POAMA: Bureau of Meteorology Coupled Model Seasonal Forecasting System

1
POAMA Bureau of Meteorology Coupled Model
Seasonal Forecasting System
Oscar Alves Bureau of Meteorology Research
Centre o.alves_at_bom.gov.au www.bom.gov.au/bmrc/oce
an/JAFOOS/POAMA
2
Acknowledgements
BMRC coupled modelling Guomin Wang, Neville
Smith, Aihong Zhong, Andrew Marshall, Faina
Tseitkin BoM others Harry Hendon
BMRC Climate Forecasting Graham Warren BoM
operations BMRC Model development group CSIRO
Marine Research Andreas Schiller, Stuart Godfrey,
Gary Meyers, Peter McIntosh, Russ Fieddler
3
Plan

• POAMA coupled model system
• Hind-cast performance
• Some issues
• Summary/Future

4
POAMAPredictive Ocean Atmosphere Model for
Australia

• Global coupled model GCM seasonal forecasting
  system
• Joint project between BMRC and CSIRO Marine
  Research
• Partly funded by the Climate Variability in
  Agriculture Program (CVAP)
• Run in real-time by BoM operational section
• Operational products issued by the BoM National
  Climate Centre (NCC)
• Experimental products available on the POAMA web
  site
• www.bom.gov.au/bmrc/ocean/JAFOOS/POAMA

5
POAMA System
Ocean Model ACOM2 (developed by CSIRO
Marine Research)
Atmosphere Model BAM (Bureau Unified Atmosphere
Model) T47L17
Ocean initial conditions Optimum Interpolation
Geostrophic Currents every 3 days
Atmos. initial conditions From BoM operational
NWP model GASP
Operational set up Real time using latest ocean
obs and atmos initial conditions. Knows about
the very latest state of the real world.
Ensembles Operationally eight month forecast
every day
6
POAMA Operational System
T-10 days
Today
Main ocean assimilation (GASP forcing, GTS T obs)
Catch-up ocean assimilation
Coupled forecast
GASP atmospheric/land IC
7
POAMA Web-site
www.bom.gov.au/bmrc/ocean/JAFOOS/POAMA
8
Experimental results pages Will also include
ocean analyses
9
Samples
10
Hind-casts

• One forecast per season, 1987-2001 (60 cases)
• Did not use NWP atmosphere initial state, used
  AMIP
• Otherwise the same as operational version
• OI ocean assimilation every 3 days

11
Skill of SST Predictions
Hind-casts one forecast per season, 1987-2001
(60 cases)
rms error (solid) Standard deviation (dashed)
Anomaly correlation
Green - model, red - anomaly persistence
12
2 months
Anomaly Correlation
4 months
6 months
13
Nino 3 Predictions
14
Key issues
15
SST Drift
16
Sub-surface drift
17
Measuring skill

• Nino 3 skill eg ACC a very gross measure
• 15 years of data too short
• More serious problem changing ocean observing
  network
• Key challenge for future
• Option
• Look for scientific quality of the model
• Eg. Madden-Julian Oscillation growing consensus
  that this is important for triggering at least
  some El Ninos

18
Madden-Julian Oscillation in Forecasts (60 cases)
Frequency/wave number surface zonal wind power
spectra
Older version
NCEP re-analysis
Latest Version
30 days
60 days
MJO
Improvements to convection closure by BAM group
19
Seasonality OLR MJO variance SH summer AMIP
style run
SH Summer
NH Summer
CDC
Model
20
SST anomaly along equatorForecast starting 1st
June 1997
Obs
Old Moisture closure
New CAPE Closure
21
Zonal wind anomaly along equator
Obs
Moisture closure
CAPE Closure
22
Dynamic imbalance
23
Forecast starting Aug 1996
Model anomaly from 90-95 models
Aug96 - Apr 97
24
Surface Current anomaly
Surface wind anomaly
20C isotherm depth anomaly
Westerly wind burst in Nov/Dec
Rossby Wave
Kelvin wave

• What causes model initial Rossby and Kelvin wave
  - initial shock ? Dynamical imbalance in IC?
• What role does Rossby wave play reflection ?
• Model - warming triggered my shock or random WWB?
  Nature by MJO ?

25
Impact of current increments during assimilation
Apply correction to currents based on currents to
temperature field based on Burgers et al (2002)
26
Zonal current at 165E 0N
OBS
No Assim
Assim
Assim Geo Adj
27
Assimilation Statistics in NINO4 How well does
model background fit observations
1993
Red Assim Green Assim Geo Adj
28
Zonal current at 140W 0N
OBS
No Assim
Assim
Assim Geo Adj
29
Problem in east (Nino 3) ?
Background - obs error Red - assim Green assim
geo adj
30
Thermocline depth - 1993 Hovmuller
Problem in east (Nino 3) ?
Assim
Assimgeo adj
Correct dynamical balance during assimilation key
issue Geo adj just one step in the right
direction
31
Atmos IC
32
Role of Atmospheric Initial ConditionsSST
anomaly along equator
Obs
AMIP IC
NWP IC
33
Zonal wind anomaly along equator
Obs
AMIP IC
NWP IC
34
20C isotherm depth anomaly along equator
Obs
AMIP IC
NWP IC
35
Instability waves
36
Tropical instability waves Collaboration with
CMR/CAR
37
Summary

• POAMA (version 1.0)
• Latest versions of BAM and ACOM2
• Operationally 8 month forecast per day using very
  latest ocean/atmos. initial conditions
• Nino 3 skill significantly better than
  persistence at all lead times - internationally
  competitive
• Good simulation of intra-seasonal variability
• Web site - details of forecasts and hind-casts
  (dynamical diagnostics)

38
Summary Key issues for future

• Local climate precip./temperature e.g. over
  Australia
• Direct from model coupled/2tier
• Statistical/dynamical downscaling
• Improve
• BAM
• ACOM2 (with CSIRO)
• Ocean data assimilation
• atmospheric initialisation
• Land surface
• Understand modes of variability in model ( real
  world) - link between intra-seasonal and
  inter-annual
• Ensembles - How to generate and how to use ?
• Measuring skill ???