga's soil-to-satellite approach

Faminga's public platform vision connects field information with farm management. The practical standard for that vision is a record that lets the farmer or adviser answer four questions: what happened, what evidence was available, what was decided and what followed?

Begin with one field and one decision. Make the records understandable, then repeat the process before expanding. Our precision agriculture guide for Rwanda shows how to organise a first-season trial. Better farm data is the information that helps people act and learn, not simply the largest collection of numbers.