I got an estimate in the mid-16s from yesterday's GFS 12z, but the model has again failed to resolve the low minimum temps under the inversion this morning, so I think it would be more like low 16s now. Apply the typical downward adjustment from provisional Hadley CET onto that and you're at GW's latest estimate .
It sure has been an extraordinary month. Looks like it will join the three previous 2.0+ CET anomaly Septembers since 1950: 1999, 2006 and 2016.
Studying previous 1.5+ CET anomaly Septembers, a very warm North Atlantic was present in 7 out of 8 cases. 2021 has made it 8 out of 9. Unsurprising given that temps in the north-eastern N. Atlantic reach their peak during this month while dependence on sunshine diminishes and the impact of heat plumes does a bit too (though remains significant).
I now find myself wishing I'd put more stock behind that rather than what the models were showing for the opening third or so, as of the end of August! They saw the warm signal but were well down on its magnitude. To be fair, though, it was a truly exceptional spell of high temps that took place in the 2nd week.