Location Accuracy too easily bypassed?

Today someone kindly asked me to improve my location accuracy. I had not idea what they were talking about at first. I’ve done over 600 observations now, all of them being (I thought) highly accurate because I spent many, many hours marking their GPS coordinates in my photo editing software (this was with my older camera, before built-in GPS), which gets written to metadata, which iNaturalist reads upon upload.

Now I know people also want accuracy data. I would be happy to add that. I don’t think there is EXIF metadata for that (please correct me if there is). But I somehow got through all of these observations without the system notifying me or asking me if I’d like to enter something in for location accuracy.

It seems important enough to be “required” or at least asked for when submitting an observation. Am I missing something? Am I the only one who didn’t know we were supposed to enter location accuracy?

I’d like to hear what others say and then perhaps make a feature request.

4 Likes

You don’t need to enter it manually even though you can, you just adjust the circle of observation when you choose the location while uploading, you’re the one who decides what is the needed accuracy, but it should be reasonable and not 100+ km wide as it happens sometimes. The manual accuracy setting presented both on app and website. Requiring accuracy would prevent people from fast uploading with an app as gps system of smartphones register the spot of where you made a photo as a dot with accuracy 0 metres, so you’ll have to change it for hundreds of thousands of observations as far from always people change the dot to circle.

There is zero requirement to add accuracy if you don’t want to

I reckon accuracy is required generally, and then pinpoint accuracy for unique species.

It’s necessary, so you wouldn’t be required to do same later.

It’s not required, you can choose 0 or half of Eurasia as accuracy, but in the latter example someone can mark the observation as “location accurate - no”. Vulnerable species location is hidden automatically.

3 Likes

I would say the accuracy info is definitely useful for some data users, so it’s worth adding if you can. Some geospatial analyses will require an accuracy/precision estimate.

However, even though I use iNat data for research and want that accuracy data to be there, it’s definitely not required, and I don’t think it should be. I think having it as a requirement (or even a consistent pop up) would be frustrating for a lot of users and maybe discourage them from use. The primary purpose of iNat is to encourage and document people’s interactions with nature, not build the best database.

That said, promoting better awareness of how accuracy works is probably a good goal (not sure what that would entail) as I’m sure a lot of folks would enter that info if they knew how/understood how it worked better. I don’t know how to enter it into EXIF to allow for but this would also be a good option. The app does generally pull a value for accuracy from smartphone GPS (at least the versions I’ve used) so this helps a fair amount in getting those data.

I make most of my observations manually and enter the accuracy as I go, but manually adding accuracy to hundreds of observations would be quite a grind. I wouldn’t worry too much about it unless there are some really “high value” observations (like the one where a commenter asked for it).

7 Likes

welcome to the forum @dotun55 :)

whilst I certainly agree that accuracy is useful, that’s not the same as it being necessary. I’m just clarifying that users are not obligated to attach any accuracy to their observations

4 Likes

Hi @dotun55, welcome to the Forum :)

Thanks! Joined in just to reply to this post, and because a similar issue was raised on a recent special observation of mine, which originally was set to a general location >> https://www.inaturalist.org/observations/66085876

The accurate geolocation may be difficult to get w/out a dedicated GPS device, etc. My best resort is to extract the details from a mobile photo. The area has no mobile telecoms reception.

As an experienced iNat user, I’ve grown to realize the importance of affixing geolocation. And sometimes, more accurate geolocation is needed to support accurate ID by other users (like @cthawley mentioned). I do same myself on here, and in Bird Clubs, Wildlife Group on WhatsApp.
The extra effort to give as much details as required for unidentified/unknown species, soon becomes a critical aspect of iNat observations for user seeking ID success and (I believe) quality contribution.

3 Likes

ah I think we’re talking about two different things! (let me know if I’ve interpreted things correctly here)

So certainly the accuracy you’re talking about (i.e. that the coordinates are where they should be, not erroneously mapped elsewhere due to e.g. errors with google maps names) is 100% important + necessary

as far as I interpreted it, the accuracy referred to by OP is the accuracy circle, i.e. the estimate of accuracy around the pinpoint on the map. In many cases (e.g. when I take pics using my iPhone), the generated GPS coordinates are precise/give an exact point on the map, but a measure of accuracy (i.e. a measure of the coordinates’ potential error) is not generated.

