HIGHLIGHT: The seasonal mean NINO3.4 anomaly was about -0.38 C for DJF2012/13, down from NDJ2012 (0.0).

The CA method, acting on global NDJ2012/13 seasonal means as latest input, but using a string of global SST going back one year, has all of its members staying in the neutral zone until spring. By MAM a few members are in weak negative (<-0.5) territory and by the fall of 2013 a large majority of members is <-.5 and three out twelve members <-1.0. A weak cold event!

Weights have changed a lot from last month. Big changes may and do occur when one year of data is added and the covariance matrix increases in size from 56X56 to 57X57. The co-linearity makes every participating year jockey for position in this changed environment. Even without a change in matrix size weights would have moved to the next year (keep in mind that the weight is assigned to a year; this is the year of the middle month of the most recent predictor season, i.e. we went from December to January. Examples are the big +ve weight of +12 having moved from 1994 (last month) to 1995 (this month) where it landed as +13. Likewise 1986 (-14) one month ago, has become -17 registered at 1987.

OCN flavor is much reduced compared to a few years ago, but not completely gone.1986-1992 is solidly negative. The relative abundance of high –ve weights is puzzling. Note that 1997 has an ominous weight of -14 – this contributed to a colder forecast obviously.

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 JFM2012 thru DJF2013). Years in the table are labeled by the year of the middle month of the most recent predictor season. Weights for 1956 and 2012 are less (by 50%) because only half the members use these years ("early" members use 1956, "late" members 2012). Data currently thru February 2013.

Year	IP	Wgt	Year	IP	Wgt	Year	IP	Wgt	Year	IP	Wgt	Year	IP	Wgt	Year	IP	Weigt
1956	xx	0	1966	xx	5	1976	Xx	-3	1986	xx	-7	1996	xx	1	2006	xx	10
1957	xx	4	1967	xx	0	1977	Xx	0	1987	xx	-17	1997	xx	-14	2007	xx	6
1958	xx	5	1968	xx	1	1978	Xx	7	1988	xx	-5	1998	xx	-2	2008	xx	8
1959	xx	0	1969	xx	-6	1979	Xx	4	1989	xx	-10	1999	xx	-2	2009	xx	6
1960	xx	-5	1970	xx	1	1980	Xx	3	1990	xx	0	2000	xx	8	2010	xx	3
1961	xx	-3	1971	xx	-2	1981	Xx	-2	1991	xx	3	2001	xx	6	2011	xx	-2
1962	xx	6	1972	xx	-3	1982	Xx	-6	1992	xx	-10	2002	xx	8	2012	xx	8
1963	xx	6	1973	xx	0	1983	xx	-5	1993	xx	-3	2003	xx	3	2013	xx	NA
1964	xx	7	1974	xx	-10	1984	Xx	8	1994	xx	-5	2004	xx	-1		xx	Xx
1965	xx	-3	1975	xx	-1	1985	Xx	8	1995	xx	13	2005	xx	-2		xx	Xx

Weights are high +ve (>+0.10) for 2006, 1995

Weights are high -ve (< - 0.10) for 1997, 1992,1989,1987,1974

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

-0.38	-0.38	-0.42	-0.5	-0.52	-0.5	-0.49	-0.59	-0.71	-0.77	-0.77	-0.79	-0.85	-0.75	-0.6	-0.36	-0.04	0.31	0.59	0.64	0.67
	JFM13		MAM		MJJ		JAS		SON13		NDJ		JFM14		MAM		MJJ		JAS14

One month ago

-0.01	-0.03	-0.11	-0.22	-0.22	-0.16	-0.11	-0.1	-0.19	-0.28	-0.25	-0.19	-0.15	-0.14	-0.17	-0.18	-0.08	0.1	0.32	0.48	0.48
NDJ		JFM13		MAM		MJJ		JAS		SON13		NDJ		JFM14		MAM		MJJ		JAS

Two months ago:

0.18	0.2	0.16	0.05	-0.02	-0.03	-0.02	-0.02	-0.06	-0.18	-0.27	-0.24	-0.16	-0.14	-0.14	-0.15	-0.12	0.02	0.19	0.34	0.41
	NDJ		JFM13		MAM		MJJ		JAS		SON13		NDJ		JFM14		MAM		MJJ

Three months ago

0.38	0.43	0.51	0.45	0.31	0.18	0.08	0	-0.07	-0.18	-0.34	-0.47	-0.5	-0.6	-0.5	-0.42	-0.34	-0.25	-0.07	0.15	0.33
SON12		NDJ		JFM13		MAM		MJJ		JAS		SON13		NDJ		JFM14		MAM		MJJ

Four months ago:

0.49	0.52	0.62	0.66	0.49	0.33	0.17	0.01	-0.11	-0.16	-0.24	-0.37	-0.54	-0.67	-0.66	-0.57	-0.48	-0.4	-0.33	-0.2	0.02
	SON12		NDJ		JFM13		MAM		MJJ		JAS		SON		NDJ		JFM14		MAM

Anomalies relative to 1981-2010 climo, as of the May 2011 release. Older forecasts (previous to May 2011 release) are relative to 71-2000