ASP ACM Colloquium 4 June 2006 An Introduction to CCSM William D. Collins National Center for Atmospheric Research Boulder, Colorado USA

ASP ACM Colloquium 4 June 2006 The CCSM Program Scientific Objectives: Develop a comprehensive climate model to study the Earth’s Climate. Investigate seasonal and interannual variability in the climate. Explore the history of Earth’s climate. Estimate the future of the environment for policy formulation. Recent Accomplishments: Release of a new version (CCSM3) to the climate community. Studies linking SST fluctuations, droughts, and extratropical variability. Simulations of last 1000 years, Holocene, and Last Glacial Maximum. Creation of largest ensemble of simulations for the IPCC AR4.

ASP ACM Colloquium 4 June 2006 Evolution of CCSM CSM 1.0 CCSM 2.0 New ocean, land, sea-ice models New physics in atmosphere June 1996 May 2002 CCSM 3.0 New physics in all models June 2004 J. Climate, 1998 Kiehl & Gent, 2004 J. Climate, 2006 IJHPCA, 2005

ASP ACM Colloquium 4 June 2006 Configuration of NCAR CCSM3 Atmosphere (CAM 3.0) T85 (1.4 o ) Ocean (POP 1.4.3) (  1 o ) Coupler (CPL 6) Sea Ice (CSIM 4) (  1 o ) Land (CLM2.2) T85 (1.4 o ) Forcings: Greenhouse gases Manmade aerosols Volcanic eruptions Solar variability

ASP ACM Colloquium 4 June 2006 The CCSM Community Atmosphere (CAM 3) Ocean ( POP 1) Coupler (CPL 6) Sea Ice (CSIM 4) Land (CLM 3) CCSM3 Model UsersClimate Community Development Group NCARUniversitiesLabs PhysicsApplicationsChemistry Current Users: Institutions: ~200 Downloads of CCSM3: Users:~600 Publications: NCAR: 87 Universities: 94 Labs/Foreign: 48 Total:229

ASP ACM Colloquium 4 June 2006 Climate Simulations for the IPCC AR4 (IPCC = Intergovernmental Panel on Climate Change) IPCC Emissions Scenarios Climate Change Simulations IPCC 4 th Assessment Results: 10,000 simulated years Largest submission to IPCC 100 TB of model output 2007

ASP ACM Colloquium 4 June 2006 CCSM Process Flow (slide courtesy of CSEG/NCAR)

ASP ACM Colloquium 4 June 2006 Computational Characteristics of CCSM3 CCSM3 has been climate-validated and run on: –IBM Power systems –SGI Altix and Origin systems –Cray X1/X1E vector and XT3 scalar systems –NEC SX vector systems (the Earth Simulator) –Xeon and Itanium linux systems Computational requirements for CCSM3 on IBM Power 4s: –T31 land/atm  3 o ocean/ice: 62CPU hrs/sim. year –T42 land/atm  1 o ocean/ice: 292CPU hrs/sim. year –T85 land/atm  1 o ocean/ice:1146CPU hrs/sim. year T85x1 T42x1 FV 2  2.5 T31x3 …. Dynamics Resolution Assessment Model Working Model Test Model

ASP ACM Colloquium 4 June 2006 Each component has different performance characteristics: –Atmosphere: 3D computational grid, currently latitude/longitude/vertical tensor product grid 1D horizontal domain decomposition for spectral dynamics or 2D tensor product horizontal and 2D latitude/vertical domain decomposition for finite volume dynamics 2D arbitrary horizontal domain decomposition for physics –Ocean: 3D grid using displaced pole, locally orthogonal horizontal grid; 2D horizontal decomposition using Cartesian blocks. –Land: 2D horizontal grid (same as Atm.); 2D arbitrary horizontal domain decomposition –Sea-Ice: 2D horizontal grid (same as Ocean); 2D horizontal decomposition using Cartesian blocks At current resolutions, either the atmosphere or ocean component model (depending on the science investigation) is the performance bottleneck. CCSM Grid Characteristics

ASP ACM Colloquium 4 June 2006 Momentum equation: dV/dt = -  p -2  ^V –gk +F +D m Where  =1/  (  is density), p is pressure,  is rotation rate of the Earth, g is acceleration due to gravity (including effects of rotation), k is a unit vector in the vertical, F is friction and D m is vertical diffusion of momentum Thermodynamic equation: dT/dt = Q/c p + (RT/p)  + D H where c p is the specific heat at constant pressure, R is the gas constant,  is the vertical velocity, D H is the vertical diffusion of heat and Q is the internal heating from radiation and condensation/evaporation; Q = Q rad + Q con Continuity equation for moisture (similar for other tracers): dq/dt = E – C + D q where E is the evaporation, C is the condensation and D q is the vertical diffusion of moisture Basic Equations for the Atmosphere Slingo

ASP ACM Colloquium 4 June 2006 Horizontal Discretization of Equations T31T42 T85T170 Strand

