NonStationarity in the circulationclimate relationship

1
Non-Stationarity in the circulation-climate
relationship Stability of NAO-Influence on the
Regional Climate of the Baltic Sea Area Possible
Effects on NAO-Reconstructions?
(1) Frederik Schenk, Sebastian Wagner, Eduardo
Zorita (2) Daniel Hansson
(1) GKSS Research Center Geesthacht Institute
for Coastal Research System Analysis and
Modeling (2) Göteborg University
Earth Sciences Center Ocean
Climate Group
2
Outline

• 1 Non-Stationarity in Observations
• - spatiotemporal changes of NAO-control on
  regional climate
• 2 Non-Stationarity in Climate Model Simulations
• - temporal evolution of the NAO-t2m-relationship
  over 990 years
• 3 Idealized pseudo-proxy reconstruction of NAO
  from local-scale
• 4 Summary

3
Assumption of Stationarity in Climate
Reconstructions

• Most statistical reconstructions assume
  stationarity between climate circulation and
  regional climate impact (proxy location)
• i.e. linear relationship between NAO and
  near-surface climate

Local climate F(large-scale x)
physical assumption
Residuum not captured by linear equation
Regional climate or local proxy
Proportional constant
Large scale i.g. PC
4
First leading EOFs of 1000-1990
PCA calculates covariability matrix of SLP
field anomalies
5
1 Non-Stationarity in observations

• The NAO temperature relationship

6
http//www.baltex-research.eu/BACC/media/
Definition of circulation indices and T-Baltic
from Echo-G and Luterbacher-SLP-reconstruction
7
Detection of Non-Stationarity

• Non-Stationarity
• changes in strength of a relationship between
  two climate variables
• - expressed as Running Correlation coefficients
  over time (Pearson)
• - window size of 31 years RC30

8
Data

• Surface Temperature (t2m)
• Long historical station temperatures
• T-Baltic (t2m) of different AOGCM simulations
  from ECHO-G
• MIB (Max. sea-Ice extent of the Baltic Sea)
• MIB (obs.) (Seinä Palosuo 1996)
• MIB (mod.) box-model PROBE-Baltic (Hansson
  Omstedt 2007)
• Circulation indices
• NAO of Azores minus Iceland (Jones et al. 1997)
• NAO from 500 year SLP-reconstruction
  (Luterbacher et al. 2002)
• NAO from SLP of different simulations from
  ECHO-G
• with different forcings and initial conditions

9
Time evolution of NAO-temperature relation
10
Time evolution of NAO-temperature relation
64
9
11
NAO and sea-ice (MIB)
Hansson, D. A. Omstedt (2007) Modelling the
Baltic Sea ocean climate on centennial time
scale temperature and sea ice. Climate Dynamics
30, 763-778
12
2 Non-Stationarity in Climate Model Simulations

• 990 year model study from AOGCM Echo-G

13
Climate Model Simulations

• Climate simulation as a surrogate climate
• - Model simplified representation of real
  processes
• Idealized pseudo-reconstruction-approach
• - comparison of NAO and CEZI
• - use of area weighted t2m of the Baltic
  catchment area for
• reconstructing the NAO by simple linear
  regression (without
• adding white noise)
• - comparison with real model NAO
• - estimation of non-stationarity for
  reconstructions within the model

14
Model description of Echo-G
15
Settings of Echo-G simulations

• Control-run of 1000 model years with fixed
  present conditions
• External forced simulations I solar volcanic
  greenhouse Gases
• ERIK1 990-1990 A.D. starting with warm ocean as
  initial condition
• ERIK2 990-1990 A.D. starting with cold ocean as
  initial condition
• External forced simulations II orbital forcing
• Oetzi1 7000 B.P. 1998 A.D. with orbital
  forcing only
• Oetzi2 7000 B.P. 1998 A.D. with orbital, solar
  and greenhouse gases (no volcanic)

16
NAO vs. Baltic Sea climateexternal forced
(solar, volcanic, GHG)
17
NAO vs. Baltic Sea climateControl-Run
18
3 Idealized pseudo-proxy reconstruction of NAO
from local scale
19
Idealized pseudo-reconstruction
estimation of NAO from pseudo-proxy
20
Idealized pseudo-reconstruction
21
Idealized pseudo-reconstruction
22
4 Summary

