Water level bias observed after full reingest using current rads.xml

[mariegitter]

We recently performed a full reingest of our historical altimetry archive and are observing a systematic ~25 cm positive bias in derived water levels compared to a regional hydrodynamic model (NordicHD), affecting missions 3a, 3b, 6a, e1, e2, j3, pn, sa but not g1, j1, or n1.

For 3a, bias went from -0.04 m (previous processing) to +0.25 m (after reingest) — scatter plot attached. The QQ fit y = 0.97x + 0.23 confirms a clean constant offset, not a scale or noise issue.

Our pipeline downloads rads.xml fresh from master on each run. Diffing our cached January 2025 version against current master, we found some things that have changed:

<!-- January 2025 -->
<alias name="alt" sat="6a 6b 3a 3b j3 sa sw tx pn">alt_gdrf</alias>

<!-- Current master -->
<alias name="alt" sat="6a 6b 3a 3b j3">alt_poeg alt_gdrf</alias>

This switches the preferred orbit for the affected missions from GDR-F to POE-G (GDR-G). Since our water_level is derived directly from alt, the difference between the two orbit standards could explain the bias.

We compute water level as:

alt - range - dry_tropo - wet_tropo - iono - tide_solid - tide_pole - ssb - ref_frame_offset - geoid

Questions

1. Is the GDR-F to POE-G switch the likely cause, and is the expected radial offset between the two standards documented?
2. What is the intended approach for historical reprocessing - always use current master, or pin to a release for consistency?

Thank you for maintaining RADS :)

[remkos]

Thanks for your report, @mariegitter

While you technically are correct that changing the orbit will change the water level, this can certainly not account for the 25 bias. The difference in orbits between the older POE-F standards and the new POE-G standards is on average below the millimeter level and in RMS below the centimeter level. So this should not see in your outcome.

Also over time, we've been tweeking the ref_frame_offset to keep on improving the alignment between missions, but that too have always been at the millimeter level.

There is another change in that we updated the EGM2008 geoid (see #211). However, over the North Atlantic the impact should be marginal. But it does appear to line up with the missions where you do and don't see a bias.