SST Forced Atmospheric Variability in an AGCM

1
SST Forced Atmospheric Variability in an AGCM
Arun Kumar Qin Zhang Peitao Peng Bhaskar
Jha Climate Prediction Center NCEP
2
Outline

• Motivation
• Data and Methodology
• Results
• Summary and Conclusions

3
Motivation
Horel and Wallace, 1981 Planetary Scale
Atmospheric Phenomenon Associated with the SO
4
Motivation

• What, then, are the prospects of utilizing
  information on equatorial SST anomalies to
  improve the quality of long-range forecasts for
  middle latitudes?

-- If the strength of correlations is limited
by the high noise level inherent in seasonal
averages then the prospects of seasonal
predictions are not encouraging
-- On the other hand, if these patterns
constitute blurred images resulting from our
inadvertent superposition of an ensemble of
shaper patterns, , then there is hope that
(seasonal prediction of) midlatitude climate
anomalies with higher degree of detail and
accuracy than is now (will be) possible.
5
Motivation
6
Motivation

• Question How much does the atmospheric response
  in the extratropical latitudes depend on details
  of the ENSO SST anomalies, or to SST anomalies in
  different ocean basins?

7
Data and Methodology

• For each DJF seasonal mean from 1980-2000, we
  have access to an 80-member ensemble of AGCM
  simulations forced with the observed SSTs
• Ensemble mean for each DJF provides a good
  estimate of atmospheric response to that years
  SST forcing
• For this data set, we analyze how the ensemble
  mean 200-mb height response varies with SSTs

8
Data and Methodology

• Data is from Seasonal Forecast Model archives
  from 2002-2003
• 10-member ensemble from different atmospheric
  initial conditions each month
• Lagged ensemble from different ICs

9
Data and Methodology
Difference in 200-mb eddy height climatology from
December and September ICs
200-mb eddy height climatology for December ICs
10
Data and Methodology
Difference in 200-mb height variance from
December and September ICs
200-mb height variance for December ICs
11
Results
12
Results
EOF1 53
13
Results
14
Results
EOF2 19
15
Results
EOF3 12
16
Results
Fraction of Variance Explained by Modes 1-3
17
Results
Z a?SST b?SST2 if ?SST ? Z ?SST-
?Z- then a (Z - Z-) / (2 ?SSTavg) and b (Z
Z-) / (2 ?SSTavg) (Monahan Dai 2004)
18
Results
DJF 1998
Ensemble mean (shaded) EOF1 (contour)
Ensemble mean EOF1
19
Results
DJF 1999
Ensemble mean (shaded) EOF1 (contour)
Ensemble mean EOF1
20
Results
Strong Cold
Cold
EOF1
- Warm
-Strong Warm
21
Results
Composite based on 1980, 81, 82 86
22
Results
Anomaly Correlation
Ensemble Mean
EOF1
EOF1 EOF2
EOF1EOF3
23
Results
AC(EOF1EOF2) AC(EOF1)
AC(EOF1EOF2) AC(EOF1EOF2)
24
Results
AC (EOF1EOF3)
AC (EOF1)
25
Summary and Conclusions

• A large fraction of extratropical variability is
  indeed related to high noise level inherent in
  seasonal averages and the prospects of seasonal
  predictions are limited.
• There are other modes of atmospheric response
  that are related to non-ENSO SSTs (e.g., EOF2),
  but this could be specific to the analysis
  period.
• This (and previous) analysis has shown higher
  order response to ENSO extremes, but it is hard
  to show any definite influence averaged over all
  SST years. This is either because of the rarity
  of such events, or because of incorrect
  simulation by the AGCM
• Should be repeated with other AGCMs