[ecco-support] About the uncertainties of mixing parameters in v4r4

[NA Yongsu]

Dear ECCO support team,

I wonder how the uncertainty fields for the three mixing parameters were produced, and how the bounds (specified in data.ctrl) for them were decided. Are they based on a similar way with producing the uncertainties of the atmospheric forcings and the initial conditions? All I could find is a short information in Forget et al. (2015). I'd be thankful if you could direct me to some related materials.

If they are given rather empirically/arbitrarily, will there be any reasonable uncertainty fields for 3d-varying mixing parameters? I've been testing a 3d diffkr field produced based on (sparse) observations for my regional state estimate, but using v4r4's uncertainty field (with interpolation) seems inappropriate if I expect diffkr to be adjusted in log scale along the iterative optimization (the issue reported in Trossman et al. (2022)). Using larger uncertainties (e.g. x10 on the 3d diffkr), with bounds to avoid diffkr from going minus, actually allows diffkr to be adjusted in log scale. But with those large uncertainties, there is an issue that the state estimation doesn't show consistent convergence towards observational constraints. Any comments would be appreciated!

Best wishes,
Yongsu Na
-----
PhD student
Department of Ocean Science
Hong Kong University of Science and Technology
-----

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.mit.edu/pipermail/ecco-support/attachments/20240319/e9addfa6/attachment.htm>