Yes, AI crop monitoring works on a small farm without drones or sensors, because it reads free satellite images, but only for fields large enough for the picture to resolve. One leading service's plans work out at about $0.63 to $1 per hectare a month. Fields under about one hectare, narrow strips and covered crops give unreliable results.
The services do not diagnose anything. They colour a map of your field by how green and dense the crop looks, and AI turns that into alerts such as "the north-east corner is lagging". You still walk the field to find out why. Used that way, it tells a farmer where to look first, which is the real saving.
How it works without anything in your field
Satellites such as the Sentinel-2 pair photograph the same ground every few days. They record more than visible light: bands in near-infrared and others, which show how much healthy leaf is present. From those bands software calculates indices, the best known being NDVI (normalised difference vegetation index), a number that is higher where plants are green and dense and lower over bare soil or stressed crop.
Facts worth knowing, from the Sentinel-2 mission's own pages: the satellites sample some bands at 10 metres, some at 20 and some at 60, and the pair is designed to revisit the same spot every five days at the equator. The data is published free of charge. Everything a paid service adds is built on top of imagery like that: boundary drawing, cloud filtering, history, alerts and the AI summary. If you want a related view of the economics, Farmonaut pricing against FieldView Plus compares two paid options.
What "AI" adds
The AI layer does three jobs. It cleans the imagery (dropping clouded images, adjusting for the sun's angle), it compares this week's map with earlier ones and with the rest of your field, and it writes a plain-language note. Better products also compare your field's progress with a typical growth curve for the crop. None of that is a diagnosis; it is pattern comparison. A pale patch could be waterlogging, compaction, a nutrient shortfall, a poor seedbed, pests or a different soil type. The satellite sees the paleness, not the cause.
The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy nowThe size problem for small fields
A 10-metre pixel is a square of 10 by 10 metres, or 100 square metres. One hectare, 10,000 square metres, is therefore about 100 pixels. That is enough for a rough picture, and not enough for much detail.
| Field | Approx. pixels at 10 m | How useful |
|---|---|---|
| 20 ha arable field | about 2,000 | Good: clear patterns, trends, zone maps |
| 5 ha grass paddock | about 500 | Useful for growth rate and grazing cover |
| 1 ha plot | about 100 | Rough: edge pixels mix crop with hedge, track and ditch |
| 0.2 ha bed block | about 20 | Poor: mostly mixed pixels, little within-field detail |
| Polytunnel or glasshouse | any | Not useful: cover hides the crop |
The problem is the edges. A pixel that is half crop and half hedge averages the two, so a narrow field is mostly edge. Rules of thumb vary by product, and vendors do not all publish a minimum field size, so ask what the smallest field they support is and look at a map of your own smallest field before paying. A field that is long and thin, say 20 metres wide, has only two pixels across and is likely to be mostly mixed.
What works at a small scale
Fields of several hectares, grass, cereals and oilseeds in open ground, orchards with a continuous canopy and large vegetable fields all work. A farm with ten small fields that average 3 hectares is a different case from a market garden with beds a few metres wide: the first gets useful maps, the second would be better served by walking the rows and keeping a log, as in planning beds and margins for a market garden.
What it costs per hectare
Pricing is mostly per hectare per month, with plan tiers that cap the area. As an example, Farmonaut's pricing page (fetched in November 2026) lists plans from $4.99 a month for up to 5 hectares (about $1 per hectare per month) and $29.99 for up to 30 hectares, through $64.99 for 80 hectares, $224.99 for 300 and $949.99 for 1,500, with per-hectare rates falling from $1.00 to $0.63. It shows an annual option described as two months free, and says prices are in each plan's billing currency, so confirm the currency at checkout. FieldView, another well-known platform, lists Basic from $0 and Plus from $649 a year; its plan comparison puts satellite Field Health Imagery in Plus, so check what each plan includes before comparing.
