Support says:
Source: https://doc.arcgis.com/en/field-maps/ios/help/high-accuracy-data-collection.htm
But the lat/long metadata fields filled in by using Field Maps on an iPad with the Eos Arrow seem to be in WGS84? When I bring the layer into ArcMap and calculate geometry for X and Y they are slightly off from the lat/long metadata fields, and using an online converter for WGS84 to NAD83 matches the difference. My point data is in a projected NAD83 coordinate system, with a transformation to WGS84 Web Merc applied before publishing to AGOL. The web map uses an Esri basemap so I know that the web map coordinate system is Web Merc. I believe the location profile in Field Maps is set up correctly for the Arrow (GCS ITRF 2008 -> Web Merc -> transformation based on location). Am I missing a transformation somewhere?
Have you tried setting up a field maps web map in your native projected coordinate system? I had a similar issue in the UK but set up a web map in British National Grid, that resolved the issue.
I also double checked the coordinate system coming out of my GNSS to double check the reference frame being used. But seemed to be fine. ETRF2000
Also, the transformation in use by the web map might be different to the transformation you've picked in arcmap. (You should probably switch to Pro now). To double check, you can add a web map to arcmap or Pro and then check the transformation in the data frame to compare.
Also, how much is it out by in metres?
Greetings Molly:
I just took a look at your post from 2021 and noted that I have been experiencing the same issue. Did you ever get a satisfactory answer to your inquiry? Based on my own experimentation, it seems that the positional data that are written to the attribute table by Field Maps are truly the un-transformed coordinates received directly from your Arrow GPS. However, it appears to me that the positional data written to the feature geometry (the actual mapped points) by field maps may actually be referenced to the map coordinate system (WGS84 Web Mercator) rather than your external GNSS coordinate system (NAD83). Therefore, when you calculate geometry for X and Y, the lat/lon data that then populates the fields are referenced to WGS84 Web Mercator rather than NAD 83. This is just my somewhat educated guess as to what is happening, but it seems to explain your own findings. It would be nice if Esri could confirm this.
https://doc.arcgis.com/en/field-maps/latest/prepare-maps/high-accuracy-data-collection.htm
If using a single GPS location to set a point's location, the following GPS metadata values are recorded:
This would indicate the Lat/Long values stored in the metadata are in whatever datum you have configured your GNSS device to output in!
The geometry of the points is controlled by the map/layer's datum in AGOL, and is affected by whichever transformation you have configured in your Field Maps profile to display your GNSS data on the map. If you are using ESRI basemaps, the map will be WGS84 web aux. Unless you are publishing hosted feature layers to AGOL directly through Pro, your point layer is likely also WGS84 web aux. (you can verify the datum in AGOL, by links in the lower right corner of the layer's item description page)
It is vital to configure and document your GNSS device's output datum (if real time corrections are in use it's the datum the corrections are provided in!), and then carefully configure your Field Maps profile so the GPS output gets correctly transformed to the spatial reference of your map. From there the point geometry is either the map's or the layer's (I haven't had to test that yet because my maps and layers always match spatial reference)
Hope this helps, definitely the kind of stuff that keeps one up at night! 😄