An Introduction to Coupled Models of the Atmosphere Ocean System Jonathon S. Wright jswright@tsinghua.edu.cn
Atmosphere Ocean Coupling 1. Important to climate on a wide range of time scales Diurnal to seasonal: coastal climates Interannual to decadal: El Niño and other oscillations Long time scales: deep ocean uptake of heat and carbon dioxide 2. Strongest in the tropics, where the circulation is thermally direct Tropical wind stress is controlled by SST, leading to a strong positive wind thermocline SST feedback In mid-latitudes, much weaker relationship between SST and surface wind stress 3. Many key uncertainties and challenges remain
Outline Part 1 1. A hierarchy of coupled atmosphere ocean models 2. Constructing coupled energy balance models 3. Constructing coupled general circulation models Part 2 4. Simple models of coupled atmosphere ocean variability 5. Can general circulation models simulate this variability?
Part 1: Coupled Atmosphere Ocean Models
A Hierarchy of Models 3-dimensional 2-dimensional (, ', z) (, '); (', z); (, z) 1-dimensional 0-dimensional ( ); ('); (z) global mean
A Hierarchy of Models 3-dimensional 2-dimensional more complex (, ', z) (, '); (', z); (, z) 1-dimensional 0-dimensional less complex ( ); ('); (z) global mean
A Hierarchy of Models 3-dimensional 2-dimensional less parameterization (, ', z) (, '); (', z); (, z) 1-dimensional 0-dimensional more parameterization ( ); ('); (z) global mean
The Simplest Climate Model 0-dimensional T s = 4 r Q(1 ) 4 Q 4 (1 ) T 4 s surface energy balance model
Adding an Atmosphere 1-dimensional T s = 4 r Q(1 ) 4 + T 4 e T 4 e atmosphere T 4 e Q 4 (1 ) T 4 s surface
Adding an Atmosphere 1-dimensional T s = 4 r 2Q(1 ) 4 T 4 e = Q 4 (1 ) atmosphere T 4 e Q 4 (1 ) T 4 s surface
Adding an Atmosphere 1-dimensional T s = 4 r 3Q(1 ) 4 T 4 e T 4 e upper atmosphere 2 T 4 e T 4 e Q 4 (1 ) T 4 s lower atmosphere 2 T 4 e surface
Adding an Atmosphere 1-dimensional T s = 4 r (n + 1)Q(1 ) 4 T 4 e 1 2 3 4 n-1 Q 4 (1 ) T 4 s n T 4 e n surface
Adding an Ocean 1-dimensional =(1 c) T 4 thermal capacity of the ocean D µ dt dt = S µ = c p D OLR(T,c) absorbed solar radiation absorption by the atmosphere
Adding an Ocean 1-dimensional =(1 c) T 4 D µ = c p D radiative forcing µ dt dt = S OLR(T,c) equilibrium T still balances S and OLR
Adding an Ocean 1-dimensional D deviation from radiative forcing equilibrium: F = S OLR T 0 = T T eq µ dt 0 1 dt = T 0 + F = @OLR @T = 4(1 c) T 3
Response to Climate Forcing
Response to Climate Forcing T 0 (t) =T 0 (0)e ( t/ ) = µ
Simulating Climate Variability full climate model one-box ocean model Held et al., J. Climate 2010
A Layered Ocean 1-dimensional µ 1 dt 0 1 dt = T 0 1 apple(t 0 1 T 0 2)+ F µ 2 dt 0 2 dt = apple(t 0 1 T 0 2) diffusion to / from the deep ocean
Response to Climate Forcing
Response to Climate Forcing fast response
Response to Climate Forcing slow response fast response
Response to Climate Forcing global warming climate projection abrupt return to pre-industrial Held et al., J. Climate 2010
Response to Climate Forcing global warming climate projection abrupt return to pre-industrial Held et al., J. Climate 2010
A Layered Ocean 1-dimensional can be modified to study CO2 uptake by the ocean
From Global to Zonal Mean
Meridional Energy Transport 1-dimensional c p @T(') @t = S(') OLR(')+F(') S OLR ocean south pole F ' north pole
Meridional Energy Transport 2-dimensional c p @T(') @t = S(') OLR(')+F(') atmosphere ocean south pole ' north pole
Meridional Energy Transport 2-dimensional c p @T(') @t = S(') OLR(')+F(') atmosphere ocean south pole ' north pole
Coupled Energy Balance Models 1. 0- to 2-dimensional (global mean to latitude height) 2. The atmosphere... Determines the radiation balance Contributes to horizontal energy transport 3. The ocean... Provides thermal inertia and/or CO2 storage, stabilizing the climate Contributes to horizontal energy transport 4. Models are suitable for studying climate sensitivity over a wide range of parameters and over long time scales. 5. Can supplement fully coupled model simulations
conservation of energy conservation of momentum conservation of water & salt ATMOSPHERE thermodynamic s of motion water conservation radiation turbulence clouds
conservation of energy conservation of momentum conservation of water & salt ATMOSPHERE thermodynamic s of motion water conservation radiation turbulence clouds thermodynamic s of motion turbulence salt conservation sea ice OCEAN