A quick sum for a small farm
An illustrative 25-hectare arable farm in six fields. On a plan like the 30-hectare tier at $29.99 a month, running it for the 7 months from March to September costs 7 × $29.99, or about $210, which works out to $8.40 a hectare for the season. The same farm on a 5-hectare plan would not cover the area at all. An 80-hectare farm on the $64.99 tier pays about $0.81 per hectare per month, around $455 for 7 months, or $5.69 a hectare for the season.
Now the benefit side. The value is in decisions made earlier or avoided: a trip to the one field showing a problem rather than all six, a fertiliser or fungicide decision made with better information, a spot where drainage failed seen in time to act. If it saves two field trips of an hour a week across 24 weeks, that is 48 hours; at a modest $20 an hour, about $960. The figures are illustrations, not promises: if you already walk every field twice a week as a matter of routine for other reasons, the saving shrinks, and a free viewer might be enough.
Hidden costs
- Your time. Drawing field boundaries, reading the maps, deciding what to do. About an hour to set up, then 15 to 30 minutes a week.
- Plan caps. Paying for 30 hectares when you farm 12 raises your cost per hectare to $2.50 a month on that example.
- Add-ons. Variable-rate application maps, scouting tools or sensors may cost more than the base plan. Read what is included.
- Auto-renewal. Annual plans bill again in a year. Put a reminder in your calendar.
Blind spots: cloud, bare soil and misread maps
Cloud
Optical satellites cannot see through cloud. A nominal five-day revisit gives you, in a wet spring, perhaps one clear image in two or three weeks. Two things help: tools that show the date of the last clear image on every map, and the habit of looking at that date first. A map that is 17 days old describes the field 17 days ago.
Misreading a good-looking map
Imagine a grower who sees a pale band across a field of winter wheat and assumes nitrogen deficiency. They apply extra fertiliser and the band stays. A walk would have shown a compacted headland where roots had not penetrated. The lesson is to treat the map as a question, not an answer. Check the cause on foot with a spade, and photograph and date what you find.
Early season and bare soil
Early in the season, when plants are small, much of each pixel is soil. A vegetation index then reflects soil colour and moisture as much as crop, so maps can be noisy. Many products are most reliable once the crop has covered a fair share of the ground. If a tool gives confident verdicts on a field just drilled, be sceptical.
Treating all crops alike
A cereal's growth curve differs from a potato's, and grass changes after every cut or graze. If the tool does not know your crop, its "behind schedule" alerts may be wrong. Enter the crop and sowing date accurately, and record cuts and grazing dates so the sudden drop after a cut does not trigger an alarm.
One season on a 25-hectare farm, week by week
An illustrative arable farm has six fields: 8, 6, 4, 3, 2.5 and 1.5 hectares, a total of 25. The owner starts a monitoring plan in March. This is how it might go, with all figures invented to show the pattern.
Setting up (March)
The owner draws each field boundary, sets the crop and sowing date, and finds that the 1.5-hectare field is long and narrow, about 25 metres wide. Its map is blocky and mostly edge pixels. The decision is to treat it as "visit by eye" and rely on the other five fields, covering 23.5 hectares, for the maps. That is a sensible reading of the table above, made on day one instead of in June.
The first alert (late April)
The 8-hectare field shows a lower-scoring patch in the south-west corner on the first clear image for 12 days. The owner walks it the next morning and finds standing water where a field drain has collapsed. A drain repair in the following dry week costs money but saves a patch of lost crop. Without the map the problem might have been spotted in June, when the patch was obvious and the damage done.
The false alarm (mid-May)
The 4-hectare field is flagged "behind". The owner checks the notes: it was sown 10 days later than the others. The flag was the tool comparing it to a typical curve for the earlier sowing date, which the owner had entered wrongly. Fixing the sowing date removes the alert. The lesson, which the owner records, is to check the inputs before acting on any alert.
