Drone Photogrammetry for Tree Counting and Crown Mapping

Updated 2026-11-02 · 4 min read · By the akar.earth specialist team

Drone tree counting with photogrammetry: flight settings, ground control, crown detection and its limits for ARR, agroforestry and mangroves in Indonesia.

Drone photogrammetry turns hundreds of overlapping photos into an orthomosaic and a surface model, from which you can count trees and map crowns at centimetre resolution. It works best for young plantings, agroforestry and open stands; under closed tropical canopy it undercounts and cannot see the ground, so it must be paired with field plots.

What photogrammetry produces

A mapping flight with a standard RGB or multispectral drone produces:

• Orthomosaic: a geometrically corrected image, often 2–5 cm ground sampling distance (GSD).
• Digital surface model (DSM): the elevation of the top surface (canopy or ground).
• Point cloud: 3D points from structure-from-motion matching.

From these, analysts detect individual trees (by local maxima in the DSM or by deep-learning object detection on the orthomosaic), delineate crowns, and, where a ground model exists, estimate height. See canopy height models.

Step by step

1. Plan the flight. Set altitude for the GSD you need. High forward and side overlap (commonly around 75–80% and 65–75%) helps matching over repetitive canopy texture. Fly higher above tall forest to avoid parallax gaps.
2. Position accurately. Use an RTK or PPK drone, or lay ground control points measured with RTK GNSS. Without this, repeat surveys will not line up and crowns will shift between dates.
3. Fly in stable light. Mid-morning, overcast-bright conditions reduce harsh shadows. Avoid wind, which blurs leaves and ruins matching.
4. Process. Generate orthomosaic and DSM with photogrammetry software. Record software version and settings.
5. Detect trees. Use a tested algorithm or trained model. Tune it on a subset with known counts.
6. Validate in the field. Count trees and measure diameter in sample plots inside the flight area. Compare counts, compute omission and commission errors.
7. Link to carbon. Use field diameter-height data and allometric equations to convert counted trees or crown sizes to biomass, with uncertainty.

Where drones work well

For ARR, drones give a fast survival census that feeds replanting budgets. For smallholders, a single flight can cover many farms that would take days to walk.

Limits to plan around

• Hidden trees. Photogrammetry sees only the top surface. Seedlings under grass or crowns under taller trees are missed.
• No ground under canopy. Height needs a terrain model. Use drone lidar, field heights or a reliable external DTM.
• Merged crowns. Touching crowns are split or merged wrongly, biasing counts.
• Area. A typical multirotor covers tens to a few hundred hectares per day depending on altitude and battery logistics; fixed-wing or VTOL drones cover more.
• Weather and logistics. Rain, wind and remote access in Papua or Maluku limit flying days.

Sampling with drones instead of full coverage

You do not always need to fly the whole project. For large or dispersed areas, fly a statistically designed sample of blocks or farms, validate counts with field plots inside them, and scale up with the same sampling logic used for inventory plots. This keeps flight days manageable while giving verifiers a clear design-based estimate with a confidence interval.

Common mistakes

• Reporting drone counts without field validation.
• Using different altitudes or cameras between monitoring rounds, so results are not comparable.
• Treating crown area alone as a carbon estimate without a calibrated relationship.
• Flying without community notice. In villages and customary land, drones over homes and gardens cause distrust. Build it into community engagement.
• Losing raw images. Verifiers may ask to reprocess.

What good evidence looks like

Provide flight logs, camera and altitude, overlap, GSD, positioning method and ground control accuracy, processing reports, the detection method and its validation statistics against field plots, and the final count or crown dataset with a clear link to field-measured trees. Archive raw images with dates.

Indonesia specifics

Drone rules are set by the Ministry of Transportation’s civil aviation authority, and some areas (airports, military zones, some conservation areas) need extra permission. Check current rules before each campaign. Rainy-season flying windows are short in much of western Indonesia. In Java and Bali, small and fragmented parcels suit drones; in Kalimantan and Papua, travel time often dominates cost.

For the airborne options compared, see lidar vs photogrammetry. We plan and run drone and plot campaigns through our field data campaigns service. For how this fits the full monitoring chain, visit the dMRV hub.