conservation of energy conservation of momentum conservation of water & salt ATMOSPHERE thermodynamic s of motion water conservation radiation turbulence clouds sensible heat radiation friction wind stress evaporation precipitation thermodynamic s of motion turbulence salt conservation sea ice OCEAN
Different Requirements ATMOSPHERE requires high temporal resolution requires high spatial resolution OCEAN
Simplify One, Simulate the Other ATMOSPHERE thermodynamic s of motion water conservation radiation turbulence clouds swamp ocean: infinite source of water vapor OCEAN
Simplify One, Simulate the Other ATMOSPHERE thermodynamic s of motion water conservation radiation turbulence clouds slab ocean: horizontally diffusive mixed layer stores heat and supplies water vapor OCEAN
Simplify One, Simulate the Other ATMOSPHERE thermodynamic s of motion water conservation radiation turbulence clouds dynamical model of the surface layer OCEAN
Surface Layer Dynamics Precipitation Solar Radiation Wind Stress Sea Spray Wave Breaking Evaporation Ekman Currents Langmuir Circulation Wave Current Interactions Penetrating Solar Radiation Mixed Layer Depth Internal Wave Radiation Wells, 2012
Surface Layer Dynamics Precipitation Solar Radiation Wind Stress Sea Spray Wave Breaking Evaporation Ekman Currents Langmuir Circulation Wave Current Interactions Penetrating Solar Radiation Mixed Layer Depth Horizontal Heat Transport Internal Wave Radiation Wells, 2012
Simplify One, Simulate the Other ATMOSPHERE stochastic atmosphere thermodynamic s of motion turbulence salt conservation sea ice OCEAN
conservation of energy conservation of momentum conservation of water & salt ATMOSPHERE thermodynamic s of motion water conservation radiation turbulence clouds thermodynamic s of motion turbulence salt conservation sea ice OCEAN
conservation of energy conservation of momentum conservation of water & salt ATMOSPHERE thermodynamic s of motion water conservation radiation turbulence clouds earliest fully-coupled models: alternating time steps thermodynamic s of motion turbulence salt conservation sea ice OCEAN
conservation of energy conservation of momentum conservation of water & salt ATMOSPHERE thermodynamic s of motion water conservation radiation turbulence clouds aquaplanet: no lateral boundary conditions thermodynamic s of motion turbulence salt conservation sea ice OCEAN
The Coupling increasing pressure ATMOSPHERE x ~ 100 300km fluxes of heat, momentum, and water at the surface ~10m sea ice ~10m x ~ 30 100km increasing depth OCEAN
conservation of energy conservation of momentum conservation of water & salt ATMOSPHERE thermodynamic s of motion water conservation radiation turbulence clouds sensible heat radiation friction wind stress evaporation precipitation thermodynamic s of motion turbulence salt conservation sea ice OCEAN
conservation of energy conservation of momentum conservation of water & salt ATMOSPHERE thermodynamic s of motion water conservation radiation turbulence clouds sensible heat radiation friction wind stress evaporation coupling shock followed by gradual equilibration precipitation thermodynamic s of motion turbulence salt conservation sea ice OCEAN
conservation of energy conservation of momentum conservation of water & salt ATMOSPHERE thermodynamic s of motion water conservation radiation turbulence clouds sensible heat radiation friction wind stress evaporation coupling shock followed climate by drift gradual equilibration precipitation thermodynamic s of motion turbulence salt conservation sea ice OCEAN
Climate Drift 1. Can complicate studies of climate change signal 2. Careful initialization crucial for coupled models Run each component model several times Observationally constrain variables at the interface 3. Empirical flux corrections Calibration of coupled model with surface variables (temperature, salinity, momentum, etc.) constrained to observed climatologies Apply calculated corrections as artificial fluxes during coupled simulations to prevent drift away from a realistic climate state Requires very long (~1000 yr) initialization runs of the ocean component Gradually being replaced by direct flux coupling techniques
conservation of energy conservation of momentum conservation of water & salt ATMOSPHERE thermodynamic s of motion water conservation radiation turbulence clouds sensible heat radiation friction wind stress evaporation precipitation thermodynamic s of motion turbulence salt conservation sea ice OCEAN
conservation of energy conservation of momentum conservation of water & salt ATMOSPHERE thermodynamic s of motion water conservation can be applied radiation turbulence clouds regionally as well as sensible heat radiation friction wind stress globally. evaporation precipitation thermodynamic s of motion turbulence salt conservation sea ice OCEAN
