On the website including the upload page and observation page, using Firefox for Mac,
this observation, https://www.inaturalist.org/observations/65454941
and other images from the same trail camera scheme using exif metadata
yield a correct textual description of the location (Sonoma, California, USA) but show up on the map off the coast of North Korea. In other words, the metadata are being read correctly to describe where it is, but then it is being mapped in the Eastern rather than Western Hemisphere.
Step 1:
I was given tens of thousands of images from trail cameras in the park taken over several years. A program from http://www.colorpilot.com was used to embed exif metadata in these images prior to my acquiring them.
Step 2:
I dragged and dropped a few of these into the upload page, including this bear observation:
Step 3: Looking at the observation both on the upload page and at https://www.inaturalist.org/observations/65454941 I see that the location is listed correctly in the text, but is mapped in the wrong hemisphere.
Step 4: Opening the photo in Preview or Photos on my Mac, it maps in the correct location.
I searched the forum for topics describing this problem and could not find one. I hope that I am not wasting your time. Thank you.
I thought that at first also, but if that were the case, why would the location be described correctly in the text, and why would it map correctly in other application? It may be that there is something in the metadata that is not a problem with other applications, but becomes a problem when it meets iNat’s code, even though that code is not itself problematic.
Looking at the EXIF data here: https://www.inaturalist.org/photos/105356158@fffffffff is correct the ‘-’ is missing from ‘GPS longitude’ BUT the ‘GPS longitude ref’ is set to W which appears to have been missed by iNat.
Either the geocoder (google?) is sent the lat/long refs or it decides the middle of the ocean is unlikely and returns the ‘corrected’ more likely location name
Lat/lon coordinate in decimal degrees are written as +/- rather than E/W/N/S as degrees-minutes-seconds coordinates are. I don’t know why, it just seems to be standard. If the program is coding it as W rather than - then you’ll probably have to manually change it before other things will understand, or use a different program. I use Geosetter, though it can be slow and finicky so dealing with tens of thousands of images may be difficult.
For a variety of reasons, I think we can discard the idea that iNat has algorithmic correction of locations in the wrong hemisphere.
Another possibility is that iNat is using a different service to generate the text location than the map location, and one picks up on the W in the exif location and the other doesn’t. The thing that strikes me as strange about that is that iNat uses Google Maps for mapping, and when I drop the photo into any Google product it maps it correctly. So iNat could be passing along the decimal degrees but not the W to Google Maps for mapping purposes, but passing along both the decimal degrees AND the W to wherever the text location is generated? That sort of inconsistency seems unlikely, but possible.
I agree that is the standard, and it makes sense. Unfortunately for me, the Exif Pilot program that was used to geotag these photos doesn’t follow that standard. Which is probably why I am hitting this unusual error. Nevertheless, if iNat describing the location correctly but mapping it incorrectly, based on the same input, that seems like a bug.
I was going to be somewhat selective. The majority of these images are deer and turkey wandering back and forth in front of the camera. What I’ve done in the past is made a limit of one observation per species per camera in each hour. I imagine this rule will bring it down to a couple of thousand observations I that I want to put on iNat.