ASP ACM Colloquium 4 June 2006 Properties of the Atmosphere Dynamics: –Previous: Spectral Eulerian –Current: Finite Volume –Future: Spectral element, discontinuous Galerkin, … ? Physics: –Deep and shallow convection –Prognostic condensate and precipitation parameterizations –Diagnostic parameterization of cloud fraction –Shortwave and longwave band-model radiative transfer –Prognostic and diagnostic aerosols –Surface exchange –Vertical diffusion and boundary layer processes

ASP ACM Colloquium 4 June 2006 The current version includes: Biogeophysics Hydrology River routing The next version will include: Natural and human-mediated changes in land cover Natural and human-mediated changes in ecosystem functions Coupling to biogeogeochemistry Drainage Hydrology Canopy Water Evaporation Interception Snow Melt Sublimation Throughfall Stemflow Infiltration Surface Runoff Evaporation Transpiration Precipitation Soil Water Redistribution Direct Solar Radiation Absorbed Solar Radiation Diffuse Solar Radiation Longwave Radiation Reflected Solar Radiation Emitted Long- wave Radiation Sensible Heat Flux Latent Heat Flux uaua 0 Momentum Flux Wind Speed Soil Heat Flux Heat Transfer Photosynthesis Biogeophysics Processes Included in the CCSM Land Model

ASP ACM Colloquium 4 June 2006 CLM Subgrid Structure Gridcell GlacierWetlandLake Landunits Columns PFTs UrbanVegetated Soil Type 1

ASP ACM Colloquium 4 June 2006 Radiatively Active Species Forcing20 th Century21 st -23 rd Century Greenhouse GasesObservedSRES Ozone Trop: MOZART Strat: Solomon Trop:MOZART scaled by O 3 TAR forcing Strat: Solomon Sulfate AerosolsSO 2 : Smith/WigleySO 2 : SRES Carbon AerosolsPopulation ScalingSO 2 Scaling Sea-salt & DustYear 2000 values Volcanic AerosolsAmmann (2003)Year 2000 values Solar VariationLean (1995)Year 2000 values Indirect EffectsNone

ASP ACM Colloquium 4 June 2006 Equilibrium Sensitivity from CAM3 + SOM T85 2 x CO 2 T85 1 x CO 2 T42 2 x CO 2 T42 1 x CO 2 T31 2 x CO 2 T31 1 x CO 2 Kiehl and Shields

ASP ACM Colloquium 4 June 2006 Climate Sensitivity of CCSM3 Kiehl et al, 2006

ASP ACM Colloquium 4 June 2006 Fidelity of the 20 th Century Simulations Criteria for the ocean/atmosphere system –Realistic prediction of sea-surface temperature given realistic forcing –Realistic estimates of ocean heat uptake Effects of ocean on transient climate response –Realism of ocean mixed layer and ventilation Ocean uptake of CO 2 and passive tracers (CFCs)

ASP ACM Colloquium 4 June 2006 Simulation of Sea-Surface Temperature

ASP ACM Colloquium 4 June 2006 Increases in Global Ocean Temperatures (Results from CCSM3 Ensemble) Gent et al, 2005 L = Levitus et al (2005) Ensemble Members Relative Model Error < 25%

ASP ACM Colloquium 4 June 2006 Simulation of Ocean Heat Variability Gent et al, < Range < Levitus0.3 = Range = Levitus 0.1 = Range < Levitus

ASP ACM Colloquium 4 June 2006 Global Ocean Inventory of CFC-11 (Passive tracer proxy for CO 2 ) Gent et al, 2005 Ensemble Members } Data CCSM3

ASP ACM Colloquium 4 June 2006 Regional Ocean Inventory of CFC-11 Moles/km 2 WOCE data (Willey et al, 2004) Gent et al, 2005 CCSM3 Simulation

ASP ACM Colloquium 4 June 2006 Penetration Depth of CFC-11 WOCE data (Willey et al, 2004) Gent et al, 2005 CCSM3 Simulation m

ASP ACM Colloquium 4 June 2006 Vertical Distribution of CFC-11 in the Atlantic Data from 20 o W, August 1993 CCSM3 Ocean-Only Simulation Data CCSM3 Gent et al, 2005

ASP ACM Colloquium 4 June 2006 Some major issues in CCSM Simulations Temperature and Precipitation –Biases in mid-lat. continental temp. and precipitation –SSTs in coastal stratus regions –Semi-annual cycle in SST for the E. Pacific –Polar temperature bias and tropical tropopause biases –Double ITCZ in the Pacific Representation of major modes of variability –El Nino / Southern Oscillation –The Madden-Julian oscillation Underestimation of surface insolation in polar regions Surface stress in the storm tracks

ASP ACM Colloquium 4 June 2006 Continental Temperature and Precipitation Collins et al, 2005

ASP ACM Colloquium 4 June 2006 Simulation of Sea-Surface Temperature Collins et al, 2005

ASP ACM Colloquium 4 June 2006 Polar Stratospheric Temperatures Collins et al, 2005

ASP ACM Colloquium 4 June 2006 Double Pacific ITCZ Collins et al, 2005

ASP ACM Colloquium 4 June 2006 ENSO Variance (120W-170W, 5N-5S) Collins et al, 2005

ASP ACM Colloquium 4 June 2006 ENSO Power Spectra Gent and Kiehl, 2004; Collins et al, 2005

