C-DRONE GUIDE · 10 SEPTEMBER 2026
Potatoes and drones: detecting late blight, mapping pre-harvest defoliation, price
In 2024, France harvested 7.6 million tonnes of potatoes for consumption, up 8% year-on-year — a sector in steady expansion, with national acreage passing 197,000 hectares in 2025. Hauts-de-France alone accounts for six hectares out of every ten, across the broad plains of the Vermandois, the Amiénois and the Santerre. On a crop with dense foliage and mechanised harvesting, two drone uses stand out: spotting the first outbreaks of late blight, the sector's most feared disease, and mapping plot heterogeneity ahead of the pre-harvest defoliation that precedes lifting. Here is what the research documents, and what a flight does not replace.
Published on 10 September 2026, reviewed on 10 September 2026 — regulations in force as of September 2026.
A sector in expansion, led by Hauts-de-France
The potato occupies a distinctive place among France's arable crops: unlike wheat or oilseed rape, it is planted in ridges — broad mounds of soil spaced roughly 75 centimetres apart — and harvested mechanically with a lifter, the whole tuber staying underground until harvest day. The sector is going through a clear expansion phase: 7.6 million tonnes of potatoes for consumption harvested in 2024, up 8% on 2023, for a national gross yield of 43.1 tonnes per hectare. The momentum continued into 2025, with national acreage passing 197,000 hectares — roughly 25% more than in 2023 — driven by demand from the processing industry (frozen fries, crisps, flakes) and by exports.
Hauts-de-France accounts for the bulk of this expansion: the region ranks first nationally with around six hectares out of every ten grown in France, spread across the Vermandois (Aisne), the Amiénois and the Santerre (Somme) and the Cambrésis (Nord). This basin is largely organised around contracts with processors and cooperatives, with size and skin-quality requirements that tie fine-grained plot management — irrigation, crop protection, defoliation timing — directly to how the lot is valued.
Late blight, the sector's top threat: what drone research documents
Late blight (Phytophthora infestans) remains potatoes' most feared disease: this oomycete, historically responsible for the 19th-century Irish famine, can destroy an entire canopy within days under mild, humid conditions, with a direct impact on yield and on the storage life of infected tubers. To limit the number of spray passes while keeping production secure, the industry leans heavily on decision-support tools such as Mileos, developed by Arvalis: this model calculates infection risk from local weather, variety and crop stage, and only triggers a treatment alert when genuinely needed — cutting, on average, three fungicide treatments a year compared with a calendar-based programme, depending on the season.
Aerial detection of late blight by drone is an active research field. A study by Duarte-Carvajalino, Alzate, Ramirez, Santa-Sepulveda, Fajardo-Rojas and Soto-Suárez, published in 2018 in Remote Sensing, trained several machine-learning models (multilayer perceptron, convolutional neural network, random forests) on drone-captured multispectral data over Colombian potato plots to quantitatively predict infection severity: the resulting mean error was 11.72%, judged acceptable by the authors for steering a targeted treatment (see the study on Google Scholar). Carried out under Colombian conditions, the study does not cover Hauts-de-France varieties or climate: the underlying principle — a measurable spectral shift before infection is visible to the eye from the field edge — nonetheless carries over, pending local validation.
A second study, led by Franceschini, Bartholomeus, Van Apeldoorn, Suomalainen and Kooistra at Wageningen University and published in 2019 in Remote Sensing, tracked Dutch plots on four occasions between 37 and 78 days after planting, before and after late blight appeared in the field, to establish the feasibility of early detection through drone-borne optical imagery (see the study on Google Scholar). The authors conclude that a usable optical signal exists before the disease can be spotted on the ground — a result that echoes the logic of existing decision-support tools: gaining a few days on spotting an outbreak, not replacing ground monitoring.
Mapping heterogeneity ahead of defoliation
Defoliation (défanage) is the operation that rapidly destroys the haulm (stems and foliage) two to three weeks before lifting: mechanical flailing, topping or chemical desiccation depending on the farm. It serves two precise technical goals — stopping tuber growth to lock in the size specified by the buyer's contract, and giving the skin time to mature and set against the flesh, a condition essential to limit bruising and "skinning" during mechanised lifting and for good storage life. Defoliating too early costs yield; defoliating too late risks a delayed harvest and greater exposure to weather.
On a large plain plot, maturity and vigour are rarely uniform: differences in soil, residual nitrogen, a localised water stress episode or disease pressure create zones that reach senescence at different rates. A multispectral flight ahead of the defoliation window produces a heterogeneity map of the plot — useful for deciding whether to split the pass across the most uneven zones, or for confirming that a single pass will do on a homogeneous field. Our guide comparing multispectral and RGB cameras in agriculture covers choosing the right sensor for the indicator sought — canopy vigour in multispectral, spotting lodged patches or visible outbreaks in high-resolution RGB.
One methodological point worth keeping in mind: unlike vineyard downy mildew (Plasmopara viticola), covered in our guide to downy and powdery mildew in vineyards, potato late blight is caused by a distinct species, Phytophthora infestans, with its own spread dynamics and risk thresholds — the two crops share the English name for the disease family, not the pathogen or the monitoring protocol.
Where to fly in Hauts-de-France, and 2026 prices
The Picardy basin and the Vermandois offer flat terrain with no notable obstacles: most standalone flights over a private plot fall under the open category, with no prior process required. The point worth checking before any flight is the presence of a nearby aerodrome — around Amiens, the Amiens-Glisy airfield calls for simple but real coordination; around Saint-Quentin, the Saint-Quentin-Roupy aerodrome (LFOW) plays the same role for plots in the Vermandois. A systematic check on the Géoportail drone zone map remains the first step before any campaign; our regional guide to drone missions in Hauts-de-France covers the full airspace picture. Missions around Amiens are run from Amiens, and those in the Vermandois from Saint-Quentin.
Ranges observed in France in 2026, excl. VAT:
| Service | Range (excl. VAT) |
| Multispectral late-blight scouting flight (up to 10 hectares) | €280 to €500 |
| Additional pass per extra 10-hectare block | €80 to €140 |
| Pre-defoliation heterogeneity mapping (up to 15 hectares) | €320 to €550 |
| Combined seasonal monitoring (2 to 3 passes, blight + defoliation) | €650 to €1,050 |
These amounts cover the flight and delivery of a usable map within 24 to 48 hours; they do not include crop-protection treatment or the defoliation-date decision, which remain the grower's, their cooperative's or their Arvalis adviser's call. For a growers' group under contract with the same processor, a shared campaign across several neighbouring plots meaningfully cuts the cost per hectare. Request a quote stating the area, the number of plots and the intended period: during the season for blight scouting, two to three weeks before lifting for defoliation mapping.