ANNUAL JOINT WMO TECHNICAL PROGRESS REPORT ON THE GLOBAL DATA
PROCESSING AND FORECASTING SYSTEM (GDPFS) INCLUDING NUMERICAL
WEATHER PREDICTION (NWP) RESEARCH ACTIVITIES FOR 2012
CANADA
Meteorological Service of Canada, Environment Canada
Science and Technology Branch, Environment Canada
1.
Summary of highlights
3
2.
Equipment in use at the centre
4
3.
Data and Products from GTS in use
5
3.1
Data
5
3.2
Products
5
4.
Forecasting system
6
4.1
System run schedule and forecast ranges
6
4.2
Medium range forecasting system (4-10 days)
7
4.2.1
Data assimilation and objective analysis
7
4.2.2
Model
9
4.2.3
Operationally available Numerical Weather Prediction (NWP) Products
11
4.2.4
Operational techniques for application of NWP products (MOS, PPM, KF,
Expert Systems, etc.)
11
Ensemble Prediction System (EPS) ( Number of members, initial state,
perturbation method, model(s) and number of models used, number of levels,
main physics used, perturbation of physics, post-processing, calculation of
indices, clustering)
12
4.2.5
4.3
Short-range forecasting system (0-72 hrs)
15
4.3.1
Data assimilation and objective analysis
15
4.3.2
Model
16
4.3.3
Operationally available NWP products
17
4.3.4
Operational techniques for application of NWP products (MOS, PPM, KF,
Expert Systems, etc.)
18
Ensemble Prediction System (EPS) ( Number of members, initial state,
perturbation method, model(s) and number of models used, number of levels,
main physics used, perturbation of physics, post-processing, calculation of
indices, clustering)
19
4.3.5
4.4
Nowcasting and Very Short-range Forecasting Systems (0-6 hrs)
20
4.4.1
Nowcasting system
20
4.4.2
Models for Very Short-range Forecasting Systems
21
4.5
Specialized numerical predictions (on sea wave, storm surge, sea ice, marine
pollution transport and weathering, tropical cyclones, air pollution transport and
dispersion, solar ultraviolet (UV) radiation, air quality forecasting, smoke, sand
and dust, etc.)
22
4.5.1
Assimilation of specific data, analysis and initialization (where applicable)
22
4.5.2
Specific Models (as appropriate related to 4.5)
24
4.5.3
Specific products operationally available
30
1
4.5.4
4.6
Operational techniques for application of specialized numerical prediction
products (MOS, PPM, KF, Expert Systems, etc.) (as appropriate related to 4.5)
Extended range forecasts (10 days to 30 days) (Models, Ensemble and
Methodology)
31
32
4.6.1
In operation
32
4.6.2
Research performed in this field
32
4.6.3
Operationally available EPS products
32
4.7
Long range forecasts (30 days up to 2 years) (Models, Ensemble and
Methodology)
33
4.7.1
In operation
33
4.7.2
Research performed in this field
34
4.7.3
Operationally available products
34
Verification of prognostic products
35
5.
5.1
Annual verification summary
35
5.2
Research performed in this field
35
6.
6.1
Plans for the future (next 4 years)
36
Development of the GDPFS
36
6.1.1
Major changes in the operational DPFS which are expected in the next year
36
6.1.2
Major changes in the operational DPFS which are envisaged in the next 4
years
39
Planned research Activities in NWP, Nowcasting, Long-range Forecasting and
Specialized Numerical Predictions
44
6.2
6.2.1
Planned Research Activities in NWP
44
6.2.2
Planned Research Activities in Nowcasting
51
6.2.3
Planned Research Activities in Long-range Forecasting
53
6.2.4
Planned Research Activities in Specialized Numerical Predictions
54
7.
References
55
1.
2
1. Summary of highlights

May 2012 – Operational runs switched to the new P7 IBM computer
For details about the new supercomputer, see section 2.

May 2012 – Upgrade of the Regional Deterministic Wave Prediction System (RDWPS)
The model brought improvements to wave dissipation, modeling of wave heights in very high wind
conditions, and the addition of a time-evolving ice mask. New prediction domains were added over the
Arctic and the Gulf of St. Lawrence waters and the spatial resolution of an expanded North Atlantic
domain was increased from 0.5 to 0.15 degrees. Additional model runs were also added at 06 and 18
UTC for all regional domains forced by winds from the Regional Deterministic Prediction System (RDPS).
For details see section 4.5.2.1.

October 2012 – Upgrade of the Regional Deterministic Prediction System (RDPS)
o
Introduction of a 4D-Var analysis procedure
o
Increase in horizontal resolution of the piloting model from 55km (720x360) to 33km (800x600)
o
Limited Area Model







Increase in horizontal resolution from 15km (649x672) to 10km (996x1028)
Changes in PBL scheme: hysteresis effect in turbulent kinetic energy prognostic equation
Modifications to the scalar roughness length over water surfaces
Increase of the turbulent Prandtl number from 0.85 to 1.0
Implicit treatment of surface fluxes
Inclusion of salinity effect for the calculation of saturation vapor pressure over the ocean
Use of a new sea ice thickness climatology
For details see sections 4.3.1.1 and 4.3.2.1.

