Does walking near tall buildings affect GPS accuracy?

The short answer

Yes. Tall buildings on both sides of a street degrade a GPS fix for two separate reasons at once — part of the sky is physically blocked, and the signals that do get through often arrive after bouncing off glass or steel rather than by the shortest path. Both effects are well understood and neither is unique to any phone or app; every receiver on the same street sees a version of the same distortion. It clears up as soon as the buildings do.

Two different problems, not one

Less sky to see

A receiver needs a reasonably clear view of several satellites spread across the sky to work out a tight position. Walk down a street with tall buildings on either side and a large share of that sky is simply gone, blocked by masonry and glass the same way a canyon wall or a cliff face blocks it in open country. What is left visible tends to be a narrow strip directly overhead, which gives the receiver fewer satellites to work with and less spread between them — both of which loosen the fix even before anything gets reflected.

Multipath off glass and steel

The signals that do reach the phone often do not arrive by the straight path from satellite to receiver. A wide glass facade or a metal-clad tower reflects a GPS signal much the way a mirror reflects light, and a receiver that picks up a bounced copy has no way to know it took the long way round — it treats the extra distance the signal travelled as if it were extra distance to the satellite, which shifts the calculated position. This is the same multipath effect that happens off a tree trunk under forest cover or off a canyon wall, just with harder, flatter surfaces producing a stronger, cleaner bounce.

What it actually looks like on a recorded track

The two effects together are why an urban stretch of a walk or a ride can look far rougher than the same distance covered on an open trail: a line that hops to the wrong side of a street, a corner that gets rounded off or cut short, or a point that lands squarely inside a building the route never went near. A brief elevation blip can show up too, for the same reason it shows up at the mouth of a tunnel — the first fix after a bad geometric stretch is often the least precise one a receiver produces all day, before it settles again.

None of it is a sign that anything is wrong with the phone, the app, or the rest of the recording. It is confined to the stretch where the sky was actually obstructed, and the points before and after it are exactly as reliable as any other point in the file.

Why the open trail does not have this problem

This effect is entirely about what physically surrounds the receiver, not about anything to do with being in a city as opposed to a trail. A wide plaza, a park, or a bridge over open water gives a phone the same clear view of the sky it would get on a ridge, tall buildings nearby or not. It is specifically a narrow street with tall structures pressed close on both sides — the same geometry as a canyon, just built rather than carved — that causes the worst of it. Most walks and rides that start or end in a city only pass through that geometry for a few minutes at the start or finish, on the way to or from somewhere more open, so it tends to show up as a rough patch at one end of an otherwise clean track rather than a problem running through the whole thing.

What actually helps

• Getting a fix in an open plaza or a park before heading between tall buildings gives the receiver a solid starting estimate, the same advice that helps under tree cover or before a canyon stretch.
• Expecting the roughest points to sit right at the narrowest part of a street, between the tallest buildings — the effect tracks the geometry directly, so it is rarely evenly spread across an entire urban stretch.
• Not reading a jump to the wrong side of the street as a sign that recording stopped or failed. It is the same receiver, still producing a position every few seconds, just from a much worse view of the sky than it had a few minutes earlier.
• Comparing a distance measured through a dense street grid against a straight-line map distance is not quite a fair test, for the same reason it is not a fair test under heavy tree cover — the GPS-recorded line has picked up extra length from the wobble, on top of whatever the street itself actually adds.

None of this is something software can correct after the fact, because the receiver only ever reports what the antenna managed to see. TrailCam records a route in the background the whole way through, open ridge or narrow street, and the recording keeps running regardless of how rough that one stretch turns out to be. The finished route exports afterwards as GPX 1.1 or CSV. TrailCam is free to download.