• Magnitude of non-stationarity for NAO-impact is
  high for Baltic Sea area
• - NAO vs. station-temperature 1824-2008 (DJF)
  RC30 10 - 65
• - NAO vs. sea-ice (MIB) since 1500 RC30 0
  64
• - NAO vs. t2m (AOGCM) (DJF) since 1000 RC30
  0 64

23
4 Summary

• Comparison of external forced simulations with
  control-run (990 years)
• - same magnitude of non-stationarity over time
  with all/no forcings
• - no relationship between forcing and
  non-stationarity
• ? non-stationarity is mainly result of internal
  climate variability
• ? possible external influence on longer time
  scales (orbital changes)?
• e.g. Groll et al. (2005) Changes in
  AO-regional-climate relationship during Eemian
  (125 kyr BP) compared with pre-industrial (1800
  A.D.)
• - significantly lower AO-t2m signal for NH
  winter during Eemian
• - also stronger NH winter westerlies towards
  Europe, warmer CET

Groll, N., Widmann, M., Jones, J., Kaspar, F.
S. Lorenz (2005) Simulated relationship
between regional temperatures and large-scale
circulation 125 kyr BP (Eemain) and the
preindustrial period Journal of Climate 2005,
18(19), 4032-4045
24
References

• Cassou, C., L. Terray, J.W. Hurrell and C. Deser
  (2004) North Atlantic winter climate regimes
  spatial asymmetry, stationarity with time and
  oceanic forcing, J. Climate, 17, 1055-1068.
• Hansson, D. A. Omstedt (2007) Modelling the
  Baltic Sea ocean climate on centennial time
  scale temperature and sea ice. Climate Dynamics
  30, 763-778
• Jacobeit, J., Beck, C. A. Philipp (1998)
  Annual to Decadal Variability in Climate in
  Europe. Würzburger Geographische Mauskripte, Vol.
  43.
• Luterbacher, J., Xoplaki, E., Dietrich, D.,
  Rickli, R., Jacobeit, J., Beck, C., Gyalistris,
  D., Schmutz, C. H. Wanner (2002)
  Reconstruction of sea level pressure fields over
  Eastern North Atlantic and Europe back to 1500.
  Clim. Dyn. 18 545-561.
• Osborn, T.J., Briffa, K.R., Tett, S.F.B., Jones,
  P.D. and R.M. Trigo (1999) Evaluation of the
  North Atlantic Oscillation as simulated by a
  coupled climate model. Climate Dynamics 15,
  685-702.
• Vicente-Serrano, S. M., and J. I. López-Moreno
  (2008), Nonstationary influence of the North
  Atlantic Oscillation on European precipitation,
  J. Geophys. Res., 113.
• Zorita, E. and F. González-Rouco (2002) Are
  temperature-sensitive proxies adequate for North
  Atlantic Oscillation reconstructions? Geophysical
  Research Letters, 29 (14), 48-1 - 48-4.
• Zorita, E., Gonzalez-Rouco, F. and S. Wagner
  (2009) Low-frequency response of the Arctic
  Oscillation to external forcing in the past
  millennium. Geophysical Research Letters
  (submitted).

25
Thank you for your attention!

• Climate is what we expect,
• Weather is what we get.
• (after Lorenz)

26
5 Outlook Principal Component Analysis

• teleconnection patterns describe the
  low-frequency extratropical atmosphere generally
  in terms of space-stationary and time-fluctuating
  structures

27
Stability of SLP-patterns over timeRunning EOF

• Moving-EOF-analysis with window size a 31years
• Comparison of reference patterns (EOFs over
  1000-1998) with subperiods
• Field-correlation detected by scalarproduct of
  reference pattern R of the whole time period with
  each subperiod-EOF S ? yields rR,S 0,1
• 
  with
• r 0,1 due to orthogonality of EOFs
• Field correlations like RunCor(X,Y) of anomaly
  field with mean

28
(No Transcript)
29
Slides for Discussion
30
Time evolution of NAO-temperature relation
31
(No Transcript)
32
__Changes of westerly winds in the North Atlantic
region
_Temporal evolution of the DJF North Atlantic
Oscillation Index
Zorita, E., Gonzalez-Rouco, F. and S. Wagner
(2009) Low-frequency response of the Arctic
Oscillation to external forcing in the past
millennium. Geophysical Research Letters
(submitted).