WP4'4: Sources of predictability in current and future climates Laurent Terray CERFACS

1
WP4.4 Sources of predictability in current and
future climatesLaurent Terray (CERFACS)

• Participants CERFACS, CGAM, CNRM, DMI(nc),
  ECMWF, IfM, IPSL, ISAC(nc)

Thanks to R.Sutton (CGAM), H.Douville (CNRM),
F.Doblas-reyes (ECMWF), S.Corti (ISAC),
B.Christiansen (DMI), J.P.Duvel (IPSL)
2
Main objectives

• To develop methodologies and tools to exploit
  existing seasonal to decadal hindcasts for
  identifying and understanding the sources of
  predictability in current and future climates
• To assess and understand the main factors which
  influence the predictability of the climate
  system at different time scales
• To improve the understanding of the interaction
  between anthropogenic climate change and natural
  climate variability modes and of the possible
  changes in predictability at all time scales

3
Work Plan for months 1-18

• T4.4a design a general framework for the
  analysis of participating models (ALL)
• T4.4b assess the role of snow and soil moisture
  in the predictability of climate (CNRM)
• T4.4c assess the potential predictability of the
  North Atlantic region at seasonal to decadal
  timescales (ALL)
• T4.4d investigate the vertical structure of
  weather and climate regimes in several
  re-analysis products and the potential role of
  the stratosphere (DMI, ISAC, CERFACS)
• D4.4.1 Synthesis of current estimates and
  mechanisms of predictability on seasonal to
  decadal timescales, including understanding the
  influence of ocean initial conditions, and with a
  focus on the North Atlantic European sector
  (month 18)
• M4.4.1 development of methodologies to explore
  climate variability and predictability, for use
  with the ENSEMBLES system
• (month 18)
• M4.4.2 Assessment of climate variability and
  predictability in exixting simulations to provide
  benchmark against which the ENSEMBLES system can
  be judged (month 18)

4
CGAM contribution to WP4.4
5
Initial condition information is ignored in
current climate forecasts
Northern European temperatures
R.Sutton
Source Anne Pardaens, Hadley Centre / PREDICATE
6
Evidence from FP5 PREDICATE project of decadal
predictability in the THC
Control simulations
Perturbed runs
Source Mat Collins,
What mechanisms determine the extent of
predictability in ocean and atmosphere variables?
R.Sutton
7
Questions and Methods

• What mechanisms determine the predictability of
  the Atlantic THC, and related aspects of climate,
  in current climate models?
• To which aspects of the ocean initial conditions
  are forecasts of the THC, and related aspects of
  climate, most sensitive?
• and later in the project
• How do initial conditions and changing external
  forcings combine to determine the evolution of
  climate on decadal timescales?
• Methods
• Further analysis of PREDICATE ensemble
  integrations
• New ensemble integrations with HadCM3 model
  (larger ensembles)
• A new methodology to estimate empirical singular
  vectors for the THC. (addresses question 2.)

R.Sutton
8
CNRM contribution to WP4.4
9
Questions and Methods

• Explore the predictability associated to land
  surface anomalies
• What is the influence of soil moisture
  conditions on atmospheric seasonal
  predictability?
• And later in the project
• Assess the influence of snow conditions on
  seasonal (to interannual?) predictability
• Methods
• Preliminary step produce a 10-yr global monthly
  mean soil moisture climatology using the 3-hourly
  atmospheric forcing provided by GSWP-2.
• run ensembles of global atmospheric simulations
  with the ARPEGE AGCM(prescribed observed SSTs
  from 1986 to 1995 and with GSWP-2 vs
  climatological initial conditions).

10
Influence of soil moisture relaxation towards
GSWP-1 on the JJAS Z500 stationary eddy anomalies
simulated by the ARPEGE AGCM
Douville Chauvin (2000), Climate
Dyn.,16,719-736 Douville H. (2OO2),
J.Climate,15,701-720
11
(No Transcript)
12
ECMWF contribution to WP4.4
13
Questions and Methods

• Focus on predictability of current climates
• Influence of anthropogenic forcing upon the
  seasonal-to interannual predictability of natural
  modes of variability (ENSO, NAO, PNA) to explain
  the latest results (see below)
• ECMWFs effort will take place after month 18
• Links to WP5.3 (Assessment of forecast quality)

2-4 months lead time (DJF)
14
Southern Europe DEMETER hindcasts
Precipitation
T2m
Nov start date
2-4 (DJF)
4-6 (FMA)
15
IPSL contribution to WP4.4
16
Questions and Methods

• Intraseasonal convective and dynamical
  perturbations have a large impact on the Asian
  monsoon activity and on the triggering of ENSO
• What is the predictability of the intra-seasonal
  activity in the Indo-Pacific region
• Study the seasonal predictability of the
  intra-seasonal oscillation in the Indo-Pacific
  region in current and future climates
• Methods
• Develop an operational tool to test the seasonal
  forecast of the intraseasonal oscillation in the
  tropics and use this tool to assess the skill of
  the different global ESMs
• First 18 months (RT5) Use DEMETER simulations
  to develop a diagnostic tool (based on the Local
  Mode Analysis) to infer the skill of seasonal
  hindcasts in describing the intraseasonal
  oscillation in the Indo-Pacific region.
• Remaining time up to 5 years (WP4.4) Analysis
  of the seasonal predictability in current and
  future climates using the core ENSEMBLES
  simulations (links with potential changes in ENSO
  activity)

