You are paying close attention to these SST datasets!  This is a very impressive accounting of them.  However, one thing I’ve noticed is that there is some redundancy in these datasets, so the values you are presenting aren’t independent of each other.   

My understanding is that CDAS-1 (another name for it is "NCEP R-1") assimilates an old version of Reynolds OISST. The same is true of Kaplan Extended, which is reliant on real time updates of OISST.  So from that list above, the only datasets with *recent values* that are truly unique in how they are constructed are:  OISST, ERSST, and the Hadley SST.  

Hopefully we will have a blog post in the future on this subject, but we strongly recommend using continuous “historical reconstructions” (i.e. ERSST, HadSST) to place the current El Nino in historical context.  The primary reason for this is b/c the dataset creators spend a lot of time and effort to ensuring that the different observing systems, which are changing through history, are placed on equal footing.  In other words, there is effort made to make sure the SST values today are adjusted so they can be compared to SST values of the past.  

OISST is only used at NCEP because it is updated on a sub-monthly basis which the historical reconstructions do not provide.  We require datasets like OISST to initialize our models or provide some information on a sub-monthly basis.  But OISST has data ingests that are not corrected for, which means biases change over time.  For example, a satellite in 1997 is different from today’s satellite (and keep in mind there are also corrections needed to get from the top-of-the-atmosphere to the surface).

As a result, we will be using the latest ERSST (v4) dataset to put this El Nino in historical context.  However, I am very supportive of comparing v4 to other independent SST datasets like HadSST (provided they make the same effort at homogeneity).  As you probably already realize SST values are only estimates because the tropical Pacific Ocean, and Nino-3.4 region, is imperfectly observed.  

Thanks for sharing your analysis and bringing up this very important topic!