What
is GEM LAM?
What is
the goal of GEM LAM?
Which GEM LAM model runs are available?
Where
can I get more info on GEM LAM?
Changes to GEM LAM 2.5 km and
other LAM versions
- Jan 2009
Milbrandt-Yau double moment precipitation scheme for GEM LAM 1km version used in 2010 olympics
Change to LAM
file names in CMC databases
- Apr 2008
Milbrandt-Yau microphysics scheme replaces
Kong and Yau
- Dec 2006
Spin Up issues in some Arctic air masses
Return to original GEM LAM 2.5 east and west windows
What
is GEM LAM?
GEM-LAM (Limited Area Model)
is essentially a one way nested version of GEM. The core of GEM-LAM is
identical to the global version of GEM used operationally. For a
detailed description of the operational model GEM-15 please refer to this
technical note by clicking here
Top of page
What
is the goal of GEM LAM?
The aim of the GEM-LAM project is to develop a high resolution model
that offers a better representation of local conditions (orography,
vegetations, etc), physical processes (cloud microphysics, radiation,
etc) and dynamical organization of weather systems at all scales (from
synoptic scale to mesoscale). The objective of the current work
is to assess the potential and shortcomings of the model at
meso-γ-scale. This task is being performed by CMC/RPN and a number of
regional collaborators.
Top of page
Which model runs are available?
At this time two experimental GEM LAM model runs are executed in
operational mode once daily. One configuration covers southern British
Colombia and a second configuration covers southern Ontario and
southern Quebec. The horizontal and vertical resolutions of the two
models are set at 0.0225 deg (~2.5 km) and 58 levels respectively. The
output of the operational GEM-15 run at 00Z is being used to provide
the initial and boundary conditions for the two runs. They are being
initiated at 12Z and provide hourly forecasts out to 24 hours. The two
model runs are non-hydrostatic, use a fully explicit physics
condensation scheme (Kong and Yau 1997) and a shallow convection scheme
(Kuo-transient) but no deep convective parameterization . A more
detailed description of the dynamic and physical configuration of the
models will be provided at a later date. Other model configurations
(for the Prairie and Northern region and for the Maritime region) are
under development for future experimental runs.
Top of page
Improvements to the snow analysis on
the GEM 2.5 grid
Prior to December 2006, the operational snow analysis was produced on a
coarser grid (37 km grid spacing) and was interpolated directly on the
GEM 2.5 km grid. This interpolation made the snow field to appear
unsmooth, especially over the large mountainous regions, causing large
changes in snow depth in very short distances. One effect of this
initial approach was on the surface temperature which then
affected the winds, precipitation typing, etc.
Bruce Brasnett of CMC
development has developed an algorithm to obtain a much more realistic
snow analysis for the GEM 2.5 given the same source of real data. He
uses the 6 hr forecast of the surface temperature and the accumulated
precipitation of the GEM 15. Then the topography of the GEM 2.5 is used
to adjust the temperature from the 15 km elevations to the one at 2.5
km, using the standard atmosphere lapse rate 0.006 deg/m.
Here is the comparison of the 15km
snow analysis and the 2.5km analysis for April 15, 2006 at 00z.
Top of page
Freezing rain
issue in snow squalls.
During the winter of 2005-2006, some unrealistic patches of freezing
rain were forecast by the GEM 2.5 which should have been snow. It was
noted that the problem resulted from the shallow convective scheme (Kuo
Transient). This scheme divided the liquid and solid partitioning of
the precipitation unrealistically. Therefore the excessive production
of the liquid precipitation was then translated into freezing rain once
reaching the sub zero degrees surface temperatures.
We corrected the
problem by defining the precipitation typing using a diagnostic of the
temperature profile of the model. This diagnostic, called Bourgouin's
method, was first developed by Pierre Bourgouin and later modified
(Extended Bourgouin's method) by Andre Methot and Andre Plante of the
CMC development branch. The extended Bourgouin's method is used in the
regional operational model (GEM 15) for the precipitation typing of all
the precipitating schemes; that is the convective and the explicit
(Sundqvist) schemes. In the case of the GEM 2.5, which is run using the
Kong & Yau explicit cloud and condensation scheme, the extended
Bourgouin's diagnostic is only applied to the shallow convective scheme
and not the Kong & Yau scheme.
Top of page
Spin up of up to
8 hours in some Arctic winter air masses in the absence of strong local
forcing.