October 2012 - Upgrade of the Regional Air Quality Deterministic Prediction System (RAQDPS)
o
Moved from 15km to 10km horizontal resolution, and from 58 to 80 vertical levels
o
Now Piloted by the new RDPS (10km, 4D-Var)
o
Reprocessed hourly anthropogenic and biogenic emissions files on the new grid
o
Advection timestep now at 5 minutes, but chemical timestep remains at 15 minutes
For details see sections 4.5.2.1.

October 2012 – Upgrade of the Regional Deterministic Precipitation Analysis (RDPA)
An upgrade to the RDPA, which provides a quantitative precipitation estimate for North-America, was
implemented in October 2012. The horizontal resolution was improved from 15 to 10 km. Both bias and
skill improved slightly. The most significant improvements in skill were observed over the United States
in summer.
For details see section 4.5.1.1.

October 2012 – Upgrade of the Regional coupled atmosphere-ocean-ice forecasting
In October 2012, the horizontal resolution was improved from 15 to 10 km.
For details see section 4.5.2.1.
3
2. Equipment in use at the centre

Supercomputer platform
Two clusters each being an IBM P Series 775, 8192 Power7 cores, 32TB of main memory, 450TB of
high-performance GPFS parallel disk capacity. Operating system: AIX 7. From January to May 2012, the
old P5 supercomputer was also running in parallel with the new P7.

Front-end platform
Being phased out were two debian Linux clusters with 82 nodes (PowerEdge 1950) total, each node
having 8 processors and 16 GB of memory. They use about 94 TB of disk space (SAS and SATA)
through Infiniband.
Being phased in are two Linux clusters, each with 80 compute nodes (Dell PowerEdge M610, 8 cores,
48GB of memory). They use about 300 TB of disk space (SATA, SAS, SSD, fast I/O) through Infiniband.
High performance, networked, parallel file system consisting of 8 IBM System x3650 M2 I/O servers,
each with 16 processors and 24 GB of memory. There is 2 PB worth of SAS and SATA disks attached
over FC. The data is shared via GPFS/CNFS.
Summary of equipment in use at the Canadian Meteorological Centre
Computer
Memory
(GB)
Disk Space
65536
900 TB
32
1.4 PB
Dell M610 blade, Intel E5530 @ 2.4 Ghz,
1280 cores
7680
300 TB
82 PowerEdge 1950, 1300 cpu
1312
94 TB
8 IBM System x3650 M2 I/O servers, 128
cpu
192
2 PB
IBM P Series 775, 16384 cores
1 SGI ALTIX 350, 16 cpu

