Model Systems at MPI-M

Marco Giorgetta

Content

What is a model in climate research as used here for IPCC?

What is a model?Components of the climate systemClimate models – What is inside?Climate models – Resources

Models available at the MPI-MModels used for IPCC simulations

Summary

What is a model?

A model is:an idealized representation or abstraction of an object with the purpose to demonstrate the most relevant or selected aspects of the object or to make the object accessible to studies

Purpose of modelsTo reduce the complexityTo avoid details that are not relevant for a specific considerationTo obtain a theoretically or practically manageable system

What is a model?Example 1: Architecture

http://www.werk-plan.de/

Representation of a building at the scales of a shoe boxWood+acryl instead of concrete, glas, etc.Overview of the object and its relation to the environmentNo details

What is a model?Example 2: Fashion

http://www.stern.de/lifestyle/mode/

The ideal lady to present a dress in the spirit of the designerTrue scaleFlawless/perfect

What is a model?Example 3: Climate model

Climate model are a mathematical abstraction of the observed real worldClimate models use quantitative methods to simulate the interactions of the atmosphere, oceans, land surface, and ice = the Earth system.They follow theoretical principles and observed relationshipsModel = simplified image representing the relevant features http://www.solarviews.com/cap/earth/

Components of the climate systemAtmosphere

DynamicsPhysicsChemistryAerosols

Ocean

DynamicsPhysicsBiogeochem.

Land

HydrologyVegetation

Society Economics Land use

Climate models and Earth system modelsExample 1

0-dimensional model of the radiative equilibrium of the Earth

(1 - a)S πr2 = 4πr2 σT4 π = 3.14159 Piσ = 5.67⋅10-8 JK-4m-2s-1 Stefan-Boltzmann constant S = 1367 Wm-2 incoming solar radiationa = 0.37 to 0.39 fraction reflected back to spacer = 6371 km Earth radius

emitted thermal

radiation

absorbed solar

radiation

→ T= ~ –26ºC = effective emission temperature of Earth→ 35ºC colder than the observed average surface temperature

Problem: Greenhouse gas effect is neglected

Climate models and Earth system models Example 2

3-dimensional comprehensive general circulation models or Earth System Models

GCM's discretize the equations for fluid motion and integrate these forward in time.

They also contain parametrisations for processes - such as convection - that occur on scales too small to be resolved directly.

Earth system models represent the pinnacle of complexity in climate models and internalise as many processes as possible, including chemistry in the atmosphere, marine biogeochemistry, land vegetation etc.

Limits: computer power

Climate models and Earth system models 1. System of continuous equations

Equations describing the evolution of a set of variablesdescribing the state of the modeled systemScale analysis include the “relevant” processes, neglect others

Equation for temperature in the atmosphere,as used for the Hamburg atmospheric GCM (ECHAM5):

Cannot be solved directly on a computer

Climate models and Earth system models 2. System of discretized equations

Equation must be discretized in time and spaceto be accessible to computational solutionsDiscretized equations

time time stepsHamburg atmosphere model for IPCC: 15 minutesHamburg ocean model for IPCC: 1 day

Space horizontal grids or spectra, vertical gridsProblems of too low resolution:

Inaccurate numerical solution of equationsMissing details in the description of the surface: Mountain ranges

Horizontal grid of the MPI ocean model[Every 5th grid line shown]

“North pole”

“South pole”

Resolution matters1. Spatial resolution in ECHAM5 and errors

Dimensionless error in ECHAM5Horizontal res.: T21 to T159 Vertical res.: L19 to L31 Reference: ERA-15 re-analysis.

