Accuracy of location measurements (Android phone) and why it is important


When I first started using inaturalist, I did not take the photographs through the inaturalist phone app. I used my (android) phone's camera directly, and then loaded the photos into observations in the phone app, by using the 'choose image' button, which then took me to the 'Gallery' page on my phone, from where I chose images for that observation. When taking the photos, I always made sure the gps function is switched on, so the phone was recording the gps location of each photo, and as such I uploaded them into inaturalist with their locations added, i.e. the pin one sees on the map.

What I did not realize at the time is that in doing it that way (not taking the photos through the app), my phone is not recording the 'accuracy of the measurement' of the gps location of that photograph with the metadata of that photo, it is recording the location, but without recording also the accuracy of the measurement of that location. For that matter, I was not even aware and took for granted that just like any measurement tool used to do any measurement, my phone does not give an 'absolute' gps location, and there will always be an inherent degree of error in the measurement of the location point, that is, the gps location is correct within a range only. This range, or degree of error, e.g. +-2m or +- 5m or whatever is relevant for that photo, is what is referred to as the 'accuracy of the measurement' of the gps location, another word used is 'precision' of measurement.

However, when I take the photo for a specific observation through the inat phone app, using the 'take photo' button, the app has been designed to also record the accuracy of the measurement of the gps location of that photo for that observation. I have noticed that it does it on the the 1st photo of each observation. After taking the 1st photo, I have noticed the app takes a couple of seconds to narrow down this accuracy value. You will see it in the location box: Lat... Long... Acc... It is this 'acc' value I am referring to. So I have learnt to pause until it has it at the narrowest value, before continuing with the 2nd photo of that observation. When one looks at the map, and if this accuracy value is recorded, it shows up as a little circle around the gps location pin.

In actual fact, and important to understand, is that the observation is located not in the centre where the pin is, but somewhere/anywhere inside the circle, measured as accurately as the phone technology is able to do at that point in time. As gps location measurement technology improves in time, the degree of error will certainly reduce, and the circle should become smaller and smaller, but for the moment, the narrowest I have seen my own phone able to do, is around +-2m, but generally +-8m. Sometimes, e.g. when I am in a narrow gorge, it seems lack of line of sight with overhead satellites, or proximity of wifi towers, has the effect that the error is quite large, up to +-160m or more. In such cases, the phone records a gps location pin, but it is not highly accurate, so the circle is larger, how can I put it.. the phone is struggling to 'find itself'.., and can be located anywhere within the circle. Off course, human intervention is possible, so if I can see via the satellite map and identifiable landmarks around me, more accurately than the phone, where the observation is, I can reduce the accuracy manually, more about that a bit further down).

I found an interesting and useful article on the matter describing the technical detail of how phone apps record gps location accuracies:

If you have observations without the accuracy recorded, it makes them less useful to the specialists, so really important to add it. Fortunately it can be added manually afterwards. You can find more information in this journal post about how to add accuracy afterwards:

The accuracy data is important for adding species into 'places'. For example Table Mountain National park is a 'place' within inaturalist, with a pre-defined boundary: Zoom into the map at the top of this page, and you will see the boundary defined for this specific 'place'

So to put it simply, if the error of the gps location measurement, i.e. the accuracy circle or a portion of it, falls outside the place boundary, then, because the observation does not lie at the pin itself, but somewhere inside the circle, and a part of the circle lies outside the place boundary, then there is a chance that that observation is in the portion of the circle falling outside the place boundary, and therefor might actually not be inside the place at all. Inaturalist has been set up to then exclude that observation from that place, and if that was the only observation of that specific species, then it will exclude that species from that place.

The same thing happens if no accuracy is recorded. The observation, and potentially the species, is excluded from that place.

You can read more about it here:

Posted by dryfveer dryfveer, August 12, 2021 08:19


thanks Tony, for the additional info. I see many people struggle grasping this issue, as I did.

Posted by dryfveer 4 months ago (Flag)

Oh: it is quite simple.
Many people think that making a locality is simply placing a point. But there are various levels of uncertainty, that result in a "location error". These include:
.* not being sure of exactly where you were on the ground
.* being careless where one puts the marker on the map.
.* not zooming in enough, so that what looks spot on is km away from were we thought,
All these contribute to a combined error.

On iNat the Location Error is inexplicably called "Location Accuracy" (it should be Location Inaccuracy). It is measured in metres radius, and corresponds to a circle shown around your peg displaying where your peg might be. If you zoom in enough, you can sometimes actually see your bush and pinpoint it precisely, with the error circle precisely around the bush. This does not work so well with animals, which are too small to show up on google, or that inconveniently move around. So most of the time we are guessing to some degree.

The important thing to remember is to be as accurate as possible. It is easy to know that you were in Kruger park, and put your Error as the entire park (500km), but that data is not of much use to anyone. This is especially important for small reserves. The secret is not to think about describing where you where, but where a scientist or manager might think that you were. They are not looking at your central peg: they are imagining the entire circle filled with thousands of pegs, any of which could be where you were. Ideally your error circle should be to within a metre or two (or more realistically a metre or ten). If not try and remember what path or road you were on and where along it: you can then refine your circle to just that section.

But if you take your pictures via the iNaturalist apps, this is all taken care of. Just remember to allow the Error to go down to less than 10m before saving your observation.

Always fill in your accuracy. No accuracy is worse than a vague accuracy, because no one knows if it can be trusted or not - so the data cannot be used. You can check for your observations without an accuracy by clicking here, and in the url bar, typing your user name instead of USER_NAME:

Note that if you dont take your smartphone pictures using the app, then you MUST fill in the Location Accuracy field manually.

Posted by tonyrebelo 4 months ago (Flag)

Excellent, thank you for the comprehensive additional notes.

Posted by dryfveer 4 months ago (Flag)

Add a Comment

Sign In or Sign Up to add comments