I’ve started obscuring the locations of some of my observations for privacy, but I still want to make sure they can get identified as research grade. Does making the observation very broad make it less likely that my observations will get to that level? I would love some advice on this.
Obscuring the location shouldn’t affect this. The rectangle your observation is randomly placed in is small enough to not significantly change the location for identification.
Setting the location to “private”, though, will affect the chances of getting to RG, as the identifiers will not know anything about the location.
What do you mean by this? Are you changing the accuracy circle to very broad? If so, I wouldn’t do that, as the location data is completely lost that way. Obscuring the observation via the “obscure” setting is preferable.
Private definitely affects ability to ID in many cases.
Using iNat’s obscuring feature won’t affect ability to ID in many cases, but can in some where there are two similar species that cooccur broadly but can be separated by location in some areas. If there are cases like this, you can obscure but provide additional targeted info to help ID (eg, found south of river XYZ).
It isn’t clear what you mean by
so adding more detail may help. If this refers to manually adjusting the accuracy circle, then smaller accuracy circles allow for better IDing and as accuracy circles get larger, there’s going to be a gradual decrease in ability to ID confidently/correctly, but the actual impacts will vary widely from species to species. For some species, it won’t matter at all, for others it will.
Got it, thanks! I’ll stick with location obscuring instead of private to keep it possible for IDs to be research-grade. Appreciate the help! For clarification, I’ve noticed that obscuring can make the observation’s location so broad that it is almost out of my province. So my question is, will obscuring make it less likely for identifiers to know what the species is if the observation is like 20 miles from where it was actually taken.
Oh okay. I doubt that it will have any effect. Species ranges are usually so large (with exceptions like islands) and usually don’t have abrupt boundaries that 20 miles don’t matter. As cthawley said, if there are any natural boundaries (like large rivers) and the observation was placed on the wrong side of it, it may be good to provide more info in the comments. :)
They can be brought to research grade, but one should consider what that means to be an observation with 20-miles of uncertainty. It becomes surprisingly less useful for geographical and other applications; coordinate uncertainty is a heavily reviewed issue in scientific literature, and in most cases I’d ask, why not just manually obscure 1-5km, which seems plenty;
A 20-mile obscuring radius encompasses A LOT of ecosystems and climatic conditions, especially depending where you are in the world. Sometimes it is useful to simply know an organisms exists within a 20-mi radius, or to consider species ranges at a viewpoint much coarser than 20 mile resolution… but for commonly observed species, it is often obscuring a redundant locale, making including its uncertainty an even more unfortunate mistake. I don’t find 20-mi a valuable resolution to report in near any circumstances, so I do not obscure my observations more than 1-km, if possible.
There are also cases where it is useful for the observer to detect whether their observation requires geography to ID appropriately, or they are contributing possibly unhelpful data by overly obscuring; short range endemic species depend on geography and tight range edges, 20-mi will comfortably throw congeners about, and the iNaturalist app does not make it obvious (to me) how obscured a locale is. I conjecture there is a concern of misleading observations, and have seen the issue in a few resesrch grade Hypochilus observations.
“Obscured observations display the 0.2 x 0.2 degree rectangular cell encompassing the hidden true coordinates.” So the area varies by latitude, but it can be figured out (not by me).
Size of an “obscuration cell”: about
22×22 km (14×14 mi) at the equator
22×16 km (14×10 mi) at mid latitudes (NYC, Rome, Melbourne…)
22×9 km (14×6 mi) close to polar circles
= Cell area wide enough to make poachers unhappy. Slightly annoying to ecologists, too.
The benefit of using the built-in obscuring feature is that precise location data is still stored, it just isn’t displayed.
The observer can choose to trust researchers (or anyone else) with it.
So there is actually not a 20-mile uncertainty, it’s just that the actual location and uncertainty radius are hidden from “the public”.
As far as I can tell, that is not true for data accessed from GBIF (the way iNaturalist suggests their data be used for researchers and cited in publications).
I’m not sure exactly what you comment means, but for obscured observations the true coordinates are not sent to GBIF. The coordinates on GBIF are the “display” coordinates from iNat (randomly chosen in the 0.2 x 0.2 degree box that contains the observation), and the accuracy value sent to GBIF if for that box, I believe it’s diagonal (the max distance the point might be from the true observation location).
The data (as accessed from GBIF) show an obscured location, as can be seen when you look at “Coordinate Uncertainty.”
So I don’t understand what is unclear about my comment: if a species is on GBIF, and if it includes an obscured data entry, then that data point is still obscured (by my assessment of the coordinate uncertainty data accessible in the dataset). Am I incorrect?
This was in reference to the point that obscured coordinates are accessible to researchers. They are not immediately accessible by users of GBIF, the method of use iNaturalist suggests for citations and accessibility.
Yep, a researcher willing to access more “precise” coords can’t contact GBIF and ask them to un-obscure things; they’ll have to chase the original author on iNaturalist, create an account there, and kindly ask for access to the true location.
For taxa having a lot of precise data available on GBIF, I suppose most researchers would singlehandedly filter out any point with such a large imprecision, anyway.