HIGHLIGHT: The seasonal mean NINO3.4 anomaly was 0.0C for MJJ2016, down from AMJ (0.5) and the first time a 3-mo mean was zero in a long time. We have ended the 15/16 warm event.

The CA method, acting on global MJJ2016 seasonal means as latest input, but using a string of global SST going back one year, shows a significant cool down predicted in 2016 by half the members. Half the members are colder than -0.5 by late 2016. But the solution is like a bifurcation, with all members that look back only 1 season staying positive or near zero, and all members that look back 4 non-overlapping seasons going negative.

There are 6 (4) large +ve (-ve) weights in excess of abs(0.10), listed below.

Weights changed modestly from last month. The weight for 1998, in the aftermath of a previous big warm event in a large 0.23. The weight for 2015 (the string ending with MJJ2015) is still high even if it has decreased some, and this puts a damper (according to CA-SST as designed) on the possibility for a La Nina for later on in 2016, but only for the 6 “late” members of the ensemble of 12. The weight for 2015 should go down dramatically if neutral or cold conditions develop and persist. Some sort of non-linearity and an oddity related to the spring transition. 2014 (a failed warm event of sort) also has a high weight and even higher when 2015 is not included. Some of the high weight for 2014 and 2015 is non-ENSO, but interdecadal or global warming like.

Table 1. Inner products (IP; scaled such that the sum of absolute values is 100; not shown) and weights (Wgt multiplied by 100.) of each of the years to construct an analogue to the sequence of 12 consecutive overlapping 3-month periods defined as the base (currently the 12 overlapping seasons string JJA2015 thru MJJ2016). Years in the table are labeled by the year of the middle month of the most recent predictor season. Weights for 1956 and 2015 are less (by 50%) because only half the members use these years ("early" members use 1956, "late" members 2015). Data currently thru July 2016.

Year	IP	Wgt	Year	IP	Wgt	Year	IP	Wgt	Year	IP	Wgt	Year	IP	Wgt	Year	IP	Weigt
1956	xx	-6	1966	Xx	-8	1976	Xx	-7	1986	xx	3	1996	xx	-7	2006	xx	-4
1957	xx	6	1967	Xx	-11	1977	Xx	1	1987	xx	-5	1997	xx	1	2007	xx	3
1958	xx	10	1968	Xx	9	1978	Xx	3	1988	xx	5	1998	xx	23	2008	xx	-6
1959	xx	9	1969	Xx	3	1979	Xx	-2	1989	xx	-4	1999	xx	-5	2009	xx	-2
1960	xx	15	1970	Xx	1	1980	Xx	1	1990	xx	-4	2000	xx	-7	2010	xx	2
1961	xx	-6	1971	Xx	-4	1981	Xx	5	1991	xx	-1	2001	xx	6	2011	xx	-11
1962	xx	-5	1972	Xx	-7	1982	Xx	-11	1992	xx	-2	2002	xx	11	2012	xx	3
1963	xx	-2	1973	Xx	6	1983	xx	1	1993	xx	-6	2003	xx	8	2013	xx	7
1964	xx	0	1974	Xx	-6	1984	Xx	-2	1994	xx	-5	2004	xx	-1	2014	xx	17
1965	xx	-13	1975	Xx	-2	1985	Xx	-4	1995	xx	6	2005	xx	0	2015	xx	11

Weights are high +ve (>+0.10) for 2015, 2014,2002,1998,1960,1958

Weights are high -ve (< - 0.10) for 2011,1982,1967,1965

Ens ave Nino34 anomaly forecasts, for leads -3, -2, -1, 0 etc: in plain numbers (ensemble mean of 12):

0.04	-0.04	-0.15	-0.24	-0.2	-0.23	-0.2	-0.16	-0.08	-0.01	0.03	0	-0.05	-0.09	0.04	0.18	0.29	0.28	0.24	0.18	0.12
MJJ16		JAS		SON		NDJ		JFM17		MAM		MJJ		JAS		SON		NDJ		JFM18

One month ago:

0.51	0.39	0.1	-0.2	-0.21	-0.16	-0.19	-0.19	-0.17	-0.12	-0.03	0	-0.01	-0.01	0	0.16	0.32	0.41	0.38	0.35	0.3
	MJJ16		JAS		SON		NDJ		JFM17		MAM		MJJ		JAS		SON17		NDJ

Two months ago

1.11	0.94	0.49	-0.18	-0.34	-0.39	-0.38	-0.42	-0.43	-0.38	-0.3	-0.17	-0.08	-0.07	-0.06	-0.07	0.04	0.15	0.19	0.13	0.14
MAM16		MJJ		JAS		SON16		NDJ		JFM17		MAM		MJJ		JAS17		SON		NDJ

Three months ago:

1.67	1.49	0.97	0.21	-0.18	-0.37	-0.47	-0.51	-0.58	-0.58	-0.51	-0.41	-0.29	-0.27	-0.19	-0.17	-0.19	-0.11	-0.07	-0.06	-0.12
	MAM16		MJJ		JAS		SON16		NDJ		JFM17		MAM		MJJ		JAS		SON

Four months ago:

2.32	2.08	1.53	0.61	0.19	-0.21	-0.42	-0.59	-0.7	-0.81	-0.84	-0.78	-0.64	-0.59	-0.36	-0.22	-0.21	-0.25	-0.2	-0.18	-0.17
JFM16		MAM		MJJ		JAS		SON16		NDJ		JFM17		MAM		MJJ		JAS		SON

Anomalies relative to 1981-2010 climo.