The quiet fields (June)
For three weeks nothing is flagged in four fields. The owner visits them anyway on the usual schedule, because maps only show what a satellite can see. In one a rust outbreak is just starting on the leaves, invisible to the index at that stage. That is the honest limit of the method: it highlights large-scale differences in vigour, and early disease on leaves, a small weed patch or a nutrient issue that has not yet changed the canopy can pass unseen.
The tally (September)
Over the season the owner reckons the map prompted seven targeted visits, found two real problems (the drain and a headland compaction patch) and gave one false alarm. Costs were about $210 for the subscription and some hours of time. Whether that was worth it depends on what the drain patch would have cost, and the owner can judge from their own yield and price figures. This is the kind of one-paragraph review that turns a tool into a decision: keep it, downgrade it, or drop it.
Other crops, other answers
Grass and grazing
Grass is where satellite monitoring is easiest to act on, because growth shows quickly. A map that shows which paddocks have covered enough ground tells you where to move stock next. The caution is that cutting and grazing reset the picture, so log those dates. An illustrative 12-hectare dairy block split into six paddocks can use the weekly map as a second opinion beside a walk with a plate meter or a stick, and the maps cost less than missing a good growth week.
Orchards and vineyards
Rows are narrow relative to a 10-metre pixel, and the soil between them shows through, so a single pixel mixes tree, grass and bare ground. Monitoring can still pick out big differences across a block, such as an area of dying trees, but it cannot assess a single row. Finer detail is a drone or ground-level job.
Vegetable fields
Large field vegetables with a full canopy work. Small successional plantings do not. If a field holds six crops at different stages in strips, the pixels blend them and the map mostly describes the mix, which is useful to no one.
Questions to put to any vendor
- What is the smallest field you support, and can I see a map of my own smallest field before paying?
- Which satellite data do you use, and how often will I really get a clear image in a wet month?
- Does every map show the date of the image?
- Which crops do you have growth models for, and does my crop appear?
- Is the plan priced by area monitored or area owned, and what happens if I add a field mid-season?
- Can I export my field boundaries and history if I leave?
- What does "AI" do in your product, exactly: alerts, forecasts, advice? Who checks the advice?
- Does the product give treatment or fertiliser recommendations, and on what basis? Treat these as suggestions for a conversation with your adviser.
A vendor who cannot answer questions 1 to 3 plainly is a vendor to avoid. Many farm apps bundle monitoring with records and compliance, and comparing Herdwatch, AgriWebb and Farmonaut is a way of seeing monitoring in that wider setting.
A weekly routine that makes it useful
- Check the date of the latest clear image for each field. If it is more than ten days old, skip the field this week.
- Look for within-field differences first, not the average score. A field that is evenly lower than last year could just be a later sowing; a field with a patch is a job for a visit.
- Mark two or three places to visit and walk them within a few days, with a phone to photograph and a spade.
- Write what you found against each place: cause, action, date. A line a week is plenty.
- Ask a general assistant to tidy the log at month end, using only your notes, and to list recurring problem areas. More on that kind of note-keeping is in how an agronomist writes up a crop walk.
A sample prompt for the log
Here are my crop-walk notes for the last six weeks (field, date, what I found, what I did). Group them by field. For each field, list recurring problems, anything I flagged to check again, and gaps (fields I did not visit). Use only these notes. Do not suggest treatments or rates.An illustrative reply notes that the "east field" had a wet patch noted on three dates, that the "top barley" was flagged for a re-check after the headland was resown and never revisited, and that two fields have no entries for three weeks. That is a to-do list drawn from your own records, which is a safe and useful job for a general assistant.
Free and cheap routes
You do not have to start with a paid plan. Free map viewers built on the open Sentinel-2 data let you draw a field and look at a vegetation index over time. The trade-off is manual work: no alerts, no AI summary, and you must learn to read the cloud and date information yourself. A reasonable path is to use a free viewer for one season, on your largest field, noting how often you actually look and whether the map changes a decision. If it does, a paid plan buys time. If it does not, you have saved the subscription. For wider context on what free tools can do, free AI for agriculture on a small farm sets out the limits.
