OpenStreetMap

Make sure that the source mentions a date

Posted by Jean-Marc Liotier on 12 December 2009 in English (English).

When I started, I tagged my additions with sources such as "GPS track". But I realized that in the future this might be a problem because ways change over time. So now I tag with something like "GPS track 2008" but I'm considering going even further in precision with source tags such as "GPS track Foo on bicycle on 20081204".

Now what is missing from JOSM is a way to filter the GPS tracks by date - for example, only show me tracks from the last two years. In areas that change, that would let us ignore obsolete data.

Comment from chillly on 12 December 2009 at 16:00

Every change in the database gets a date and time stamp, so you don't need to.

Comment from Jean-Marc Liotier on 12 December 2009 at 16:51

Modification date does not tell you how old is the specific source you chose. But as you suggest, maybe modification date is enough.

Comment from Ævar Arnfjörð Bjarmason on 12 December 2009 at 17:08

Just upload the trace and add a tag to the changeset pointing to it.

Comment from RichardB on 12 December 2009 at 17:14

If you make your GPX tracks identifiable, then they are downloaded as ordered points with timestamps, so tagging with a source date is not necessary.

Secondly, in most areas, the geometry of roads/other features will not have changed at all over the last several years, so it's not clear why traces from, say, 2 years ago, are any less reliable than those taken today.

Thirdly, in many areas, there are multiple traces, and most people will use some degree of "averaging out" of GPS errors in getting the alignment. These traces could be many months or even years apart - so tagging source dates are either wrong - or will require many dates to be added.

Comment from Jean-Marc Liotier on 12 December 2009 at 21:36

@Ævar : Linking to an uploaded GPS trace sounds like a good idea to me.

@Richard : The use case I had in mind was when in an area with multiple GPS traces I want to relate to a specific one that I trust because I carefully surveyed it.

Comment from RichardB on 12 December 2009 at 22:46

Unless you have professional GPS surveying equipment, then there is little reason to trust a single trace over multiple others, especially if yours is significantly different to the others. Use all of the traces and find an "average position" which hopefully will remove significant bias such as; GPS 'multi-path' errors, poor signal strength, having to move out to pass a parked vehicle etc. etc.

Having said that, you may often find traces uploaded where the signal strength was obviously quite poor and the traces jump around seemingly randomly. It's fine to put less "weight" on these traces, but I still wouldn't completely ignore them.

Comment from Ævar Arnfjörð Bjarmason on 13 December 2009 at 12:45

@RichardB: The point isn't that one trace is more reliable than any other, but supplying information about what traces you used for reference when making an edit.

Comment from datalogg on 13 December 2009 at 13:47

I'd like the option to discard points with a HDOP value above a certain value. People who sample five points a second and indiscriminately upload (i.e. without filtering the data first) are the bane of using GPS points, since their point cloud mess obscures the better measurements.


Login to leave a comment