Mass storage system
The Meteorological Service of Canada has been using a robotized storage/archive facility for
Environment Canada (operated out of CMC Dorval) since 1994 in order to store and secure critical
services and departmental data including: Numerical Weather Prediction data (essential to improve
forecasts); Climate change scenarios (including IPCC run results), the Climate Archive Database;
computer backups, logs and ECONET router and firewall logs/data (essential in the investigation of
security incidents, performance statistics, etc).
The system comprises a SGI Altix 350 with 16 Itanium processors, 32GB of internal memory and 1.4 PB
of high-performance disks. The two tape libraries are Quantum Scalar i6000 with 4000 LTO tape slots
and 16 LTO-5 drives each. The hierarchical software manager is StorNext. As of December 31 2012,
12.3 PB of data was stored (primary copy).
4
3. Data and Products from GTS in use
3.1 Data
The following types of observations are presently used at the Centre. The numbers indicate the typical
amount of data (reports or pixels) received during a 24-hour period:
SYNOP/SHIP
67,600
TEMP (500 hPa GZ)
1,240
TEMP/PILOT (300 hPa UV)
1,300
DRIFTER/BUOYS
40,000
AIREP/ADS
15,200
AMV’s (BUFR)
1,750,000
MCSST (US Navy)
4,750,000
SA/METAR
185,000
AMDAR/ACARS
380,000
9501
PIREP
PROFILER
240
GEO radiances
2,800,0002
ATOVS (AMSU-A)
2,025,0003
16,400,0003
ATOVS (AMSU-B/MHS)
SSM/I
1,400,0004
SSM/IS
8,160,0005
A/ATSR
06
AIRS (AQUA)
320,000
IASI (Metop-2)
325,000
ASCAT (Metop-1/2)
1,920,000
2,1507
GPS-RO
3.2 Products
GRIB ECMF
GRIB KWBC
GRIB EGRR
FDCN KWBC
FDUS KWBC
U.S. Difax products
Significant weather forecasts
Winds/Temperature forecasts for various flight levels
1
2
3
4
5
6
7
Not assimilated
Clear sky radiances for 5 GEO satellites
Four NOAA satellites are assimilated, plus AMSU on AQUA, and Metop-2; obtained by ftp
A third of these are used for ice analyses; only F15 available; obtained by ftp
Three DMSP satellites (F16, F17, F18); obtained by ftp
This instrument is on ENVISAT; no longer available, was used for SST analyses
Data from COSMIC, GRACE-A, TERRASAR-X and METOP-1/2 GRAS
5
4. Forecasting system
4.1 System run schedule and forecast ranges
Description
Global assimilation
Regional assimilation
Regional final analysis
Global ensemble
assimilation
CanSIPS assimilation
Description
Name
Assimilation and final analysis run schedule
(all times in UTC)
Name
Time
Remarks
G2
R2
R3
E2
00, 06, 12, 18
00, 06, 12, 18
00, 12
00, 06, 12, 18
Details section 4.2.1.1
Details section 4.3.1.1
Cut-off: T+7:00.
Details section 4.2.5.1
M2
00
Details section 4.7.1
Time
Forecast run schedule
(all times in UTC)
Forecast period
Remarks
Global
G1
00, 12
Regional
R1
Local
high resolution
WH
EH
AH, MH
E1
00, 12
06,18
06, 18
12
06
00, 12
240 hours at 00
Details section 4.2.2.1
360 hours at 00 on All products available at T+5:00.
Sundays
144 hours at 12
48 hours
Details section 4.3.2.1
54 hours
All products available at T+3:30.
42 hours
Details section 4.3.2.2
24 hours
(experimental GEM-LAM 2.5 km)
24 hours
16 days
Details section 4.2.5.1
ER
00, 12
72 hours
Details section 4.3.5.1
GM
WG
00, 12
00, 12
48 hours
120 hours
Details section 4.5.2.1
Details section 4.5.2.1
WR
48 hours
54 hours
One month
Details section 4.5.2.1
M1
00, 12
06, 18
00
MA
00
3 month periods
covering 1 year
Global
ensemble
Regional
ensemble
Air quality
WAM fed by the
GDPS
WAM fed by the
RDPS
Monthly
Seasonal
6
Details section 4.6.1
Produced at the beginning and middle of
every month.
Details section 4.7.1
Produced at the beginning of every month.
4.2 Medium range forecasting system (4-10 days)
In this section, we refer to the Global Deterministic Prediction System (GDPS) (sections 4.2.1 to 4.2.4) and to
the Global Ensemble Prediction System (GEPS) (section 4.2.5).
4.2.1 Data assimilation and objective analysis
4.2.1.1
In operation
Method
Variables
Levels
Domain
Grid
Simplified
Physics
Frequency
Cut-off time
Processing
time
Data used
4.2.1.2

A 4D-VAR multivariate analysis, at the appropriate time, to the
9-hour forecast of a 80-level 0.25 uniform resolution GEM (Charron
et al., 2011). The incremental approach is used for 4D-Var. (Gauthier
et al., 1999). The GEM tangent-linear model and its adjoint with
simplified physics are used to propagate the analysis increments and
the gradient of the cost function over the assimilation window. The
length of the assimilation window is 6 hours with a time step of 15
min.
T, Ps, U, V and log q (specific humidity).
80 hybrid levels of GEM model.
Global
0.25o resolution for the outer loops and 1.0o for inner loops (T180).
Vertical diffusion
Subgrid scale orographic effects8
Large-scale precipitation
Deep moist convection
Every 6 hours using data ±3 hours from 00 UTC, 06 UTC, 12 UTC
and 18 UTC.
3 hours for forecast runs. 9 hours for final analyses at 00/12 UTC
and 6 hours at 06/18 UTC.
75 minutes plus 11 minutes for trial field model integration on the
IBM P7.
TEMP, PILOT, SYNOP/SHIP, AWV’s, ATOVS level 1b (AMSU-A;
AMSU-B/MHS, GLOBAL and RARS), BUOY/DRIFTER, PROFILER,
AIRS, IASI, GPS-RO, ASCAT, AIREP/AMDAR/ACARS/ADS,
SSM/IS and clear sky radiance data from geostationary imagers.
Research performed in this field
Ensemble-Variational (En-Var) data assimilation
Building on previous research in which an intercomparison of the global 4D-Var and ensemble Kalman
filter (EnKF) data assimilation systems was conducted (Buehner et al 2010a,b), new approaches for
using EnKF ensemble members to specify the background-error covariances in the variational data
assimilation system were examined. A new approach of efficiently applying spectral and spatial
localization to ensemble covariances was examined and showed some potential (Buehner, 2011).
Recently, the focus of this research is on using the 4D ensemble covariances obtained from the EnKF to
perform a 4D deterministic analysis with the variational data assimilation system. This has been named
the “Ensemble-Variational” or “En-Var” approach. By making full use of the already available EnKF
ensemble members, the En-Var approach avoids the computational and development costs associated
with the tangent-linear/adjoint versions of the forecast model. En-Var experiments, in which the
ensemble covariances are blended with the stationary covariances currently used operationally, are
currently being performed as a possible replacement for 4D-Var in the global deterministic prediction
system.

