How do you handle observations where the “Accuracy” is so large that it takes in a whole continent? I frequently see observations that are put on the map in Central Australia, with a circle of accuracy that takes in the whole country - giving an “accuracy” of 4 or 5 million metres. The circles often also take in New Zealand, PNG, a lot of Indonesia, and multiple smaller island countries. This is meaningless, really.
Do you ignore this and ID as usual? Ask the user to improve the accuracy? Or mark the observation as “No” for Location is Accurate?
I tend to assume that the real location is close to the centre of the circle, as this is probably how most people use it..
Marking tge location as inaccurate is wrong, as probably it is within the circle, so it is technically correct.
If you feel you cannot make an accurate ID without a more concise location, I would go as far as you can, maybe '@tag the observer and see if they are going to make the circle smaller and if not you might want to think about using the DQA “good as ot can be” ic you are sure there a indistinguishable species within this circle
Please refine your location. Check the point (zoom in deep) and reduce the Location Accuracy [= error or precision] as to as small a value or circle as is valid.
The app should be accurate to 5m, but this is 14km radius.
When first switching on the app, please make sure the GPS is on, and give the app about a minute to find out where you are. Once found, it should remain accurate to metres for the rest of the period until you close the app.
If you take the pictures with the camera and then add them to the app, you will need to manage the Location Accuracy manually.
The phrase “Location is Accurate” in the Data Quality Assessment is misleading. It can’t possibly refer to the accuracy of the GPS coordinates since many observations are inaccurate in that sense. In fact, many observations have no GPS accuracy at all! Since that is not visible in the user interface, we usually let it slide without comment. Out of sight, out of mind.
I don’t usually make a comment on observations with a poor or non-existing GPS accuracy unless the observation is of special interest. Since the platform has ready access to this data, it seems like the user interface is responsible for this. A simple approach would be to make the GPS accuracy visible to the user in both upload interfaces. Currently it is hidden in the web upload interface (I don’t know about the iPhone interface). Again, out of sight, out of mind.
Maybe a solution, and not saying it is the solution, the system should have a default accuracy.
For example 50 meters.
That will prevent accuracies greater than the country it was taken in.
I think most general users do not realize how to set the accuracy, or the importance thereof.
Also, possibly in your personal settings add the option to set your default accuracy. I know my camera GPS is good to within 10 meters. I could set this as my default, and then rather adjust the specific observations accuracy I needed.
If I have gone into the observation far enough to see the accuracy circle, it must be an observation which interests me. So I would ask the observer to amend the circle to something more useful.
I filter out user obscured observations, so I haven’t paid much attention to how the accuracy filtering might be affected by the obscuration.
(in other words, my filter requires that geoprivacy be open, so the accuracy filter doesn’t really come into play)
The URL I have bookmarked for my identification work is actually quite long at this point (filtering out problem people, problem projects, etc. etc.). I wish there was a less fiddly way of setting up my identification filtering. But since the URL filtering mostly works, enhancements to that function are pretty low on my wish list vs other things that would help me work more efficiently (for example).
Please do not automatically pester the user when you see this. Large accuracy circles can result, for example, when entering historical observations and all the user can remember is that something was seen in Australia, so he types “Australia” in the Google search box, and an accuracy circle such as you describe is automatically created. Some people have lots of such observations, and it can get really annoying to receive multiple comments asking for greater location accuracy. Sometimes the cue for this is in the description of the location – if it is a broad description, then it follows that the large accuracy circle is (likely) intentional and cannot be improved.
Also, marking the location as “not accurate” is untrue and improper. As noted in the other threads linked to, hiding observations with large accuracy circles from range maps makes sense, but can be done independently of research grade status.
Finally, be aware that lots of people don’t have GPS coordinates embedded in their photos – many high end cameras don’t even have this functionality. So assuming that a pin is the exact location of an observation and/or that the accuracy circle can necessarily be improved is jumping to conclusions.
And I have no illusions that a low value for the “accuracy” (which should really be called “coordinate uncertainty”) guarantees that those coordinates are actually accurate.
In many cases, I know with a high degree of certainty that they are not. But filtering out observations that are obscured and/or have large accuracy figures is a start.
Most observations I see with large location circles are not intentional, and the user is generally not aware that it covers thousands of kms. So it often does makes sense to check with the user, particularly if the observation is recent and particularly if it is not a consistent pattern in their observations.
Identifiers have to do the best we can with the info an observer provides. If you feel ‘pestered’ we feel irritated. Glad I don’t have to ID something ‘somewhere in Australia’ All those different climates ! wow
Accuracy and precision are two different things. The “location is accurate” question is asking if it’s accurate, not if it’s precise - it has to be correct, but doesn’t have to be a small circle.
Thank you, I think that is probably the cause of the ones that I see so often. So lots of things that couldn’t possibly occur in Central Australia end up plonked there, with huge circles. For anyone not familiar with Australia, there is no coastline, or rainforest, in the middle of the country, but photos taken in those environments get put there. When identifying, if I see a photo of coastline with the location showing as somewhere a few thousand Kms from any coast, my initial thought is that the location is wrong. But then I check the accuracy circle to discover that it contains the whole coastline, and don’t know what to do about it. So at least now I have a bit more understanding of how that happens.
I think maybe from now on I will just ignore such observations.
I generally consider the location to be inaccurate as the observation could be anywhere within the bounds, the situation you described happens when someone selects “Australia” as the location rather than a specific place, the observation is rarely from the centre of the circle
I hope this isn’t too much of a digression from the intended topic, but I do always feel badly when someone objects to “obscured” locations. 100% of my observations are on our property. Most are in our backyard. And… I just don’t want to make that universally available. My location accuracy is typically down to the meter, and I have “unobscured” for several people and all project moderators. But the obscured range itself is so small as to have no effect on what species would be appropriate here.
Am I missing something? Is there an objectionable consequence of obscuring that I haven’t realized? I’m not sure there are any consequences that would persuade me to un-obscure my house, but I’d still like to know…. :-)
The boxes I see on obscured records are about 12 km x 25 km. In this country, UK, most mapping schemes are at the 10 km scale or finer, so the obscuring is enough to put false dots on a map. But I expect mapping schemes would just exclude records with the location obscured.
Even county-wide accuracy circles are a problem with some of the species I ID, as it means potential range overlap makes a species unidentifiable. I have generally had very good luck with letting a user know that they have to be a little more specific if they want an ID to species on these organisms, how to make the accuracy circle smaller on different platforms, and if they don’t know a more specific location, letting them know it can only be IDed to genus.
If the accuracy is large but the species is still IDable, I just ID it.
However, I have encountered a user who consistently uses a county-sized accuracy that includes multiple potential species. When asked to lower the accuracy, said user has said where exactly the observation was made. They do not correct the accuracy, simply put the location in a note. I have so far considered this good enough evidence in terms of data to ID to species, though I’m not sure that is the correct move.
However, they continue to use the same sized accuracy circle, and when I have made the same note of needing a more precise location to ID (forgetting that it’s the same person because there are a lot of observations), they have responded with irritation (“as I say every time, it’s at x location” sort of thing) .
My question in this case, if I do remember who this user is and that they always post observations from the same general location which is species IDable, but their accuracy circle is always too big to ID to species, from a data standpoint, is it better to leave it at genus, or, since they have told me in the past it’s always at x location, and that location is within the accuracy circle, ID to species?
All those different climates ! wow