Error of T21L19 = 100Error of T63L31 = 50

Error of ERA-40 = 20

IPCC, carbonT63L31

Aerosol T63L19

T21: 5.6°×5.6°T63: 1.9°×1.9°T159: 0.75°×0.75 °

L19: 19 levels, 30km L31: 31 levels, 30km

RMS error of seasonal average patterns of T, Z, U at 200, 500, 850 hPa and SLP compared to Era-15

Resolution matters2. Precipitation REMO 1/2 ° (1979REMO 1/2 ° (1979--93)93)

Annual precipitation in the Alps in observations and in the REMO regional model

ObservationsObservations (1971(1971--90)90) REMO 1/6 ° (1979REMO 1/6 ° (1979--88)88)

source: ETH, source: ETH, ZürichZürich

General circulation models3. Parameterizations

Processes on unresolved scales have effects on resolved scalesThese effects must be described as functions of the resolved fields

ParameterizationsUnresolved dynamics

TurbulenceConvectionGravity waves in the atmosphere

Processes at microphysical or molecular levelRadiationClouds and precipitation in the atmospherePhotosynthesis in plants or plankton…

General circulation models4. Translation to a computer program

Climate models have grown over time with the increasing knowledge on details of processes and the improved power of computers High performance computing requires additional complex structures in the codes

Example: MPI-M climate model for IPCCca. 140000 lines of code = 1700 printed pages

Errors / Bugs : unavoidableLarge bugs are easily detectableMinor bugs are not obvious, hence not easily found

General circulation models5. Verification of Climate models

Careful testing is necessaryTo justify the choice of equationsTo understand effects of the discretization and selected resolutionTo decide if the model is fit for specific applications, e.g. IPCC

Validation of models in standard tests for which acceptable references are known from observations or proxy data

Tests of individual components: atmosphere, ocean, landTests of the coupled system, which may drift to unrealistic statesExample of a coupled phenomenon: El Nino/La Nina

Validation is based on the observed climate of the pastCredibility of simulations for the future (IPCC) is based on the “success” of modeling the climate of the past

General circulation models6. Climate simulations and resources

Resources consumed by IPCC simulations:Simulated years for all IPCC related experiments computed at the DKRZ: 3500 yearsConsumed computer time: 400000 CPU hours =

550 CPU months =~12 months on ¼ DKRZ SX-6

Generated raw data: 400 TBData in IPCC data bank: 80 TB

(c.f. hard disk in a PC: 200 GB)

Substantial effort in work and money

Models available at the MPI-M

Atmosphere GCM

Dynamics global: ECHAM5, regional: REMO+PhysicsAerosols HAM(M7)

Ocean+Ice GCM

Dynamics MPI-OM+PhysicsBiogeochem. HAMOCC/DMS

Land model

Hydrology HDVegetation JSBACH

The IPCC model

MomentumEnergy

H2O

Sun/Spaceconst. Irrad.

Energy AtmosphereECHAM5 T63 L31

OceanMPIOM 1.5°L40

LandECHAM5/HD

IPCCA1B, B1, A2

GHG conc.SO4 conc.

PRISM

The aerosol model

MomentumEnergy

H2O

Sun/Space11y cycle

Energy

AtmosphereECHAM5 T63 L19

HAM

OceanMPIOM 1.5°L40HAMOCC+DMS

LandECHAM5/HD

IPCC,NIESA1B

GHG conc.SO2 em.BC em.OC em.

PRISM

DMS

dust

Volcanoes

The carbon cycle model

MomentumEnergy

H2O

Sun/Space

Energy

AtmosphereECHAM5 T63 L31

OceanMPIOM 1.5°L40

HAMOCC

LandHD, JSACH

IPCCA1B

CO2 em.CH4, N2O conc.

SO4 conc.

PRISM

The regional model for IPCC

Atmosphere+LandREMO

SunGlobal data from IPCC model simulations

Global MPI-M model systems for IPCC

IPCC model Most presentations

Aerosol model Johann Feichter

Carbon cycle model Christian Reick

Daniela JacobRegional model

ConclusionsClimate models are the only tool to explore the climate systematically

Models as used for IPCC are comprehensive but still simplifications of the real world

Their success in the simulation of the climate of the well observed past is the basis for simulations into the future.

3 global models + 1 regional model have been used at the MPI-M:Physical system IPCC computations IPCC data baseAerosol systemCarbon cycle system

The End