I’ve recently noticed that the shape_dist_traveled fields in realtime bus GTFS bundle (both shapes.txt and stop_times.txt) significantly deviates over the course of the route.
I think it may have something to do with the change in June when the TSN was removed from the shape.
As an example, I compared the shape_dist_traveled field for shape_id 177 with the cumulative distance travelled calculated using the WGS84 coordinates provided. The shape_dist_traveled field is always larger – I get a deviation of over 300 meters at the end of the route.
I’m trying to use the shape_dist_traveled from stop_times to work out approximate trip distances and it seems like this doesn’t represent real life travel distances, as per Reference | Static Transit | Google Developers.
For example - this is showing maximum trip distances for Mondays. There are a significant amount showing trip distances about 400km. When looking at some of the trips >800km, their real life distance is closer to < 200km. (I’m assuming shape_dist_traveled is measured in meters, so have already divided by 1000).
Even when you take out regional trips, and only consider the Sydney Bus Network they’re still too high. Cross referencing those outliers (>400km), the real distance shown on the map should be closer to ~25km.
And then when you take those outliers above 200km out, there seems to be a divide between 10km and 100km values. When you consider Penrith → Sydney CBD ~55km this has got to be wrong.
I don’t think this is just an issue of increasing error with increasing stop sequence as some 2nd stops jump from 0km to >20km in under 5 minutes (over 200km/hr!).
Hoping this is just my misunderstanding of what shape_dist_traveled can be used for, but would appreciate some clarification.
