Really good civilian GPS units, e.g., used for surveying and field engineering applications, are capable of accuracy to within half a millimeter. But they use a technique called "real time kinematic" measurement, which relies on the carrier wave of the transmission rather than by matching psuedorandum transaction identification numbers. They take anywhere from forty-five minutes to three days, while they remain absolutely unmoved, to make that kind of measurement.
Garmin (the only specific brand of handheld device I'm familiar with) can get within a few inches by taking repeated measurements while remaining absolutely unmoved, over a course of a few minutes. The process can be repeated to improve accuracy over time. Ordinarily, just walking around, the Garmin devices are accurate to within about eight to ten feet; that improves if you're using a device-powered external antenna to about three or four feet.
So, if someone were charged with the crime of walking too close to a "gun free school zone", and I were hired to represent him, I could use my device in location-averaging mode to identify the specific location of the property boundaries and his location at the time of the incident. That takes hours, mainly because of all the walking around and time-consuming measurements. That data can be plotted on a USGS topographical map, which is admissible into evidence because it's prepared by the government. The points plotted can be separately authenticated by my "paralegal/research assistant" (since he doesn't work for me on a W-2 basis, it's illegal for me to call him an "investigator", because he doesn't have a license from DCJS) as a witness to the process (i.e., I do the investigation and he observes the process).
I have such a case right now, the charge is reckless handling of a firearm while hunting. The complaining witness was over a quarter mile away at the time of the incident, and on the other side of some forested areas. The collected GPS data can establish where my client's shotgun was, where the complainant was, and where the relevant structures on the land were. (Without the need for line-of-sight measurements.)
My point is that even though my unit is supposed to be optimized for hiking and such, I wouldn't trust it for walking past a GFSZ, unless I had at least a twelve foot buffer between me and the zone.