I was assuming he mean't lateral position, but now that you point it out I am not sure, since he used the word roof vs. building. That would imply he looked at both the location and elevation? Still I suppose the software could explain even that , as navigation software for cars would assume you are not only over the road, but on it?
I think users of car navigation gps aren't aware of this, except maybe when pulling into a driveway, but hikers notice this, as I did just walking around downtown. Some hiker type GPSs have an option to do street navigation. I searched for something on this, that might be more convincing, read here, Garmin calls it lock on.
I have read that GPS is bad at elevation, from what I have read it has to do with the model of the earth it uses, but more because the triangluation isn't very good since all the satellites are very high, at least for hikers, not sure about pilots who could get more of them on the horizon, or maybe WAAS etc. A scene in the movie Enemy of the state explained this pretty well, when one guy complained about the satellite not being able to take a picture of a guys face, if he wasn't looking up. The other guy said maybe you could invent a new one!
Here is his link that he posted above:
Quote:
Trustful GPS measurements | -1 NO. Through many experiments with GLO vs just iPhone 5 GPS I should tell, the probablity of yourself walking on the roofs instead of the road is pretty high with GLO. I'd expect from GLO very reliable measurements. Unfortunately iPhone's GPS track often stayed on the road while GLO was jumping all the way to the roofs of buildings one would walk around. | IT'S OK. It was closer to the real road on more occasions. Still, fantastically, iPhone 5 was doing better I should tell (probably GLONASS boost makes the difference). |