search

NOAA-EMC / godas

7 stars 4 forks source link

EnOI evaluation of static perturbations using the LETKF #430

Open guillaumevernieres opened 1 year ago

guillaumevernieres commented 1 year ago

Similar to #429 but using the LETKF. We should expect a better understanding of:

liuxiao37k commented 1 year ago

I plan to first discuss with @kbhargava to get the latest updates on soca team's LETKF development.

guillaumevernieres commented 1 year ago

I plan to first discuss with @kbhargava to get the latest updates on soca team's LETKF development.

Yes, no worries/rush, just something that we should start thinking about.

liuxiao37k commented 1 year ago

What I remember from last Thursday's discussion: Technical steps of running a LETKF cycle using the static perturbations that are currently used in EnVAR (better just work off /SCRATCH):

  1. recenter the corresponding ensemble forecasts on a single analysis restart from EnVAR;
  2. checkpoint application (or should this be the 1st step?);
  3. LETKF observer;
  4. LETKF solver;
  5. compare increment fields from the same cycle between LETKF vs EnVAR, and, if makes some sense:
  6. forecasts and postprocessing;
  7. repeat the steps for multiple cycles. @guillaumevernieres anything to add or correct? thanks!
guillaumevernieres commented 1 year ago

@liuxiao37k , for point 2, if by checkpoint you mean insert the new state in the restarts, you don't need to.