Drone Ground Sample Distance Explained

On This Page

• What a GSD number tells you
• Calculate GSD from the actual camera configuration
• Why height, terrain, and image quality change the result
• GSD does not certify mapping accuracy
• Turn the target into a commercial capture workflow
• What to request before accepting the delivery

What a GSD number tells you

With a downward-looking camera over level terrain, a 2 cm GSD means neighboring pixel centers are approximately 2 cm apart on the ground. Each pixel corresponds to roughly 2 cm by 2 cm, or 4 square centimeters. GSD is a linear sampling interval, not an area measurement. PIX4D's GSD documentation explains the relationship between ground distance and image pixels.

A hypothetical 10 cm-wide feature spans about five pixels at 2 cm GSD, compared with two pixels at 5 cm GSD. That arithmetic describes sampling only. It does not prove that an analyst can identify the feature, locate its edges accurately, or diagnose its condition.

Keep three quantities distinct:

This distinction combines PIX4D's definitions of GSD and accuracy with Esri's orthomosaic documentation, which exposes output cell size and resampling as separate processing settings.

An orthomosaic is a geometrically corrected mosaic of overlapping images. Exporting it with smaller cells does not, by itself, create new observations of the ground. Ask for the original-image GSD as well as the exported raster resolution.

Calculate GSD from the actual camera configuration

For a nadir image, with the camera looking vertically down at approximately level ground, the idealized relationship is:

GSD (m/pixel) = height above the surface (m) × sensor width (mm) ÷ [actual focal length (mm) × image width (pixels)].

Sensor width and focal length must use matching units. Use the physical sensor width corresponding to the recorded image and the actual focal length, rather than a 35 mm-equivalent focal length. Cropped or resized imagery needs inputs that describe that recording mode.

The geometry is straightforward: height multiplied by the sensor-width-to-focal-length ratio gives the ground footprint width; dividing by the pixel count gives the sampling interval. PIX4D's GSD calculator uses these camera, height, and image-dimension inputs.

Consider a hypothetical camera with a 13.2 mm-wide sensor, an 8.8 mm lens, and images 5,472 pixels wide. These are illustrative inputs, not specifications or measured results for a recommended product. At 80 m above a level surface:

GSD = 80 × 13.2 ÷ (8.8 × 5,472) = 0.02193 m/pixel, or approximately 2.19 cm/pixel.

The corresponding image footprint is 120 m wide. Using the same inputs gives this planning comparison:

These are editorial calculations using ideal nadir geometry, rounded to two decimal places for GSD. They exclude lens distortion, image blur, terrain variation, and processing effects; the heights are calculation examples, not flight permissions.

To solve for height instead, rearrange the equation:

Height (m) = target GSD (m/pixel) × focal length (mm) × image width (pixels) ÷ sensor width (mm).

For a 2 cm/pixel target with the example camera, the result is approximately 73 m above the surface. Treat this as a starting point for capture planning, with room for terrain and image-quality variation.

Why height, terrain, and image quality change the result

The relevant height is distance to the surface being photographed. A height displayed relative to takeoff is not necessarily height above every part of a site. With the example camera held at one elevation, terrain 20 m below a surface that was 80 m away increases the distance to 100 m. The calculated GSD becomes 2.74 cm/pixel, 25% coarser than at 80 m.

Roof elevations and sloping ground create similar variation. An oblique image also has changing scale across the frame, so one nadir calculation cannot describe an entire facade. PIX4D notes that terrain and camera-angle differences can produce varying GSD within a project, even when flight height appears constant.

Flying lower narrows the footprint. Keeping the same overlap then requires closer-spaced exposures and flight lines; camera triggering and aircraft speed must be planned together. PIX4D's altitude and focal-length guidance explains why smaller footprints demand a higher image rate at a given travel speed.

Lower GSD also makes blur more consequential. For a simplified example of straight, level travel over stationary ground, ground motion during exposure is speed multiplied by exposure time. At 5 m/s and 1/500 second, that is 0.01 m: half a pixel at 2 cm GSD, but a full pixel at 1 cm GSD. This calculation isolates forward motion; it excludes rotation, vibration, and shutter effects.

The practical response is to inspect sample images at their original resolution before committing to the full site. PIX4D's accuracy guidance specifically identifies blur and camera vibration as problems that become more significant at very fine GSD. A nominally finer GSD offers little help if the detail needed for the task is blurred or obscured.

