Embed Size (px)
Citation preview
Towards a Seasonal Predic0on System using MPI-‐ESM:
Descrip0on, Performance, Analysis Daniela Domeisen*, Kris/na Fröhlich (DWD), Wolfgang Müller (MPI), Michael Botzet (MPI), Luis Kornblueh (MPI), Dirk Notz (MPI), Robert Piontek (U Hamburg), Holger Pohlmann (MPI), Steffen Tietsche (U Reading), Johanna Baehr (U Hamburg)
* Ins/tute of Oceanography, University of Hamburg [email protected]
Interna/onal workshop on seasonal to decadal predic/on Toulouse, 13 – 16 May 2013
2
Ø MPI-‐ESM-‐LR in setup used for CMIP5 and IPCC AR5
Ø use of the MPI-‐ESM as a seasonal forecas/ng system
Ø planned German contribu/on (DWD, MPI, Univ. Hamburg) to EUROSIP (ECMWF opera/onal mul/-‐model seasonal forecas/ng ensemble)
§ simple ini/aliza/on: atmosphere / ocean / sea ice nudging and breeding in the ocean for ensemble genera/on
Seasonal forecasting with the MPI-ESM
3
n ECHAM6 (atmosphere) • horizontal T63 (1.9°x1.9°) • vertical L47 (47 levels up to 0.01hPa,
~24 levels above 200hPa) • land module JSBACH and
hydrological discharge model n MPIOM (ocean) plus sea ice • horizontal: 1.5° • vertical: 40 levels • sea ice model (dynamic /
thermodynamic) Literature on MPI-‐ESM: Giorgeba et al (2012), Stevens et al (2013), Jungclaus et al (2013), Schmidt et al (2013)
MPI-ESM: The Earth System Model of the Max-Planck-Institute for Meteorology Hamburg
Seasonal forecasting with the MPI-ESM: Model Setup
4
Data assimilation: Full field nudging towards reanalysis data
1. Atmosphere: ERA-INTERIM (vorticity, divergence,
temperature, surface pressure). Atmospheric temperatures are only nudged above the boundary layer. Relaxation time: 1 day
2. Ocean: ECMWF (ORA-S4) ocean (salinity, temperature, no SST nudging). Relaxation time: 10 days
3. Sea ice: NSIDC data (sea ice concentration). Effective relaxation time: 10 days
Seasonal forecasting with the MPI-ESM: Initialization
more informa/on on model ini/aliza/on: Poster 30 S2 P1, J. Baehr et al.
5
Ensemble genera0on: Breeding
Ø capture the informa/on about the fastest growing modes in the model related to flow dependent instabili/es (Toth and Kalnay, 1997)
Ø used to generate ensemble spread for free hindcast simula/ons
Ø perturba/ons of ocean only (3D temperature and salinity)
Hindcasts: November and May start dates for 1980 -‐ 2011
Ø ini/alize every November 1 and May 1 with a run/me of 1 year
Ø 9 ensemble members for each start date
Seasonal forecasting with the MPI-ESM: Ensemble generation
6
2m temperature for DJF: MPI-‐ESM vs ERA interim
surface temperature for DJF: MPI-‐ESM vs HOAPS
RMSE
ACC
Evaluation of the model skill: RMSE and ACC
Figures: Kris/na Fröhlich
7
Evaluation of the model skill: ensemble spread
Ø generally good ensemble spread, especially in Northern Hemisphere
Ø central Pacific ensemble spread underestimates variability
model underes/mates observa/ons model overes/mates observa/ons
Talagrand of bias corrected surface temperature for DJF (1982 – 2010)
Figure: Kris/na Fröhlich
8
Evaluation of ENSO predictability: RMSE and ACC
ACC
RMSE
MPI-‐ESM assimila/on run HadCRU2010 Ensemble members
Figures: Kris/na Fröhlich
9
Evaluation of ENSO predictability: El Niño 1997/98
K K K
t i m
e
t i m
e Figures: Kris/na Fröhlich
10
Evaluation of the stratosphere: ENSO vs stratospheric variability
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 20105
0
5
tem
pera
ture
anom
alie
s [C
]
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
10
5
0
5
10
tem
pera
ture
anom
alie
s [C
]
ü strong El Nino
events high polar cap
temperatures
è enhanced predictability over Europe?
Nino3.4 temperature anomalies Dec 1 – Feb 28
polar cap (60-‐90N, 10hPa) temperature anomalies Dec 1 – Feb 28
[°C]
[°C]
> 6 SSWs/winter in 8 ensemble members ensemble mean Acknowledgments: Amy Butler
11
ü variability well reproduced ü ensemble spread encompasses natural variability x predictability of specific events
Evaluation of the stratosphere: Stratospheric variability and predictability
ERA interim ensemble mean ensemble members
zonal mean zonal wind at 60°N / 10hPa for winter 2006/2007
2006 2007
12
Composite of 67 of the stratospheric sudden warmings btw 1981- 2010 ü downward coupling x precursor structure
pres
sure
lag [days]
Principal Component from EOF 1 in stddev, level by level, SSW composite (67 events)
30 20 10 0 10 20 30
1
3
10
30
100
300
900
2
1.5
1
0.5
0
0.5
lag [days]
pressure [h
Pa]
Level-‐by-‐level EOF/PCs, cp. Baldwin & Dunkerton (2001) Acknowledgments: Edwin Gerber
Evaluation of the stratosphere: Stratospheric sudden warmings
95% significance
zero line
13
Ø the MPI-‐ESM can be successfully ini/alized as a seasonal predic/on model using breeding for ensemble genera/on and nudging for data assimila/on
Ø atmosphere is well reproduced by ensemble in terms of its mean state and variability
Ongoing and future work
Ø opera/onal work: seong up the system at ECMWF within the EUROSIP framework
Ø research: predictability of § ENSO § MJO § ocean heat content § stratosphere / stratosphere – troposphere coupling
Summary and Conclusions
14
Ø Baehr, J., K. Fröhlich, W. Müller, M. Botzet, D. Domeisen, D. Notz, R. Piontek, H. Pohlmann, S.
Tietzsche (in prep.): A seasonal predic.on system based on the MPI-‐ESM coupled climate model. Ø Baldwin, M.P. and T. Dunkerton (2001): Stratospheric harbingers of anomalous weather
regimes. Science, 294. Ø GiorgeOa, M.A. et al. (2012): Climate change from 1850 to 2100 in MPI-‐ESM simula/ons for the
Coupled Model Intercomparison Project 5. JAMES. Ø Jungclaus JH, Botzet M, Haak H, Luo J, La/f M, Marotzke J, Mikolajewicz U, Roeckner E (2006):
Ocean circula.on and tropical variability in the coupled model ECHAM5/MPI-‐OM. J. Clim. 19:3952–3972
Ø Piontek, R. and J. Baehr (in prep.): Perturba.on of mul.-‐year ensemble simula.ons through bred vector implementaion into the ocean component of a coupled climate model.
Ø Schmidt, H. et al. (2013): The response of the middle atmosphere to anthropogenic and natural forcing in the CMIP5 simula/ons with the MPI-‐ESM. JAMES.
Ø Stevens et al. (2013) The atmospheric component of the MPI earth system model: ECHAM6. J. of Advances in Modeling Earth Sys., DOI 10.1002/jame.20015
Ø Toth, Z. and E. Kalnay (1997): Ensemble Forecas.ng at NCEP and the Breeding Method, Mon. Weather Rev. Vol. 125, 3297-‐ 3319.
Literature
