The issue raised in the original post is that NOAA-PSL and Hycom models are making incompatible statements about regional ice thickness. One (or both) of them
has to be wrong. DMI, Piomas and other thickness products do not make forward predictions and were not considered for reasons given many times by Oren.
Based on the brief and erroneous statement from the publicist back in Bremerhaven, 30 ice cores were collected and frozen for later lab studies. These ranged around 1.55m with very little scatter (10%), very similar to NOAA-PSL but well out of Hycom range.
Reading between the lines, the cores were collected at regular intervals along a transect: the newly established road system which goes out on level ice perhaps a km from the ship to sensor stations. Photos so far do not show notable ice jumbles or pressure ridges; cores there would be far more difficult to obtain and much thicker than reported. Melt ponds are not cored.
Mosaic has provided no information whatsoever about the site selected other than it had a lot to do with the time crunch of late arrival, match-up with their January drift location, floe stability and safety (0.7 m or more is sought). FYI, SYI, MYI ice? Nothing has been said about floe backtracking history.
Persistent thick fog may have prevented helicopter launch and so the taking of overhead photos. They did launch a drone but nothing from it has been shared. We do not know at this time whether melt ponds and rotten ice are as pervasive as at the North Pole.
How representative are these core thicknesses? Mosaic has
seven methods for accurately measuring ice thickness: en route bow/side em, helicopter-flown em-bird, Polar 6 em-bird, sled em sensor, drilling for buoys, hole excavating for oceanography, and coring for biogeochemistry.
The first four can add up to tens of thousands of km of accurately determined ice thickness along swaths and rasters. None of this data has been disclosed nor will be disclosed prior to 2023. Mosaic did release various over and under camp maps for the first floe in November but in no case were thickness scales included or maps updated over the drift.
In summary, they have considerable context for the 30 core thicknesses but we do not. It's terribly naive to think site bias wasn't mitigated (how could they wring a publication out of the data?)
These models are not entirely ab initio physical theory but rather semi-empirical: before each day's recalc they are fed fresh weather, surface temperatures, ice edge location, ice concentration etc. Consequently they always get the current shape and overall drift of the ice pack correct as it's baked in.
Unobservable fields such as frictional drag of surface to wind and keels to ocean water can be inferred from observed response to recent wind. Indeed NOAA-PSL has separate graphics for them; Hycom provides rolling 30 days of which 24 are history that inform the 6 days of future.
In winter, they could (but don't) validate against SMOS/SMAP etc. However in summer months they have no sources of observed ice thickness to assimilate (though Polarstern weather and perhaps some buoy data is used right away in ECMWF). Consequently, the models have no way to control drift in thickness accuracy (except where it melts out to zero) until the fall sensor reset.
We do not run 'operational' forums here so rather than focus on futile forecasting, we can simply wait a week, see actually happens and try to understand why. However it does look like the persistent Siberian winds that began on Aug 23rd will continue a few more days and even worsen on Sept 6th,
https://tinyurl.com/y64busr6Damaging swells remain on the table but so far no evidence for impacts on the ice pack has surfaced. The Beaufort arm is losing its loop but a patch may hold on through the season; the sheltered ice south of SevZem is finally dissipating.
The main story though changes in daily bulk movement of the ice pack to a squeeze against the CAA with reduced transgression on the Barents side. Hycom sees that as continuing through Sept 10th. We can revisit that prediction with AMSR2_AWI and Polarstern track at the time.
For now, from uniq's efforts, we have very accurate records of regional ice movement at the positions of ~50 individual buoys (these are sparse on the Siberian side). Buoys near the Polarstern show its track and recent variations in speed are shown below with 4 dp GPS precision. It can be followed loosely on the awiMet hourly report site.