Why do observations map outside county borders?

Perhaps before the change is a good time to post an observation I noticed the other day. Here I selected observations for Travis County. I see some are clearly within the Travis County boundary, but others are obviously in a square grid that approximates Travis County. It appears there may be some observations from Travis County that may be missed, and some observations that are actually from nearby counties. Care to comment on what’s really going on here?

1 Like

I would guess at obscuration, with the pin location being moved to a random location within the same “grid” as the original observation.

3 Likes

The observations associated with Travis county, TX that are mapping outside the county boundaries are obscured observations with true coordinates located inside Travis county. The boxes that emerge here are the 0.2 x 0.2 degree cells within which a random location is assigned for mapping purposes. If you zoom in further, you’ll see the markers that lie outside the county boundary are different than most of those within, indicating that they are not showing a precise location.

You can read more about different types of geoprivacy here, or see the details for a specific observation by clicking “details” below the map view, e.g. for this obscured Travis county moth.

01%20PM

Housekeeping note: I’m moving this to its own thread since it’s not directly related to the new type of map display.

4 Likes

OK. That makes sense, and I thought that’s what it might be. I would suggest that in some situations, this obscuration method would not be very significant in certain edge cases: consider a county whose outline just barely goes past a grid line. Moving to a random point within the grid would then put the marker outside the county. Then looking at the map, taking the intersection of the grid and the county, the actual location would be easy to spot. We can sort of see this on my illustration and the one posted in reply. Note the two squares on the right. The northern one has much fewer points than the southern square. This is probably mostly because the intersection between the square and the county is smaller for the north square than for the south square.

Note, for example, on my screen shot, that there is one circle just north of the Liberty Hill marker. We can presume that the actual location for that observation is at the intersection of the square containing this circle and Travis County, ie., the very northern edge of Travis County.

A better algorithm would be to shift the location by some random amount. The random number could be based on a seed so it’s always shifted to the same spot. Or it could be truly random so that it is shifted to a different spot for each viewing. In the latter case, repeat viewings could result in a good approximation to the actual location by averaging multiple views.

Anyway, this is probably not a serious consideration except where obfuscation of location is critical.

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.