17
Variability of the ISO patterns between hindcasts
members Internal Variability
OLR-NOAA

• Example for the CNRM model in January 2002
• One member (member 9) give a reasonable pattern
• One member (member 5) with low organisation (weak
  var), unrealistic pattern at too short time scale

Member 9
Member 5
18
DMI contribution to WP4.4
19
Questions and Methods

• Evidence for nonlinear regime behaviour has been
  found in both the stratosphere and the
  troposphere and strong evidence has been reported
  for a stratospheric regime shift in the late half
  of the 1970ies
• What is the atmospheric regime behaviour in the
  recent period ? What is the vertical extent of
  the regimes ?
• Are there any connections between the
  stratospheric and tropospheric regimes (polar
  vortex strength and the NAO-AO)?
• Methods
• Critical assessment of the standard algorithms
  (k-means, mixture models) used to perform
  clustering (nature of the underlying probability
  distribution) - link with WP4.3, KNMI ?
• Use of the ERA40 dataset
• Later in the project analysis of the core
  ENSEMBLES simulations for current and future
  climate (Any of the core ENSEMBLES models
• with high-res in the stratosphere ??)

20
Bimodality in the tropospheric wave amplitude
index
Christiansen JAS 2005
Wave amplitude index defined by Hansen and Sutera
Change in 1990
Bimodality in the strength of the stratospheric
vortex
Christiansen 2003 J. Climate
Change in 1979
What is the connection?
21
ISAC contribution to WP4.4
22
Questions and Methods

• What is the vertical and thermal structure of
  (global, hemispheric-scale)
• circulation regimes for the current climate?
• Explore the potential role of weather regimes and
  non-linearity in the
• emerging anthropogenic signal.
• Later in the project
• Verification of regime structure in present and
  future climate core ENSEMBLES
• simulations.
• Interaction between natural and forced
  variability
• Regime response to anthropogenic forcing and SST
  anomalies
• Troposphere-stratosphere connection
  Collaboration with DMI
• Methods
• Study of the extended winter(Oct-Apr) with
  reanalysis datasets (NCEP
• and ERA40)
• Diagnostic tools multivariate EOF analysis, Pdf
  estimators and clustering
• techniques

23
Multivariate combined EOF analysis Data NCEP
reanalysis Clustering in the first 2-EOFs phase
space K-means algorithm 3-cluster
partition positive NAM
Changes in cluster frequency and significance
when different periods (corresponding to
different external forcings ENSO and climate
signal) are considered.
It suggests that the associated tropical heating
anomalies reorganize the mid-latitude
circulation sufficiently to disrupt the normal
regime behaviour.
24
CERFACS contribution to WP4.4
25
Questions and Methods

• What are the physical processes associated to
  climate predictability of the North Atlantic
  European sector at various timescales ? (Focus on
  SST influence and interaction between the
  different ocean basins) (months 1-18)
• What is the influence of anthropogenic forcing
  upon the levels of predictability of the major
  climate modes ? (months 19-60)
• Methods
• Analyses of existing integrations (e.g PREDICATE
  and DEMETER) and coordinated experiments (to be
  discussed)
• Assess the relevance of various predictability
  measures to improve the understanding of physical
  mechanisms (e.g relative entropy Kleeman 2002
  Stephenson and Doblas-reyes 2000)
• Analyses of the core ENSEMBLES integrations

26
Weather regimes and local climate
2003 heat wave a process study
Summer (JJA) weather regimes (daily timescale)
from NCEP-NCAR Reanalysis (1950-2002)
A

represent 80 Of 2003 summer days
Tropical Atlantic forcing?
A
associated to an increase of warm days
(exceeding the 95 percentile) over France (data
from Météo-France)
OLR anomalies for June 2003
Two ensembles of 40 members with the NCAR
AGCM One CTRL and one forced with 2003 TATL
diabatic heating
Simulated changes Of warm regime occurrence For
JJA 2003 in response To the tropical Atlantic
diabatic heating forcing Cassou et al. 2004
A
Percentage of days exceeding the 95
climatological threshold for a given regime
27
Influence of anthropogenic forcing on the NAO
Perturbed climate
Current climate
GHG forcing
NAO-
NAO
PRUDENCE simulations series of Time-slice exp.
With ARPEGE (high res. Over Europe, 50 km)
forced by Observed SST and GHG (1960-1999)
And SST (from 2 CGCMs) and SRES Scenarios
(2070-2099) Terray et al. Jclimate 2004
28
Remarks

• Existing simulations PREDICATE, DEMETER, AR4,
  Others Need a list of available model data
• Coordinated experiments to be discussed soon
• Need good coordination with WP4.2 and WP5.3
• Utility and limitations of regime analysis
  algorithms (interaction with WP4.3, others e.g
  downscaling)