There is a feature request for a “geofence” to auto obscure home photos. My impression is that the developers are likely to implement it in some fashion.
3 Likes
We are having an issue with this-- the location accuracy is automatically set for a photo/observation and we need to change it! The location accuracy for observations for one of the users here is very wide (about 200m) but seems to not be under our control. <Edited: We can change the location accuracy one-by-one by editing each observation-- but this is time-consuming. Does anyone know if there is a faster way to adjust the accuracy? (The problem is that we’re having a local competition for # of observations on school campuses, and the wideness of the accuracy means that the observation is not counted for our campus.) For example: https://www.inaturalist.org/observations/23085791 Thanks for any help!!
1 Like
Only the user who submits a record may edit the location information associated with it, no other user has the authority to do so. So you need to ask the submitter to make any changes.
3 Likes
It will either be the accuracy value picked up by their gps (device dependant) or related to the zoom level when they place a manual pin. I am not aware of any account setting for a default accuracy.
3 Likes
If someone has a bunch of observations that need to have all of their Accuracy values changed to a different, single value, they can filter their observations for the subset that needs to change, then use the Batch Edit function to change them all at once. Select (check-box) the observations to change, click Edit Selected, then expand the Batch Operations tab at top. There will be a place to enter an accuracy value, apply them to all the observations being edited, then Save All at bottom.
6 Likes
I’d be cautious doing this though, if the accuracy came from a phone or GPS. One by one you can verify locations but unless they were in the same place the whole time, if you shrink the circle you may create ‘false precision’.
3 Likes
Many thanks! The batch edit is probably what we’re looking for. We’ll see if we can find out more about what was causing the initial large accuracy width-- these were camera photos uploaded by computer. The actual location of the pin seems to be perfectly correct, while the accuracy width is not! Let’s see. Thanks!!
2 Likes
My android 8.1 phone has 3 location settings. Using WiFi and cell only, GPS only, and using all. The first is what was being used when I first looked at it, which is a power saving feature.
2 Likes
Everyone should also be aware of the basic limitations of recreational grade GPS units. While it can be very accurate most of the time, it will have spikes of inaccuracy at times. Clock errors and atmospheric distortion of the signals are the main factors creating the position inaccuracy. Using a feature to average multiple position calculations will help but the iNat app is probably just using a single position calculation. Heavy canopy also greatly degrades the accuracy due to weak signal and multi-path issues. And finally, I would not trust the “error estimate” displayed by the unit itself.
I had a PDF whitepaper by USFS on the accuracy of various GPS units but I can’t find it now. I did find a handy webtool though.
https://www.fs.fed.us/database/gps/mtdcrept/accuracy/index.htm
It’s awesome technology but it’s not free of problems.
5 Likes
Important points, thanks for adding them! At the link you provided, they link to this PDF (don’t know if it’s the same one you were trying to find…) which lists all of the model results in a single table, including smartphone models.
Interesting that where it was an option (i.e., Garmin Etrex 30), using the GLONASS satellite array as part of the position calculation seemed to at least double the error size. Hmmm…
Also interesting that in many cases, the averages of multiple position calculations had larger errors than single calculations for the same model.
4 Likes
Ha, found it. http://www.christinafriedle.com/uploads/1/8/4/7/1847486/consumer-grade_gps_accuracy_and_reliability.pdf
The GLONASS additional error is unfortunate but not surprising.
I have a theory about why single or small sample averages have better accuracy than large samples. If you plot the positions of a location over time, you’ll see the occasional spike of inaccuracy. Maybe single and small average calculations miss the spikes while the longer 60 position averages almost always capture some of this noise. When I use rec GPS data for getting trails into GIS, I try to take 3 tracks and then follow a “majority rules” path when 1 track deviates too much.
Luckily, most GPS errors as easily swamped by the inherent fuzziness of biology.
4 Likes
i think a lot of us make the assumption that maps are as close to perfectly accurate as you can get, but maps are sometimes inaccurate, too. here’s an article (from 2013?) that describes how someone verified the accuracy of Google Maps: https://sites.google.com/site/wayneholder/self-driving-car---part/how-accurate-is-google-maps. i notice that Google maps satellite and map views are sometimes slightly different in my area. i don’t always know which to trust more.
3 Likes
I’m confused. Don’t we all see the same map on any given iNaturalist observation?
2 Likes
there are at least 3 different sets of map views available. there’s the street map, there’s an overhead aerial imagery view, and at closer zoom levels, there’s a slightly slanted aerial imagery view. sometimes if i start in the street map view and put a pin in the middle of, say, a street intersection, then when i switch to either imagery view, the pin might no longer be in the intersection. in such a case, the pin didn’t move, and the coordinates didn’t change, but the different map views tell me different things about where the pin is.
5 Likes
part of the problem can be perspective. When an aerial image is taken (typically from a plane) the only position on the image that is accurate as far as lat/lon positioning is the point immediately below the plane. The image can be stretched to match the curvature of the Earth, but even then it is only accurate for points that are at the same elevation. Then take into account the fact that the Earth is not a perfect sphere, and As elevation changes, the perspective changes.
Here in Gisborne, I use LINZ data to position a point on two different buildings, one at Hackfalls Arboretum in Tiniroto, and the other at Eastwoodhill Arboretum in Ngatapa, just 28km distance between them. At Eastwoodhill the elevation is 150m and at Hackfalls it is 290m. At Eastwood hill the Google maps are aligned correctly, but at Hackfalls the google maps are out by about 1m.
2 Likes
In my experience (plotting latitude/longitude coordinates generated by GPS’s attached to large towers around the globe), the satellite view of Google Maps is the most consistently accurate. When I plot a high-accuracy GPS point, I can often see the tower on satellite view, and the pin ends up right at the base of the tower. The roads on the map view, on the other hand, often veer away from the roads on the satellite view, usually only by a few meters, but occasionally much more. The tilted aerial imagery is consistently inaccurate, causing pins to point to places 10s of meters from the bases of the towers. The tilt-view makes it easier to visualize and identify landmarks on the ground, but doesn’t match up well with map pins at all.
6 Likes
More like the shape of the ground than the curvature of the Earth but yes, that’s the idea. Many of the orthophoto imagery products used by GoogleMaps and other webmaps are corrected using digital elevation models (DEMs). The quality of the DEM will affect the accuracy of the imagery. The orthophotos in NC are certified to be within 4 foot of the actual location in areas where the ground can be seen. The DEMs in NC are created with LIDAR and can reliably produce 2-5 foot contours depending on how steep the terrain gets. Areas with clouds of vegetation may not meet the standard (You can’t ground-truth if you can’t see the ground).
Your recreational grade GPS and GooleMaps are not going to line up perfectly. I’ve got 4 feet of possible error from the aerial imagery and add to it the 10-20 feet of GPS error possible in open areas. That becomes 50-100 feet of GPS error if you’re in thick canopy. But again, fuzzy biology usually doesn’t require better locations. Relax go find some cool species.
1 Like
i wonder if that’s what all these cars going through my neighborhood with radar-looking contraptions on top are doing? i think they have “nuor” written on the side of them, which i suppose might be a reference to “new orthophotography”?
1 Like
If it is “Nuro”, it is probably a self-driving car.
3 Likes
oh yeah… that’s what it is. i see people driving these cars though… so maybe the cars haven’t gotten the message that they should be driving themselves… anyway, i’m getting off topic…
2 Likes