Transcript Meteo-France EUROSIP contribution: present, future and

Météo-France EUROSIP contribution:
present, future and sensitivity experiments
Jean-François Guérémy
Michel Déqué, Jean-Philippe Piedelievre, Lauriane Batté
Outlook
• Present: EUROPSIP Syst3
- Components
- Some products
• Future: EUROPSIP Syst4
- Components and first tests
• Sensitivity experiments and specific studies
- l91 versus l31 in the frame of EUROPSIP Syst3
- Predictability of regimes and heavy precipitation events
- Skill sore over Africa
EUROPSIP Syst3
• Operational international project:
Multi-model seasonal forecast; (CNRM, ECMWF, UKMO)
• model:
ARPEGE-Climat V4 Tl63l91 + OASIS V2.4 + OPA V8.2 (ORCA 2° grid)
Since May 2008
• forecast set-up:
- Ocean ICs: MERCATOR, Kalman filter analysis including altimetry, SST
and (T,S) in situ data
Atmospheric ICs: ECMWF analyses
- 41 members (8 lagged atmospheric ICs combined to 5 lagged ocean
ICs), each month (starting the first) for 7month run
- Hindcasts from 1979 (11 members)
EUROPSIP Syst3; product examples
• Hydrology: reservoir management in Mali
• Energy: heating management in western Europe
• JFM forecast
Manantali Dam
Water release management in September-October function of the August
seasonal forecast
heating management in western Europe
Seasonal cumulation: 18°C – (Tn + Tx)
JFM forecast
Toward EUROPSIP Syst4
operational mid 2012
• Model:
ARPEGE-Climat V5 Tl127l31 + OASIS V3 + NEMO V3.2 (GELATO ice, ORCA
1°)
(new radiation and soil schemes - SURFEX)
IPCC set-up (climate projections and decadal forecasts)
• Preliminary tests without SURFEX and GELATO:
Similar scores compared to EUROSIP Syst3
44 years (NDJF and MJJA)
9 members
Month 2-4 seasonal average
Anomaly correlations
Toward EUROPSIP Syst4
operational mid 2012
• Initial Conditions:
Possibly, nudging or anomaly nudging in coupled mode toward atmospheric
analyses (ECMWF), ocean analyses (MERCATOR, from a ¼° analysis) and
possibly soil analyses (MF).
Initial and/or in-run perturbations taken from analysis departure terms
(stochastic term).
l91 versus l31 in the frame of EUROPSIP Syst3
• Time period: 1979-2007, EUROPSIP Syst3 l31 and l91
Predictability of regimes and heavy precipitation
events (MEDUP french project)
• National research project: Seasonal forecast of weather regime
and heavy precipitation event occurrence (from 01/2008 to
12/2010); (CNRM, IPSL, LTHE)
• Time period: 1960-2001 ENSEMBLES EU project (42 years)
• Sensitivity tests: 3 # models: ENS_MF, ENS_CEP and
Pro_Tl127l62 (CNRM model with a new atmospheric physics –
turbulence, convection and microphysics)
Precipitation biases (/CMAP), coupled simulations
DJF
Standard
Tl63l31
Pronostique
(convection
Guérémy)
Tl127l62
Guérémy 2011, accepted in Tellus
JJA
Teleconnections SON
(similar to Guérémy et al. 2005, Tellus)
• Composites of T2m anomalies for the years during which the
occurrence of Heavy Precipitating Events (HPE) over South-East of France
was greater than the mean + 1 standard deviation (1960-2001).
ERA40
ENS_MF
Years: 1960, 1963, 1964, 1968,
1977, 1994, 1995
>= 7 HPE per year
Pro_
Tl127l62
ENS_CEP
Teleconnections SON
• Composites of 200 hPa Velocity Potential anomalies for the years
during which the occurrence of Heavy Precipitating Events (HPE) over
South-East of France was greater than the mean + 1 standard deviation
(1960-2001).
ERA40
ENS_MF
Pro_
Tl127l62
ENS_CEP
Teleconnections SON
• Composites of Z500 anomalies for the years during which the
occurrence of Heavy Precipitating Events (HPE) over South-East of France
was greater than the mean + 1 standard deviation (1960-2001).
ERA40
ENS_MF
Pro_
Tl127l62
ENS_CEP
Regimes (stream function , velocity potential ) 200hPa
ERA40, SON
1
2
3
4
5
6


Predictability of (, )200 regime occurrence
relationship to heavy precipitation events, SON
ENS_MF
ENS_CEP
Pro_Tl127l62
Regime 1
0.45
0.38
0.41
Regime 2
0.17
0.30
0.28
Regime 3
0.12
-0.11
-0.11
Regime 4
-0.09
0.39
0.05
Regime 5
0.19
0.13
-0.08
Regime 6
0.49
0.23
0.39
Correlation regime occurrence / ERA40
Association HPE (177
over 42 years)
/ regimes (, )200
Predictability of Z500 regime occurrence
relationship to heavy precipitation events, SON
NAO1
Atlantic trough
2
ENS_MF
ENS_CEP
Pro_Tl127
l62
Regime 1
0.06
-0.05
0.08
Regime 2
-0.02
0.00
0.00
Regime 3
-0.07
-0.04
0.27
Regime 4
0.24
0.06
0.31
Correlation regime occurrence / ERA40
NAO+
3
Atlantic ridge
4
Association HPE (177 over 42
years) / regimes Z500
Predictability of heavy precipitation events (HPE)
from analogues, SON
• Following Clark and Déqué (2003, QJRMS), analogues (for each member of the
ensemble) are chosen in the model hindcasts according to a minimum distance in
terms of (, )200 regime occurrence.
The result is a distribution of analogue years.
From this distribution, forecasted HPE occurrence (larger than mean + 1
std) is calculated and compared to the observed occurrence of the
considered year.
Occurrence
correlation
Occurrence
correlation
ENS_MF
ENS_CEP
Pro_Tl127l62
-0.07
-0.13
0.23
ENS_MF
ENS_CEP
Pro_Tl127l62
0.00
0.09
0.21
From the analogues
Model forcasted
HPE occurrence
Skill sore over Africa
RPSS (Ranked probability skill score) computed for precipitation
deciles over the 1960-2005 period (ENSEMBLES), for West Africa in
JJA and South Africa in DJF; GPCC is the reference. Results from a
multi-model (5 out of 9); this muti-model provides greater scores
than 0, where most of individual model gives no information.
Batte 2010, accepted in Tellus
Bathymetry ORCA2 and ORCA1
ORCA2 (182*149)
ORCA1 (362*292)
Scores over DJF, observed SST mode
Std phys (Tl63l31)
Pro phys (Tl63l31)
ACC T850 Trop
0.52
0.52
0.56
ACC T850 NH
0.25
0.31
0.31
ACC Z500 NH
0.25
0.29
0.33
Standard physics, left
Prognostic physics, right
(31 levels top, 91 levels bottom)
 Pro91 > Pro31> Std31
Pro phys(Tl63l91)