Coupled Model Intercomparisons 1. Study the response of coupled atmosphere ocean models to idealized climate forcings 2. Main sources of model spread include Cloud processes and interactions with radiation Cryospheric processes (sea and land ice) Deep ocean processes (i.e., the slow response) Atmosphere ocean interactions 3. Often supplemented by Monte Carlo-type simulations using a single model framework
Simulation of Surface Temperature contours: observations shading: error in multimodel mean typical model error IPCC AR4
Simulation of Precipitation observations multi-model mean IPCC AR4
Simulation of Precipitation ITCZ + SPCZ observations multi-model mean double ITCZ IPCC AR4
Simulation of Sea Ice observed extent number of models (out of 14) simulating at least 15% ice cover IPCC AR4
Coupled AOGCMS 1. Essential for full representation of the climate system 2. Coupling is a major technical challenge AGCMs have different requirements than OGCMs Climate drift due to the coupling can distort the magnitude of climate feedbacks and responses to climate forcings 3. Coupled models are improving rapidly Increasing computational power enables finer resolutions Improvement in parameterizations of atmosphere ocean interactions and introduction of new coupling techniques Conversion of flux correction techniques to direct flux coupling Still substantial inter-model spread and differences relative to observations
Part 1: Modeling Coupled Atmosphere Ocean Variability: El Niño Southern Oscillation
What is ENSO? under normal conditions, convection is centered in the western Pacific noaa.gov
What is ENSO? under normal conditions, convection is centered in the western Pacific under El Niño conditions, convection shifts eastward to the central Pacific noaa.gov
What is ENSO? under normal conditions, convection is centered in the western Pacific under La Niña conditions, convection shifts even further toward the west noaa.gov
How SST Changes heat from atmosphere ocean mixed layer upwelling of cold water
Why SST Controls Precipitation warm SST ocean mixed layer
Why SST Controls Precipitation low SLP warm SST ocean mixed layer
Why SST Controls Precipitation low SLP warm SST ocean mixed layer
Why SST Controls Precipitation low SLP warm SST ocean mixed layer
Changes in the Thermocline ocean mixed layer deep ocean
Changes in the Thermocline climatological wind direction ocean mixed layer deep ocean
Changes in the Thermocline climatological wind direction ocean mixed layer deep ocean
Changes in the Thermocline the density of seawater in the mixed layer is less than the density in the deep ocean, so the bottom depth increases more than the height of the sea surface. climatological wind direction ocean mixed layer deep ocean
Changes in the Thermocline regional eastward wind anomaly ocean mixed layer deep ocean
Changes in the Thermocline regional eastward wind anomaly ocean mixed layer deep ocean
Changes in the Thermocline Kelvin waves communicate the anomaly in the depth of the thermocline across the entire ocean basin. regional eastward wind anomaly ocean mixed layer deep ocean
Bjerknes Feedback 1. Winds flow from low SST to high SST... 2....leading to a shallower thermocline under low SST and a deeper thermocline under high SST... 3....leading to cooling in the region of low SST and warming in the region of high SST... 4....reinforcing and strengthening the winds... zonal wind stress in equatorial Pacific changes in sea surface temperature + changes in depth of thermocline
Describing ENSO Sea surface temperature anomalies in various regions serve as indices noaa.gov
El Niño Southern Oscillation ENSO varies on interannual timescales, with a period of 2 7 years. noaa.gov
Typical Effects of El Niño on Winter Climate Typical Effects of La Niña on Winter Climate noaa.gov
Seasonal Climate Forecasts
Seasonal Climate Forecasts
Seasonal Climate Forecasts Even with current understanding, ENSO predictions are highly uncertain
What Is ENSO? 1. An unstable nonlinear oscillator? Delayed oscillator: changes in thermocline are out of phase with changes in wind stress. Recharge oscillator: a warm event (El Niño) leaves the equatorial thermocline shallower and the sea surface colder than normal (La Niña). The reservoir of warm water is then refilled over time. 2. A stable system with non-normality? Small disturbances grow and then decay 3. A combination of these two?
