It took some time, but friday the GPS data logger arrived. It is a Columbus V-990, produced by http://cbgps.com. It comes with some software written in Java, so the most important parts are compatible with Linux, too. The first test failed, because I tried it in motion tracker mode, which does not seem to work very well at walking speeds. Got a track file with about 6 dara points for an afternoon’s worth of walking. Luckily I also had my phone’s GPS tracking in parallel, so I got some useful data from that. Yesterday I did another experiment, this time going to the Sants quarter’s alternative summer party near Barcelona Sants station. There I did not move very much between fetching beers, which helped with collecting variations in sampled position over a couple of hours. I am currently still evaluating that. The thing is, the device produces a CSV formatted log file with values filled up with null characters. This is extremely stupid, since this way it needs processing to be imported in spreadsheet programs as OpenOffice Calc or Gnumeric. If I quit the null characters, their own processing software for exporting to standard formats TimeAlbum can’t read the file anymore. So I am currently working on a small tool to filter out all the points with low accuracy (PDOP >= 2.0, perhaps) while keeping their silly format to use their GPX export afterwards.
Well, my first impression is that under urban conditions it does not make much sense to try and collect precision data with GPS, even in sparsely built areas.