GPS-RO
The assimilation of GPS radio-occultation observations from COSMIC (constellation of 6 satellites), the
GRAS receiver onboard METOP-A, and GRACE was implemented operationally in 2009. These data are
8
Used in second outer loop only
7
assimilated as refractivity profiles, and have shown significant impact at all levels, in both summer and
winter (Aparicio and Deblonde, 2008; Aparicio et al. 2009). GPS-RO data are assimilated as an
absolutely calibrated source, i.e. without bias correction. It has been found that this requires a
demanding level of accuracy to ensure compatibility with other observations, particularly radiosondes
(Aparicio et al, 2009; Aparicio and Laroche, 2011).
The volume of COSMIC data has slowly but persistently decreased as the constellation ages. Beginning
in October 2010, additional GPS-RO data were received, from satellites TERRASAR-X, SAC-C and
C/NOFS, which were added to monitoring. TERRASAR-X and SAC-C were found ready for assimilation,
and their data were included in the assimilation in Aug 2011. They compensate for the reduction of
COSMIC data due to ageing. C/NOFS contains only tropical data and the profiles do not reach the lower
troposphere. C/NOFS radio occultation data was otherwise found to be comparable to existing tropical
data from other missions. Due to the different distribution, it was decided that further research is required
to properly benefit from this source. SAC-C, however, presented problems shortly after being approved
for assimilation, and data delivery ceased. As of 2013, operators still expect it to return online. COSMIC
has improved its performance by data volume over 2012 and 2013, reverting the decline by strongly
optimizing the onboard power use, despite ageing of batteries.
METOP-B, second in the METOP series was launched in 2012. Refractivity data was commissioned
after several months. After several weeks of monitoring, the data was found to present certain moderate
statistical differences with respect to METOP-A. This was partially expected due to a different (firmware)
parameterization of the instrument. It has better performance picking rising occultations early. The
statistical differences did not include bias, and only affect the pattern of O-B standard deviation. Since
assimilation is performed with dynamically generated a-priori error estimation, this is within the ability of
the assimilation system to handle GPSRO data. Therefore, after monitoring to verify the absence of bias,
METOP-B refractivity data were added to the assimilation pool.

GPSRO Source
Received
COSMIC-I
METOP-A (aka METOP-2)
GRACE
TERRASAR-X
SAC-C
C/NOFS
TANDEM-X
METOP-B (aka METOP-1)
Yes
Yes
Yes
Yes
No
Yes
No
Yes
Approved
assimilation
Yes
Yes
Yes
Yes
Yes
No
No
Yes
for
Assimilation of radiance data
Radiance assimilation was revised in the context of the upcoming EnVar system. Following extensive
(2-month) assimilation tests, the number of hyperspectral infrared channels will double with the
implementation of EnVar. The number of AIRS channels will increase from 85 to ~140 while that of IASI
channels will increase from 62 to ~140 as well. It is also possible to increase the density of AMSU-A
data if tests reducing the thinning (currently 150 km for all radiances) are conclusive. The same bias
predictors will be used for all radiances. The observation error will be defined by a factor times the std of
(O-P). Preliminary tests were made for the assimilation of Cris and ATMS from Suomi satellite as well
as IASI from METOP-B. Research is also conducted to test the impact of considering observation error
correlation (temporal and interchannel) in the context of the ensemble system. Research has also
started aiming at assimilating surface sensitive infrared channels over land. Research on cloudy infrared
radiance data will be pursued within EnVar.

Observing System Simulation Experiments (OSSE)
An OSSE capability was developed at Environment Canada to support future missions. The Nature Run
(NR) available from ECMWF for the period 2005-2006 was used as “truth” atmosphere. That original NR
(40 km) was interpolated to EC operational 35 km model grid. All data assimilated operationally in 20082009 were simulated (date shifted by three years w.r.t. NR). The OSSE was used to support missions
such as PREMIER (Rochon et al., 2012) and PCW (Garand et al., 2013, see Section 6 on PCW). EC
also contributed to OSSE science via simulation of all-sky infrared radiances (Heilliette et al., 2013).
That study also allowed evaluation of the expected level of cloud contamination present in the
operational assimilation system, and how it reduces as a function of the peak of the channel response
function.

Conventional observations
8
Important upgrades to the pre-processing and screening of aircarft reports and radiosonde data have
been undertaken. The 4D position of radiosonde balloons, available in BUFR code, is now supported in
the data assimilation systems. For all radiosonde bulletins in alphanumeric codes (as for most
radiosonde data currently received at CMC), the missing information about the balloon position is now
retrieved by using a mean elapsed ascending time profile (Laroche and Sarrazin, 2013). Observations
on significant levels are used, but thinned to retain one observation of temperature, wind and humidity
per model layer. Aircraft data are also thinned the same way. A temperature bias correction scheme is
now part of the data processing for aircraft reports. Observation error statistics for both radiosonde data
and aircraft reports have been revised. A verification package (against radiosonde data) that takes into
account the balloon drift has been developed. Summer and winter trials with these upgrades have been
carried out with the global En-Var system. Verification scores show significant improvements in shortterm forecast ranges, particularly in the stratosphere.

