Ground Sample Distance (GSD) is the real-world size that a single pixel of a satellite image covers on the ground. A 30 cm GSD means each pixel spans a 30 cm square — sharp enough to pick out individual cars. The smaller the GSD, the higher the resolution and the finer the detail.
Understanding GSD
How GSD is calculated
GSD follows directly from optics and orbit height: GSD = pixel size × altitude ÷ focal length. Flying lower shrinks the GSD, which is why the sharpest Earth-observation satellites sit in low orbit (roughly 450–700 km); a longer focal length — a physically larger telescope — does the same job without dropping altitude. As an illustration, an 8-micrometre detector pixel viewed through a 16 m focal length from 600 km projects to about 0.30 m on the ground. GSD is always quoted 'at nadir', looking straight down. When a satellite tilts to image a target off to one side, the pixel footprint stretches and the effective GSD grows — WorldView-3, for example, is 0.31 m at nadir but about 0.34 m at 20° off-nadir.
Why GSD is not the same as resolution
A 30 cm GSD does not mean a 30 cm object can be identified. GSD measures how finely the scene is sampled, not the smallest feature the optics can resolve — that is limited by lens sharpness (the modulation transfer function) and atmospheric blur. Reliably recognising an object takes several pixels across it, so the smallest identifiable feature is typically a few times the GSD. This is why 30 cm imagery shows that a vehicle is a car, and perhaps its rough class, but never a face or a number plate — the 'read a licence plate from space' idea is a myth. Beware, too, of '15 cm HD' marketing: those products are interpolated (super-resolved) from ~30 cm native pixels, not genuinely finer sampling. Commercial sharpness was long capped by regulators — the US limit was 0.5 m until 2014, when it was relaxed to 0.25 m panchromatic.
GSD across the imaging fleet
GSD spans nearly three orders of magnitude across today's Earth-observation satellites — from sub-metre commercial 'tasking' platforms that must be pointed at a target, to free, wide-swath public missions used for climate, agriculture and land-cover monitoring:
| Satellite / sensor | GSD at nadir | Access |
|---|---|---|
| WorldView-3, WorldView Legion (Maxar) | ~0.30 m panchromatic | Commercial |
| Pléiades Neo (Airbus) | 0.30 m panchromatic | Commercial |
| Planet SkySat | ~0.50 m | Commercial |
| Planet PlanetScope (Dove) | ~3 m | Commercial |
| Sentinel-2 (ESA / Copernicus) | 10 m multispectral | Free / open |
| Landsat 8/9 (USGS / NASA) | 15 m pan, 30 m multispectral | Free / open |
| MODIS (Terra / Aqua) | 250–1000 m | Free / open |
The resolution–coverage trade-off
Sharper pixels come at a price. Halving the GSD quarters the ground area each pixel represents, so a sensor must either accept a much narrower swath width or generate far more data per scene. Narrower swaths mean each satellite images less of Earth per pass, which lengthens revisit time — the wait before the same spot is photographed again. This is the central trade-off in Earth observation: Maxar and Airbus fly a handful of large, very sharp satellites that are tasked to point at specific targets, while Planet flies a big flock of smaller satellites at coarser GSD to image the whole landmass daily. Synthetic-aperture radar satellites play by different rules, seeing through cloud and darkness, and most optical imagers use a sun-synchronous orbit so surface lighting stays consistent between passes.