Analysis of Interannual variability in atmospheric signal and noise with SSTs Bhaskar Jha and Arun K

1
Analysis of Interannual variability in
atmospheric signal and noise with SSTs Bhaskar
Jha and Arun KumarClimate Prediction Center,
NCEP/NOAA, Camp Springs, MD-20746
P-3.4 Wed. October 26, 2005

• Motivation
• Quantify the impact of interannual SST
  variability on the mean and the spread of
  Probability Density Function (PDF) of seasonal
  atmospheric means.
• Abstract
• Our analysis is primarily based on an
  ensemble of AGCM simulation forced with the
  observed SST forcing. To make sure that the
  analysis is not unduly affected by biases in a
  particular AGCM, the analysis is based on
  simulations from eight different AGCMs. Further,
  the availability of multiple realizations of AGCM
  simulations forced with a constant SST forcing
  makes it possible to estimate the spread of PDFs
  for different SST states
• We focus on the analysis of the impact of
  the inter-annual variability in the SSTs both on
  the first and the second moments of the PDF of
  the December-January-February (DJF) seasonal
  atmospheric means. All the AGCMs show that the
  influence of the inter-annual variability in SSTs
  is much more systematic for the seasonal mean.
  The influence of the SSTs on inter-annual
  variability of seasonal means is the well known
  pattern over the Pacific-North-America (PNA)
  region that has a dominant large scale spatial
  structure. On the other hand, the impact of
  inter-annual variability of SSTs on the spread of
  the seasonal mean atmospheric state is small in
  all the AGCMs. Results are very consistent in all
  the models.
• Models
• The models used in this study are CCM3 run
  at Climate Diagnostics Center (CDC), the NASA
  Seasonal-to-Interannual-prediction Project
  (NSIPP), ECHAM4.5 run at International Research
  Institute for Climate Prediction (IRI), Two
  versions of Scripps Institution of Oceanography
  (SIO) and Three versions of Geophysical Fluid
  Dynamics (GFDL). All AGCM simulations are forced
  by observed SST variability for the 1950-2000
  period. Different simulations within an ensemble
  for each AGCM start from different atmospheric
  initial states but experience identical SST
  forcing throughout the integration period.
• Data and Analysis procedures
• The analysis is based on at least 10 member
  ensemble from eight different AGCMs and is for
  the DJF seasonal mean of 200-mb height anomalies
  for the period of 1950-2000.
• Inter-annual variability in the first (i.e. the
  mean) and second moment (i.e. the spread) of PDF
  of the seasonal mean of 200-mb height circulation
  with tropical SSTs is analyzed.
• Results

Fig. 6. (Left panel) warm composite of spread,
and (right panel) cold composite of spread.
Fig. 3. (Left panel) observed seasonal mean 200
hPa height regressed against the observed Nino
3.4 SSTs variability and (right panel) same but
for the AGCM ensemble means.
Analysis of inter-annual variability in the mean
and spread and its relative contributions of
changes in the mean and spread to seasonal
predictability
Fig. 7. (Left panel) The difference of spread of
warm composite and mean spread and (right panel)
same as left panel except for cold composite.  
Fig. 4. (Left panel) observed 200 hPa height warm
composite and (right panel) for the AGCMs. The
warm SST events are defined as those when the
Nino 3.4 SST index is at least one standard
deviation above normal.
Fig. 1. DJF total variability of 200 hPa
seasonal mean heights. Total variance is computed
from the variance of all DJF for the period of
1950-2000. (Left panel) observation and (right
panel) Eight AGCM. Units are in meters2.0.
Fig. 8. Signal to Noise ratio of 200 hPa height
for eight AGCMs.

• Summary and Conclusions
• A unique aspect of this study is use multiple
  AGCMs forced with constant SST.
• Analysis indicates that the influence of the
  inter-annual variability of tropical SSTs on the
  first moments of seasonal mean is much stronger
  over the PNA region compared to its influence on
  the spread (internal variability).
• Analysis of internal variability of 200 hPa
  heights in the extra-tropical latitudes confirmed
  some of the previous analysis that during the
  cold events, the internal variability of seasonal
  mean height tends to increase, while it decreases
  during the warm events.
• The results also suggests that the dominant
  contribution to seasonal predictability comes
  from the impact of the tropical SSTs on the first
  moment of the PDF, while the impact of SSTs on
  the second moment of the PDF is weak and noisy.

Fig. 5. As in Fig. 4, but for cold composite. The
cold SST events are defined as those when the
Nino 3.4 SST index is at least one standard
deviation below normal.
Fig. 2. Left panel shows the external variability
of 200 hPa. External variability is computed from
the ensemble mean of all DJFs. Right panel shows
the internal variability (spread). Units are in
meters2.0. Note the difference in magnitude of
the external and the internal variability.