Transcript CRCM5 description

A CRCM5 Description
Arctic System Model Workshop III
UQÀM, Montréal
Bernard Dugas & Katja Winger
Environment Canada/RPN
July 16, 2009

Where does the CRCM5 come from ?
GEM, GEM/DM, GEM-LAM
 GEMCLIM, MC2
 CGCM


Examples of runs done with
CRCM5 / GEM-LAM
 Dependence on domain size and
Using Spectral Nudging
 High resolution experiments
2
Atmospheric models
at Environment Canada (EC)
Atmospheric Science and Technology Directorate
CGCM
Model for
climate change
studies
CRD/CCCma
GEMCLIM
Monthly and
seasonal
forecasting
(Victoria)
CGCM (Canadian General Circulation Model)
CRD (Climate Research Division)
CCCma (Canadian Centre for Climate
Modelling and Analysis)
GEM
Model for weather
forecasting
MRD/RPN
(Dorval)
GEM (Global Environmental Multiscale)
MRD (Meteorological Research Division)
RPN (Recherche en prévision numérique)
3
4
GEMCLIM
GEMCLIM is the climate configuration of Environment Canada's
forecast model GEM (Global Environmental Multiscale)
global uniform
global variable
limited area
LAM
GEMCLIM was originally developed to study systematic errors of the
forecast model GEM and is routinely run at RPN as a validation tool.
The model is used operationally for monthly and seasonal
forecasting. As such it is fully supported by Environment Canada
4
GEM/CRCM5 timelines
V1.0.0 (EC, 1998) GEM
V2.0.0 (EC, 2000)
V3.0.1 (EC, 2002)
V3.1.2 (EC, 2004)
V3.3.0 (EC, 2007)
V4.0.x (EC, 2009)
First (operational forecast) version
(Côté, J. et al., MWR 1998a and b)
GEM/DM
MPI enabled GEM (inspired by MC2)
Non-hydrostatic dynamics option
(Yeh, K.-S., et al., MWR 2002)
GEM-LAM LAM configuration (largely based on MC2)
4DVarOperational 4D Var implementation
(Gauthier, P. et al., MWR 2007)
VStag
Vertical staggering (nearly complete)
GEMCLIM v3.2.2
=> CRCM5.0.0 (UQÀM, 2005; intro to GEMCLIM)
GEMCLIM v3.3.0
=> CRCM5.0.1 (UQÀM, 2007; extensively
used by students and post-docs)
GEMCLIM v4.0.x
=> CRCM5.1.0 (UQÀM, 2010; will include
support for CGCM physics)
Note that the main difference between GEM and GEMCLIM is in the
scripts used by the two models: There are few differences in their
dynamics and physics libraries and +95% of the code is shared.5
CRCM5
at UQAM / Ouranos
Global and regional, current and future
climate scenarios
CGCM
CRCM5
GEMCLIM
UQAM / Ouranos
CRD/CCCma
(Montreal)
MRD/RPN
(Dorval)
(Victoria)
dynamics
dynamics
?
physics
physics
dynamics
physics
physics
6
CRCM5
Numerics Highlights

Time stepping: Fully implicit, two-time level (with 2 Crank-Nicholson iterations)

Arakawa horizontal C-grid (U staggered in x-axis and V staggered in y-axis)

Charney-Phillips vertical staggering (Temp, moisture and W on half levels)

Semi-Lagrangian advection (absolutely stable)

Physic parameterization are treated slab-wise (2D in x-z plane)
This implies that the CRCM5 time step need only be determined by the
desired precision (of its physics, for example) and not by the numerical stability
constraints of its dynamics: Courant numbers above two or three near the upper
boundary can be easily accommodated. Note that Courant numbers below 1
should still be the norm in the troposphere. We now can use a 60 minute time
step when running global 1,5° simulations and 20 minute at 0.5°.
7
Arctic SLP, Smaller Domain, 1983-2002 (next slides from M. Qian)
Summer
CRCM5
ERA40
Diff
Winter
8
A suite of CRCM5/GEM integrations are being compared to evaluate the overall
performance of the model in representing Arctic climate processes.
Before coupling LAM model to an Arctic Regional Ocean Model.
In the next set of results 4 model runs are compared to ERA40.
1.
CRCM5-LAM (0.5°) forced by ERA40
2.
CRCM5-LAM (0.5°) forced by LBCs from a GEM-Global integration at 2°
3.
CRCM5-Variable resolution grid. With the high resolution
telescoped region of the global grid being at 0.5° resolution and covering
almost the exact same area as the CRCM5-LAM runs in 1 and 2 above.
The outer regular region of the global-variable grid is at 2° resolution
4.
CRCM5-GLOBAL, uniform resolution at 2°
All runs used AMIP SSTs for the period 1979-2004, CRCM5 version 5.0.1
9
Winter SLP, Larger Domain, 1979-2001
Piloted by
Piloted by
GEM
ERA40
Global
2.0 DEG
CRCM5
Variable
GEM
Global
2.0 DEG
ERA40
10
What happens when Spectral Nudging is used ?
No nudging
Nudging UV
Winter SLP (DJF 1978) differences between CRCM5 and ERA40
11
Example of high-res work at EC with GEM-LAM
(project UNSTABLE config, courtesy of J. Milbrandt)

Daily 12-hour forecasts

1-km driven by 2,5-km

Non-hydrostatic models

1-km: 412x482x58, 30s

2,5-km: 648x470x58, 60s

Uses proposed Vancouver
Olympic setup
Investigation of convective initiation and forecasting of deep convective
storms in south-central Alberta from July to the end of September ‘09
12
Merci / Thank you
DRAFT – Page 13 – March 31, 2016