Land surface initial conditions
A first version of the Canadian Land Data Assimilation System (CaLDAS) has been completed and is
currently being tested to improve land surface initial conditions for the medium-range forecasting
systems (deterministic and ensemble). Surface-based data includes observations of screen-level air
temperature and humidity, and snow depth. Space-based data includes L-band brightness temperatures
from the Soil Moisture and Ocean Salinity (SMOS) satellite for soil moisture, and snow water equivalent
retrievals from AMSR-E microwave emissions and from GlobSnow products. First implementation in
experimental mode is expected in summer 2013, including the assimilation of surface-based
observations. Operational implementation of these systems is expected late 2013 or early 2014.
Implementation of CaLDAS with space-based observations is expected in 2014 or 2015.
4.2.2 Model
4.2.2.1
In operation
Initialization
Formulation
Domain
Numerical technique
Diabatic Digital Filter (Fillion et al., 1995).
Hydrostatic primitive equations.
Global.
Finite differences: Arakawa C grid in the horizontal and Arakawa A grid in
the vertical.
Uniform 800 × 600 latitude-longitude horizontal grid. Horizontal resolution is
0.45o in longitude and 0.33o in latitude.
80 hybrid levels. Model lid at 0.1 hPa.
Implicit, semi-Lagrangian (3-D), 2 time-level, 900 seconds per time step
(Côté et al., 1998a and 1998b).
x, y,  and time.
E-W and N-S winds, temperature, specific humidity and logarithm of
surface pressure, liquid water content, Turbulent kinetic energy (TKE).
MSL pressure, relative humidity, QPF, precipitation rate, omega, cloud
amount, boundary layer height and many others.
Grid
Levels
Time integration
Independent variables
Prognostic variables
Derived variables
Geophysical variables:
Derived from analyses
initial time, predictive
at
Surface and deep soil temperatures, surface and deep soil moisture ISBA
scheme (Noilhan and Planton, 1989; Bélair et al. 2003a and 2003b); snow
depth, snow albedo, snow density.
Derived from climatology at
initial time, fixed in time
Sea ice thickness.
Derived from analyses, fixed
in time
Sea surface temperature, ice cover.
Derived from a variety of
geophysical
recent
data
bases
using
in
house
software, fixed in time
Orography, surface roughness length (except over water), subgrid-scale
orographic parameters for gravity wave drag and low-level blocking,
vegetation characteristics, soil thermal and hydraulic coefficients, glaciers
fraction.
Horizontal diffusion
Del-6 on momentum variables only, except del-2 applied on temperature
and momentum variables at the lid (top 6 levels) of the model.
Parameterized (McFarlane, 1987; McFarlane et al., 1987).
Parameterized (Hines, 1997a,b).
Orographic gravity wave drag
Non-orographic gravity wave
drag
9
Low level blocking
Radiation
Surface scheme
Surface roughness length
over water
Turbulent mixing (vertical
diffusion).
Shallow convection
Stable precipitation
Deep convection
4.2.2.2

Parameterized (Lott and Miller, 1997; Zadra et al., 2003).
Solar and infrared using a correlated-k distribution (CKD) (Li and Barker,
2005).
Mosaic approach with 4 types: land, water, sea ice and glacier (Bélair et al.,
2003a and 2003b).
Charnock formulation except constant in the Tropics.
Based on turbulent kinetic energy (Benoît et al., 1989; Delage, 1988a and
1988b) with mixing length from Bougeault-Lacarrère (1989; see also Bélair
et al, 1999) except near the surface and in the upper-troposphere.
1) Turbulent fluxes in partially saturated air (Girard, personal
communication).
2) Kuo Transient scheme (Bélair et al., 2005).
Sundqvist scheme (Sundqvist et al., 1989; Pudykiewicz et al., 1992. For
QPF evaluations see Bélair et al., 2009).
Kain & Fritsch scheme. (Kain and Fritsch, 1990 and 1993).
Research performed in this field
Dynamical cores
o
Advection scheme
Different mass fixing approaches are tested within a semi-Lagrangian scheme to better conserve
advected chemical species. For exemple, see Mahidjiba et al., (2008).
o Vertical discretization
After extensive testing, a formulation of vertical staggering is being introduced in the model. It is a
Charney-Phillips arrangement of the variables with an extension for a non-hydrostatic formulation;
both the global and nested versions of the model are developed simultaneously. Further
improvements are being tested, especially for the treatment of the semi-Lagrangian advection. The
position of the vertical levels, especially near the surface, is also being revisited. An improved upper
boundary condition has been introduced.
o Horizontal discretizations.
As a mean to avoid the so-called pole problem that plague lat-lon grid-point global models, research
on two alternative grid discretizations has been performed.
The first type of grid is a Yin-Yang grid. Tests with the full physics package have been conclusive.
This new model grid is expected to scale well up to a few hundred thousand processor cores. Work
on optimization is being performed. It is expected that the Yin-Yang grid for global applications will
become operational in 2014.
An approach based on finite volume and an icosahedral grid is being evaluated. This approach uses
a conservative Eulerian advection. The extension to 3D is being developed by the use of a quasiLagrangian vertical coordinate. The scalability of this approach is expected to be even better than
the Yin-Yang model.
o Solver
The use of a three-dimensional iterative solver is being developed for the GEM model. This provides
a more stable numerical model, and a potential improvement to the implicit scheme. It will in
particular stabilize the numerical model over very steep terrain at very high grid spacing (a hundred
meters).

