Recently I had a discussion with my friend who never had his accuracy radiuses added - he thought those weren’t ‘‘needed’’, because his localities were very precise, taken from GPS.
However this actually is opposite, because if accuracy radiuses were missing, we don’t know whether you’ve picked the location while looking at the map very broadly (e.g. a whole country), or you’ve assumed it was there from the aerial imagery, or if it’s taken from a GPS equipment.
Even from a GPS, different GPSes have different accuracies, so it’s always good to add specific numbers such as 1m, 5m, 100m, 1km, etc.
(1km might sound a bit silly, but it often happens to me when I walk a whole coastline and forgot to turn on GPS)
After explaining this to my friend, he started adding the accuracy reflecting the accuracy of his GPS (e.g. 1m), which was good - sometimes I see other observers who haven’t realized the importance of this, so I made this topic as a reminder.
i disagree that it’s particularly important. it’s fine to capture the information if you like, but there’s no need to ask / “remind” everyone to try to capture the information.
this has been discussed many times in other threads. here are a few of my posts that expand upon what i’ve said above:
I agree it’s a good thing to include accuracy with observations. If I were analysing iNat/GBIF data, one of the filters I would probably use is to throw away everything with accuracy values that are too large, or missing.
It’s important not to just add specific numbers uncritically, however. It’s good to check that the resulting circle does in fact include where the organism was observed. Even with a device that is capable of recording to less than 1 m, not every GPS fix will be that good, so I’d suggest erring on the side of caution and putting 5 m or 10 m - and checking some observations to make sure that seems reasonable.
Thanks for this reminder. I have probably hundreds of observations that are based on standalone camera photos that were geotagged after the fact using a GPX track, none of which have accuracies specified. Thankful that batch editing allows you to apply an accuracy reading (5m in my case seemed reasonable) en masse.
Edit: Actually, I just realized almost all of my observations have no accuracy value set – Android phones don’t seem to encode accuracy values in images’ metadata. Perhaps filtering out images with no accuracy set is not the best idea data-wise? I have a feeling less than ~5% (made-up number, but feels realistic) of Android users are going add an accuracy level to their observations… and I’m not sure that bulk adding fake accuracy values to my observations is a good idea – fake information seems worse than no information.
This is a tough one for underwater photography. I tend to use the entry location from my Garmin GPS dive computer as the basis and estimate the direction and distance traveled as the error location. When GPS is not available (underwater, underground) you have to get creative.
I tend to double-check the recorded GPS anyway, because the satellite has told my camera that it was up to a klick away from where I know we both were! It beats reconstructing my path manually, but it’s kind of funny when the pin suddenly parks me in, say, the middle of an airport runway.
Luckily, when that happens, switching to satellite view usually lets me get within three meters of the right spot.
I don’t usually do identifications on observations where the accuracy circle is missing or I feel it is too large. The resulting record is of little use and I don’t want to encourage sloppy recording.
Lately, this has been happening very often to me. My Camera phone app simply stopped forcing the Android phone to get a new location with the GPS. If I do not open a map app before taking the picture, I can have very inaccurate coordinates. To some extent, it is cured by having a mapping app to record the location in the background in regular intervals, but I find out, to my dissatisfaction, that even then, some of the coordinates are off, even if fewer of them. I fear that many observations of observers who do not specifically care about this issue might be affected, although those are more likely to use the iNaturalist or Seek apps, which should likely force the phone to get the current accurate location.
To get the real coordinates I sometimes have to download the GPX record of my walk or ride from Garmin or Strava and get the coordinates for the timestamps from there, which is tedious.
it’s up to you whether or not you want to identify a particular observation, but it’s absolutely not accurate to say that not including an accuracy value makes an observation “sloppy” or “useless”. you can have that opinion if you like, but don’t go around telling other people their observations are “sloppy” and “useless” because that’s just wrong.
I don’t think I have ever told an iNatter their observation was sloppy or useless. That would be rude. I am sure with your iNaturalist interrogation skills, you will be able to find any examples where I have insulted someone with those words and I will be happy to apologise to them.
I sometimes suggest people could be more precise with their locations and I explain how that would make their observations more useful and maybe benefit wildlife conservation. And occasionally they agree and change their ways. But mostly when I see what I consider to be sloppy practice, I just move on without giving an identification.
So I suggest you don’t go around telling other people what they must not go around telling other people unless those other people have told the other other people what you don’t want them to go around telling them.
Most of my observations (some 55,000 of them) were taken with my Olympus Tough camera, which has built-in GPS, but does not specify an accuracy radius. Should I go back and add a figure for accuracy for those observations?
Have your observations got a location name that corroborates where you put the place marker? That can be just as good in my not unanimously supported opinion.
This would be difficult to do reliably (can you be sure that whatever value you add - 10 m, 100 m - is correct and encompasses the true location?) If you can check on the map and confirm that the points are within such a radius of where you remember observing, you could do this, but for so many observations, I am not sure that it is worth your time and effort.
For some (many?) research purposes, those observations will still be usable and of value. But if it’s not too difficult to include accuracy values on future observations, it will enhance the value of those observations.
I honestly can’t say I’ve noticed the accuracy radius when I’m doing IDs… I just checked and I see now there is a field that says “Acc”, but I hadn’t even noticed it over nearly a quarter million IDs.
I’m curious if all the herbarium and museum specimen records of the world are also “of little use” and “sloppy recording” since they also generally provide GPS coordinates in their data without any radius of uncertainty specified. I agree the uncertainty radius can be useful data, but I hardly think that lacking it is a fatal flaw in a record.
My assumption is always that if a user meant for their location to be very approximate, they’d either say “location approximate” or put some big uncertainty radius on it. If they don’t do either of those things, I’d be willing to bet that 99+% of the time they mean for the location provided to be pretty precise, but their camera/phone/whatever doesn’t record an accuracy bubble. Whether “pretty precise” means within 5 m or within 50 m doesn’t really matter to most projects.
And as for the accuracy of the accuracy bubbles themselves, as someone who posts nearly all my stuff from iOS, I can say for certain that Apple’s automatic accuracy bubbles very often do not include the correct location. My GPS throws me a km away in the middle of a river with an accuracy radius nowhere near my present location on a regular basis. I check the map manually every time I make an upload and correct the location as needed, but my point is that if most users are just getting these accuracy bubbles from their phones/cameras, they’re far from perfect, and I strongly suspect that if you just artificially put a 100 m accuracy bubble on every observation lacking accuracy data, that that dataset would include the correct location in the radius at least as often as a dataset with the accuracy radius included automatically by the observer’s device. There’s no way to test this that I can think of, but anecdotally, it would certainly be true of my data if I weren’t manually altering the locations from where my GPS thinks I am.
My best overall estimate of the usual accuracy of the latitude and longitude the Olympus camera spits out is that it’s within 10 m. But there are certainly times when the camera GPS can’t find satellites or enough satellites or whatever, and the resulting location ends up on the wrong side of the river or road or is otherwise obviously wrong. If the camera can’t give me any location, I interpolate among the points I do have and my knowledge of the site and assign a point with an accuracy radius. (Which is one reason why I prefer to upload photos the day I take them, because my memory of the site fades over time.)
You make a very good point about the accuracy of pre-GPS, pre-digital-photos locations for museum specimens. For many older plant specimens, the only location data noted on the specimen is the town.