We have access to error and bias estimates for the IR SST retrievals, which we currently rely on for correction and obs error estimates. The same isn't necessarily true for the MW retrievals of SSS and SST.
There are several ways one could estimate errors and biases
collocate remote obs with insitu obs, bin and compute std and mean errors. We're NOT going to do that
run data denial experiments, method chosen here since the software already exists. Method chosen for this Epic
Deliverable
Error and bias estimate of all/most satellites
Obs error and QC estimate in the ioda files, from what was found above
Possibly a bias correction, if time permits.
A forecast sensitivity to obs system, before and after all the errors/corrections have been applied
Description
We have access to error and bias estimates for the IR SST retrievals, which we currently rely on for correction and obs error estimates. The same isn't necessarily true for the MW retrievals of SSS and SST. There are several ways one could estimate errors and biases
Deliverable