apparent bias between ATM data and single frequency GPS data compilation about gcnet_positions HOT 3 CLOSED

Great job, thanks!
I'm not surprised that the handheld GPSs have 30-50 m accuracy.
The difference between the ATM and the GEUS measurements are more puzzling!

from gcnet_positions.

Thanks Jakob ([email protected]) for teaching me some geodesy that I forgot

Now I understand the AWS and handheld GPS output "elevation" defined as height above mean sea level or 'orthometric height', here "z"

z is of course not the same as height above ellipsoid.

I found a source of geoids https://www.agisoft.com/downloads/geoids/

Using some python code from https://gis.stackexchange.com/questions/428728/get-lanlon-and-values-from-geotiff-using-python, I adjusted the NASA altimetry heights above the WGS84 ellipsoid, by subtracting the geoidal height. I used the EGM2008 geoid instead of EGM96.

here are some examples of the results, which look v promising, just want to confirm using EGM2008 is best:

Swiss Camp, before

after looks good, note the vertical axis label is now accurate

Tunu-N, before:

after looks good

At Saddle, there are some outliers in the monthly averaged GEUS elevations that come from apparently the thredds data having CP1 data for the first part of the Saddle record!

and at CP1 at the start of the record in 2022 (June), there are some 'bad' elevations that like Saddle, are some kind of data issue with the GEUS AWS data.

from gcnet_positions.

got confirmation from [email protected] tjat the correction from ellipsoidal height to orthometric height makes sense, so am proceeding

from gcnet_positions.