Multi-Model%20Ensembles%20for%20Climate%20Attribution

1
Multi-Model Ensembles for Climate Attribution

• Arun Kumar
• Climate Prediction Center
• NCEP/NOAA
• Acknowledgements Bhaskar Jha Marty Hoerling
  Ming Ji OGP Participants in the Seasonal
  Diagnostics Consortium

2
What is Attribution?

• Attribution ascribe to or regard as the effect
  of (a stated cause) (e.g., the delays were
  attributed to the heavy traffic).
• In the context of the observed climate,
  attribution refers to can we relate observed
  climate anomalies to external forcing(s)?

3

• Atmospheric anomalies ? Sea Surface Temperatures
  Soil Moisture OR a manifestation of variability
  internal to the atmosphere
• Trends in the ocean-atmosphere system? changes in
  solar forcing aerosols CO2OR are intrinsic to
  the coupled system

4
(No Transcript)
5
(No Transcript)
6
(No Transcript)
7
Why is Attribution Relevant?

• Helps in understanding what the dominant forcing
  mechanisms controlling climate variability are
• Provides an understanding of why climate is
  evolving the way it is
• Provides a basis for making long-range
  predictions and projections

8
An Approach to Attribution Analysis

• Analysis of observed anomalies (atmosphere,
  ocean, solar,)
• Conceptual separation of system into internal and
  external (or forced and forcing) components
• Formulation of hypothesis (e.g. such and such
  anomaly may be because of such and such forcing)
• Testing the hypothesis involves analyzing
  response to the external forcing
• Empirical approach
• General circulation model approach

9
DJF 1997/98 200-mb Z
10
DJF 2001/02 200-mb Z
11
Problems with attribution based on a single AGCM

• In the AGCM approach, attribution keys on the
  comparison of the observed anomalies with the
  AGCMs response to the external forcing
• Is the AGCM response to an external forcing
  correct?

12

• One possible solution is to
• Use Multi-model approach as a confidence (or
  consensus) builder in documenting response to an
  external factor
• Once a level of confidence could be placed in the
  atmospheric response to the external forcing,
  more definite statements about the causality of
  observed anomalies can be made

13
DJF 1997/98
14
DJF 2001/02
15
Seasonal Diagnostics Consortium
Model CCM3 NCEP NSIPP ECHAM4.5 (From IRI) GFDL
Type Spectral Spectral Grid Spectral Grid
Resolution T40L18 T62L64 2 Deg Lat/Lon T40L18 N45L18
of Simulations 20 18 9 24 10
Total of Simulations 81
16
2004 EOS
17
2004 EOS

• Add the line plot

18
MM Climate Attribution

• Other Application of MM attribution runs
• Documenting atmospheric responses to boundary
  forcings
• Inferring current state of climate
• Analyzing successes and failures of operational
  SI forecasts
• Generating different SI prediction scenarios
• Model validation (has long been the implicit
  basis for various MIP activities)

19
Atmospheric Response to SST Forcing
20
(2000-2004) - (1961-90)
21

• Atmospheric Climate Model Simulations
• Observed SST Forcing
• Institution Model Ensemble Size
• GFDL AM-2 10
• MPI/IRI ECHAM-4 24
• NASA GMAO 23
• Simulation production for 1950-2004
• Specified monthly varying global SSTs
• Climatological GHG/Aerosols
• 57 member multi-model ensemble
• Experiments are part of Seasonal
• Climate Diagnostics Consortium
• http//www.cpc.ncep.noaa.gov/produc
  ts/people/bjha/

22
(2000-2004) - (1961-90)
23
(No Transcript)
24
(No Transcript)
25

• Atmospheric Climate Model Simulations
• Idealized Indian Ocean SST Forcing
• Institution Model Ensemble Size
• GFDL AM-2 10
• NCEP GFS
  10
• NCAR CCM3 10
• NCAR CAM3 10
• Transient 0.1C/yr Indian Ocean warming
• Each run is 11-yrs in duration.
• Climatological SSTs elsewhere.
• Climatological GHG/Aerosols
• 40 member multi-model ensemble
• http//www.cgd.ucar.edu/cas/asphilli/Results/Ram
  ped-IndianOcean/

26
(No Transcript)
27
Inferring Current State of Climate (consistent
with the external forcing)
28
Response Forced by the Observed SSTs
29

• Coupled Ocn-Atm Climate Model Simulations
• Observed GHG/Aerosol Forcing
• Simulation production for 1895-2004
• Specified monthly varying GHG/Aerosol
  1895-2000
• A1B Scenario 2001-2004
• 18 Different Coupled Models/47 total runs
• Experiments are part of the IPCC AR-4 Suite
• http//www-pcmdi.llnl.gov/ipcc/a
  bout_ipcc.php

30
SSTs
Land Temp
31
(2000-2004) - (1961-90)
32
Validating Forecasts
33
NDJFM 2004/2005
34
Efforts in NOAA

• What currently exists An informal activity
  maintaining monthly updates in AMIP runs forced
  by Global SSTs
• Within NOAA, need to formalize attribution
  activity for different time scales could be
  centralized or a virtual activity
• Such an activity can also support other model
  based hypothesis testing efforts (e.g., impact
  of different ocean basins causality of trends
  and droughts )

35
OBS
June 1998-May 2002 (The Perfect Oceans for
Drought)
MODEL