New GCP accuracy feature

Hello!

I just processed a map with some GCPs and saw the new GCP error report. This is a particularly handy feature. I’ve been digging into why my absolute accuracy isn’t where I’d like it to be and it’s a very complex issue. I’ve been able to use a PROJ string and get similar results in different coordinate systems. I know that the output is in WGS84 / UTM. Does ODM expect WGS 84 input? If so, I can convert the coordinates over to that system.

I’m going to do some more testing on this. Ultimately what I would like to publish is a workflow for users of Emlid products and ODM to get the most accurate results possible.

2 Likes

We don’t need it, no, but not all CRS conversions are created equal, so there might be the possibility of some precision loss depending upon source CRS and the resultant conversion.

I have not dug in terribly much yet, so I’m not sure what our conversion process looks like and what conversions are available.

I was taking a look at the accuracy reports and noticed a GPS error in the area of 32 meters in all cases. This actually correlates with the difference between elevation and ellipsoidal height in my area. My guess is that any input system is fine as long as you note it properly in the gcp list. The Z value might need to be ellipsoidal height. I’m running a test on this right now.

1 Like

I want to clarify something for Emlid users working with ODM. Reachview 3 wants your base elevation in ellipsoidal height. Ohio’s NTRIP service provides lat/lon/elevation information. The elevation is mean sea level, not ellipsoidal. When actual elevation is being input for ellipsoidal height, Reachview subtracts the geoid height from the mean sea level height. GPS works from ellipsoidal height and ODM compares GPS data to your GCP input. In Ohio, the difference is about 33 meters. In order to get the correct MSL elevations and ellipsoidal heights, I have to add the difference to both numbers.

My input is EPSG 6318, NAD83(2011). I’m sure there is a little bit of an accuracy loss in the conversion but I doubt it’s in the .5 to 1m range like I’m getting. The ellipsoidal height error may account for the inaccuracy. I bet I’m confusing the software.

I’m running one now. I’ll have results in a few hours.

Use this tool to find the geoid height in your area.
https://www.ngs.noaa.gov/cgi-bin/GEOID_STUFF/geoid99_prompt1.prl

1 Like

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.