Just wondering at the situations that will have iNaturalist list an observation position accuracy of NA or No Accuracy?
The one I know if is that Android has a secondary accuracy setting that folks do not know to set, so there is no accuracy information attached to photo’s uploaded with coordinate information.
Are there other situations? Wondering if I can assume accuracy of NA on records (for those not obscured)?
Thanks.
When i edit a location in the web editor, the accuracy gets automatically removed. I dont understand how you do it in the Android app? Its not possible for me to remove accuracy there.
Not sure about when using the app itself, but when you are attaching location information to photo’s to upload on your browser, if a secondary setting is not turned on, the accuracy information is not included in the exif information so iNaturalist sets it as NA.
On web uploader or already uploaded observation you can delete the accuracy, but 99% of cases are from the app, where you need to add it.
In fact, if the photo coordinates are obtained from the phone or camera’s built-in GPS receiver, it is difficult to exactly determine their accuracy. This is a dynamic value, which the operating system can only estimate approximately. But in most cases, it can be considered equal to ~ 5-10 m. Less accuracy is rare (only if the GPS signal is low). Accuracy of 1 m and less can practically reliably provide only professional geodetic equipment.
Therefore, in cases where the accuracy of coordinates is specified as NA or “Not recorded”, we can take as the most reliable empirical value of 10 m. Therefore, coordinate values can be rounded to 1/10000 of a degree (~100 km / 10000 is ~10 m). Higher accuracy cannot be reliably provided by consumer GPS receivers. Even if they are nominally accurate to one millionth of a degree (~10 cm).
So, it seems unreasonable to me to add the accuracy of observation coordinates, if it is obtained automatically by consumer equipment, rather than estimated by some special method. For example, from a map without instrumental data.
Perhaps it would be more correct if there is an additional field “coordinates source” with value “from EXIF data” or “user defined”.
For the purposes of iNaturalist I do not think we care that much about high accuracy, 10m seems plenty, even 100m would be fine, imo. The real issue with device locations is that they are greatly impacted by the time available for the device to get a position fix before the information is recorded.
If you are using the iNaturalist App I assume it starts this process right away and has threshold requirements to get better accuracy. For those of us that attach location to photo and upload later this is not the case, it just grabs whats available when you take the picture, which could be 20km away.
I personally run an Avenza (app) track while I am hiking, keeps location refreshed so when I go to take a picture its typically <10m accuracy, however not everyone is even aware of this issue.
Yes, there is a “cold start” problem with GPS - when the receiver has no data on the estimated position of satellites after a long shutdown or transfer over a long distance. Worst of all, it sometimes shows the coordinates of the previous known position. I have several times had this issue. It takes some getting used to - “don’t check after turning it on, don’t use it.”
I don’t use the Android app. I upload nature photos mostly through the web interface. The coordinates in the EXIF are either set when I shoot, or with exiftool (https://exiftool.org/) by synchronizing with a Garmin navigator track (if the camera doesn’t have a built-in receiver). In both cases, the coordinate accuracy on the website is set as “Not recorded”.
If the coordinates are known only approximately (for example, old photos whose coordinates I can only understand from the map), or photos of the collected specimens were made ex situ, I use my software tool (https://github.com/kgbase/php-inaturalist), and in a request to the iNaturalist API I set the estimated accuracy explicitly (e.g. https://www.inaturalist.org/observations/108368171). Unfortunately, many arthropod species just cannot be reliably identified in the wild or from photographs made in situ. And sometimes species identification is impossible without dissection of a specimen. I’m sorry :-(.
We definitely care about accuracy, I set it for 3m unless it’s off then I add it as well as possible by eye. You need to make sure your first photo has correct coordinates, then turn the internet off (or it can start trying to reconnect to satilite and screw everything), check each 30 mins if the location added is still correct.
