Climate Modeling
Why build a climate model
Atmosphere H2O vapor and Clouds Absorbing gases CO2 Aerosol Land/Biota Surface vegetation Ice Sea ice Ice sheets (glaciers) Ocean
Box Model (0 D) E IN = E OUT The Energy Balance Problems we were looking at could be simulated in a 0D climate model. 1-layer atmosphere (0 D) Surface multi-layer atm (1 D) Surface 2 D
Complete 4D model Coordinates: Latitude Longitude Height Time
Z T Lat Long
Typical values for current climate models 5 variables (minimum) 8190 boxes 26 vertical levels 30 minute time step (48/day) 51.1 million equations / day 18.6 billion equations / year 300 km x 300 km
Resolution Spectral models instead of discrete box calculation (finite difference) these calculate a wave function Resolution is determined by truncation T of the wave function how waves are included. T42 128x64 = 2.8 x 2.8 deg Current: T63 192x96 = 1.9x1.9 deg T85 256x128 = 1.4 x 1.4 deg T106 320 x 160 = 1.1x1.1deg Processing increases 5-6 fold for each increase in resolution.
Land or water? Surface vegetation type? Elevation?
In numerical models, the approximate statistical representation, using a combination of theory and observations, of the relationship between a sub-grid scale process (clouds) and the large-scale computed model fields (temperature, humidity, winds).
Grid resolution ~ 300 km x 300 km Prognostic variables: T, q, u, v, w (ftns of p)
Satellite Satellite Reasons: - What is a cloud? - Resolution - Parameterization NCAR Model
Initial condition s Prescribed forcing (Sun, CO2, etc.) Atmospheric Global Climate Model Run forward in time for at least 10 to 30 years Prescribed sea surface temperatures Compare averaged model results with averaged climate results
Initial condition s Prescribed forcing (Sun, CO2, etc.) Atmosphere Coupled Global Climate Model Run forward in time for at least 100 years Ocean Compare averaged model results with averaged climate results
How do we use models Validation does the model simulate observations reasonably compare to past observations Responses Transient response change boundary condition inputs and see how model conditions evolve in time Equilibrium response run boundary condition inputs until climate is stable (typically ~1000 model year runs)
observed Model Validation With amplitude of the seasonal cycle modeled
Model hindcast - validate models with past 2 1.5 1 0.5 0-0.5-1 -1.5-2
Models used for future prediction 2040-2060 minus 1980-2000 temperature change (A1B Scenario) 2 1.5 1 0.5 0-0.5-1 -1.5 Transient response climate is not in equilibrium with forcing How long does it take for the climate to come to an equilibrium? -2
Fig 6-5 now
How models help us learn about the climate system -> can isolate variables in the climate system Model without water vapor feedback Model with water vapor feedback observed
Ensemble Runs average together results of many runs
Transient versus Equilibrium warming Transient warming is smaller Transient warming is asymmetric across hemispheres Transient warming is modest in the northern North Atlantic
Model comparison: Equilibrium warming from 2XCO2 What is equilibrium temperature response of 2XCO2? Used to compare models ΔT EQ ranges from 1.5-4.5 C Gives an idea of the climate sensitivity of the model The range is awfully large (factor of three!) Hasn t narrowed in 30 years Are predictions even useful for policy-making purposes?
Why do models disagree? At Equilibrium Clouds, clouds, clouds - either positive or negative feedback depending on height and thickness Magnitude of Ice-albedo feedback In Transient run Ocean retains heat (ocean heat uptake) What else? Emissions scenario including aerosols
IPCC Intergovernmental Panel on Climate Change Formed by UNEP and WMO to help synthesize understanding of climate changes and climate science Fourth Assessment Report (AR4) 2007 IPCC and Al Gore awarded Nobel Prize Increasingly stronger language attributing observed climate change signal to human activities. First Assessment Report (FAR) 1991 Second Assessment Report (SAR) 1997 Third Assessment Report (TAR) 2001 Scenarios lay out a range of projected changes in greenhouse gases, population, adoption of technology, etc. A1B middle of the road; BAU Business as Usual
Robust patterns similar in most models Example: precipitation at equator
Sometimes all the models are missing something robust result but missing a key underlying physical mechanism?
2007 4.3x10^6
Summary Climate Modeling Models based on fundamental physical first principles Conservation of energy (Ein = Eout) Conservation of mass Conservation of momentum Allow investigation of complicated interactions and feedbacks Limited by resolution (computer processing) Parameterization represent sub-gridscale processes that are impossible to model Inherent uncertainties in feedbacks
Summary Climate Modeling Validation with observations Transient (short term response 10-30 yrs) vs Equilibrium response (climate becomes stable) Sensitivity experiments allows comparison among models - ie 2xCO2 Models differ in sensitivity to different feedbacks Many results are robust features in most climate simulations
Summary Climate Modeling GCMs do a very good job of simulating Large scale behavior of the climate system Seasonal cycle Ocean-atmosphere coupling GCMs do less well in simulating Hydrologic cycle (clouds and precipitation) Climate variability GCMs do a poor job of simulating Regional climate variations