ASP ACM Colloquium 4 June 2006 Madden-Julian Oscillation Propagation Collins et al, 2005

ASP ACM Colloquium 4 June 2006 Surface Insolation in Arctic Regions Collins et al, 2005

ASP ACM Colloquium 4 June 2006 Surface Stress on Southern Oceans Collins et al, 2005

ASP ACM Colloquium 4 June 2006 Scientific objectives for the near future Major objective: Develop, characterize, and understand the most realistic and comprehensive model of the observed climate system possible. Subsidiary objectives: –Analyze and reduce the principal biases in our physical climate simulations using state-of-the-art theory and observations. –Simulate the observed climate record with as much fidelity as possible. –Simulate the interaction of chemistry, biogeochemistry, and climate with a focus on climate forcing and feedbacks.

ASP ACM Colloquium 4 June 2006 Recent evolution of climate forcing Hansen and Sato, 2001

ASP ACM Colloquium 4 June 2006 Simulating the chemical state of the climate system Emissions Chemistry + BGC + Physics + Ecosystems Chemical Reservoirs ConcentrationsForcing Climate Response Feedbacks In the past, we have generally used offline models to predict concentrations and read these into CCSM. This approach is simple to implement, but  It cuts the feedback loops.  It eliminates the chemical reservoirs. The next CCSM will include these interactions. Offline models Feedbacks Chemical Reservoirs

ASP ACM Colloquium 4 June 2006 CCSM4: a 1 st generation Earth System Model Atmosphere Ocean Coupler Sea IceLand C/N Cycle Dyn. Veg. Ecosystem & BGC Gas chem. Prognostic Aerosols Upper Atm. Land Use Ice Sheets

ASP ACM Colloquium 4 June 2006 Computational Design of CCSM4 Application Driver CLMCICEPOP CAM Phys Dyn Couplers ATM mergerOCN mergerLND mergerICE merger Thin coupling layer Thin coupling layer Thin coupling layer Thin coupling layer

ASP ACM Colloquium 4 June 2006 Challenges: Linking Weather-Climate Interactions Coupling regional weather and global climate models: Testing climate models as weather models: Linking air quality and pollution with global change: Regional and Global Models Satellite Data Field Programs Simulations to Plan field programs Operational forecasts For field programs In situ data Synthesis / Analysis Model Evaluation Tie and Madronich, 2005 CO Simulation for MIRAGE Experiment Tests of model physics Using short-range prediction Development of regional coupled ocean/atmosphere models for 2-way nesting in CCSM.

ASP ACM Colloquium 4 June 2006 Challenges: International Climate Assessments The 5th IPCC Assessment: It is likely that the AR5 report will be issued 6 years after AR4, in Following the precedent in AR4, the simulations will have to be finished two to three years ahead, in 2010 to Therefore CCSM4 has to be ready for testing in 2008 and production in The big challenges: Make sense of the new feedbacks in Earth System Models. Make sense of the “meaning” of multi-model ensembles of simulations. Make the assessments relevant for urban and regional-level policy makers AR5 ProcessCCSM4 DevelopmentCCSM4 Release

ASP ACM Colloquium 4 June 2006 Food for thought CCSM is developed through a community process. We welcome your interest and involvement in the development and exploitation of the model. There are funds available to support visits by you and your faculty advisor to NCAR -- please do so! Some of our greatest challenges are in front of us.

ASP ACM Colloquium 4 June : Lewis Fry Richardson –Basic equations and methodology of numerical weather prediction 1950: Charney, Fjørtoft and von Neumann (1950) –First numerical weather forecast (barotropic vorticity equation model) 1956: Norman Phillips –1 st general circulation experiment (two-layer, quasi-geostrophic hemispheric model) 1963: Smagorinsky, Manabe and collaborators at GFDL, USA –Nine level primitive equation model 1960s and 1970s: Other groups and their offshoots began work –University of California Los Angeles (UCLA), National Center for Atmospheric Research (NCAR, Boulder, Colorado) and UK Meteorological Office 1990s: Atmospheric Model Intercomparison Project (AMIP) –Results from about 30 atmospheric models from around the world 2001: IPCC Third Assessment Report –Climate projections to 2100 from 9 coupled ocean-atmosphere-cryosphere models. 2007: IPCC Fourth Assessment Report –Climate projections to 2300 from 15 coupled models. Brief History of Climate Modeling Slingo

ASP ACM Colloquium 4 June 2006 Improvements in Atmospheric Fidelity D. Williamson, in Collins et al, 2005 Unconditional error Conditional error Phase error Scaled Variance Ratio

ASP ACM Colloquium 4 June 2006 Ensemble Members Trends in Ocean Temperature: Upper 300m (Results from CCSM3 Ensemble) Gent et al, 2005 }

ASP ACM Colloquium 4 June 2006 Trends in Ocean Temperature: Upper 3km (Results from CCSM3 Ensemble) Gent et al, 2005 Ensemble Members }

ASP ACM Colloquium 4 June 2006 Increases in Regional Ocean Temperatures (1996 – 1957, CCSM3 Ensemble) Gent et al, 2005