Computing with the GEM Model

Download Report

Transcript Computing with the GEM Model

The Canadian Regional Climate Modelling
and Diagnostics (CRCMD) Network
CRCMD Network (Oct2006-2010): Main research aims:
1. To develop the next-generation of the Canadian Regional Climate
Model (CRCM5), for application at high-resolution (~10-15km).
CRCM5 will be developed within the GEM dynamical core and
parallelised code architecture. This development is absolutely
necessary to allow high-resolution regional climate modelling
2. Introduce state of the art physical parameterisation packages
into CRCM5. Evaluate and improve the scalability of these
packages for combined used at multiple resolutions.
3. Increase the realism of surface –climate interactions
(specifically directed to regional processes over North America)
4. Improve our ability to analyse the large quantity of high resolution
regional climate data. With an emphasis on high-impact weather
events, quantification of the uncertainty associated with regional
climate change simulation & added-value due to increased resolution.
GEM scales well in parallelisation tests: 90% of linear scaling out
to 300 PEs on a Cray XT3, 85% of perfect on 192 SUN opteron PEs
This is important as models are applied at increasingly higher resolution
e.g. A large number of grid points & short timestep costs a lot!
Source : R. McTaggart-Cowan
GEM can be run as :
(i) a regular resolution GCM
(ii) a variable resolution GCM
(iii) a standard limited-area (RCM) configuration.
GEM supports non-hydrostatic dynamics, allowing (in principle)
application at all scales from GCM to Convection–Resolving.
An emphasis of the CRCMD Network will be to ensure
CRCM5 at high-resolution RCM becomes operational (with
the Ouranos CST) within the lifetime of the Network.
This will require extensive collaboration between
University, Environment Canada (CCCma and RPN) and
Ouranos scientists.
An important component of this will be close phasing with
Environment Canada with respect to release versions of
GEM and periodic code harmonization to prevent drift
between the individual institutes/model versions.
An equally important part of CRCMD Network activities
is to train graduate students and PDFs in the
development, use and diagnosis of high-resolution RCMs:
Key Science/research themes in the CRCMD Network
Theme 1: Science issues for high-resolution climate modelling.
Evaluating the scalability of parameterisations from RCM to GCM resolutions
Improving the simulation of the atmospheric water-cycle at high-resolution
Establishing the best modeling tools to go from GCM scales to high-resolution
regional scales (fixed GCM-RCM, variable GCM, or a combination of both)
Theme 2: Surface feedback processes in RCMs
Improved surface flux and surface wind representation
Implement and evaluate dynamic vegetation modules in GEM
Implement and test interactive permafrost treatment in GEM
Implement and evaluate interactive lake and Arctic Ocean models into GEM
Theme 3 : New diagnostic techniques for RCMs
Quantifying uncertainty in RCM climate change projections
Quantifying the added-value of high-resolution RCMs
Analysing the water-cycle and small-scale processes in RCMs
Establishing the global transferability of RCMs
Main institutes and participants:
Colin Jones
René Laprise
J-P Blanchet
Eric Girard
Changhui Peng
Adam Monahan
John Scinocca
Greg Flato
Vivek Arora
Daniel Caya
Ramon de Elia
Laxmi Sushama
Gilbert Brunet
Jean Cote
Bernard Dugas
Ayrton Zadra
Paul Vaillancourt
Caroline Chartrand
UQAM
UQAM
UQAM
UQAM
UQAM
UVIC
CCCma/UVIC
CCCma/UVIC
CCCma/UVIC
Ouranos/UQAM
Ouranos/UQAM
Ouranos/UQAM
RPN/UQAM
RPN/UQAM
RPN/UQAM
RPN/UQAM
RPN/UQAM
UQAM
Model Parameterisations
Model dynamics, BBE, diagnostics
Moist parameterisation, satellites
Cloud microphysics, Arctic climate
Terrestrial ecosystem modelling
Diagnostic analysis, surface scheme
Model paramertisations
Ocean and sea-ice modelling
Terrestrial ecosystem modelling
‘Operational’ RCM, Diagnostics
Diagnostic analysis
Surface processes
Diagnostic analysis
Model dynamics, Variable Resolution
Parameterisation, GEM structure
Parameterisation, GEM strutcure
Parameterisation, GEM structure
Network coordinator