Implementation Plan for CCSM 4 CCSM 4 needs to be ready by the end of 2008 for AR5 in early 2013.

Download Report

Transcript Implementation Plan for CCSM 4 CCSM 4 needs to be ready by the end of 2008 for AR5 in early 2013.

Implementation Plan for CCSM 4
CCSM 4 needs to be ready by the
end of 2008 for AR5 in early 2013.
Most important items to address for CCSM 4
• Physical biases in CCSM 3. Double ITCZ,
ENSO frequency, continental precipitation,
high latitude land temperatures, too large
windstress, and too much Arctic low cloud.
• CCSM 4 should have some form of carbon
cycle.
• The indirect effects of aerosols should be
included, which were omitted in CCSM 3.
HadiSST
data set
1949-2004
B31.002
Changes to
FV deep
convection
scheme and
momentum
transport
due to
convection
Stage 1 – starts on 1 March 2007
•
•
•
•
•
•
•
•
BGC land is CLM-CN; results of C-LAMP
Aerosol indirect effect scheme: NOT NOW.
Atmosphere: updated version of the FV
Land: Community Hydrology Project
Ocean: POP 2 base code plus updates
Sea Ice: merged CICE4 and CSIM4 codes
Resolution: FV 1.9x2.5, ocean x1
Significant advance on current BGC control
runs in the T31x3 CCSM 3
Stage 2 – complete by end of 2007
• Developments in all components designed
to reduce the significant CCSM 3 biases.
• Include in prognostic mode the land ice
component being worked on by Lipscomb.
• Why so early? I’m afraid if we say June
2008, then won’t be ready by end of 2008.
• CAM should just include the troposphere.
• Not include interactive chemistry. This
was controversial – include time slices?
Stage 3 – complete by end of 2008
• 2008 is year to validate and understand
CCSM 4 that includes BGC, indirect
aerosol effects, and land ice component.
• Target resolution? FV 1.9x2.5 for carbon
cycle – higher resolution for short-term
simulations: FV 1x1.25, x1 Ocean?
• Many questions: eg. should CCSM 4 have
a dynamic biogeography component?
• Low resolution Paleo version also in 2008;
this might still be the T31x3 version?
Short-Term Simulations: Proposed Form
• Start in about 1980, then run in pure simulation or
simulation/assimilation mode until 2005. The
short-term simulation would be from 2006 to 2030.
• Need an ensemble size of >10 to address extremes.
• Does it make an important difference if the CCSM
is initialized to the actual climate of 2005?
• This requires data assimilation into the ocean, and
possibly sea ice extent. Do we need to initialize the
tropical Pacific for ENSO and N Atlantic for MOC?
CCSM3: Present Day Control Runs
Maximum MOC in North Atlantic
Blue: T85, 1
Red: T42, 1
Black: T31, 3
Advantages of Short-Term Simulations
• Because the runs are short, the atmosphere model
can be run at higher resolution: produces relevant
regional information for the relatively near-term.
• Most of the climate change is already committed,
so the projections are much less dependent on the
highly uncertain future greenhouse gas scenarios.
• There is a much smaller range between models in
their transient climate response, so that the multimodel ensemble is less dependent on the quite
wide range of sensitivities among climate models.
Projections for Arctic Land Temp
Challenges of Short-Term Simulations
• No experience so far with assimilating data into
the CCSM ocean and sea ice components, or with
coupled model assimilation as at Hadley Centre.
• Should run chemistry in prognostic mode or with
time slices? Should carbon cycle be included?
• This increases the CCSM project workload as
these S-T simulations would be in addition to the
more familiar, long future scenario runs planned
for CCSM4 that includes a form of carbon cycle.