1. The ECMWF Monthly and Seasonal Forecast Systems
D. Anderson, M. Balmaseda, L. Ferranti,
F. Molteni, T. Stockdale, F. Vitart
ECMWF, Reading, UK

2. Main topics Monthly forecast system: merging with VAREPS
new atmospheric model cycle : improved simulation of tropical intra-seasonal variability
some recent results from the operational version
Seasonal forecast system: from System 2 to System 3
improved ocean data assimilation
better predictions of tropical SST
reduced systematic errors in atmospheric fields

3. Monthly Forecasting System
Real-time forecast:
Coupled ocean-atmosphere integrations: a 51-member ensemble is integrated for 32 days every Thursday.
Atmospheric component: IFS with the latest operational cycle and with a TL159L62 resolution
Oceanic component: HOPE (from Max-Planck Institute) with a zonal resolution of 1.4 degrees and 29 vertical levels
Coupling: OASIS (CERFACS). Coupling every ocean time step (1 hour)
Background statistics:
5-member ensemble integrated at the same day and same month as the real-time time forecast over the past 12 years
Yes/no, error bars, probabilities?
Users should be told about past records of forecast quality

4. Model changes during the past year
New versions of IFS :
Cycle 29R2 in June new sea ice treatment*
Cycle 30R1 in Feb change of vertical resolution (62 vertical levels)
The sea-ice cover is persisted from the atmospheric initial conditions till day 10, then relaxed towards climatology. After day 30, the sea-ice cover is the climatologocal sea-ice cover (from ERA40).
Other changes:
Archiving of probabilities
Verification web site
New products (Madden Julian Oscillation, products for the Southern Hemisphere)

5. TL399L62 twice a day uncoupled
Future developments: Merging the monthly forecasting system with VAREPS
Present situation: 2 separate systems
EPS:
TL399L62 twice a day uncoupled
Day 0
Day 10
TL159L62 once a week
MOFC:
Day 0
Day 32
Ocean model

6. Future developments: Merging the monthly forecasting system with VAREPS
Q1 2007: single system
VAREPS:
Once a week
TL399L62
TL255L62
Day 0
Day 32
Day 10
MOFC:
Day 10
Day 0
Day 32
Ocean model

7. 48 5-ensemble member cases, CY30R1
Prob. 2m temp. in upper tercile. NH. Day 12-18
VAREPS
Op. TL159
32/ % significance
ROC Area:
Tl159
VAREPS

8. Madden Julian Oscillation experiments: 15/12/92 - 31/01/93
ERA40 Analysis:
Velocity Pot. 200 hPa
OLR
U 850 hPa

9. Spectra of tropical velocity potential
MJO simulation:
Spectra of tropical velocity potential
(in seasonal exper.)
ERA-40
Cy 30R2
Cy 23R4

10. Madden-Julian oscillation Forecast starting on 31 December 1992
CY29R1
Analysis
CY30R2

11. MJO EOF analysis
The combined EOFs (vel. pot. 200-hPa, U 850 hPa, OLR) have been computed on ERA40 daily data from to 2004.

12. MJO EOF analysis
PC1
PC2
30R2
29R1

13. Precipitation over India
08/05-15/05
15/05-22/05
22/05-29/05
29/05-05/06
ANA
Day
5-11
Day
12-18

14. Precipitation over India
08/05-15/05
15/05-22/05
22/05-29/05
29/05-05/06
ANA
Day
19-25
Day
26-32

15. Indian monsoon Probability of precipitation in the upper tercile
27 real-time cases covering the period May-June-July-August 2002, 2003, 2004 , 2005
Monthly Forecast
Persistence of the probabilities
of the previous week
DAY 19-25
DAY 26-32
DAY 12-18
ROC score:
ROC score:
ROC score:

16. System-2 ECMWF seasonal forecast COUPLED MODEL
Atmospheric model cycle 23R4
Atmospheric resolution TL95 and 40 levels
Hope ocean model (1x1)
Oasis coupler
INITIALIZATION
ERA-15 data to initialize ocean and atmosphere
Assimilation of subsurface temperature only
“Multivariate” corrections to the salinity and velocity fields
Ensemble of 5 ocean analyses back to 1987.
ENSEMBLE GENERATION
Real time FC: 40 ensemble members (SST and wind perturbations)
Back integrations:
5 members, for calibration.
40 members (Nov and May starts) for skill assessment.