Physical parameterizations
o
Improvements to the land surface processes, with improved surface fields (orography, subgrid-scale
parameters, land use / land cover, soil texture, and albedo) and surface processes (essentially
related to a better coupling between the surface and the atmosphere, including the use of a
distributed drag scheme).
o
Revision to the orographic blocking scheme: modulation of the drag coefficient as a function of 1) the
Froude number (stability), and 2) a non-linear directional amplification factor. See Wells et al. (2008)
and Vosper et al. (2009).
10
o
Improvement of planetary boundary layer (PBL) processes: (1) the inclusion of a hysteresis effect;
(2) a coupled-PBL approach using higher vertical resolution; (3) implementation of the distributed
drag formulation (see Beljaars et al., 2004).
o
Revision of parameterizations related to moist processes.
o
Sensitivity studies related to the tropopause cold bias. Aspects looked at are: optical properties of
cirrus clouds, upper-level moisture and thermodynamic functions.
o
Sensitivity studies related to the use of a multi-component 3D monthly climatology of aerosols.
o
Sensitivity studies related to surface fluxes in order to evaluate the impact of new formulations of
momentum roughness length and thermal roughness length, over land and over water.
o
Sensitivity studies of an implicit treatment of surface fluxes.
4.2.3 Operationally available Numerical Weather Prediction (NWP) Products
4.2.3.1
Analyses
A series of classic analysis products are available in electronic or chart form (snow depth and snow cover,
sea surface temperature, ice coverage, MSLP and fronts, 1000-500 hPa thickness, geopotential height,
temperature and winds at different pressure levels).
4.2.3.2
Forecasts
A series of classic forecast products are available in electronic or chart form (MSLP and 1000-500 hPa
thickness, 500 hPa geopotential height and absolute vorticity, cumulative precipitation over given periods
and vertical velocity, 700 hPa geopotential height and relative humidity). A wide range of bulletins containing
spot forecasts for many locations are produced. As well, other specialized products such as precipitation
type and probability of precipitation forecasts, temperature and temperature anomaly forecasts are produced.
4.2.4 Operational techniques for application of NWP products (MOS, PPM, KF, Expert
Systems, etc.)
4.2.4.1
In operation
Perfect
Prog
6 h and 12 h probability of precipitation forecasts at the 0.2, 2 and 10 mm
thresholds, at all projection times between 0 and 144 hours (Verret, 1987). An error
feedback system is applied on the probability of precipitation forecasts to remove
the biases (Verret, 1989). Consistency is forced between the 6 h and the 12 h
probability of precipitation forecasts using a rule based system which favors
forecasts sharpness.
Spot time total cloud opacity at three-hour intervals between 0 and 144 hours
projection times (Verret, 1987). An error feedback system is applied on the
forecasts to remove the biases and to force the forecasts to show the typical Ushaped frequency distribution like the one observed (Verret, 1989).
Spot time surface temperatures at three-hour intervals between 0 and 144 hours
projection times (Brunet, 1987). An anomaly reduction scheme is applied on the
forecasts so that they converge toward climatology at the longer projection times.
All weather element guidances mentioned above are also produced off each
member of the Ensemble Prediction System at all projection times between 0 and
240 hours.
Stratospheric ozone is used to calculate the Canadian UV Index (Burrows et al.,
1994).
11
Model
Output
Statistics
(MOS)
Automated
computer
worded
forecasts
SCRIBE
Weather
element
matrices
4.2.4.2
For the global system, the 2-m temperature is post-processed using the UMOS
(Wilson and Vallée, 2001 and 2002) package. This is done at three-hour intervals
from 0 to 144 hours. Note that the other weather elements from the global model
(winds, probability of precipitation and cloud cover) are statistically post-processed
using the Perfect-Prog method.
Before implementing a new version of the numerical model, the statistics are
updated using R&D final tests.
A system, named SCRIBE, is running at all the Regional Weather Centres in
Canada to generate a set of automated plain language forecast products from a set
of weather element matrices for days 3 to 7 inc. for the public forecast and for days
3 to 5 for the marine forecast (Verret et al., 1993; 1995; 1997). See the following
section Weather element matrices. SCRIBE is the main tool for operational public
and marine forecast preparation. Operational meteorologists use an interface to
add value to the automated forecast as required. Once the meteorologist has
reviewed the weather elements, SCRIBE generates the forecast products
automatically.
An ensemble of weather element matrices including statistical weather element
guidance, direct model output parameters and climatological values are prepared at
a 3-hour time resolution at 1399 points in Canada and over adjacent waters. The
data is valid at the projection times between 0 h and 144 h. A set of matrices
developed in 2008 based on the Global Ensemble Prediction System (GEPS)
extends the range out to day 7. Included in the weather element matrices are:
climatological maximum / minimum temperatures on a local time window; statistical
spot time temperature forecasts; maximum / minimum temperature forecasts
calculated from the spot temperatures on a local time window; climatological
frequencies of a trace or more of precipitation over 6-h and 12-hour periods;
climatological frequencies of 10 mm or more of precipitation over 12-hour periods;
statistical spot cloud opacity; statistical forecasts of probability of precipitation over
6-hour and 12-hour periods at the trace and 10 mm thresholds; model precipitation
amounts; model cloud height in three categories high, middle and low, Showalter
index; vertical motion at 850 hPa; conditional precipitation type; various
thicknesses; wind direction and wind speed at the surface; model surface dew-point
depression; Canadian UV index; model total clouds; 6-hour and 12-hour diagnostic
probability of precipitation; model surface temperature, model temperature and
dew-point depression near -level 0.97; sea surface temperature; ice cover; snow
depth; wave height forecasts and freezing spray accumulation forecasts. These
matrices are disseminated to the Regional Weather Offices where they are used to
feed an interactive system for composition of meteorological forecasts called
SCRIBE (Verret et al., 1993; 1995 and 1997).
Research performed in this field
Tests are being conducted to implement a 2-dimensional kriging interpolation procedure of UMOS
innovations using trial computed from the Global Ensemble Prediction System (GEPS). This new procedure
will improve the forecasts from numerical models at sites without observations but in vicinity of sites with
UMOS forecasts.
4.2.5 Ensemble Prediction System (EPS) ( Number of members, initial state, perturbation
method, model(s) and number of models used, number of levels, main physics used,
perturbation of physics, post-processing, calculation of indices, clustering)
This system is referred to as the Global Ensemble Prediction System (GEPS).
4.2.5.1
In operation
Since 2005, the Canadian Meteorological Centre (CMC) Global Ensemble Prediction System (GEPS) is
making use of an Ensemble Kalman Filter (EnKF) as data assimilation scheme to produce the initial
conditions for the forecast model (see for details Houtekamer and Mitchell, 2005 and Houtekamer et al.
2005).
Twice per day 16-day global forecasts are produced using the CMC NWP model (named GEM, version 4.2)
run with different physics packages. Here are the main characteristics:
12

