Gettelman: November 2006 An Introduction to Climate Modeling Andrew Gettelman National Center for Atmospheric Research Boulder, Colorado USA Assistance.

Gettelman: November 2006

An Introduction to An Introduction to Climate ModelingClimate Modeling

Andrew GettelmanAndrew GettelmanNational Center for Atmospheric Research National Center for Atmospheric Research

Boulder, Colorado USABoulder, Colorado USA

Assistance from: Assistance from: J. J. Hack (NCAR)J. J. Hack (NCAR)


Climate Modeling

A. Gettelman& J. HackReal NCAR Scientists

Science, Statistics,

Parameterization, Results

It’s all in here!

Simulate the future on your desktop


OutlineOutline

• What is ClimateWhat is Climate– Why is climate different from weather and forecasting

• Hierarchy of atmospheric modeling strategiesHierarchy of atmospheric modeling strategies– Focus on 3D General Circulation models (GCMs)

• Conceptual Framework for General Circulation Conceptual Framework for General Circulation ModelsModels

• Parameterization of physical processesParameterization of physical processes– concept of resolvable and unresolvable scales of motion

– approaches rooted in budgets of conserved variables

• Model Validation and Model SolutionsModel Validation and Model Solutions


Question 1: What is Climate?Question 1: What is Climate?

A.A. Average/Expected ‘Weather’Average/Expected ‘Weather’

B.B. The temperature & precipitation rangeThe temperature & precipitation range

C.C. Distribution of all possible weatherDistribution of all possible weather

D.D. Record of Extreme eventsRecord of Extreme events


Climate changeand its

manifestation in terms of weather

(climate extremes)

(1) What is Climate?


Climate changeand its manifestation in terms of weather(climate extremes)


Climate changeand its manifestation in terms of weather(climate extremes)


Impacts of Climate ChangeImpacts of Climate ChangeMote et al 2005

Observed Change 1950-1997Observed Change 1950-1997SnowpackSnowpack TemperatureTemperature

(- +)

(- +)


Observed Temperature Records

IPCC, 3rd Assessment, Summary For Policymakers


‘‘Anthropogenic’ ChangesAnthropogenic’ Changes

Rad

iativ

e F

orci

ng (

Wm

-2)

1000 1200 1400 1600 1800 2000


‘‘Anthropogenic’ Changes (2)Anthropogenic’ Changes (2) 560ppmv 560ppmv COCO22

~2060~2060


Question 2Question 2

• What is the difference between What is the difference between Numerical Weather Prediction and Numerical Weather Prediction and Climate prediction?Climate prediction?


Climate v. Numerical Weather Climate v. Numerical Weather PredictionPrediction• NWP: NWP:

– Initial state is CRITICAL– Don’t really care about whole PDF, just probable phase space

– Non-conservation of mass/energy to match observed state

• ClimateClimate– Get rid of any dependence on initial state– Conservation of mass & energy critical– Want to know the PDF of all possible states– Don’t really care where we are on the PDF– Really want to know tails (extreme events)


Question 3Question 3

How can we predict Climate (50 How can we predict Climate (50 yrs)yrs)

if we can’t predict Weather if we can’t predict Weather (10 days)?(10 days)?Statistics!Statistics!


Conceptual Framework for Conceptual Framework for ModelingModeling

• Can’t resolve all scales, so have to represent Can’t resolve all scales, so have to represent themthem

• Energy Balance / Reduced ModelsEnergy Balance / Reduced Models– Mean State of the System– Energy Budget, conservation, Radiative transfer

• Dynamical ModelsDynamical Models– Finite element representation of system– Fluid Dynamics on a rotating sphere– Basic equations of motion– Advection of mass, trace species– Physical Parameterizations for moving energy

• Scales: Cloud Resolving/Mesoscale/Regional/GlobalScales: Cloud Resolving/Mesoscale/Regional/Global– Global= General Circulation Models (GCM’s)


Physical processes regulating Physical processes regulating climateclimate


2000 2005

Earth System Model ‘Evolution’Earth System Model ‘Evolution’


Modeling the Atmospheric General Modeling the Atmospheric General CirculationCirculation

Requires Requires understanding of :understanding of :– atmospheric predictability/basic fluid dynamics– physics/dynamics of phase change– radiative transfer (aerosols, chemical constituents, etc.)

– interactions between the atmosphere and ocean (El Nino, etc.)

– solar physics (solar-terrestrial interactions, solar dynamics, etc.)

