Skip to content

Flight planning (overlap, GSD)

The quality of the result is 80% the quality of the flight. Even the best processor cannot save a poorly planned mission. This page gathers the key planning parameters.

  1. High overlap — at least 75% along-track, 70% cross-track
  2. Constant AGL altitude — avoid manual altitude changes
  3. Overcast or thin clouds — direct sun shadows add noise
  4. Calibration panel on P4M / M3M — before and after the flight

Video: DJI M3M mission planning — how to fly a good survey with the Mavic 3 Multispectral, and the settings that matter:

Overlap is the area of the field that two adjacent images share.

  • Along-track overlap (along the direction of flight): aim for 80%
  • Cross-track / side overlap: aim for 75%

Lower overlap means:

  • Weaker feature detection
  • Holes in the orthomosaic in some areas
  • The BA optimisation sees fewer tie points

In DJI Pilot 2’s 2D Mapping mission, overlap is configurable. Defaults are 75% / 70%; for agricultural surveying, raise to 80% / 75%.

Flight altitude (AGL — above ground level)

Section titled “Flight altitude (AGL — above ground level)”

Altitude directly determines the GSD. DroneField’s supported GSD range is 3–20 cm/px, default 5 cm/px.

AGL altitudeGSD (M3M)GSD (P4M)Images / ha (~75% overlap)
80 m~3 cm/px~4 cm/px~20
100 m~4 cm/px~5 cm/px~12
120 m~5 cm/px~6 cm/px~9
150 m~6 cm/px~8 cm/px~6
200 m~8 cm/px~10 cm/px~4

Recommended altitudes by purpose:

  • NDVI / detailed analysis (3–5 cm/px): 80–120 m AGL
  • Drone-applied application map (5–7 cm/px): 120–150 m AGL
  • Ground (tractor) application (10 cm/px): ~200 m AGL
  • Quick overview (15–20 cm/px): 250+ m AGL (check airspace restrictions)

On hilly terrain it’s worth setting the planned altitude to the average elevation of the field — survey drones at typical mapping heights do not actively follow the terrain; they fly at constant absolute altitude.

Coverage — fly over what you want to see

Section titled “Coverage — fly over what you want to see”

At the edges of an aerial image, parallax tilts tall objects (trees, terrain features, poles) — the further from image centre, the more they distort. Therefore:

Whatever you want to see in detail in the result, fly over it. Don’t just observe an obstacle from the side of the field — fly above it.

If you need NDVI on a tree at the field edge, or to clearly see a corner wire obstacle, the drone should fly over the tree / wire — not only up and down the field’s centre.

  • 5–8 m/s — optimal; minimal motion blur
  • 10 m/s — upper boundary, slight blur
  • >10 m/s — avoid; image alignment degrades
WeatherSuitable?
Overcast✓ Ideal — no stochastic shadows
Thin clouds (1–2 octa)✓ Suitable
Clear sky, midday△ Strong shadows — avoid near trees
Dawn / dusk, low sun△ Colours skewed, NDVI uncertain
Rain, fog✗ Don’t fly
Strong wind (>10 m/s)✗ Drone can’t hold altitude

For multispectral indices (NDVI, NDRE), it’s especially important that all bands experience the same illumination — hence overcast is ideal.

The lawnmower (back-and-forth parallel strips) pattern is standard. Avoid grid patterns (perpendicular passes) — they take longer and DroneField’s BA is tuned for lawnmower.

  • Nadir (straight down, −90°) — standard surveying pose
  • Oblique (45° forward) — not handled by DroneField for agricultural processing (the M3 Elevation 45° photos are filtered out at import)

The gimbal must remain fixed — manual intervention or gimbal oscillation injects noise.

P4M and M3M drones ship with a white / grey calibration panel in the box. Photograph it before take-off and after landing. DJI Pilot 2 prompts for this automatically.

DroneField detects calibration shots automatically and excludes them from the processing pool (but uses them for reflectance scaling).

Optimal is 2 hours before sunset — 1 PM. Too early or too late amplifies long-shadow noise.

For multispectral, 10:00–14:00 is ideal (in Europe).

NDVI shows parallel bands that line up with the flight strips:

  • Weather changed mid-flight (a cloud reached half the field)
  • The drone’s automatic exposure (M3M) operated under different parameters on the two halves

Fix: re-fly under stable weather, or exclude the bad strips from processing.

“Holes” in the orthomosaic (blank patches):

  • Overlap was insufficient in that region
  • The drone went around an obstacle
  • Firewall or wind issues led to bad GPS

Fix: fly a smaller pass over the affected region and add the new imagery to the existing project.

If your imagery is >50 GB in a project:

  • Unnecessarily high resolution — perhaps you didn’t need to fly at 60 m over a 200-hectare field (10–15 cm/px would have been fine)
  • Excessive overlap (90%+) — 80% is usually enough

DroneField reads everything from the XMP and MRK, but it’s worth recording on paper:

  • Date, start and end time
  • Weather (overcast / sunny / cloud cover)
  • Drone and pilot name
  • AGL altitude, overlap percentages
  • Notes (on-site problems, calibration panel success)

You can put these in the project’s description field — they will appear in the report too.