CCSM3s IPCC Simulations and Lessons Learned

1
CCSM3s IPCC Simulations and Lessons Learned

• Bill Collins
• http//www.cgd.ucar.edu/wcollins
• National Center for Atmospheric Research
• Boulder, Colorado

• IPCC Production and Data Transfer
• Scientific results from IPCC Simulations
• Lessons Learned

2
Contributions from CCSM to IPCC
IPCC Emissions Scenarios
CCSM3 Simulations
IPCC 4th Assessment

• Results
• 10,000 simulated years
• Largest submission to IPCC
• 100 TB of model output

3
The Experimental Configuration(from IPCC TAR)
A2
A1B
B1
4
Production Schedule of IPCC Runs
5
IPCC Ensembles on US Machines
http//www.cgd.ucar.edu/ccr/ipcc_status/
6
US Simulations for IPCC FAR
7
Computational Resources for Simulations

• Execution speed 3.5 years/day on 192 IBM
  Power-4 CPUs
• Time required to simulate 5370 years5670 / 3.5
  1620 wall-clock days
• CPU time for 1 simulated year1146 CPU hours
• Equivalent number of CPU hours for IPCC
  runs5670 ? 1146 6.5 ? 106 CPU hours

8
IPCC Ensembles on the Earth Simulator
http//www.cgd.ucar.edu/oce/weese/status/ccsm_data
.html
All data has been transferred to NCAR mass store.
9
Statistics of the IPCC Ensembles

• Ensemble size
• 20th Century 8
• A1B Scenario 8
• B1 Scenario 8
• Commitment 8
• A2 Scenario 5
• Total integration 10,800 years
• Data volume 110 TB

10
Resolutions for all Modeling Groups
Karl Taylor, PCMDI
11
Model Simulations by all Groups
Karl Taylor, PCMDI
12
Transfer of the Data to PCMDI

• First 1-TB disk transferred to PCMDI in October.
  Three Disks are now in use.
• CCSM IPCC data is now on the PCMDI IPCC website
  http//www-pcmdi.llnl.gov/
• Estimated scheduling of data transfers
• Details http//swiki.ucar.edu/cgd-ccr/48
• Sequence
• 20thC
• CAM monthly
• CAM daily/6hr
• CLM monthly
• Controls
• CAM monthly
• CAM daily/6hr
• CLM monthly
• Scenarios
• CAM monthly
• CLM monthly
• POP and CSIM monthly to follow.

13
Papers on CCSM IPCC Scenario Runs

• Meehl et al, 2005 How much more global warming
  and sea level rise? In press, Science
• Meehl et al, 2005Climate change in the 20th and
  21st centuries andclimate change commitment in
  the CCSM3 Submitted to J. Climate
• Bryan et al, 2005Response of the North Atlantic
  Thermohaline Circulation and Ventilation to
  Increasing Carbon Dioxide in CCSM3 Submitted to
  J. Climate
• Kyosei Consortium, 2005Analysis of CO2
  overshoot experiments

14
Global Temperature and Sea Level
15
Ocean Temperature and Overturning
16
Increases in Surface Temperature
17
Increases in Precipitation
18
Arctic Sea Ice Concentrations
19
Changes in Sea Ice Coverage
20
Lesson 1 Experimental Design Plan

• Contents goals of the experiments
• Description of all forcing agents vs. time
• Description of pre-industrial spin-up
• Pre-industrial configuration tuned for energy
  balance?
• Simulation for year 2000 integrated backwards to
  19th C.?
• Experimental basis for pre-industrial spin-up
• Description and plan for CCSM resource
  requirements
• CCSM software engineering support
• Development efforts by working groups and WG
  liaisons
• Plan and timeline for dissemination of
  experiments
• Distribution within the CCSM community
• Distribution to wider climate community

21
Lesson 2 Development of Forcings
22
Aerosol Optical Depths during 20th C.
Sulfates
Black Organic Carbon
Strat. Volcanics
23
Present-day Aerosol Optical Depths
24
Comparison of CCSM3 and Global Surface
Temperatures

• Eight-member ensemble
• T85 atmosphere land
• 1-degree ocean sea-ice

Meehl et al, 2004
25
Recommendations on Forcings

• Coordination with other groups
• For example, Bouchers shared SO4 data set
• Efforts for AR4 to develop common forcings
• Develop components of state-of-art forcings
• Emissions as functions of space and time
• Chemical transport framework to simulate agents
• Parameterization development for indirect effects
• Exploration of indirect forcings, for example
• Effects of irrigation
• Effects of land-use change

26
Lesson 3 Design of Model Spin-up
27
Effects of Energy Balance on Ocean Temperature
in 1870 Control
Trend ?0.17 K/Century
28
Lesson 4 (Ideally) Start Early

• Sequential, not concurrent, mile-stones
• Release of CCSM3 to the community
• Peer-reviewed documentation of CCSM3
• Execution of IPCC experiments
• We did all of these in 2004!

29
Lesson 5 Check Throughput of Data Transfer
30
Lesson 6 Plan for all End Users

• Issue CCSM output 100 TB
• This is orders of magnitude more data than our
  colleagues in IPCC WG2 and WG3 are used to
  analyzing.
• How do we make the data accessible usable?
• Recommendations for steps forward
• NCARs GIS tool for model output?