gmao-msienkie
commented
2 years ago The MetOp-A non-radiance data is controlled by the gmao_global_convinfo file. We had previously determined that there were problems with having both MetOp-A and MetOp-B ASCAT turned on at the same time, so the action taken in f5271_metop was to turn off (iuse=0) MetOp-A ASCAT and turn on MetOp-B ASCAT. This should also be done here - however we do not currently have a time-dependence configured for the convinfo file so turning off MetOp-A ASCAT will affect all assimilations. This will probably not be detrimental to assimilations conducted during the time that MetOp-B ASCAT is available but will impact assimilations prior to when MetOp-B ASCAT became available (4/8/2014 06z). The MetOp-A GPS GRAS data has been assimilated with all the other GPS types so it isn't necessary to turn it off in order to be able to turn data from a different satellite on. It is probably OK to leave MetOp-A GRAS data turned on in the convinfo since, again, we don't have a time-dependent convinfo file implemented.
Modifications to the satellite radiance configuration as well as conventional data configuration need to be incorporated in the DAS to accommodate the decommissioning of MetOp-A on 15 November 2021 https://www.eumetsat.int/plans-metop-end-life The configuration for MetOp-A IASI has already been modified to add an end date for the data so will not be discussed here.
The experiment f5271_metop was conducted to test the effect of turning off MetOp-A instruments Instruments turned off as of 18 Sep 2021, 12z: Metop-A AMSU-A, AVHRR, ASCAT Instruments turned on as of 18 Sep 2021, 12z: N19 and MetOp-B AVHRR, MetOp-B ASCAT, CrIS-FSR NPP
The radiance instruments that were turned on used the existing radiance bias coefficients from the FP assimilation that were generated during the time the instruments were set to passive. At a later date the radiance bias coefficients for CrIS-FSR NPP were replaced with ones from FP just prior to the switch from side-2 back to side-1, because during the period right after the switch to side-1 the coefficients adjusted under passive mode to a different state and Will suggested resetting the radiance bias by putting the prior bias coefficeints into the assimilation. The bias coefficient transplant may not be feasible for running assimilation experiments, there could be other ways to prevent the coefficient drift that could be implemented. To keep this issue simple, the CrIS-FSR NPP restart will be deferred until after further study.