All techniques › Novel & technical
Satellite drought monitoring
Using satellite imagery to see drought and crop stress. The free European products are good at regional scale and cannot see your field. The paid ones resell that same free imagery, and no European trial shows one changing a farm decision for the better. The satellite that matters most is the one your paying agency already points at your parcels.
The short answer
The free European products are good, and they cannot see your field. The drought indicator runs at 5 km resolution on a 10-day step. One pixel is about 2,500 hectares. Use it for regional early warning and for a dated record of a drought year, and do not expect it to tell you anything about a particular parcel.
The paid platforms resell free imagery, and the evidence for them is missing. Sentinel data is free, including for commercial use. We searched specifically for a European trial in which satellite advice changed a farm decision and measurably improved an outcome, and found none. The index tells you something is wrong. It never tells you what.
The satellite worth understanding is the one already pointed at you. Paying agencies began replacing inspections with satellite checks in 2018; our most recent count is 15 agencies in five member states in 2019. It reads crop type, tillage, mowing, catch crops and fallow, which is most of what this library recommends.
Start with what is free, and with what scale it works at
Two European public products matter, both free, both open in their methods.
The European Drought Observatory's Combined Drought Indicator treats drought as a cascade and reports which stage a region has reached (factsheet; method paper):
| Stage | What has happened |
|---|---|
| Watch | Rainfall shortage, measured as a standardised precipitation index |
| Warning | The shortage has become a soil-water deficit |
| Alert | The deficit has reached the vegetation, seen as a greenness anomaly |
There are three matching recovery classes. It runs at 5 km resolution on a 10-day step across Europe. That is the number to hold on to: one pixel is about 2,500 hectares. The indicator is genuinely useful for knowing whether your region is in drought, for watching one develop, and above all for having a dated, independent, defensible record of a drought year when you are talking to an insurer, a lender or a landlord. It is not a field diagnosis and does not claim to be.
The JRC crop yield forecasts are issued monthly through the season at EU and member-state level, and have been since 1993. Their accuracy has been measured against what was actually harvested (open access):
| Crop | Median absolute error | Range across member states |
|---|---|---|
| Soft wheat | 0.379 t/ha | 0.156–0.679 |
| Durum wheat | 0.368 t/ha | 0.11–0.923 |
| Grain maize | 0.570 t/ha | 0.279–1.388 |
| Potato | 1.563 t/ha | 0.929–3.878 |
| Sugar beet | 3.858 t/ha | 1.948–16.03 |
Error falls as the season goes on, and the pre-harvest forecast is "generally considerably better than the first forecast of the season". The system makes no claim of farm-level usefulness; it is built for policy. What a farmer gets from the bulletin is an early read on what the market will hear about supply, which is a price signal rather than an agronomic one.
What it is
Four different things travel under satellite monitoring, and this page grades them separately.
- Free public drought products: the drought observatory and the crop bulletins above.
- Commercial field-monitoring platforms: a subscription showing a vegetation-index map of your own parcels, almost always built on free imagery.
- The CAP area monitoring system: your paying agency verifying your claim from orbit. Not optional and not a service you buy.
- Satellite index insurance: a policy that pays on an index rather than on your assessed loss.
This is the only entry in this library where the technology is pointed at the farmer as well as offered to them.
How it works, and the limit that matters
Optical sensors measure how much red and near-infrared light a canopy reflects. Healthy, well-watered vegetation absorbs red and reflects near-infrared, and stressed vegetation does the opposite. The ratio becomes a vegetation index. Radar sees through cloud and responds to structure and surface wetness instead of greenness. Drought indicators combine a rainfall index, a modelled soil-moisture anomaly and a vegetation anomaly.
The index detects that something is wrong. It does not tell you what. A depressed reading can be drought, nitrogen shortage, disease, compaction, a weed patch, a sprayer miss or a different sowing date. That is the single most important limitation for any farm decision and it is rarely stated in the marketing.
The imagery underneath is free. Sentinel-2 gives 10 m resolution at its finest with a 5-day revisit from the two-satellite constellation, under a licence that permits commercial use with attribution (Copernicus). Some bands are 20 m, which limits it in fragmented landscapes, and Sentinel-2 carries no thermal sensor, so canopy-temperature stress detection is not available from this source.
What the evidence shows
The gap under the paid platforms
Platforms sell a parcel-level index map, usually weekly, built on that free imagery. The recurring claim is that this finds stress seven to ten days before visible symptoms. That figure traces to vendor and vendor-adjacent material. We found no independent European trial behind it, so we do not repeat it as fact. The claim we checked sets out what was and was not found.
What we searched for specifically and did not find: a European trial in which satellite advice changed a farm decision and measurably improved an outcome. The literature is full of studies correlating an index with yield, which answers a different question. The randomised evidence that exists for digital agricultural advice is non-European and is about agronomic advice in general rather than satellite imagery. That absence is the finding, and it is why this page is graded weak.
Small parcels are the weak case. Under the CAP's own satellite checks, parcels below 0.5 ha produce unresolved flags for want of clean pixels — between 1.0% and 5.7% in the three regions the auditors report, with no larger-parcel rate published to compare them against — and in Hungary 8% of monitoring claims could not be resolved by Sentinel-2 at all (as reported). In Poland, over half of all parcels do not exceed five hectares (as reported). A 10 m pixel on a small or narrow parcel is mostly edge.
The satellite already pointed at your parcels
Since 2018, member states may replace on-the-spot inspections with checks by monitoring, using satellite time series to verify claims parcel by parcel. The result is a traffic light: green confirms compliance, red shows non-compliance, and yellow is unresolved and needs follow-up.
The European Court of Auditors examined the system in Special Report 04/2020, which is the best independent source on this page.
| First paying agency | May 2018, Foggia, Puglia, Italy |
| In use in 2019 | 15 paying agencies in 5 member states (Belgium, Denmark, Italy, Malta, Spain) |
| Projected for 2020 | 28 agencies across 13 member states |
| Can see | crop type, tillage, mowing of grassland, catch crops, nitrogen-fixing crops, fallow land |
| Cannot see | hedges and trees under 20 m wide, extensive grazing, crops under glass, land abandonment, food safety, animal identification and welfare |
| Cross-compliance | only one GAEC standard found fully monitorable |
| Parcels under 0.5 ha, unresolved flags | 1.9% of parcels in Flanders, 5.7% in Denmark, 1.0% in Castile and León |
On money, the Commission spent €480,000 subsidising cloud platform access for agencies in 2019, and the auditors wrote that its "value added is yet to be established". About €94 million of EU research funding went to agriculture-monitoring projects, as counted by the same 2020 audit. No direct savings were quantified.
What this means in practice. A yellow flag is a request for evidence, increasingly through a geo-tagged photo app that records location, timestamp and device. The app built in Ireland is also used in Greece, Estonia and Austria. Farmers and member states told a 2024 consultation that the requirement was "burdensome" and the technology "not always reliable or user-friendly", and the Commission moved in July 2024 to make geo-tagged photos voluntary for member states (EU CAP Network).
This matters here because the practices this library recommends, a catch crop, a mown understory, a ley, are precisely what the system is trying to confirm. Knowing how it reads your field is worth more than any subscription on this page. The funding page has the claim side.
Index insurance: better than the alternatives, and still mostly your risk
An index policy pays on a measured index rather than an assessed loss. No loss adjuster comes out, which is the attraction. The flaw is basis risk: the payout may not match what happened on your farm, in either direction.
A systematic review of 89 peer-reviewed articles across 34 countries (NHESS 25:913, 2025) found that NDVI dominates, used in 30 of 49 studies that used satellite imagery. Vegetation-health indices outperformed weather indices, with hedging effectiveness of 38%, 38% and 29% for three of them against rainfall- and temperature-based contracts. Correlation between index and actual loss ran from 43.8% to 62%, with one pasture index reaching almost 90% in some years. Europe is 14 of the 93 studies, and only five European studies used satellite data directly. Only two studies anywhere examined whether higher-resolution data reduces basis risk.