Drones, sensors and why you may not need them
Drones give much finer detail, sometimes a few centimetres per pixel, and can fly below cloud and on a day you choose. Soil moisture probes and weather stations measure what a satellite cannot. For a few high-value or very small areas they can be worth it. For a typical farm with fields of several hectares, a satellite service covers the first question, "where should I look?", for a modest monthly cost, and a drone is something to add only if a particular problem needs the detail. A crop that sits under cover is the clearest exception, and reviewing nutrient and pH logs each week in hydroponics shows what monitoring looks like there instead.
Which fields suit satellite monitoring
- Most fields over 3 hectares, open ground, a crop that covers the soil for months: try a free viewer, then a paid plan if you act on what you see.
- Fields of 1 to 3 hectares: try the free route and check the maps against what you see on the ground. Expect rough results at the edges.
- Fields under 1 hectare, narrow strips, beds, polytunnels: skip satellites; use records, photos and walking the crop.
- Any field where a mistake is costly: use the map to decide where to look, never what to apply.
Satellite monitoring is a good fit for a small farm that has the right sort of fields, and a poor fit for one that does not. Check your own smallest field's map first, note the date of the last clear image, and compare what the map suggests with a walk. That single comparison tells you more about whether to pay than any vendor page.
More questions about satellite crop monitoring
What is NDVI and should I care?
NDVI is a score from the way plants reflect red and near-infrared light. Healthy, dense leaves score higher, bare soil and stressed crops lower. It shows where a field differs, not why. Treat it as a map of places to walk to, and confirm the cause on foot.
Can I do this for free?
Partly. The Sentinel-2 satellites' data is published free of charge, and free viewers can show a vegetation index for your field. What you lose by going free is the convenience: alerts, history, field boundaries saved, and plain-language summaries. Decide whether your time is worth more than the subscription.
Does cloud stop it working?
Cloud blocks the view for that pass, so a nominal five-day revisit can turn into one usable image in three weeks in a wet spell. Good tools skip cloudy images and show the date of the last clear one. Always check the image date before acting on a map.
Is a polytunnel or glasshouse crop visible?
Generally not in a useful way. Covers reflect light that hides the crop, so a vegetation index tells you little. Use ground-level checks, a camera or sensors there. Satellite monitoring suits open-field crops and grass. Open fields beside the tunnels still show normally, so a mixed holding can monitor its outdoor ground and log covered crops by hand.
Further reads
- Crop Planning With AI: Rotations and Sowing Dates From Field Lists — Plan rotations and sowing dates from your field list.
- Spray Records With AI: Voice Notes to Logs, Then Check Doses — Keep spray records from voice notes.
- AI for Farm Business Management: A Monthly Margin Review by Enterprise — Check whether a tool paid for itself, by enterprise.
- Why Farmers Still Don't Use AI and What a Small Farm Can Try First — First steps for a farm that has not tried AI.
- What Is Automation in Agriculture? Farm Jobs Ranked by Cost and Payback — Farm jobs ranked by cost and payback.
- Flower Farm Business Plan: Bed Maps, Stem Pricing and Wholesale With AI — Plan a small flower farm's beds and pricing.
- Greenhouse Management Software: What AI Climate Alerts Cost and Add — What sensor-alert greenhouse software costs against a spreadsheet and chat-assistant routine, with a break-even sum for a small growing operation.
- Do Farmers Use AI? Eight Uses of Artificial Intelligence, With Costs — How many farmers use AI, and eight small-farm uses with indicative costs, example prompts and the checks that keep each one honest.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Sentinel-2 mission page on the Copernicus Data Space Ecosystem (bands, revisit, free data); NASA Earth Observatory explainer on NDVI; Farmonaut and FieldView pricing pages (all fetched November 2026).