Blog · October 14, 2026

What ground sample distance means for building damage grades

A satellite ground station with large dish antennas under a dark, cloudy sky beside water.

Ground sample distance shows up in every imagery spec sheet, usually as a number like "0.5 m GSD," and it's easy to skim past if you're not the one ordering the imagery. For anyone deciding where search teams go first after an earthquake, it's the number that decides whether a damage grade means anything at all.

What ground sample distance measures

GSD is the ground distance one pixel covers. A 0.5 m GSD image means each pixel represents a 0.5 by 0.5 meter patch of ground. Finer GSD (a smaller number) means more detail per structure. Coarser GSD means each pixel is averaging more of the scene together, and detail gets lost in that averaging.

It's a different number from "resolution" in the way people use that word casually, and different again from image file size or zoom level in whatever viewer you're using. You can zoom into a coarse-GSD image all day and it will just show you bigger, blurrier pixels. The ground detail was never captured in the first place.

For earthquake response, GSD is the thing that determines whether an analyst (or an automated classifier) can tell a partial roof collapse from an intact roof with a skylight glare, or a pancaked structure from a parking lot full of debris that was already there before the shake. At coarse GSD, a city block is a handful of smudges. At fine GSD, each building has enough pixels to read clearly.

Why GSD sets the floor for a damage grade

A building damage grade is a judgment call made from what's visible in the image: roof condition, wall collapse, debris spread around the footprint, shadow patterns that suggest a leaning structure. Every one of those visual cues needs a minimum number of pixels across the structure to show up reliably. Coarser GSD doesn't make the damage invisible. It makes the evidence of damage ambiguous, which is worse for triage than having no image at all, because it produces a confident-looking grade the image can't support.

This is why vertical very-high-resolution (VHR) imagery, GSD of 0.5 meters or finer, is the baseline for any serious per-structure damage pass. Coarser public datasets are fine for a regional overview of where the shaking was worst. They are not fine for deciding which specific block gets a search team sent to it this afternoon.

A few practical things follow from that:

  • Satellite vs. aerial isn't the deciding factor, GSD is. Both VHR satellite and aerial platforms can deliver sub-0.5 m imagery; what matters is the spec the tasking comes back at, not which platform flew it.
  • RGB is enough for a visual grade. You don't need multispectral or thermal to read roof collapse and debris spread. Plain color imagery at fine GSD does the job a damage grading pass needs.
  • Cadence matters as much as resolution. A fine-GSD image from before the event tells you what the building looked like last year. What you need is a tasking triggered by the event itself, flown and delivered on a timeline that matches your operational window, not a scheduled pass that happens to land near the right date.

What this means when you're ordering imagery

If a vendor or a public source hands you post-event imagery and doesn't quote GSD at all, ask for it. "High resolution" is a marketing phrase that stands in for a number nobody gave you. "0.5 m or finer" is a spec you can hold a grading pass accountable to.

And if you're staring at a dataset trying to decide if it's good enough to send a damage read off of: zoom in on one building you already know the condition of, by radio report or by eye. If you can't confirm the roof state at that zoom, the GSD isn't fine enough for the rest of the city either.

A structure-by-structure damage read built on this kind of VHR, event-driven imagery is what Building Damage Classification delivers once the tasking comes back, so your triage call is resting on pixels that show you what's broken.

If your next imagery order is going to drive a search-and-rescue sequence, it's worth getting the GSD question answered before the aircraft or satellite is even tasked.