Simple Models 1. Coupled shallow-water atmosphere and ocean on an equatorial ß-plane Gill-type atmospheric model (quasi-geostrophic with a simple thermal forcing) Ocean model assumes a well-mixed surface layer, no mean currents, and a deep ocean at rest (a one and a half layer model). 2. Results Coupling of the tropical atmosphere and ocean can produce unstable coupled modes with interannual periods (Hirst, 1986) Propagating signals on the equatorial thermocline are an important part of the ENSO response (Wakata and Sarachik, 1991) The propagating signals strongly depend on the shape of the thermocline and the distribution of the upwelling Can achieve unstable coupled modes under constant (annual mean) conditions
The Zebiak Cane Model An anomaly model the climatological annual cycle is specified for both atmosphere & ocean! modified Gill-type shallow water model atmosphere surface winds respond to SST wind-driven convergence and divergence surface layer linear reduced-gravity model upper layer thermocline depth responds to wind stress; determines temperature of entrained water u = v = w =0 deep ocean
Simulated SST Anomalies noaa.gov Zebiak and Cane, Mon. Wea. Rev., 1987
Development of El Niño 1 2 December March 3 4 June December Zebiak and Cane, Mon. Wea. Rev., 1987
Development of El Niño Zonal wind stress anomaly Thermocline depth anomaly Zebiak and Cane, Mon. Wea. Rev., 1987
Development of El Niño Zonal wind stress anomaly Thermocline depth anomaly El Niño Zebiak and Cane, Mon. Wea. Rev., 1987
Development of El Niño La Niña Zonal wind stress anomaly Thermocline depth anomaly El Niño Zebiak and Cane, Mon. Wea. Rev., 1987
The Zebiak Cane Model 1. ENSO is an oscillation of the coupled atmosphere ocean system 2. All of the necessary interactions take place in the tropical Pacific 3. The rapid response of the surface layer to the atmosphere is crucial 4. The basin-wide response down to the thermocline is the core of the interannual variability 5. ENSO is a combination of a positive (Bjerknes) feedback and the basin-scale dynamic response
Simulation of the Tropical Pacific simulations of SST Sun et al., J. Climate, 2006
Simulation of the Tropical Pacific model constrained by observations (reanalysis) older models newer models Bellenger et al., Clim. Dyn., submitted
Simulation of SST Variability standard deviation of SST Guilyardi et al., BAMS, 2009
Simulation of ENSO Amplitude preindustrial 2xCO2 Guilyardi et al., BAMS, 2009
Simulation of ENSO Seasonality older models observations newer models month Bellenger et al., Clim. Dyn., submitted
Simulation of ENSO Bellenger et al., Clim. Dyn., submitted
Modeling ENSO 1. Simple coupled models can produce unstable modes 2. Surface layer dynamics play a key-role in generating ENSO variability, which extends across the tropical Pacific to the depth of the thermocline 3. The exact ENSO mechanisms are still uncertain, and ENSO is difficult to predict at seasonal timescales 4. Coupled models still have difficulty simulating ENSO Problems remain not only in simulations of coupled variability, but even in simulations of the mean climate of the tropical Pacific The CMIP5 model ensemble offers some improvement relative to the CMIP3 model ensemble