Due to the tremendous impacts of ENSO on global climate, the prediction of the tropical Pacific SST

1
SST Anomalies of ENSO and the Madden-Julian
Oscillation A Study with NCEP Global Ocean Data
Assimilation System Yan Xue and Kyong-Hwan
Seo Climate Prediction Center/NCEP/NOAA,
yan.xue_at_noaa.gov, Kyong-Hwan.Seo_at_noaa.gov
CDWS2004 P4.11
INTRODUCTION Due to the tremendous impacts of
ENSO on global climate, the prediction of the
tropical Pacific SST anomalies associated with
ENSO has always been the most critical and
important task within operational centers.
However, the forecast skill for the past two
decades is moderate, and there are a lot of
uncertainties in forecasting the transition phase
of ENSO. There is a consensus that wind anomalies
associated with the Madden-Julian Oscillation
(MJO) play a critical role during the onset and
termination phases of ENSO. Here we attempt to
quantify the relationship between SST anomalies
of ENSO and the MJO-related oceanic variability.
The National Environmental Prediction Center
(NCEP) has been producing a quasi-global ocean
reanalysis using a global ocean data assimilation
system (GODAS) since October 2003 (Behringer and
Xue, 2004). The retrospective ocean reanalysis
for 1979-present in pentad time scale provides us
an unique opportunity to study intraseasonal
oceanic variability and its association with the
MJO and SST anomalies of ENSO.

Figure Group 3 Lag Correlation

Figure Group 2 EEOF of Depth 20OC
Lag Correlation of filtered OLR, zonal stress,
D20 and SST averaged in 2OS-2ON with the first PC
(top row) and third PC (bottom row) of EEOF
modes. Shaded areas are the regions where the
correlation is significant at 95 level.

• DATA
• GODAS temperature, 1982-2003, pentad
• TAO mooring temperature, 1986-2003, pentad
• OLR outgoing long wave radiation, 1982-2003,
  pentad
• Surface stress from CDAS2, 1982-2003, pentad
• OI SST, 1982-2003, weekly

The first two EEOF modes form a pair and their
PCs have a lag correlation of 0.95 at 4 pentads,
describing an eastward propagating oceanic Kelvin
wave at a period of 80 days. The 3rd and 4th EEOF
modes form a pair and their PCs have a lag
correlation of 0.9 at 3 pentads, describing an
eastward propagating oceanic Kelvin wave at a
period of 60 days. Unit is meter, and PCs have
standard deviation of 1.

Figure Group 4 Variance Ratio of Reconstructed
fields
Figure Group 5 ENSO v.s. Seasonal Variance of
Oceanic Kelvin Waves

• METHODOLOGY
• Lanczos filter 20-120 days, weight 100
• Extended EOF for filtered depth 20OC averaged in
  2OS-2ON
• Lag correlation of filtered fields with PCs of
  EEOF
• Reconstruction using regression patterns onto the
  first two pairs of PCs

Time series of intraseasonal variance explained
by multiple linear regression onto the PCs of the
four leading EEOFs of D20. The explained variance
was computed for D20 (red) in the domain 2OS-2ON,
167OE-102OW, and OLR (thin black) and zonal
stress (blue) in the domain 10OS-10ON,
130OE-165OW, and SST (green) in the domain
10OS-10ON, 130OE-100OW. Also shown is the time
series of reconstructed variance, normalized to a
maximum value of 100 units, of D20 (thick black).
It shows that the four leading PCs explain about
55 of variance in D20, 10 in OLR and zonal
stress and 5 in SST.
Figure Group 1 Comparison with TAO data
Top figure shows the NINO3.4 SST (red) and
seasonal variance of reconstructed D20 at 130OW
(blue). Heavy arrows point to enhanced seasonal
variance of Kelvin waves that lead ENSO events by
9-12 months.

• CONCLUSIONS
• Extended EOF analysis of filtered (20-120 days)
  depth 20OC anomaly is used to extract the
  dominant intraseasonal oceanic variability
  associated with the MJO.
• Two pairs of EEOF modes, accounting for 60 of
  variance in filtered D20 anomaly, are used to
  represent oceanic Kelvin waves.
• Regression patterns onto the PCs of EEOF modes
  suggest that oceanic Kevin waves selectively
  respond to the low frequency tail of
  intraseasonal wind variations related to the MJO.
  
• Enhanced seasonal variance of oceanic Kelvin
  waves leads all El Nino events (1986/87, 1991/92,
  1994, 1997/98, 2002/03) and the 1988/89 La Nina
  event by 9-12 months, but enhanced seasonal
  variance of oceanic Kelvin waves does not always
  lead to ENSO events (e.g. the aborted 1990/91
  event).

Left figure shows Hovmoeller diagram of SST
anomalies (contour) and seasonal variance of
Kelvin waves (shaded) averaged over 2OS-2ON. It
shows that seasonal variance of Kelvin waves has
largest amplitude in 180OW-105OW.
Pentad temperature (3-point-running mean) at the
TAO mooring site 140OW, 0ON and at the depth of
120 meter. TAO observation (red) and GODAS
(green). It shows that GODAS simulates
intraseasonal variability well but it tends to
smear the signal due to a four week window
smoothing technique used in the data assimilation
scheme.
Behringer and Xue, Evaluation of the global ocean
data assimilation system at NCEP The Pacific
Ocean. Proceedings of Eighth Symposium on
Integrated Observing and Assimilation Systems for
Atmosphere, Oceans, and Land Surface, Seattle,
Jan. 2004.