Summer in the city

Well, even with selective availability now “permanently” disabled, there’s always going to be a margin of error when GPS devices cannot track satellites adequately.

Since GPS signals are fairly weak microwaves, line of sight is fairly important, so cloud cover can distort signals slightly, as can trees. Sources of reflection can also warp signals in a nasty way, such as tall buildings. Most GPS devices cannot compensate for this.

Since GPS is essentially based on very accurate clocks, using calculations based on announced position of the satellite in orbit and the time taken to receive the same time-stamp from several satellites, errors can be easily introduced. GPS satellite clocks can be wrong causing errors up to two metres (less than a microsecond out) courtesy of relativity (thanks Albert), although they are set to run a little slower than clocks on Earth, and often get corrected from ground stations. Clouds and trees can introduce about another metre or so of error. Reflected radio sources can introduce from one to twenty meters error depending on how many satellites (“channels”) the receiver can use at one time.

If Galileo ever happens, it will to have a lot of new technologies that will fix these errors. It’s no wonder, since it’s a 30 year jump ahead in technology.

Google have issues with terrain and roads because of several different cartography systems in use. Some treat the earth as a ball, some as the more correct squished-ball shape, and some as if it’s simply one big rectangle. Different countries normally have their own standards, and calculations to correct errors between the systems can be really complex and still inaccurate. Look at steep cliffs or roads compared to the texture mapping.

Further to this, Google have further issues with their texture mapping not necessarily matching the co-ordinates on the sphere being rendered because they’re mostly based on aerial photography instead of satellite photography. Many defects can be introduced from humans, despite the incredible sophistication involved these days. Follow a lot of highways and eventually you’ll come across one which fades out at one place, but fades in a few kilometres away, where the next aerial photo takes over. In particular, look at cross-over points between aerial and satellite photos.

Personally, I think the 5–10 metre accuracy of something slightly bigger than a watch that depends on a system designed in the late-1970’s, being used on an overcast day, under heavy foliage, is really impressive!

Technology these days is amazing if you think about what’s behind it, but I too fall into the trap of moaning about how things aren’t quite right yet.