Hindcast Skill in the new coupled NCEP Ocean-Atmosphere Model
Suranjana Saha, Wanqiu Wang, Hua-Lu Pan
and the NCEP/EMC Climate and Weather Modeling Branch
Environmental Modeling Center, NCEP/NWS/NOAASpecial Acknowledgements : Sudhir Nadiga, Jiande Wang, Qin Zhang, Shrinivas Moorthi, Huug van den Dool
E
M
C
Introduction
A new global coupled atmosphere-ocean model has recently been developed at NCEP/EMC.
Componentsa) the T62/64-layer version of the current NCEP atmospheric GFS (Global Forecast System) model and
b) the 40-level GFDL Modular Ocean Model (version 3)
Note:Direct coupling with no flux correction
This model will replace the current operational NCEP coupled model (CMP14) for SST prediction in 2004.
Hindcast Skill Assessment
• 5-member ensemble over 22 years from 1981-2002
• January and April initial conditions• Other months to follow
• 9 month runs
• Initial atmospheric states 0000 GMT 19, 20, 21, 22, and 23 for each month• Reanalysis-2 archive
.• Initial ocean states NCEP GODAS (Global Ocean Data
Assimilation System) 0000 GMT 21st of each month • Same for all runs• GODAS operational September 2003
Hindcast Skill Assessment (cont)
• So far 220 runs have been made
• Hindcast skill • Estimated after doing a bias correction for each year• Uses model climatology based on the other years
• Anomaly correlation skill score for Nino 3.4 region SST prediction
• Skill maps • Global SST • U.S. temperature and precipitation.
• Comparisons with CMP14 and CASST
CoupledRed: monthly bias
Observed
Composite Warm and Cold Events
• Events exceed ERSST variance by
• 1.0 SD (warm)
• 0.75 SD (cold)
• Heavy black line is mean
+36 moPeak- 36 mo
Red: coupled model
SST Climatology on Equator
Ensemble Mean
CMP14CASST
April IC
January IC
CASST
CMP14
Ensemble Mean
RMS ErrorApril
RMS ErrorJanuary
6 Month Lead (November)
from April IC
SST anomaly for 1981-2002
Note Amplitudes
Observed
Observed
6 Month Lead (August)
from January IC
SST anomaly for 1981-2002
Note Amplitudes
HindcastSeasonally (3 month) Averaged
SST Anomaly CorrelationApril IC
Note: large & persistent skill in tropics
HindcastMonthly Averaged
SST Anomaly CorrelationApril IC
June-September
Left: New Coupled SystemRight: CMP14
HindcastMonthly Averaged
SST Anomaly CorrelationApril IC
October-January
Left: New Coupled SystemRight: CMP14
HindcastSeasonally (3 month) Averaged
SST Anomaly CorrelationJanuary IC
Note: large & persistent skill in tropics
HindcastSeasonally Averaged
SST Anomaly CorrelationJanuary IC
Left: New Coupled SystemRight: CMP14
Hindcast3 month Averaged
U. S. Surface Temperature Anomaly Correlation
April IC
Note: areas of persistent skill > 60% at up to 6 month lead
U. S. Surface TemperatureHindcast Skill
3 Month AveragesApril IC
Comparison with CPCCCA Method
Note: Coupled System skillcomplementary to CCA
U. S. Surface TemperatureHindcast Skill
3 Month AveragesJanuary IC
Comparison with CPCCCA Method
Note: Coupled System skillcomplementary to CCA
Hindcast3 month AveragedU. S. Precipitation
Anomaly CorrelationApril IC
Note: areas of persistent skill > 60% at up to 6 month lead
U. S. PrecipitationHindcast Skill
3 Month AveragesApril IC
Comparison with CPCCCA Method
Note: Coupled System skillcomplementary to CCA
U. S. PrecipitationHindcast Skill
3 Month AveragesJanuary IC
Comparison with CPCCCA Method
Note: Coupled System skillcomplementary to CCA
MJO Forecasts (W. Wang)
Experiments damp: GFS03 with damped SST anomalies clim: GFS03 with climatological SSTs amip: GFS03 with observed SSTs coup: CFS03 with MOM3 ocean analysis
All forecasts to 45 daysComposite results
Phase 1
Phase 2
Phase 4
Phase 3
(Initiation)
(Max pos. ampl. Over IO)
(Max pos. ampl.Over WPAC)
(Decay)
Note: coupling necessary for propagation in Phases 1-3
Days 1-30
Coupled
AMIP
Climo
Damped
Observed
SST Expt.
Summary and Conclusions• CFS03 hindcast skill for January and April initial
conditions (1981-2002 ) have been evaluated
• For April, the SST AC skill over Nino 3.4 is better than CMP14 and CASST at all leads
• For January, the SST AC skill over Nino-3.4 is better than CMP14 and CASST for all leads, except lead 2
• Ensemble mean forecasts for U.S. temperature and precipitation show comparable skill to CPC’s CCA method.
• This skill is complementary to CCA as it manifests itself in different geographical areas and can be used in CPC’s operational seasonal consolidated forecast.
• Hindcasts for the rest of the calendar months are being performed
• Implementation is being planned for late 2004
Summary and Conclusions (cont)
New Climate Positions at NCEP/EMC
• UCAR Visiting Scientist Position at NCEP/EMC – Work with NCEP Coupled Model– http://www.earthworks-jobs.com/climate/ucar3101.htm
l• NCEP Climate Team Leader (GS-15)
– Coordinate development activities with community– Provide strategic guidance on NCEP’s Climate
Numerical Modeling activities– Participate actively in development activities with
EMC staff