Missing Location Accuracy data?

Hi all

I received a message from another iNat user informing me that a large proportion of my >20,000 observations are useless because they lack “Location Accuracy” data.

I mostly use an interchangeable lens camera to record images (DSLR and mirrorless cameras). I only infrequently use the camera app on my phone. Phone images are geotagged automatically. For images taken with other cameras, I have used various methods over the years to geotag them - (i) recording GPS tracks with phone or a GPS device and synching GPS locations with time stamps, (ii) a Bluetooth connection between camera and phone that allows the camera to grab GPS coordinates from the phone in real time, and, most recently, (iii) a mirrorless camera with built-in GPS capability. I then upload images through the web interface.

Regardless of what camera or phone was used and the method of GPS tagging, the “Location Accuracy” field is by default left unpopulated the vast majority of the time. On occasional phone images, I sometimes see a value in the “location accuracy” field.

I’ve been told I need to go back and add some arbitrary value to the “location accuracy” field for these observations in order to make them useful, but I’m unlikely to find the time to do that. If I did, what value would I use - some arbitrary number? Doesn’t seem very scientific. I have two questions:

  1. Are my data really of no use without the location accuracy information?

  2. Is there something I should be doing differently moving forward?

Thanks

2 Likes

That doesn’t seem like a valid complaint. Many observations have just a single GPS point as their location, with no “accuracy” value.

5 Likes

whoever is telling you this is simply wrong. some people may like positional accuracy to be included with coordinates, but frankly, i think many of those people misunderstand what that information means because they have no idea how it is collected.

if someone is actually using the word “useless” to describe your observations, i think it may be worth writing back to them that they are just wrong and to stop misinforming others. if i got a message like that, i would almost even consider flagging that person’s note as abusive to further discourage them from doing this to others.

8 Likes

It’s not useless, but please check as many as you can to add it, one point can mean it was roght there or in a radius of five kilometers, it’s not hard to add accuracy later on, so please add it.

4 Likes

Might the message be from someone trying to use your obs as data for research. Needing the location accuracy?

2 Likes

You should definitely not add an arbitrary value to the location accuracy field (or any field really). Incorrect data is much worse than missing data overall. If the user wants to add an arbitrary value, they can do that themselves! And then they will at least know what it means (to them) and how to interpret it.

I do not think that your observations are useless without location accuracy entered. In general, I think that observations with good quality location accuracy data are more useful than those without for some applications. However, on iNat, because of the variety of ways that field gets populated and the differing ways that users enter location accuracy data, it needs to be interpreted with caution. Observations without accuracy data certainly still provide value.

Overall, I don’t think that you should feel obligated in any way to go back and edit your old observations. If you want to, you certainly can. If there are particularly high value observations that you can also make a reasonable assessment of accuracy for, it might be worth it. I would say that if you are estimating accuracy, you should always “round up”, ie, err on the side of a larger value so that you are sure that the accuracy circle includes the true location rather than having a lower value for accuracy that could be incorrectly precise.

Going forward, I think it’s potentially worth seeing if there’s an efficient way for you to add accuracy data when you upload. If you are uploading “manually”, it isn’t too difficult to add these values as you upload. There are some threads on the forum about getting accuracy data for different tech setups, but you could also post a question asking for advice from other users about your use case in particular and see if there are any good suggestions for how to make this process work.

9 Likes

lack of accuracy data would make the data unusable for some things, but not for other things and certainly not for other citizen scientists to use. The data is still much better on the site versus not adding it. And as others have said please don’t add made up numbers.

If someone is downloading the data for bulk use it’s easy to filter out anything without location confidence filled in, if they don’t want it.

3 Likes

Thank you so much for this input.

To quote from the message I received:

“Too large an accuracy, or missing accuracy, means that the data are not incorporated into checklists for the smaller nature reserves and places, and cannot be used for conservation planning and red listing purposes. And researchers who may need to revisit populations cannot know where exactly the localities are: at the point?, nearby? (how nearby?), or is this just a general midpoint for some reserve? (and for Kruger or Table Mountain that is quite useless).”

This seems to miss the point that the coordinates of each image were determined using a GPS-capable device. How accurate the GPS measurement was, I have no idea of knowing, as none of the GPS-capable devices I’ve used over the years seem to record that information. And surely there’s a certain level of error that’s inherent in all GPS measurements, however they are collected?

I also don’t understand why lacking “location accuracy” means the observation can’t be included in checklists for nature reserves etc. iNat Projects seem to have no problem including observations that lack accuracy data, provided the coordinates provided fall within the geographic limits of the project.

Any insights? Thanks again!

4 Likes

Thanks for the input. I appreciate your point. But what accuracy do you give? None of the GPS devices I’ve used over the years seem to record accuracy. And surely there’s a certain amount of error that’s assumed to occur with any GPS-capable device you use?

1 Like

I think that might be part of their issue, yes. Pardon my ignorance, but why is the location accuracy needed in this context? Isn’t there an inherent level of inaccuracy in all GPS readings?

1 Like

Thanks so much for this detailed response. I definitely won’t be making up accuracy data for these observations! As a professional biologist myself, that is not something I will ever do.

I would be interested in any pointers you have for determining “location accuracy”, so I can add that data moving forward. It doesn’t appear that any of the GPS devices I have used over the years record location accuracy.