– impacts of anthropogenic and other biological activity

Basic Process: Basic Process: – iterate finite element versions of dynamics on a rotating sphere

– Incorporate representation of physical processes


Meteorological Primitive Meteorological Primitive EquationsEquations

• Applicable to wide scale of motions; > Applicable to wide scale of motions; > 1hour, >100km1hour, >100km


Global Climate Model PhysicsGlobal Climate Model Physics

Terms Terms F, Q,F, Q, and and SSq q represent physical processesrepresent physical processes

• Equations of motion, Equations of motion, FF– turbulent transport, generation, and dissipation of momentum

• Thermodynamic energy equation, Thermodynamic energy equation, QQ– convective-scale transport of heat– convective-scale sources/sinks of heat (phase change)– radiative sources/sinks of heat

• Water vapor mass continuity equationWater vapor mass continuity equation– convective-scale transport of water substance– convective-scale water sources/sinks (phase change)


Grid DiscretizationsGrid Discretizations

Equations are distributed on a sphereEquations are distributed on a sphere

• Different grid approaches: Different grid approaches: – Rectilinear (lat-lon)– Reduced grids– ‘equal area grids’: icosahedral, cubed sphere– Spectral transforms

• Different numerical methods for solution:Different numerical methods for solution:– Spectral Transforms– Finite element– Lagrangian (semi-lagrangian)

• Vertical DiscretizationVertical Discretization– Terrain following (sigma)– Pressure– Isentropic– Hybrid Sigma-pressure (most common)


Model Physical ParameterizationsModel Physical ParameterizationsPhysical processes breakdown:Physical processes breakdown:

• Moist ProcessesMoist Processes– Moist convection, shallow convection, large scale condensation

• Radiation and CloudsRadiation and Clouds– Cloud parameterization, radiation

• Surface FluxesSurface Fluxes– Fluxes from land, ocean and sea ice (from data or models)

• Turbulent mixingTurbulent mixing– Planetary boundary layer parameterization, vertical diffusion, gravity wave drag


Basic Logic in a GCM (Time-Basic Logic in a GCM (Time-step Loop)step Loop)

For a grid of atmospheric columns:For a grid of atmospheric columns:

1.1. ‘‘Dynamics’: Iterate Basic Equations Dynamics’: Iterate Basic Equations Horizontal momentum, Thermodynamic energy, Mass conservation, Hydrostatic equilibrium, Water vapor mass conservation

2.2. Transport ‘constituents’ (water vapor, Transport ‘constituents’ (water vapor, aerosol, etc)aerosol, etc)

3.3. Calculate forcing terms (“Physics”) for Calculate forcing terms (“Physics”) for each columneach columnClouds & Precipitation, Radiation, etc

4.4. Update dynamics fields with physics Update dynamics fields with physics forcingsforcings

5.5. Gravity Waves, Diffusion (fastest last)Gravity Waves, Diffusion (fastest last)

6.6. Next time step (repeat)Next time step (repeat)


Physical ParameterizationPhysical Parameterization

• Physical parameterizationPhysical parameterization– express unresolved physical processes in terms of

resolved processes– generally empirical techniques

• Examples of parameterized physicsExamples of parameterized physics– dry and moist convection– cloud amount/cloud optical properties– radiative transfer– planetary boundary layer transports– surface energy exchanges– horizontal and vertical dissipation processes– ...

To close the governing equations, it is necessary to To close the governing equations, it is necessary to incorporate the effects of physical processes that incorporate the effects of physical processes that occur on scales below the numerical truncation limitoccur on scales below the numerical truncation limit


Q

F

F SqSq


Atmospheric Energy TransportAtmospheric Energy TransportSynoptic-scale mechanisms

• hurricanes • extratropical storms

http://www.earth.nasa.gov


Process Models and ParameterizationProcess Models and Parameterization

•Boundary Layer•Clouds

StratiformConvective

•Microphysics


RadiationRadiation


Other Energy Budget Impacts From Other Energy Budget Impacts From CloudsClouds



Energy Budget Impacts of Energy Budget Impacts of Atmospheric AerosolAtmospheric Aerosol



Scales of Atmospheric Scales of Atmospheric Motions/ProcessesMotions/Processes

Anthes et al. (1975)

Resolved Scales

Global ModelsGlobal Models

Future Global ModelsFuture Global Models

Cloud/Mesoscale/Turbulence ModelsCloud/Mesoscale/Turbulence Models

Cloud DropsCloud DropsMicrophysicsMicrophysicsCHEMISTRYCHEMISTRY


Global Modeling and Horizontal Global Modeling and Horizontal ResolutionResolution


Examples of Global Model Examples of Global Model ResolutionResolution

Typical Climate Application Next Generation Climate Applications

~300km 50-100km


High Resolution Art Global Model SimulationHigh Resolution Art Global Model Simulation

100km x 100km Global Model Precipitation

NCAR CCM3 run on Earth Simulator, Japan


Key Uncertainties for Climate Key Uncertainties for Climate (1):(1):

1. Low Clouds over the ocean: Reflect Sunlight (cool) : Dominant EffectTrap heat (warm)

More Clouds=Cooling Fewer Clouds=Warming


Marine Stratus: Low Clouds over the OceanMarine Stratus: Low Clouds over the Ocean


Parameterization of CloudsParameterization of Clouds

Weare and Mokhov (1995)