This is mainly a larger concern during the winter time for the eastern
window as the western window maintains widespread local forcing due to
large presence of the high terrain and the ocean.
The problem is due to the different condensation scheme used in the two models. The GEM 15 uses a more implicit condensation scheme after Sundqvist et al. (1978, 1981) where as the GEM 2.5 uses a more explicit scheme after Kong & Yau (1997). In general due to the implicit nature of the Sundqvist's scheme the total cloud content may be as much as one order of magnitude less than the more explicit scheme Kong & Yau.
Currently the GEM 2.5 obtains all the necessary lateral boundary conditions by the GEM 15. Once the cloud and moisture is passed from the GEM 15 to GEM 2.5, the Kong & Yau works to generate a higher cloud quantity it requires. In the context of Kong & Yau the rate of cloud generation in short is dependent on the available moisture and vertical velocity. Therefore in the absence of instability and wide spread topography, the rate of cloud production is in general lower. This problem is traditionally resolved by having a spin up cycle using the same physics on the required grid.
Given the limitation of
our current computer resources the most efficient and the most
effective solution so far is to have another Limited Area Model at 15
km resolution (LAM 15) with the Kong & Yau scheme. This model is
integrated by obtaining its boundary condition from the forecast of the
GEM 15. Then the LAM 15 provides the lateral boundary conditions for
the GEM 2.5. This nesting strategy will mitigate this issue at
relatively low cost.
Top of page
Reverting back to
east and west windows for GEM LAM 2.5 km
During the summer of 2006, the GEM LAM western window had been extended east to cover most of Alberta to allow users to help evaluate the performance of GEM LAM 2.5 during the summer convective season.
In December 2006, we have reverted back to the
original east and west GEM LAM 2.5 km windows with each window
integrating forecasts out to 24 hours. .
Top of page
Update to GEM LAM 2.5 microphysics scheme
As of April 2, 2008, the Kong and Yau
microsphysics scheme has been replaced by the single-moment version of
the Milbrandt-Yau multi-moment bulk microphysics scheme in all
instances of the GEM LAM 2.5km model. For more technical
information on this version of the scheme click here.
Top of page
Update to double moment microsphyics scheme
in 1km version of GEM LAM used in 2010 Olympics
In January 2009, a double moment
Milbrandt-Yau bulk microphysics scheme was installed in the 1km version
of the GEM LAM proposed for the 2010 Olympics.
For more
technical
information on this scheme click here.
Top of page
Where can I get more technical information?
For further technical
information about GEM-LAM please refer to
http://notos.cmc.ec.gc.ca/mrb/rpn/eng/gemdm/gemdm.html
http://notos.cmc.ec.gc.ca/mrb/rpn/eng/gemdm/lam_seminar.pdf
MSC employees can also access this page for more information
Top of page
Change to LAM file names in CMC database
Users of LAM model data from the CMC gridpt
database or CMC archives will notice that the LAM filenames have
changed.
As of Janunary 1, 2009, the grid spacing of
the model (in km) will be encoded at the end of the LAM file names.
Also, all files for each domain will share a common gridpt
subdirectory.
For example, the west window model data in eta
coordinate that was:
lam.spinup/west.eta/YYYYMMDDHH_FFF
(15km)
lam/west.eta/YYYYMMDDHH_FFF
(2.5km)
will BECOME:
lam/west.eta/YYYYMMDDHH_FFF_15km
(15km)
lam/west.eta/YYYYMMDDHH_FFF_2.5km
(2.5km)
Files in the pressure coordinate directory and diag directories will
also be changed to reflect the new file naming approach.
Model data generated by the olympics grids
will reside in:
lam/olympics.eta/YYYYMMDDHH_FFF_15km
(15km)
lam/olympics.eta/YYYYMMDDHH_FFF_2.5km
(2.5km)
lam/olympics.eta/YYYYMMDDHH_FFF_1km
(1km)
Users who access LAM files using the utility
fgen+, will have to add the -x argument (extension) with grid size in
combination with the -s (start) and -e (end) time arguments. So,
for example, to get the file name for the 6h forecast of the lam west
2.5 km GEM, you would use the command:
fgen+ prog/lam/west.eta -t wh12 -x _2.5km -s 006 -e 006
This change will impact archiving and all
users that rely on the current gemlam gridpt database structure.
Users of LAM data are advised to arrange their
processes to reflect the new file naming convention by or before
Feburary 2009.