HIGHLIGHT: In retrospect the middle of May 2002 may well be the real onset time for what suddenly appears to be a more convincing warm event. Around that time the trade winds across the entire Pacific basin suddenly became quite weak. Why this happened? There was no associated convection. One might call it a weatherless dry MJO, as though the upper level zonal wavenumber 1 disturbance manifested itself strongly and unusually in boundary layer winds, however without creating any great convection. A case of atmospheric forcing.
The seasonal mean NINO3.4 is positive, but at only +0.26 degree C anomaly (wrt 71-2000) for MAM 2002, the individual months showing some development with +0.15(Mar), +0.24(Apr) and +0.38 (May). Most recent weekly SST fields show >+1.0 C anomalies along the equator from 160E to 110W, a sensational increase in the last 2 weeks, indicating Nino34 to be near +1.0C at the moment. The weak trades have forced a new Kelvin wave which may warm up the waters closer to the South American coast, once again!. The sea-level height anomalies are generally still very small across the Pacific indicating a certain shallowness to this warm event. On a global scale the seasonal mean SST anomalies continue to be fairly weak, making the construction of an analogue a difficult task. Small anomalies imply a relative dominance of noise relative to the signal and a lower than average % of variance explained by 5, 10 or 20 EOFs.
The CA method, acting on MAM 2002 seasonal mean as latest input, although slightly warmer than last month, does not see any great development in Nino3.4 for the next 12 months. Actually, the forecast is about +0.3 for the rest of the year, with a maximum of +.45 in NDJ. The forecast shown this time is based on MAM only (no 'history' used), using all 25 EOFs on the most recent season (a method referred to as CA-1). This makes CA a little more responsive to current conditions, although we have verified that in the long run forecast skill is better when we take the development over the last 4 seasons into account (the CA-4 method). With reference to ftp://ftpprd.ncep.noaa.gov/pub/cpc/wd51hd/index.html one can see that the projections onto the first 4 modes are not only small, but a truncation at 4 modes would leave the tropical Pacific actually cold, an eye-sore under the present warm conditions. The forecast for JJA, SON 2002 and DJF 2002/03 by CA-4 would be -.06 -.21 +.04, while for CA-1 it is +.36, +.35 and +.40.
Table 1. Inner products (IP; scaled such that sum of absolute values is 100) and weights (Wgt multiplied by 100.) of each of the years to construct an analogue to the sequence of 4 consecutive 3-month periods defined as the base (currently the string JJA01, SON01, DJF01/02 and MAM02). Years are labeled by the middle month of the last of the four consecutive predictor seasons. 2001 is not yet used as a candidate analogue because long lead forecasts would not be possible beyond the latest observations. Data currently thru May 2002.
Warning: Only MAM is used (25 EOFs was used for construction on MAM2002; CA-1).
| Year | IP | Wgt | Year | IP | Wgt | Year | IP | Wgt | Year | IP | Wgt | Year | IP | Wgt |
| 1956 | -3 | -2 | 1966 | -3 | -12 | 1976 | -4 | -6 | 1986 | -2 | -2 | 1996 | 0 | 7 |
| 1957 | 2 | 14 | 1967 | -4 | 1 | 1977 | -6 | 0 | 1987 | 2 | 8 | 1997 | 3 | 7 |
| 1958 | 2 | -2 | 1968 | -3 | -17 | 1978 | -5 | -14 | 1988 | 0 | -4 | 1998 | 1 | 1 |
| 1959 | 2 | -7 | 1969 | 3 | 15 | 1979 | -2 | -2 | 1989 | -1 | 11 | 1999 | 0 | 6 |
| 1960 | -3 | 0 | 1970 | -2 | 6 | 1980 | 2 | 2 | 1990 | 4 | 19 | 2000 | -1 | 3 |
| 1961 | -2 | -1 | 1971 | -3 | -3 | 1981 | -5 | -6 | 1991 | 3 | 9 | 2001 | NA | NA |
| 1962 | -3 | 1 | 1972 | 1 | 7 | 1982 | -1 | -8 | 1992 | 1 | 4 | |||
| 1963 | -1 | 6 | 1973 | -1 | 0 | 1983 | 1 | -1 | 1993 | 1 | -23 | |||
| 1964 | -2 | 3 | 1974 | -3 | -10 | 1984 | -4 | -15 | 1994 | -2 | 2 | |||
| 1965 | -1 | -10 | 1975 | -3 | -3 | 1985 | -2 | 4 | 1995 | 1 | 7 |