Cloud amount (fraction) as simulated by 25 Cloud amount (fraction) as simulated by 25 atmospheric GCMsatmospheric GCMs


Low Clouds Over the OceanLow Clouds Over the Ocean

Change in low cloud with 2xCO2

2 Models: Changesare OPPOSITE!



2. High Clouds: Dominant effect is that they Trap heat (warm)

More Clouds=Warming Fewer Clouds=Cooling



3. Water Vapor: largest greenhouse gasIncreasing Temp=Increasing water Vapor (more greenhouse)Effect is expected to ‘amplify’ warming through a ‘feedback’

1D Radiative-Convective Model:Higher humidity=>warmer surface


SummarySummary

• Global Climate ModelingGlobal Climate Modeling– complex and evolving scientific problem – parameterization of physical processes pacing progress

– observational limitations pacing process understanding

• Parameterization of physical processesParameterization of physical processes– opportunities to explore alternative formulations

– exploit higher-order statistical relationships?

– exploration of scale interactions using modeling and observation– high-resolution process modeling to supplement observations– e.g., identify optimal truncation strategies for capturing major scale interactions

– better characterize statistical relationships between resolved and unresolved scales


How can we evaluate simulation How can we evaluate simulation quality?quality?

• Compare long term mean climatologyCompare long term mean climatology– average mass, energy, and momentum balances– tells you where the physical approximations take you– but you don’t necessarily know how you get there!

• Consider dominant modes of variabilityConsider dominant modes of variability– provides the opportunity to evaluate climate sensitivity– response of the climate system to a specific forcing factor

– exploit natural forcing factors to test model response– diurnal and seasonal cycles, El Niño Southern Oscillation (ENSO), solar variability


Comparison of Mean Simulation Comparison of Mean Simulation Properties 1Properties 1

ObservedPrecipitation

SimulatedPrecipitation


Comparison of Mean Simulation Comparison of Mean Simulation Properties 1 Properties 1

Difference:Sim- Observed

SimulatedPrecipitation



ObservedLand Temp

SimulatedLand Temp



SimulatedLand Temp

Difference:Sim- Observed


Testing AGCM SensitivityTesting AGCM Sensitivity

Cloud (OLR) Anomalies and ENSOCloud (OLR) Anomalies and ENSO

Hack (1998)

Observed

Simulated

More Cloud Less Cloud


Turning The Crank: ResultsTurning The Crank: Results

• Simulations of Atmospheric Model Simulations of Atmospheric Model Coupled to OceanCoupled to Ocean

• Present Day ClimatePresent Day Climate• Simulations into the future with Simulations into the future with ‘Scenarios’‘Scenarios’

• Different Models=Different Different Models=Different ‘Sensitivity’‘Sensitivity’

• Potential Changes in Temp, PrecipPotential Changes in Temp, Precip


Kicking the System: Radiative Kicking the System: Radiative ForcingForcing


Observations: 20th Century Warming Model Solutions with Human Forcing


Surface Temperature Variations 1000-Surface Temperature Variations 1000-21002100


CCSM Past: Last Millennium to CCSM Past: Last Millennium to 21002100


Atmospheric CO2 (input) Temperature (output)

CCSM Future: Next 100+ yearsCCSM Future: Next 100+ years


CMIP 2001: Temperature and CMIP 2001: Temperature and PrecipitationPrecipitation

Covey et al. (2001)


Impacts of Climate ChangeImpacts of Climate ChangeMote et al 2005

Observed Change 1950-1997Observed Change 1950-1997

SnowpackSnowpack TemperatureTemperature

(- +)

(- +)


The FutureThe Future

Regardless of Scale: Still need parameterizations for most things

Resolved Scales

Global ModelsGlobal Models

Future Global ModelsFuture Global Models

Cloud/Mesoscale/Turbulence ModelsCloud/Mesoscale/Turbulence Models

Goal: get interactions right (Mesoscale). Also extreme events


Example of State of the Art Global Model Example of State of the Art Global Model SimulationSimulation

10 X 10 km Global Model Precipitation

NEIS AGCM for the Earth Simulator, Japan


Example of State of the Art Global Model Example of State of the Art Global Model SimulationSimulation

10 X 10 km Global Model Precipitation: Mid Latitude Cyclone over Japan


‘‘Nested’ Models inside a GCMNested’ Models inside a GCM

Another Approach: Nested Modeling (GCM forces Cloud or Mesoscale Model)NCAR NRCM: Outgoing Longwave Radiation, Jan1: 36km

QuickTime™ and aPNG decompressor

are needed to see this picture.

Recall Scales: Still need parameterizations for most things (Radiation, Convection, Microphysics).

Goal is to do small scale interactions better


The End

Gettelman: November 2006 An Introduction to Climate Modeling Andrew Gettelman National Center for Atmospheric Research Boulder, Colorado USA Assistance.

Documents

climate slide

climate prediction

climate different

desktop slide

hack ncar slide

observed change

energy scales

possible weather d