The July JAMSTEC predicted a cold DJF for the SE US, ~2 F colder than normal averaged out and the coldest of any region on the entire map:
View attachment 5535
I looked at the prior 10 years of JAMSTEC July SE US winter predictions to see how it did to give me an idea of how to treat the current cold prediction:
Winter..........Prediction...........Actual...........Missed by
18-19...............-2......................?...................?
17-18...............-1.....................+2.5...............-3.5
16-17................0......................+5..................-5
15-16................-1.....................+3..................-4
14-15...............-3.....................-1.5..............-1.5
13-14................+1....................-0.5................+1.5
12-13...............-0.5...................+3..................-3.5
11-12................+1.....................+4..................-3
10-11................+4.....................-3...................+7
09-10...............-2.....................-4.5..............+2.5
08-09................+2.....................+1..................+1
Avg....................0.......................+1..................-1
So, JAMSTEC has on average been 1 F too cold (indicative of a possible small cold bias). Both the average and median of the absolute value of misses have been 3.3 F (not good accuracy).
So, only one winter of the prior 10 was predicted by the July JAMSTEC to be colder than that for 2018-9: 2014-5, which did turn out to be cool but was 1.5 warmer than the prediction. OTOH, 2009-10 was predicted to be about as cold as that for 2018-19 but it ended up being 2.5 colder than the prediction.
What do 2009-10, 2014-5, and 2018-9 have in common? The July JAMSTEC DJF prediction for Nino 3.4 was +1.0. Furthermore, 6 of the 11 predictions have been for colder than normal and the 6 were all within the range of +0.4 to +2.4 in Nino 3.4 for DJF. The other 5 predictions were for either right at normal or warmer than normal and all 5 of these were within the range of -0.2 to -2.3 in Nino 3.4. So, there is a pretty strong negative correlation between its DJF Nino 3.4 forecast and its SE US DJF temperature forecast...i.e.,
it tends to predict cool for El Nino and mild for La Nina.
As mentioned,
the July 2018 prediction for DJF Nino 3.4 is for +1.0. For the other two times it predicted +1.0 in 3.4, DJF ended up being weak El Nino (0.6 in 2014-5) and strong El Nino (+1.5 in 2009-10), respectively. On average, the July JAMSTEC missed 3.4's DJF only 0.2 too warm (likely indicative of only a small warm bias, if at all). The average and median of the absolute value of DJF 3.4 misses have been 0.5 and 0.4, respectively.
**Editing to make table more readable.