GSD does not certify mapping accuracy

Relative accuracy concerns dimensions and relationships within a model. Absolute accuracy concerns where the model sits in a defined coordinate system. A model can preserve local distances while being displaced from the correct site coordinates. PIX4D's accuracy guidance distinguishes these cases and recommends checkpoints to assess absolute accuracy.

Ground control points constrain the reconstruction. Independent checkpoints are reserved for evaluating it. Real-time kinematic (RTK) or post-processed kinematic (PPK) positioning can improve image geolocation, but the optical GSD equation remains unchanged for the same camera and imaging geometry.

USGS's 2023 imagery-calibration guidelines identify ground-control quality and tie-point quality as contributors to geometric accuracy regardless of GSD. Tie points are features matched across images. Their distribution, camera calibration, and capture geometry influence the reconstruction.

Consequently, a quoted “2 cm GSD” is not a promise of 2 cm horizontal or vertical accuracy. Request separate horizontal and vertical results, the statistic used, checkpoint locations, and the accuracy of the reference measurements. Report checks on points excluded from model adjustment; the fit to the control used to build the model is not an independent test. A single average should not conceal poor performance in the part of the site that matters.

Nor does GSD certify a defect diagnosis. If the purpose is inspection rather than mapping, the next question is what drone inspection evidence can establish, including whether a visible indication needs closer examination or another measurement.

Turn the target into a commercial capture workflow

Start with the deliverable and the smallest feature that must be interpreted. Specify whether the job needs a visual site record, dimensions, elevations, or a repeatable comparison with a previous survey. Have the provider demonstrate suitable detail on representative surfaces; there is no single GSD that serves every task.

Before capture, agree on these inputs:

• Camera configuration: sensor dimensions, actual focal length, image dimensions, and recording mode.
• Site geometry: boundary, terrain and structure heights, obstructions, and required viewing directions.
• Image acquisition: planned GSD range, overlap, exposure settings, trigger interval, and treatment of difficult surfaces.
• Positioning and reference: coordinate system, vertical reference, control strategy, and independent checkpoints appropriate to the measurement requirement.

Overlap is a reconstruction requirement as well as a coverage choice. PIX4D's image-acquisition guidance recommends at least 75% forward and 60% side overlap for its general mapping case, with different treatment for vegetation, thermal imagery, and buildings. Those are provider planning recommendations, not universal guarantees.

During capture, check sharpness, exposure, coverage, and any departure from the planned distance to the surface. After processing, inspect the final output for holes, doubled edges, and local failures. Record affected areas rather than presenting a project-wide GSD as evidence that they are usable.

What to request before accepting the delivery

Ask the supplier to state exactly what its resolution number refers to. Then make the handover specific:

1. Capture record: camera and recording mode, acquisition date, height reference, and original-image GSD range across the required area.
2. Usable output: agreed raster or model format, pixel size where applicable, coordinate system, units, vertical reference, and coverage boundary. Confirm that the receiving software opens a sample correctly.
3. Accuracy report: independent checkpoint results, horizontal and vertical statistics, reference-measurement quality, and a map showing where checks were made.
4. Exceptions: missing or unreliable areas, blur or reconstruction problems, and any resampling that changes the delivered pixel grid.
5. Remedy: an agreed process for correcting or recapturing areas that fail the specified detail, coverage, or measurement requirements.

These are suggested buyer requirements derived from the distinctions above, not a certification standard. The useful purchasing question is whether the delivered data supports the intended decision. Select a GSD that captures the necessary detail, then verify coverage, image quality, and measurement accuracy on their own terms.

Source notes

• PIX4D: Ground sampling distance in photogrammetry. Technical documentation defining GSD and explaining terrain and camera-angle variation.
• PIX4D: GSD calculator. Provider tool specifying camera, flight-height, and image-dimension inputs; the worked examples here use hypothetical inputs.
• PIX4D: Selecting focal length and flight altitude. Technical guidance on footprint, GSD, and image-rate tradeoffs.
• PIX4D: Relative and absolute accuracy of drone mapping. Provider guidance on reconstruction, positioning, and checkpoints.
• PIX4D: Image acquisition. Terrain-dependent flight planning and overlap recommendations.
• USGS: Guidelines for Calibration of Uncrewed Aircraft Systems Imagery. Government technical report, 2023, on calibration and geometric data quality.
• Esri: Generate Orthomosaic. Official software documentation distinguishing output cell size and resampling settings.

Last checked: September 7, 2026.