17. System-3 (expected late 2006)
COUPLED MODEL
New cycle of atmospheric model (Cy31R1)
Higher atmospheric resolution TL159 and 62 levels
Green house gasses and new aerosols.
New sea-ice specification algorithm
Include ocean currents in wave model
INITIALIZATION
ERA-40 data to initialize ocean and atmosphere
Include bias correction in ocean assimilation.
Include assimilation of salinity and altimeter data.
Ocean reanalysis back to 1959, using ENACT/ENSEMBLES ocean data
ENSEMBLE GENERATION
Extended range of back integrations: 11 members,
Revised wind and SST perturbations.
Use EPS Singular Vector perturbations in atmospheric initial conditions.
Forecasts out to 12 months (4x per year)

18. EUROSIP: EUROpean multi-model Seasonal-to-Interannual Prediction system
Currently composed of ECMWF (System 2), Meteo-France (Arpege/ORCA) and UK MetOffice (GloSea) coupled systems
Ensemble integrations performed at ECMWF
Multi-model products to be computed and made available to member states
Graphical products available on the ECMWF web site
Data access/distribution policy to be agreed

19. New ECMWF operational ocean (re)analysis
Basic (existing) Setup:
Ocean model: HOPE (~1x1 going to 1x.3 at the equator)
Assimilation Method OI
Assimilation of T + Balanced relationships (T-S, ρ-U)
10 days assimilation windows, increment spread in time
Ensemble of 5 ocean analyses to represent uncertainty
System-3
+New Features
ERA-40 fluxes to initialize ocean
Retrospective Ocean Reanalysis back to 1959.
Multivariate on-line Bias Correction .
Assimilation of salinity data.
Assimilation of altimeter-derived sea level anomalies.
3D OI
ECMWF has a new operational ocean (re)analysis to initialize seasonal and monthly forecasts
To the basic setup (existing SYSTEM 2), we have added new features. The new system assimilates salinity and altimeter.

20. Reanalysis time series : trends and variability
(with uncertainty from 5 ocean analysis)
North Atlantic:
T300 anomaly
Climate signals….
North Atlantic:
S300 anomaly
The shading in the figures is a representation of the uncertainty in our analysis. It comes from the ensemble of 5 analysis in which the wind stress has been perturbed.

21. THC: Atlantic Meridional Transport (30N)
Values from Bryden et al 2005

22. Tropical Pacific SST indices: bias
---- System 2
---- System 3 (exp.)

23. Tropical Pacific SST indices: skill scores
---- System 2
---- System 3 (exp.)

24. Tropical Pacific SST indices: error time series
---- System 2
---- System 3 (exp.)

25. JFM rainfall standard deviation and regression against the Nino3
JFM rainfall standard deviation and regression against the Nino3.4 index: GPCP

26. JFM rainfall standard deviation and regression against the Nino3
JFM rainfall standard deviation and regression against the Nino3.4 index: System 2

27. JFM rainfall standard deviation and regression against the Nino3
JFM rainfall standard deviation and regression against the Nino3.4 index: System 3

28. Systematic errors in atmospheric fields: 500-hPa geopotential height in JFM (m.4-6)

29. Systematic errors in atmospheric fields: sea level pressure in JAS (m

30. Systematic errors in atmospheric fields: 2-m. temperature in JAS (m

31. Latest Niño 3.4 Forecast EuroSIP ECMWF
The EUROSIP NINO plumes for the forecast started in March06 indicate a fast return to near-normal conditions by May and the tendency of warm after that.
Here is the latest collection of forecasts of 20 ENSO models. Time Series of predicted sea surface temperature anomalies (SSTA) for the Niño 3.4 index (°C) from various dynamical and statistical models for nine overlapping 3-month periods. Figure provided by the Forecasting and Prediction Research Group of the IRI.
Plume of 20 ENSO forecast models from March 2006.
There is quite a range of forecasts from later northern spring onward. Some clustering of model forecasts is apparent. Four models (NCEP/CFS, ECHAM/MOM, CPC Markov and Zhang intermediate coupled model) see the mild cold even continuing into and
beyond the middle of 2006, while the others generally call for a return to neutral by around June, with a few warming to weak El Nino anomaly levels by July or August.
IRI defines La Nina as SSTs in the Nino3.4 region being more than the following anomaly amounts, varying as a function of the season:
DJF AMJ ASO
JFM MJJ SON
FMA JJA OND
MAM JAS NDJ
The above anomaly thresholds approximately mark the 25%iles for the distributions of the respective seasons.

32. ECMWF: Prob. 2m temp. > upper tercile

33. EUROSIP: Prob. 2m temp. > upper tercile

34. Seasonal forecasting of tropical storms
ECMWF
Forecasts starting on 1st June 2005: JASON
OBS: JASON 2005
EuroSIP
WNP
ENP
Atl

35. Tropical storm seasonal forecast for JASON 2006 ECMWF forecast issued in May

36. Tropical storm seasonal forecast for JASON 2006 ECMWF forecast issued in June

37. Summary
Integration of VarEPS and monthly forecast: a step towards a “seamless prediction” system.
Improvements in physical parametrizations (to be included in cy 31R1) reduce systematic errors and improve tropical intra-seasonal variability in monthly and seasonal forecasts.
New ocean (re-)analysis beneficial to both seasonal forecasting and climate research
Multi-model predictions moving from experimental (DEMETER, ENSEMBLES) to operational phase (EUROSIP).