1. Climate prediction activities at Météo-France
Hervé Douville
Météo-France/CNRM
Acknowledgements:
L. Batté, C. Ardilouze, M. Déqué,
J-F. Guérémy, J-P. Piedelièvre
and E. Maisonnave
WGSIP 16th session, Exeter,, March 2014

2. Outline Sensitivity tests to resolution Towards System 5
Recent global warming hiatus
Summary and prospects

3. Sensitivity tests to resolution
Experiment design
hindcasts
ARPEGE V5.1 AGCM / NEMO V3.2 OGCM (climatological sea-ice)
Low Res: Tl127 (1.4°) L31 (or L71 for high-top) / 1°L42
High Res: Tl359 (0.5°) L31 / 0.25°L75 (with CERFACS)
4 seasons (start 1 Feb, May, Aug, Nov) till month 4
30 members (initial atmosphere perturbation)
10 members till month 7
60 members in high resolution (30 selected, sampling intervals)
Atmosphere start: Era interim
Ocean start: Glorys/Psi2g3

4. Horizontal resolution
Nino 3.4 SST
« spring barrier »
(By courtesy of M. Déqué)

5. Horizontal or vertical resolution
Nino 3.4 SST (+71l)
(By courtesy of M. Déqué)

6. Horizontal resolution Impact on spatial mean temporal correlations
Scores for extratropical Z500
90S-30S
30N-90N
(By courtesy of M. Déqué)

7. T2M P Horizontal resolution
Impact on spatial mean temporal correlations
90S-30S
30S-30N
30N-90N
T2M
30N-90N
90S-30S
30S-30N P
(By courtesy of M. Déqué)

8. High Res vs Low Res: summary
Worse in the tropics in Spring
Better in other seasons
Significant differences (60 members)
Summary of temporal correlations for extratropical circulation indices
Low res
High res
High res A
NAO (DJF)
0.41
0.46 (0.52 with 71l)
0.34
NAM (DJF)
0,39
0.60 (0.55 with 71l)
0.48
(By courtesy of M. Déqué)

9. Towards System 5
ACC DJF; IT=[-30,+30] ; NH=[+30,+90] ; SH=[-30,-90] ; N34=Niño3.4
32 years ( ), 20 members among 30, (perturbed ICs):
N8=Tl127l91, CMIP5 Physics, ajc, jpp (dt=1200s, SBU=290/mois ?),
N16=Tl127l91, dyns, New Physics, ajc, SFX7.3, gelato, jpp (dt=900s, SBU=590/mois),
N17=id N16 Tl159l91, jpp; NP_HTg159=id N17-dyns-O3 pro, jfg N8
N16
N17
NP_HTg159
T2m .IT
Z500.IT
PREC.IT
T2m .NH =
= =
Z500.NH =
PREC.NH = =
T2m .SH =
= + +
Z500.SH =
= =
= = =
PREC.SH =
= - -
T2m.N34 =
= - -
In red, comparison vs neighbouring left column, blue (green) vs first (second) column
(By courtesy of J-F. Guérémy)

10. Cor DJF (30 members) Towards System 5 N8 N16 N17 NP_HTg159
(By courtesy of J-F. Guérémy)

11. Cor DJF (30 members) Towards System 5 N8 N16 N17 NP_HTg159
(By courtesy of J-F. Guérémy)

12. ENSO contribution to the recent global warming hiatus
Experiment design
integrations
ARPEGE V5.2 AGCM / NEMO V3.2 OGCM (Gelato sea-ice)
Tl127 (1.4°) L31 / 1°L42
5 members (initial states from historical CMIP5 simulations)
HISCTL: CMIP5 simulations (ALL forcings)
HISSST: HISCTL + SST anomaly nudging in central and eastern tropical Pacific (as in Kosaka and Xie 2013)
HISTAU: HISCTL + prescribed wind stress in the tropical Pacific (not exactly as in England et al. 2014)
All anomalies relative to the climatology
Question not addressed here and still a matter of debate:
Should we expect more la Niña events in a warmer climate?

13. Correlation with observed annual mean SST
1997
Domain with SST nudging in HISSST
Domain with prescribed wind stress in HISTAU
1997
(Douville and Voldoire, in preparation)

14. Global annual mean T2M anomalies
Free OAGCM
Constrained OAGCM
HISSST results in line with Kosaka and Xie (2013) despite a lower (more realistic) ENSO influence on global mean temperature in CNRM-CM5:
ENSO multi-decadal variability (mostly internal and not predictable by state-of-the-art OAGCMs) is sufficient to explain the recent global warming hiatus;
HISTAU results in line England et al. (2014) despite a reduced domain with prescribed wind stress; global warming hiatus less pronounced than in HISSST due to a too narrow ENSO signal in Pacific SST (+ cold start).
(Douville and Voldoire, in preparation)

15. 1998-2012 annual mean T2M linear trends
Correlation with ERAI pattern
Stippling denotes significant trends at the 5% level
(i.e. weak ensemble spread)
K/decade
(Douville and Voldoire, in preparation)

16. Recent global warming hiatus
(Douville and Voldoire, in preparation)

17. End

18. Stochastic dynamics: principle
X(t + ∆t) = X(t) + M(X(t), t) + δX
Perturbed variables : T , Ψ, Q
δX : random draw every 6 hours of a coherent initial tendency error correction term from a given population {δX}
derived from a 32-winter 4-member coupled model run weakly (10-day relaxation time for Ψ, 1 month for T and Q) nudged towards ERA-Interim
Classification of the {δX} population according to :
- actual month (« perfect sampling »): SD_OPT
- current month (november to february): SD_RAND
- other criteria : ongoing research !
Courtesy of L. Batté

19. Stochastic dynamics: results
Pattern ACC of DJF Z500 (northern extratropics)
Mean Pattern ACC in DJF
Courtesy of L. Batté

20. Other sensitivity tests
Additional experiments (in progress)
Low res initialized by Nemovar (not shown)
High res with 3 hour coupling (winter only)
Tl359 with stratosphere (not shown)
Tl359 coupled with Nemo 1° (in progress)

21. Horizontal or vertical resolution
Spatial mean temporal correlations for precipitation (DJF only)
90S-30S
30S-30N
30N-90N