Analyses over Oceans

CORe assimilates fewer observations over the oceans so the
quality of the its analyses should be worse.  One important
source of observations is those from ships, buoys and more 
recently unmanned vehicles.  These observations include 
near surface winds, temperature, surface pressure and 
perhaps water temperature. CORe directly assimilates the 
near surface winds and surface pressure.  

How good should the CORe analyses be over the ocean?  At
minimum one expects it to be as good as the 20th Century Reanalysis
(20CR) because both analyses use an EnKF with incremental update 
data assimilation system.  20CR only ingests the surface pressure
observation while CORe assimilated the same surface pressure
as well as near-surface winds and other observations such
as AMVs (Atmospheric Motion Vectors).  The winds are strongly related 
to the gradient of the surface pressure, so if you consider the
gradient information equivalent to another surface pressure observation,
then CORe would be like the 20CR with twice as many observations 
ignoring other the other observations.  This is the rough
estimate for the early years when suface ships were the only source
of oceanic observations.

The impact of the AMVs is less clear.  With modern data assimilation
systems, the AMVs are considered to be of low quality. The clouds do 
not move with the ambient flow, and there is a considerable height 
undercertanity. Given surface pressure observations, the system will 
draw towards the AMVs as long as it was consistent with the surface 
pressure field.