So in the case of the observation you linked, you can see that the point on the map is surrounded by a blue circle with a radius of 211m, i.e. it’s effectively saying your observation was made at that point, but with a potential location error of up to 211m in any direction. Conversely, there are many observations that do not have that blue circle, i.e. the photo’s metadata does not contain an estimate of location accuracy/error.

The issue with this is that, unless you’re using some kind of military grade ultra-satellite/GPS, there will always be some kind of accuracy error (and even then, the uber GPS will still have error, even if it’s a few cm). So the argument often presented in these cases is that observations with no accuracy circle are actually misleading/‘wrong’.

4 Likes

Noted. Thanks!

2 Likes

i do mapping using iNaturalist data for work, and i definitely find the accuracy (precision) circle helpful. I don’t think it should be required, that sort of requirement just sets up more burdens to access and share data for whatever purpose. Others have use of low-precision data, though generally i do not. What I do wish is that there were a better way to filter for observations by accuracy (say, a filter to only show observations with accuracy recorded and 100m or better accuracy). I’ve got my poor or unknown precision observations too, but when i download my data for my own use, i always delete those records before using it for mapping.

5 Likes

Even when using extremely accurate GPS units there is still a margin for error that’s over and above the few cm that the GPS thinks it is. For truly accurate GPS measurements you need to measure both from a set of satellites (as all GPS units do) and from a known fixed location on the ground that’s also been running a continual set of measurements for a long time and correlate both sets of measurements with the exact time.

This is what high end systems like Trimbles do, and what we did with the German military Wildcat 2000 units we used for glacier research back in the mid-90s.

The accuracy measurements in even high end stand-alone GPS units is not quite as accurate as it pretends to be.

4 Likes

I see entering in location accuracy is not required. But it seems to be useful for some people and I wouldn’t mind entering it (especially if it were easy). I also appreciate that we want people to get observations uploaded, and if it’s too taxing, people wouldn’t do it. But it was very easy for me to not even be aware if it. Other webpages do things like have a popup or flag of some sort to call people’s attention to something, e.g. “Would you like to add location accuracy?” or something.

2 Likes

When I see a large accuracy ring, I wonder how accurate is the location. I find that GPS data reported by my camera (Nikon 5300) or my Android phone, can be miles off.
I try getting myself (not 100% reliable) to double check location on satellite view and then tighten up the accuracy range as much as I can. I try to have the observation findable by myself or others in the future in case someone wants to verify ID or collect additional data, e.g., seeds or flowers, a more magnified photo, . . . .

2 Likes

It is easy, there’s a window with it.

I totally agree with faerout. While location accuracy data are not required, they are an essential part of a good quality observation. It’s easy to enter the accuracy data when you are uploading observations on the web, but I still forget to do it most of the time. I use ExifTool to geotag my photos, and I tried to figure out if there was a way to add an accuracy estimate during this step, but I couldn’t find anything.

I really appreciate the warning pop-up in the web uploader that alerts me when I try to upload observations without location data or taxon IDs. I would love to have the location accuracy added to this validation step. If that’s too intrusive, it would be great to have an easy way to add location accuracy data after the fact, for example by adding the option to search for and update observations missing accuracy data in the batch editor.

3 Likes

https://www.inaturalist.org/observations?acc=false&place_id=any&subview=grid&user_id=treegrow
Try adjusting it for the bulk edit page, don’t know if it worth it, accuracy 0 is not a problem at all comparing with too large accuracy.

1 Like

Thanks, it looks like adding the acc=false parameter to the batch editor doesn’t work. I see plenty of observations with accuracy data in the search results using:
https://www.inaturalist.org/observations/treegrow?utf8=✓&q=&search_on=&quality_grade=any&reviewed=&geoprivacy=&identifications=any&captive=&place_id=&swlat=&swlng=&nelat=&nelng=&taxon_name=&taxon_id=&day=&month=&year=&order_by=observations.id&order=desc&rank=&hrank=&lrank=&taxon_ids[]=&d1=&d2=&created_on=&site=&tdate=&list_id=&filters_open=true&acc=false&view=table

If I could update observations in batches of 200, I would probably do that occasionally. But fixing them one at a time is just not feasible.

I recently learned that my phone doesn’t. I’ve been adding them manually. I check the map looking at several at a time to get an idea of how accurate they are then using batch editing add in a single accuracy number for all of them. I figure it is better to have an accuracy circle that is a bit too large than none at all.

3 Likes