Assimilation system
The ensemble Kalman filter (EnKF) uses four ensembles of 48 members for data assimilation for a total
of 192 members. The model top is located at 2 hPa. It has to be noted that the vertical localization of
levels above 6 hPa is now more restrictive then before to prevent the channel 12 of the ASMU-A satellite
from having an influence at the model top. The model configurations differ in the choice of the physical
parameterizations (convection, surface processes, gravity wave drag, mixing length) used. This is a
similar approach (Monte-Carlo like) to the one applied to the forecast component since the introduction
of the EPS in the CMC operational suite (see for example Houtekamer et al. 1996).
The EnKF assimilates the observations at their times of validity using the time interpolated trial fields
from those at 3, 4.5, 6, 7.5 and 9-hours. The control variables are wind, temperature, specific humidity,
surface pressure, radiative surface temperature. The initial perturbations have a global coverage and
include perturbations to observations. Isotropic perturbations are added to the “pure” EnKF perturbations
before the medium-range forecasts are started. No perturbation is done on SST, soil moisture or surface
wind stress.
The 20 initial conditions for the medium-range ensemble forecasts are obtained in the following manner:

o
First, twice a day, at 00 and 12 UTC, twenty representative members are chosen among the 192
analyses of the EnKF.
o
Then, the average of this subset of analyses is constrained to be equal to the 192 member analyses
ensemble mean.
Forecast system
The 20 initial conditions are then provided to the 20 configurations of the GEM model for the calculation
of 16 day forecasts.
The main model characteristics of the forecast component of the EPS system are:
o
Only one dynamical core is used: GEM
o
Horizontal resolution: 600x300 uniform global grid
o
Time step: 30 minutes
o
Number of levels: 40 (it is 58 for the assimilation model)
o
Top of model: 2 hPa
o
Stochastic perturbations of the physical tendencies inspired from Buizza et al. (1999) (Markov chains
with random number between 0.5 and 1.5 described in Li et al. 2008).
o
Stochastic kinetic energy back-scattering parameterization is used as in Shutts (2005).
4.2.5.2
Research performed in this field
In 2012, the EnKF system has been tested regarding:

The horizontal localization length scale (increasing with height instead of being a constant also known as
multiscale algorithm)

