Tested that the observation operator in NEMO works.
Tested that the IAU (to add in the DA increments) code in NEMO works.
Tested the use of the decomposed horizontal/vertical normalisation factors in the old version of the model since we want to use that functionality in CO9p2 and demonstrated that it has little impact (so we will use this option).
Regenerated the horizontal normalisation factors to be used in CO9p2.
Interpolated the existing AMM15 error covariance ancillary files from the old version to the new one.
Almost everything is working now except a final (hopefully minor) issue with the SST bias correction.
Once everything is working, the plan is to:
do some focussed investigation into the way the DA is working near wetting-and-drying points. This might lead to changes in the DA settings near the coast.
run a full year with the DA and compare results to the previous version of the model.
do some further tuning of the DA, including recalculation of the error covariances, and test some other changes which were shown in the latest global system to lead to improved results.
James has:
Almost everything is working now except a final (hopefully minor) issue with the SST bias correction.
Once everything is working, the plan is to: