Drone Flight GSD, Overlap, and Image Count

The flight height times the sensor's pixel pitch, over the focal length. It scales LINEARLY with height, which is the whole flight-planning trade -- and the cost of improving it does not. HALVING THE HEIGHT HALVES THE GSD AND QUADRUPLES THE IMAGE COUNT, because both the along-track and the cross-track footprints halve at once. Four times the images means four times the flight time and battery swaps, and processing that scales worse than linearly, so promising a GSD without running that multiplication is how a one-day flight becomes a three-day flight. THE MOST IMPORTANT SENTENCE IN THIS SUBJECT IS THAT GSD IS NOT ACCURACY. Ground sample distance is RESOLUTION -- how finely the ground is sampled -- and accuracy is how close the resulting coordinates are to truth, which comes from ground control, from the camera calibration, and from the geometry of the block. A survey flown at half a centimetre per pixel with no ground control can be metres out in position, and it will look magnificent while it is wrong. Overlap is the other lever and it is cheap by comparison. Mapping practice runs about seventy five to eighty percent forward and sixty five to seventy percent side, and it goes HIGHER over vegetation, water, and uniform surfaces, because the matching algorithms need texture to tie images together and those surfaces have little. A flight planned at the minimum overlap over a forest canopy produces holes in the model that no amount of processing recovers, and reflying is the only fix. Nadir photogrammetry over flat terrain from a rectangular sensor. It assumes the ground is level at the entered height: terrain relief changes the GSD across the frame, and a flight planned at a single height over rolling ground has a varying GSD and a varying overlap, which is what terrain-following flight modes exist to fix. It does not plan the mission, place ground control, evaluate the block geometry, or address oblique imagery, corridor mapping, or the crossing lines that stabilise a self-calibrating bundle adjustment. It does not compute accuracy at all, and no image count buys it. It does not address airspace, authorisation, visual line of sight, or the applicable aviation rules, which govern whether the flight happens. The applicable civil aviation rules, the project's accuracy specification, and the surveyor of record govern.

GSD = flight height x sensor pixel pitch / focal length; the image footprint is GSD x the sensor pixel dimensions; line spacing = footprint width x (1 - side overlap); shot interval = footprint height x (1 - forward overlap); image count = area / (line spacing x shot interval).

The photogrammetric ground sample distance relation by name. GSD IS RESOLUTION, NOT ACCURACY: accuracy comes from ground control, camera calibration and block geometry, and is not computed here. Nadir imagery over level ground; terrain relief varies both the GSD and the overlap across a frame. Airspace, authorisation and visual line of sight are outside it. The applicable civil aviation rules, the project's accuracy specification, and the surveyor of record govern.

One similar-triangles ratio on numbers off a camera spec sheet; no flight planning software output is reproduced.

Estimate. AHJ and licensed professional govern.

Field names used by the API: flight_height_ft, focal_length_mm, pixel_pitch_um, sensor_width_px, sensor_height_px, forward_overlap_pct, side_overlap_pct, area_acres, gsd_cm_px, footprint_width_ft, footprint_height_ft, line_spacing_ft, shot_interval_ft, image_count, half_height_gsd_cm_px

• GSD is resolution, not accuracy accuracy comes from ground control and block geometrythe project's accuracy specification
• Nadir over level ground terrain relief varies the GSD and overlap across the frameterrain-following flight planning
• Airspace and authorisation are outside it and they govern whether the flight happensthe applicable civil aviation rules