240 likes | 413 Views
Coupled GCM. The Challenges of linking the atmosphere and ocean circulation. Brief History of LRF. Statistical and Analog - earliest Simple models and Teleconnections Coupled Models with dynamic and statistical components Dynamic Coupled Atmosphere-Ocean Models. Grid Spacing
E N D
Coupled GCM The Challenges of linking the atmosphere and ocean circulation
Brief History of LRF • Statistical and Analog - earliest • Simple models and Teleconnections • Coupled Models with dynamic and statistical components • Dynamic Coupled Atmosphere-Ocean Models
Grid Spacing dependent on coordinate system for globe dependent on computer space Time Step dependent on resolution and length of forecst Terrain and Ocean Mapping generally rough with little detail Parameterize solar radiation convection heat flux wind stress GCM Model Matters
Oceans Sea Surface Temperature evaporation Mixed Layer heat flux/transport Annual cycles upwelling Pacific circulation Atmosphere Solar Energy sun angle cloud cover Wind Stress mixing layer estimate profile Heat & Moisture Transport shallow and deep convection The Interface
Coupled GCM’s Focus • Tropical Oceans - Pacific • Initial Conditions • Atmosphere • inferred from spotty observations and detailed satellite analysis • Ocean • uses a data set developed by Florida State University which shows climatology of temperature and wind stress
Center for Ocean-Land-Atmosphere Std Geophysical Fluid Dynamics Lab (GFDL) NASA-Lamont Doherty (Columbia Univ) Scripps Institute UCLA NCEP Max Plank Institute Bureau of Meteorology Research Centre CGCM’s - Many Models
The COLA’s Model • The Ocean Portion • Adapted from GFDL - for Pacific Domain from 30S-45N &130E-80W • Resolution: x=1.5 y=.5 (20S-20N) 1.5 degrees elsewhere • 20 vertical levels to 4000m - 1-16 are within the top 40m • non-linear vertical mixing of heat, salinity and momentum
The COLA’s Model • The Atmosphere Portion • Global Spectral Model with 30 wave limit • 18 layers on a sigma coordinate • Solar radiation is parameterized • Deep convection - modified Kuo • Shallow convection - Tiedtke • Complex scheme for exchange of heat, moisture and momentum
Coupling Strategy • Several Methods • Interpolated Exchange • Anomaly Coupling • Mixed Methods • Significance of Ocean-Atmosphere Exchange is especially important in the Tropics
Coupling Strategy • Interpolated Exchange • Daily mean values are exchanged • OGCM produces SST for Atmosphere • AGCM produces surface heat flux, momentum and freshwater (rainfall) for the Oceans • These values are parameterized and interpolated for grid points in each model
Coupling Strategy • Anomaly Coupling • Each part of the model predicts and anomaly component compared with a set model climatology. • Atmosphere climatology - 45 years (1949-94) • Ocean climatology - 30 years (1964-94)
Coupling Strategy • Start with Atmosphere (AGCM) predicts solar-radiation to estimate SST for Ocean • SST is used to predict a wind profile in the tropical boundary layer - the anomaly component of this profile is used for adding to the wind stress on the ocean.
Coupled GCM from COLA - now uses anomaly of initial conditions from an in-house ocean data assimilation analysis Coupled GCM from COLA using interpolated values from AGCM and OCM Hybrid Coupled Ocean-Atmos Model - Scripps-Max Plank Experimental Long Lead Models
2004 Model Forecast • CPC/EMC • GFDL Ocean • MRF reduced • Ensemble-16 • updated wkly • http://www.emc.ncep.noaa.gov/cmb/sst-forecasts/
2004 Model Forecasts • Scripps • Plank • Hybrid • 30S-30N • 13 vertical • AGM - Stat • mainly wind stress
2004 Model Forecasts • Japan Meteo Agency • AGCM (T42/40 levels) • OGCM (T 20 levels) • 2.5 x 2.0 • Flux Exchanges every 24 hrs for mean values
2004 Model Forecast • LDEO Model - • wind stress • Focus on initialization • Ensemble of 3 wind stresses • FSU,NCEP,QUIKSCAT
2004 Model Forecast • Markov Model of SST - CPC • Linear Statistics trained 1980-95 • Verified by 1964-1979
2004 Model Forecast • LIM (Linear Inverse Model) from CIRES/CDC - Boulder • Uses a specific Stat function (Green)
2004 Model Forecast • Constructed Analog (Van Den Dool) • Uses past anomalies as predictors
2004 Model Forecast • IRI Summary • All Models; Statistical & Dynamic
Long Lead Predictions • Summary of 2004 Model Forecasts
Forecast of SST in Tropical Pacific with a Markov Model - NCEP (linear statistical) Tropical SST’s using a Linear Inverse Model- CIRES - Boulder Tropical Pacific SST using and intermediate ocean and statistical atmosphere model - Earth Environmental Studies - Seoul Long Lead Predictions
Further Readings • http://grads.iges.org/ellfb/contents.htm • - (updated every 3 months)