I’ve just noticed this issue with observations on the edge of a place, and I’m not sure what’s caused it. One the edge of the place, some observations are included and some excluded, and I cannot work out why.
The place in question is this one, although I haven’t checked any others to see if they have the same issue. On the very edge of the place, some of the observations are simply excluded (i.e. if you narrow searches by place, they will not show up) and I cannot work out why.
Here is the map when no location restrictions are in place:
And here is the map when I restrict it to just this place:
Why are these observations excluded?? It’s clearly not proximity to the border, because the far left and right observations are closer than most of the ones that were excluded. I thought it might have been something to do with location uncertainty, but having a look at the sightings this is not the case either. E.g. this sighting is included while this sighting is excluded, and yet both have the same coordinate uncertainty, and both have some of their uncertainty area outside of the place.
So I’m lost as to why some are excluded and some included. My understanding was that all non-obscured observations that have their ‘primary’ coordinates inside the place should be included. Am I just missing something obvious??
i’m fairly certain this isn’t a bug. the exclusion of an observation from a community-created place occurs when the accuracy circle breaks the bounding box of the place, not the boundary polygon of the place. in this case, the southeast corner your place is very close to the bottom of the bounding box and is less close to the eastern extent of the bounding box. see this for more info: https://www.inaturalist.org/pages/help#placeindex.
I’m not 100% sure whether it is the location polygon or bounding box, but it definitely looks to be an accuracy circle issue.
See these two observations which are the same location
https://www.inaturalist.org/observations/100061053 (50m accuracy)
https://www.inaturalist.org/observations/109745362 (100m accuracy)
The one with 50 m accuracy comes up as inside the place, while the one with 100m does not (even though it is farther from the boundary). So if that accuracy circle crosses the place boundary, the observation might not be from the place. The system takes a conservative approach and doesn’t register the observation as from the place.
FWIW, the red box here is an approximation of the place’s bounding box:
Looks like the two observations @cthawley pointed out were close to the southern edge of the bounding box in that weird corner, so it makes sense that the differences in radii would keep one out of hte place.
On our test server I changed the accuracy radius of https://www.inaturalist.org/observations/109745362 to 50m and it now appears in that place. So I don’t believe there’s a bug here, the system is working as designed. It’s being conservative when it comes to accuracy and places.