27/07/2026
303 drone flights over the same patch of rainforest. Four years. One goal: watch individual trees change with the seasons.
Tropical forests hold a huge share of the planet's carbon, but they're the hardest to track. Clouds block satellites most of the year, and even on a clear day the image is too coarse to pick out a single tree.
Drones can get close enough. The catch: to compare one visit to the next, every flight has to land the camera over the same ground at the same resolution. Miss that, and the dates don't line up.
A team from the University of Montpellier did it 303 times, over 3.5 years, at Paracou in French Guiana. Three different consumer DJI drones, each flying at a different height to hit the same 5 cm detail on the ground.
They planned every flight in . The software held the settings steady: same resolution, same overlap, same flight lines, across every date and every drone.
That consistency is what made the whole dataset comparable.
The payoff: a clear four-year record of the forest greening and shedding leaves, down to individual tree crowns. They even caught one branch cycling out of step with the rest of its tree.
The takeaway for anyone doing long-term monitoring: cheap cameras work fine, as long as the flights stay consistent.
Full case study:
IRD researchers flew 303 autonomous UgCS missions over 50 ha of French Guiana rainforest with three DJI drones, holding 5 cm GSD for a 4-year tree phenology record.
