# Accuracy polygon

Hypothetical situation: I know I hiked along a straight trail and that I saw a bird somewhere along a 200m stretch. I set the accuracy circle to 100m radius but, on the sides, that 100m takes the circle over a cliff on one side, and onto a road on the other. I know that I saw it somewhere on the trail, but without a polygon I can’t simultaneously set the observation to being on the trail AND not on the road/cliff.

A polygon tool just like for creating places, or even just an option to draw a rectangle would be helpful.

There are many situations where you can look at the pin location and accuracy circle on observations by others, and deduce where it was more likely to be seen based on what you know of the species. I think it would introduce a significant level of complexity to the system without much in the way of benefit, but it would be interesting to hear the thoughts of others.

1 Like

I’m not sure that there is any meaningful benefit for accuracies that tight, especially with highly mobile species like birds. Even with sessile species (like plants) I don’t think it’s necessary unless you want other people to be able to go to the exact spot it was found in.

While this site does have the ‘research grade’ aspect, it’s more ‘presence/absence’ and less “where to the nearest decimeter was this seen” that you have in certain other research activities.

If you really need a finer accuracy you can scale the location circle down smaller.

3 Likes

Fair enough, you guys make good points. I was thinking of functionality mostly for maintaining accuracy while ensuring obs don’t leave the confines of places and miss out on projects.

2 Likes

That would be significant benefit… good point :)

1 Like

I think the project point is a good one as far as iNat goes. For use of data outside iNat, there would still need to be a single point estimate of accuracy (how GIS and other analyses work). I supposed you could just calculate the maximal distance from the edge of whatever shape was drawn to its centroid. But as folks have pointed out, this would add a fair amount of complexity.

I also see a small subset of observations with accuracy circles which are already obviously incorrect (like ones that encompass multiple countries, etc.) These are often from infrequent or new users, and they often don’t come back to edit the accuracy when there are requests in the comments. Having a more complex polygon tool for accuracy might be confusing for these types of users or result in messier data than the current point-based approach.

1 Like

Mine are mostly plants along the trail. Using a circle as a pin means most of that location area is wrong. I use the satellite view of the map to work out where That plant was. No GPS. It would be more accurate if my pin was a ‘rectangle along the trail’.

Perhaps the polygon pin as an opt in option?

Yeah, in my day job I am all over the accurate spatial representation of field data and its associated uncertainty. In the iNaturalist context, though, I can only imagine 25M ad hoc accuracy polygons interacting with several 100K place polygons to determine project membership, etc. (While those numbers may not be there yet, they would be coming fast…) Would be wonderful if the infrastructure could handle that and still keep the rest of the site functioning responsively, but I’m guessing that would be very expensive.

1 Like

Polygons are taxing on our system, and after talking with our developers we won’t be moving forward with this functionality.

2 Likes