It depends a lot on the motivations of each iNaturalist user. Coming from a rare plant surveying background, I always feel compelled to be as accurate as possible. And I have been very grateful for the 5-10 meter accuracy of other people’s observations in cases where I have needed to find them again in the field and collect a scientific specimen to document a new species, a range extension, or some such.
Definitely agree here, and when using the phone app I do the same thing you do, keep a GPS tracking app active in the background so that the GPS fix is always fresh. I also permanently turn off the auto-sync option in the iNat app settings, and keep the phone in airplane mode while in the field. Observations are uploaded later with good wifi, battery drain is less, and any chance of positional interference from cell tower triangulation is eliminated.
The Internet connection cannot affect the receiver’s connection to the satellites. This communication works over a separate radio channel using NMEA or similar protocol. Essentially, satellite positioning is reverse triangulation by radio signal. We determine our (receiver’s) position by getting the position of the transmitters (satellites). The more transmitters, the higher the accuracy (theoretically). So turning off the internet connection may reduce power consumption, but it will have no effect on communication with GPS satellites (Galileo, GLONASS, Baidu, whatever).
There is also a difference between theoretically possible and reliable positioning accuracy. For example, my navigator CAN determine the coordinates within one meter (I checked by repeated measurements). But it requires some knowledge, time and luck :-). If I use only one measurement, I cannot be sure that it is more accurate than 10m. So 3m seems to me a very optimistic estimate of accuracy for any camera or phone.
It definitely affects it, because it will start to load the map if you open it, you can’t check how accurate the position is without opening google maps or other map app, with internet off it will just show you the part of map it already loaded, with internet on it will restart the process and will try to load more and your next photos will have no gps at all or in another town, you will either spend then next 5 mins trying to get it to the correct spot by turning off gps, internet, map, restarting it in different combinations to make it work, or will have no gps data at all if internet has no connection, why saying it’s not linked when it’s literally what happens?
My choice is not something to question, I won’t ever change it to 10m on default.
It seems to me that we should separate the two questions:
Interaction with satellites and data calculation. In any device, this is performed by a separate microchip (in fact, a separate microcomputer). Internet cannot affect this even theoretically.
interpretation and visualization of these data by the operating system. Of course, at this stage, data obtained from the Internet may be used and somehow affected.
Probably, fortunately, I do not use the phone to define the coordinates. I can understand the reliability of positioning by the number of available satellites and signal quality. In Android it is also possible to do this with third-party apps. Perhaps that can provide even more understanding than maps.
I just would like to point out that the communication of any device with satellites is NOT via the Internet. The satellites just don’t have any connection to it.
I’ve yet to use any external apps, it’s working well, just need to not try to do something via internet if connection is weak, which sounds easy and straightforward but from time to time I make this mistake again. I know those are separate thing, it just happens this way, gps doesn’t start working without internet too, it needs maps, I don’t know if it happens with every android, because my last phones were all from the same brand.
There are applications that allow to use offline maps - stored in the phone and do not require a connection to the Internet. Maybe it would be better to use them, if updating the map through the mobile network causes such difficulties.
Its not about effecting communication with GPS satellites, its about phone’s prioritization process. Putting phone in airplane mode removes the phones ability to get a quick, low accuracy, location based on tower triangulation.
This is not accurate. An app may need maps, but the GPS doesnt care. I use androids in the field that do not even have a sim card, no issues.
I’m not sure how it’s not accurate when just turning on gps won’t get me photos with location added, unless I have internet and open maps, it won’t get any, so your experience may be different, but it’s wrong to say it’s not accurate.
(This entire discussion is extremely over my head. I just came to say “Welcome to the forums and thank you” to @kgrebennikov to whom I am indebted for helping with my Corythucha and Corythaica.)
Another thing to remember: A lot of times, but not always the accuracy given for observations is for the location of the device, not necessarily the subject.
Your device could show an accuracy of 4 m. yet the subject could be 10 meters away.
If the taxon is scarce enough for location accuracy to matter that much, it probably auto-obscures anyway.