I guess “location error” would be a better term, since uncertainty increases as the number gets bigger :)

Thanks again!

1 Like

Yes, “error” or “precision” might both be improvements!

As far as specific ways to figure out accuracy, I do all my observations (which aren’t many) by hand and err on the conservative side (ie, larger values) - sometimes taken with a GPS, sometimes with a friend’s phone, sometimes done from memory+maps.

There are some other forum threads that would be more useful for using devices/bulk adding accuracy:
https://forum.inaturalist.org/t/improving-location-accuracy-on-observations/1767
https://forum.inaturalist.org/t/set-location-accuracy-for-all-observations-together-when-uploading/15148/1
https://forum.inaturalist.org/t/gps-accuracy-not-captured-when-uploading-photos-taken-with-a-camera-with-a-built-in-gps/18861
https://forum.inaturalist.org/t/how-should-you-set-accuracy/16040
https://forum.inaturalist.org/t/geotagging-photos/66

2 Likes

Thank you! I will go read these now.

Thanks again for your help - I really appreciate it.

It should contain the area you think it was in, so, if it a tree, make circle big enough to contain that tree on the map. Depending on how good your GPS in it can be a couple metres or 10, just add it the way all possible locations of where the organism was at that moment were in the circle.

2 Likes

the “missing accuracy” part of this quote is wrong at best, or else the statement is constructed in a way that is misleading. in either case, it’s overstating the importance of the accuracy values. capturing good lat / long coordinates is far, far more important.

if the positional accuracy value is too large, it’s possible that that could cause observations to be excluded when filtering within some community-created places (when the accuracy circle extends beyond a place’s bounding box). if the positional accuracy is too large (in the neighborhood of >40km), that could cause the system to treat the observation as unmappable, making it not show up on certain map views, such as in the Explore page.

observations where positional accuracy is not recorded are absolutely not affected by either of these issues. the system effectively treats null positional accuracy as accuracy = 0m when it comes to mapping and filtering.

now, someone working with iNat data might have their own rule that they will effectively treat null accuracy as accuracy approaching infinity, and as far as i can tell, where i’ve heard about folks encouraging others to add positional accuracy, it’s usually for the benefit of some researcher who adopts this kind of blanket technical rule, usually ignoring the fact that the accuracy value is recorded using very different methods on different observations.

that somebody working with iNat data decides not to use perfectly good data just because it lacks a positional accuracy value is their own choice. it is not the observer’s burden to record that information for anyone else’s benefit, and and it’s especially unproductive to go back after the fact and try to guess at that information.

if you want to record positional accuracy just for completeness, that’s totally fine. when an accuracy value is recorded, it does provide some extra useful information about the observation. but as i noted before, there’s absolutely no consistency to how this kind of information is recorded in the system, and in most cases, the system does not provide any reliable way to figure this out.

again, the issue here is that these values are recorded using very different methods and mean very different things. in many cases, “(positional) accuracy” is absolutely the correct term. in some cases, you could make a case that you’re capturing some sort of error or precision or something else entirely. it just depends on how you recorded this value.

3 Likes

If you can add a reasonable accuracy estimate when you upload new observations, that’s probably good, but as a professional botanist, I’m used to dealing with herbarium collections that say things like “Campbell County. A couple of miles S of Alexandria along Highway 27”. Animals move, so I don’t see how having a really specific accuracy estimate is critical, as long as your location data was reasonable in the first place. For plants, getting to the right habitat will generally work for finding some, so coordinates specific to a few meters aren’t really necessary, though of course they could be useful if you had them. Overall, I’d say do what you can to provide good location data and don’t stress about it otherwise.

3 Likes

I don’t have GPS so my own obs are pinned to the satellite view of map.
Not a scientist and I defer to answers you have received here.

Some obs are set in such a huge circle they cover half of Africa and a swathe of Atlantic Ocean - those are clearly not SO useful.

3 Likes

There must be a valid reason iNat has the option to set the accuracy.
If it is a redundant field, then should the argument not be to request that it be removed from the system?

As @dianastuder says

And as for saying

and also

maybe understand why they are asking for refined information first.

I also received the message mentioned by stephen220 and do not see anything abusive in it. On the contrary, I am glad that my observations are important enough that someone checks and adds their valuable input to improve the quality of my endeavours. As I am for every person that provides an ID or comments on my work.

The same message does go on to say “Ideally we like observations to be 2-5m accuracy, but often one is uncertain or does not know quite where one was, in which case it might be 100s of metres. but please if you genuinely don’t know exactly where you were don’t make them too precise, even if it is to the nearest km or two.”
Said with that I have seen observations with a radius of hundreds or even thousands of kilometres for accuracy.

It goes on to say that in some biomes a course accuracy is useless due to the high diversity thereof and knowing the Fynbos biome mentioned it can change in meters.

As @m_whitson correctly states. Some organisms can move, and you could see the exact same specimen kilometres away. As to plants, I am sure many a botanist has cursed a collector for “not providing that little more detail” while trying to relocate a species and spending days or years walking up the wrong ridge.

I like this:

4 Likes

Thanks so much - this is very helpful and reasonable!

1 Like

I don’t know what you mean by “good”. Isn’t location accuracy a measure of “goodness”?

1 Like