First of all, thanks for a splendid service!
We have used the isochrone feature to determine the population inside a catchment area defined by an isochrone. However, there seems to be a bug somewhere, that was quite possibly fixed a few months back.
When running the isochrone for a specific set of coordinates a few months back, an estimated population was returned together with the polygon (I saved the geojson file)
When running the same set of coordinates now with the same settings, the exact same polygon is returned. However, the population figure is significantly lower (+/- 20% depending on location). It seems like the new number if the correct one however.
Has something been done on this functionality or the dataset that could explain this behaviour?
Thanks!
Hi @sveinung,
indeed, about a month ago we got a notice via email, that for a 1h catchment area the population data returned was way above the actual number, which was also cross-checked with other data.
This wasn’t happening on all calculations but mostly with larger isochrones.
It turned out, that the the calculation error was introduced by the PostGIS 2 version we were running on, which had some errors in an aggregation function.
After an update to PostGIS 3, the population data is now returned with the correct calculation.
Please note that this is still an estimate and not to be treated as an exact number.
As this came in via mail, there was no notice in the forum up until now. I’ll add an announcement for other people to catch.
Best regards
@amandus
Thank you very much for a prompt and good reply. Good to hear the error has been fixed.
One other peculiarity I have discovered however is that the isochrone for 60 minutes returns a lower population number than for 54 minutes for a few coordinate pairs. Take e.g. this:
https://classic-maps.openrouteservice.org/reach?n1=52.577356&n2=12.74818&n3=8&a=52.562428,12.956019&b=0&i=0&j1=60&j2=6&o1=1&k1=en-US&k2=km
Population within 54 minutes: 3,0 million
Population within 60 minutes: 2,1 million
Could something still be a little off?
Kind regards,
Hi @sveinung,
that’s a different issue…
If you hide all smaller polygons you will see the issue 
The lowest isochrone is actually a polygon where a large part is missing, which explains the lower population value.
This is somthing for a backend issue.
If there is none for this proplem, please create one.
Best regards