Flight Planning for a 10 km Block Against a Tight Planimetric Accuracy Requirement
An aerial survey is planned with a 23 cm x 23 cm frame camera, focal length 152.0 mm and 5.2 micrometre pixels. The aircraft will fly at 1,450 m above the project datum with 60% forward overlap and 30% sidelap. Ground elevations in a 10 km x 10 km block range from 250 m to 420 m. The deliverables are a 1:1,000 orthophoto and a planimetric feature layer specified as accurate to 0.05 m. The project manager has already told the client that the flight will produce the required 5 cm layer because the pixel size on the ground at the datum is about 5 cm.
1. Compute the ground sample distance at the highest and lowest terrain and assess whether the 0.05 m feature accuracy is achievable.
Read the worked answer
Scale number = H / f, and GSD = (H/f) x pixel size.
- At datum: H = 1,450 m, H/f = 1,450/0.152 = 9,539.5; GSD = 9,539.5 x 5.2e-6 = 0.0496 m.
- Over the 250 m ground: H above ground = 1,200 m, H/f = 7,894.7; GSD = 0.0411 m.
- Over the 420 m ground: H above ground = 1,030 m, H/f = 6,776.3; GSD = 0.0352 m.
So GSD ranges roughly 3.5-5.0 cm across the block; the datum value is the coarsest case, not the representative one.
Is 5 cm feature accuracy achievable? No, not reliably. The project manager has conflated GSD with accuracy. A pixel is a sampling interval, not an error budget: identification of a feature edge is typically good to about 1 GSD at best and 2-3 GSD under real conditions (poor contrast, shadow, vegetation, oblique surfaces). At 3.5-5.0 cm GSD a defensible expectation is roughly 7-15 cm, not 5 cm - and that ignores georeferencing error, which is additive in the RSS.
To reach 5 cm the flight must be lower: at 700 m above ground, GSD = (700/0.152) x 5.2e-6 = 0.024 m, giving a realistic margin - plus a deliberately designed control/checkpoint scheme.
2. Using the mean ground elevation of the block, compute ground coverage per photograph, the spacing between exposures, the spacing between flight lines, and the approximate number of photographs required.
Read the worked answer
Take the mean ground elevation (250 + 420)/2 = 335 m, so H above ground = 1,450 - 335 = 1,115 m, and H/f = 1,115/0.152 = 7,335.5.
- Ground coverage per frame: 0.23 m x 7,335.5 = 1,687 m on a side.
- Exposure spacing (60% forward overlap, so 40% new ground): 0.40 x 1,687 = 675 m.
- Flight-line spacing (30% sidelap, so 70% new ground): 0.70 x 1,687 = 1,181 m.
Block of 10 km x 10 km:
- Lines across: 10,000/1,181 = 8.47, so 9 intervals, i.e. 10 flight lines.
- Photos per line: 10,000/675 = 14.8, so 15 intervals, i.e. 16 photographs.
- Total: 10 x 16 = about 160 photographs.
State the assumptions: flat mean terrain for the spacing, no allowance for terrain-induced scale change, crab, wind, turns, or extra lines for stereo coverage at the block edges. Because the actual AGL varies by 170 m (+/-8%), exposure spacing should be flown on a constant AGL rather than a constant altitude above datum, or the overlap will fall short over the high ground and waste film over the low ground. To limit GSD variation, plan a nominal AGL and accept a scale range of about +/-8%, or split the block into terrain bands.
3. Beyond flying lower, what must be controlled to deliver a 1:1,000 orthophoto and a 5 cm feature layer, and how would you verify the result?
Read the worked answer
The flight is only one link. The judgement points:
- Georeferencing: an airborne GNSS/IMU direct-georeferencing solution alone rarely reaches 5 cm; plan a block adjustment with ground control points around the perimeter and in the centre, and hold back independent checkpoints (typically 10-20% of the control, deliberately not used in the adjustment).
- DEM/DSM for orthorectification: the orthophoto is only as good as the elevation model used to remove relief displacement. A coarse DEM leaves building lean and terrain displacement that can exceed several times the GSD; use a DSM/DTM matched to the site, and plan a separate DTM for bare-earth work.
- Ground truth is the deliverable check: report RMSE at checkpoints in X, Y and Z separately, against the stated specification, and state the confidence level. A 1:1,000 orthophoto conventionally needs GSD of about 0.10 m or finer, so the planned 3.5-5 cm comfortably supports the orthophoto - the binding constraint is the 5 cm feature accuracy, and the report should say so.
- Metadata and honesty: record flying height, camera calibration, GSD range, control and checkpoint statistics, datum and projection. Never present the datum-scale GSD as an accuracy figure; the client bought an accuracy, not a pixel size, and the professional obligation is to say plainly, before the flight, that the plan as briefed will not meet it.