Yes, there are definitely observations that were not within the accuracy circle. I meant many can’t be determined to be or not be, though there are exceptions like the duck. I agree though, I don’t think they should be thrown out entirely and I also don’t think people should go around aggressively culling these observations to casual, as that is not helpful to researchers.
This is why I have a lot of pinned locations that started with the Google Maps location but that I widened the circle to include the whole park, for if I can’t remember where it was taken and my GPS connect app isn’t being any more helpful. (Honestly, the only reason I use it is that it will add at least a general location to my photos so that I will have that information if I come across the photo ten years in the future; it’s not useful for iNat at all.)
I gave up that battle on another obs.
But I do push to Casual random obs ‘somewhere in Africa’ when their other obs that day are clearly out of Africa.
I would push back on the “no visible body of water” criteria. There are many shallow wetland habitats that completely dry up for part of the year. Depending on when the Google satellite image was taken, you might see only a dry field or meadow, but that area could be underwater other times of the year. Plus, a heavy rainstorm could easily flood a low-lying field to the point where you have ducks swimming there, even if it is normally just grass. The same thing goes for rivers. Some of them will greatly expand their size in flood stage, so you can’t just look at the Google satellite image to determine where the riverbanks should be. At least not at a precise level.
to be clear, my judgment that the example duck in the water could not have been in that accuracy circle is not based on Google satellite images. i’ve been on the ground in that Park, and i can guarantee to you that you would not have been able to capture a photo like that at any time of that year within that circle. additionally, if you look at the user’s other observations from that day, it’s even more clear that they all have used the same method for setting their location.
I see this all the time (multiple users having multiple observations at a single spot within a public park or similar) and assume it’s because they enter the location name and this is where the marker ends up by default.
I’m also seeing an increase of these popping into casual marked as “location inaccurate” even though many have a large circle and quite likely the true location falls within that circle. Rarely is this preceded by a question about accuracy or accompanied by a note to the observer - these things seem to go casual silently.
I find the second thing here more alarming. I usually lean towards not marking things casual without first checking with the observer or at least leaving a comment so they get a note and can check and if necessary correct the marker. I’ve started making comments when I come across these in casual and quite a few have been promptly fixed by the observer.
Camera and “GPS Tracks” phone app. The app logs a track and stores waypoints. I can email both to me and, at the desktop, I can download both to Google Earth and grab the lat/longs for iNat and add my photos.Best accuracy.
Camera and phone. I take photos with my camera, but also one with my phone. At the site, I upload the single photo and the lat/long that comes along with it to iNat (if I have connection). At the desktop, I’ll add better images.
The phone images are good enough and I have a connection - upload directly and no post-processing (which can get tedious, although more accurate).
I have tried to figure this out, and it was not simple. (And I’m pretty tech savvy!)
One of the issues is that specific app suggestions depend on your phone and most of the recommendations are for paid apps. I looked up Geotag Photos Pro and while they do have an android app, I can’t even find the price on their website, which does not inspire confidence.
Currently I use the Nikon app, which is erratic, and I don’t know if it’s the app or my phone that doesn’t know where it is. But the entire workflow is as follows:
1- Turn on the app on my phone.
2- Take photos.
3- Check the locations when I upload the photos and correct the ones that are wrong.
For researchers, it would be very helpful to know which locations have been entered in this way, i.e. using a Google location name and not adjusting. Wouldn’t it be possible to have this information recorded with the observation ? That would certainly help automatize the sorting of thousands of data when needed.
Just a quick note that I see the location accuracy as 1.19 km, which is much more reasonable than a
This park is a good example of where the Google provided circle captures the central park area reasonably well but totally misses an outholding and sticky-outy bit because of the non-standard geometry.
i think iNaturalist takes the main point and uses that as the base coordinate, and then i think it uses the viewport somehow – maybe half the (width + height)? – to determine the accuracy. (i’m sure someone could look at the code to the exact algorithm, but i’m not going to do it today.)
you can see that in this case, not only is the viewport in Google too small to encompass the entire park to begin with, but the resulting accuracy circle (1.19km) in iNat is also too small.
I’m in the same boat–no GPS in the camera and tying with an iPhone is a hassle. I’ll note also that in a lot of instances I’m uncomfortable giving exact coordinates of an observation, e.g. a rare plant or reptile. Just using a general location with an accuracy circle can help protect organisms from those who might exploit them.
I do wonder how much it matters to a researcher who knows enough to pay attention to such the observation metadata if the location accuracy is set to 1,000m or 10,000m, or 100,000m. After a certain point, it is just ‘bad’, and I don’t think I would trust that exactly how bad it is has been precisely quantified?
When I see accuracy circles that encompass a whole state, several states, or, in a few cases, a whole continent, then I know that the OP was not paying attention to that particular detail of the upload process! In my oldest observations, dating long before GPS was in practical use, I have had to place some locations with an accuracy of up to about 5 km radius, but I view those as still appropriate and useful observations.
The tolerance for accuracy in natural history data is necessarily a function of the context in which it will be used. IF someone is interested in the occurrence of a particular butterfly or dragonfly species in northern Taiwan back in 1969, then my observations have utility. IF some researcher wants to refind the precise location to collect a modern specimen, then my observations will be of no use.
Geotag Photos Pro also has poor reviews in the Play Store, I noticed. Maybe their iPhone app is better, but the Android one definitely doesn’t seem like something I’d pay for.
I’m using Nikon SnapBridge, and it is erratic. Today I used it and it recorded locations for 11 out of 468 photos (and the accuracy was “point where I opened the app” not where the photos were taken)… no idea why. Usually it’s better than THAT at least. I just did my best to make a mental note where each photo was taken; luckily I could edit and upload them today or it would’ve been more difficult to remember.
I’ve had the opposite problem. I found a spider I know where was within about 30 mm, and tried to put 2 dm as the accuracy because I converted it to degrees with six digits after the decimal point. The site takes only integers for the accuracy. I have a GPS system for surveying which can get within a few millimeters if I cook a point for three minutes. That traverse point is in the woods, so I had to traverse to it, which makes its position a little less accurate than the two points on the road I started from.
The iPhone version seems to have much better reviews, also.
If anyone knows of a good app specifically for Android (camera is Nikon) that’s better than SnapBridge I would love to try it… it’s such a pain to have to go through and manually enter all the location data.