Read those numbers plainly. A hedging effectiveness of 38% is better than the alternatives and still leaves most of the risk with you. The claim we checked has the rest.
Where it fits
| Context | Fit | Why |
|---|---|---|
| Knowing whether your region is in drought, and proving it later | Good, and free | The 5 km indicator is dated, independent and open |
| Any farm claiming CAP support | Learn how it works | Your parcels may already be checked from orbit |
| Large, regular parcels a manager cannot walk | Worth a trial season | The imagery is real; the interpretation is yours |
| Parcels under about 0.5 to 1 hectare | Not worth paying for | A 10 m pixel on a small parcel is mostly edge, and the CAP's own checks struggle below this size too. The free products and the CAP row above still apply |
| Expecting a diagnosis of what is wrong | No | The index shows stress, never its cause |
| Replacing crop insurance with an index policy | Partial cover at best | Hedging effectiveness 29–38% |
What it costs
| Item | Figure | Country, year | Status |
|---|---|---|---|
| Sentinel imagery | free, including commercial use, with attribution | EU, 2025 | Copernicus licence |
| Drought indicator and crop bulletins | free | EU | public products |
| Commercial field monitoring | from about €5/ha a year | EU, 2025–26 | one vendor's list price; no independent European figure found |
| Commission subsidy for agency cloud access | €480,000, value added "yet to be established" | EU, 2019 | Court of Auditors |
| Savings from replacing inspections | not quantified | EU, 2020 | Court of Auditors |
| Responding to a yellow flag | no figure found; reported as burdensome | EU, 2024 | public consultation |
There is no capital cost anywhere on this page, which is unusual for this library, and no evidence of a return either. Every figure is also on the costs page. No entry in the stress test: there is no evidenced European margin effect to model, and at €5 a hectare the question is not whether the farm survives it.
Commitment
Fully reversible and close to free. No land change, no capital, no permit, and a subscription can be dropped at the end of a season. The only part you cannot opt out of is the CAP monitoring system, which is not a choice at all.
How to start
- Look at the free drought products and learn what scale they work at. Bookmark the drought map. Use it for regional context and to document a drought year while it is happening.
- Find out whether your paying agency uses checks by monitoring, and what it will ask of you. A yellow flag is a photo request, not an accusation.
- Before subscribing, ask what decision you would change. If the answer is "I would walk that part of the field", the map is worth little on ground you already walk.
- If you subscribe, trial one season on parcels over a hectare and check every flagged patch on foot. The index says something is wrong, never what.
- Treat index insurance as partial cover. Ask the seller for the measured correlation between the index and assessed losses in your region, and read 38% as "most of the risk is still mine".
Risks and failure modes
The index does not say why. Drought, nitrogen, disease, compaction and a sprayer miss all look alike from orbit.
Your parcels may be too small. Below half a hectare the CAP's own system struggles.
Cloud. Optical revisit of five days is a theoretical maximum. We found no European figure for usable cloud-free acquisitions per season, so we make no claim about it.
Paying for free data. Every platform here resells public imagery. Whether the interface is worth the subscription is a question no independent European study answers.
Basis risk. Paid when unharmed, unpaid when harmed.
Compliance burden. Yellow flags cost time, and farmers said so in 2024.
Confusing regional with local. An Alert on a 5 km map does not mean your crop is failing.
Who is talking. Platform vendors and space-industry promotion supply most of what a farmer finds, including the seven-to-ten-days claim. The independent sources here are the Court of Auditors, the research centre's own accuracy assessment and a systematic review, and all three are more modest.
Claims checked. Satellites spot crop stress before you can and index insurance pays out when drought hits. The same free-versus-resold pattern runs through precision irrigation.