The importance of location precision

I’ve recently been geotagging my photos “automatically” rather than selecting location for each photo individualy.
I use an app to collect a track of GPS data while I photograph and then merge those GPS locations to my photos. However, there are two problems that emerge from this process:

1 - The app records the location where I am and not where the subject of the photo is. So for example, if I’m photographing birds in a lake, the birds are in the middle of the lake but the GPS point will be in one of the margins.

2 - When I geotag my photos, data on the accuracy of the GPS location is not transmited to the photo. So when I upload that observation to iNaturalist, the location will appear as a point with no precision associated (no circle around the point).

Both this problems can be solved by manually reviewing all observations, but I was wondering if someone has found an alternative method to tackle these small issues.

And I’d also like to hear your opinion on whether or not the issue of missing observation precision is something that a user should worry about or if it is actually better to have no value of precision associated (aka meaning that the location is quite precise)

3 Likes

I think that in your described example, “camera shot” distance is certainly close enough for the pin. Unless the part of the lake you were shooting into was in a different county/province in which case you might want to manually change it. But considering how many of us are using cameras with no GPS and are putting the pins as best we can, you don’t need to be pin-point accurate.

I do think that a certain degree of accuracy is very important especially if you are making your data available to land managers (which prairie exactly was that in?). If you are logging insects then pin-point accuracy might be desirable.

Or, perhaps ? upload the cell phone photo first using the phone app to have location/date:time meta data uploaded to an observation. Then, later edit the observation to add the photos from the other camera?

6 Likes

See also:
https://forum.inaturalist.org/t/make-users-more-aware-of-the-accuracy-field-when-uploading-observations-through-the-web-interface/23494

Also maybe something in here is relevant:
https://forum.inaturalist.org/t/geotagging-photos/66

2 Likes

I upload via web and am always careful to make sure that the “accuracy circle” encompasses the true location. I semi-recently got a camera with built-in GPS, but after using it on a few observations I decided to manually edit the location when uploading, since the automatic point has a small accuracy circle and often does not cover the true location. I’ve seen other people’s observations (presumably made via app) with an obviously wrong location that’s close enough to the likely true location that I didn’t mark them inaccurate; e.g. a photo of gull tracks on a beach, located 50-100 meters from the nearest beach. I recommend using the web uploader if possible, and using the GPS location as a starting point rather than trusting it completely.

6 Likes

I find my phone’s GPS is generally accurate maybe 8 or 9 times out of ten. tenth time it’s inaccurate enough to be noticeable. I try to remedy this by taking multiple photos of the subject when I have the storage space to reduce the chance, but when in areas like heavy forest it’s sometimes better to just make the accuracy circle large to make up for the fact that I can’t tell if it’s accurate.

3 Likes

I think it’s not really necessary to pin the exact location. You always can set the accuracy circle when adding location. I usually set it below 100m and pin wherever place that i think correct and within below 100m from the exact location.
In some remote places, like in a forest, where i can’t be sure the location is, i usually set the accuracy to 200-300m.

1 Like

Sometimes mine is way off, sometimes it is spot on.

5 Likes

From my experience, scientific specimens in research collections don’t have accuracy associated with their locality data. So I don’t think it would be critical here on iNat (except for cases where you know the observation was made in, say, Texas, but not sure where in Texas).

2 Likes

Just make sure the location of the subject is in the circle. Overestimate if necessary. If your observation does include precision, make sure that the value is accurate. If it doesn’t include precision, then let the end user of the data beware and decide for themselves. To be clear, having no value for precision does not mean that the location is precise - it means other users have no idea whether it is precise or not.

As an aside, I have worked on some projects using iNat data that exclude observations with no precision, so adding precision may make your data more useful to some.

8 Likes

Depends on the phone, environmental conditions, geology, etc. Metadata can also easily be corrupted or edited to obscure or falsify location. Combine issues with phone GPS co-ordinates with the precision issues attendant to pinning an observation manually to a satellite image and there can be a fair bit of slop.

4 Likes

I often use my cellphone to take pictures, but the GPS is sometimes off. Many times it is off by a hundred yards but sometimes it is off by miles. I’ve learned to review the location and fix them manually.

Sometimes when I use my regular camera to take pictures of birds I also create a placeholder observation (with no picture) using the phone app just to get the location. I later upload the pictures into this observation.

Regarding the original topic, I would think the location where you are when you made the observation is fine, especially for observations of birds and mammals.

3 Likes

This is true of historic (pre-GPS) collections, but more modern collections, if they have field-collected GPS coordinates, increasingly do list GPS reading accuracy.
That being said, even the lowest-accuracy GPS/map point readings on iNat can be a lot better than some historic specimen localities where the provenance of specimens is sometimes unknown or exceptionally vague (for instance, specimens simply listed as being from “Mexico” or “Sumatra”).

5 Likes

That was one of my concerns. Having worked with citizen science data before for modelling purposes, I know that having data on each observation’s precision can be important.

3 Likes

Usually that is what I do, and I always try to correct wrong locations. But my original question was whether it is crucial to have data on observation precision (the radius of the precision circle)? Just because the software I use to assign location data to my photos does not assign the value of location precision

Just pointing out that you’re talking about accuracy, not precision. Precision is how many decimal points a recorded position has. One can record a location with very high precision, but be wildly inaccurate.

8 Likes

Thank you for the clarification. I wasn’t sure what was the english word for what I wanted to say. Downsides of not being a native speaker, sometimes I confuse some words

I wonder how users of iNat data know whether the accuracy of the accuracy is accurate enough? Just because an accuracy circle is present, doesn’t mean the accuracy circle is anything but a wild guess or purely fictitious (as most of them are, I suspect–because GPS recording devices either don’t document accuracy or that value doesn’t get used to generate the accuracy circle).
see tiwane’s post here

1 Like

I agree. If I have the precise location of a bird or an insect, it will have moved somewhere else in a few minutes. I believe that some observations of sessile organisms (plants, say) benefit from more precision. And as @pfau_tarleton says, most collections don’t have exact locations, though the general location is better than a general one. I think for most observations a location of 100 to 300 m. is adequate.

I’d agree that exact location (as in, down to several meters) doesn’t matter for a majority of organisms, only for ones that won’t have moved by next time someone comes in. Descriptions like “the large tree on the right of the trail entrance” may also be helpful.

1 Like