Higher horizontal, vertical and temporal resolution

Increased volume of AMSU radiance observations

Improved observations bias correction
For the medium-range prediction system, work has been done to fix some pathological behaviour in the
precipitation fields when deep convection occurs in conjunction with large Physical Tendencies Perturbations
(PTP). Also, in support of the extension towards monthly ensemble forecast, the option of evolving surface
fields with climatological trends has been added.
A new version of the GEM model with improved physics was tested for both the trial fields and mediumrange forecasts components.
4.2.5.3
Operationally available EPS Products
The following EPS products are available on the web as forecast charts at the following address:
http://www.weatheroffice.gc.ca/ensemble/index_e.html
13

10-day mean temperature anomaly

Spaghetti plots of the 500 hPa heights

Calibrated probability of equivalent precipitation for various thresholds

Accumulated quantity of precipitation

Sea level pressure centres

500-hPa geopotential heights
Also available on the web page is the ensemble spread of the trial fields.
The EPS forecast gridded data are available in digital format (GRIB2) from a MSC server. Technical details
as well as the terms and conditions of utilization of these data are available at this address:
http://www.weatheroffice.gc.ca/grib/index_e.html
The Canadian ensemble outputs are used in the North American Ensemble System (NAEFS) project, a joint
initiative involving the MSC, the United States National Weather Service (NWS) and the National
Meteorological Service of Mexico (NMSM). The following products based on the NAEFS joint ensemble
forecasts are available on the official MSC web server:
http://www.weatheroffice.gc.ca/ensemble/naefs/index_e.html

Temperature anomaly for the second week (day 8 to 14 outlooks). This is a common product produced
by MSC and the NWS.

EPSgrams for more than 300 cities in Canada, Mexico and the USA

Ensemble means and standard deviation charts for various gridded fields

Charts of probabilities of occurrence of several weather elements
Also, as mentioned in section 4.2.4.1, the Canadian EPS is used to produce Scribe matrices form day 1 to
day 10.
14
4.3 Short-range forecasting system (0-72 hrs)
In this section, we refer to the Regional Deterministic Prediction System (RDPS) (sections 4.3.1 to 4.3.4) and
to the Regional Ensemble Prediction System (REPS) (section 4.3.5). The last upgrade of the RDPS occurred
October 3, 2012. See also the section “Summary of highlights” at the beginning of this document.
4.3.1 Data assimilation and objective analysis
4.3.1.1
In operation
A continental variational data assimilation system (called REG-LAM3D) was implemented operationally on
20th October 2010 and is documented in Fillion et al. 2010. It is based on the Limited Area Model (LAM)
version of the GEM model (GEM-LAM see section 4.3.2.1). It was updated to 4D-Var on October 3rd, 2012.
For more details see the table below.
Method
Variables
Levels
Domain
Grid
Frequency
and cut-off
time
Processing
time
Data
used
Bogus
4.3.1.2

The short-range forecasting system is driven using the analysis produced
by the Regional Data Assimilation System (RDAS). This system consists of
a 6-hour spin-up period during which a 6-hour trial field is produced by the
Regional Global Environmental Multiscale (GEM) model (80 levels), which
is initiated from the analysis of the Global 4D-Var Data Assimilation
System.
The four-dimensional multivariate variational (4D-Var) method is also used
in the RDAS and it is performed at the end of the spin-up period. The
computation of innovations for the regional analysis is performed using the
high resolution grid of the GEM model. The 4DVar analyses are done in
spectral space using the incremental approach.
The analysis fields are then supplied to the short-range forecasting model
directly on its model vertical coordinates and limited area domain. (Fillion et
al, 2010).
T, Ps, U, V and log q (specific humidity).
Same 80 hybrid levels as the GEM global model.
A LAM domain, covering North America and adjoining oceans (see section
4.3.2.1).
The analysis is done spectrally as for the Global system but at T180.
Analysis increments are then interpolated to the regional GEM-LAM3D
model's analysis grid (10 km) at the end of the analysis ste.
Four 6-hour spin-ups are produced each day (00 UTC, 06 UTC, 12 UTC
and 18 UTC). They are initiated from global analyses valid 6-hour earlier
(with cut-off times of 3:00 hrs at 00 UTC or 12 UTC; and 6:00 hrs at 06 UTC
or 18 UTC) to generate REG-LAM first guesses (as well as global piloting
first guesses). The final regional 4D-Var analysis of each spin-up cycle (00
UTC, 06 UTC, 12 UTC and 18 UTC) has a cut-off of 2:05 hours. Data within
+/- 3 hours of analysis time are used.
25 minutes for the regional analysis (9 minutes for the 3D-Var Pilot) and 7
minutes for the 6-hour GEM-LAM (and 9 minutes for the global pilot)
integration.
TEMP, PILOT, SYNOP/SHIP, AMV’s, ATOVS level 1b (AMSU-A, AMSUB/MHS, GLOBAL and RARS), BUOY/DRIFTER, PROFILER, AIRS, IASI,
SSM/IS, GPS-RO, ASCAT, AIREP/AMDAR/ACARS/ADS, and clear sky
radiance data from geostationary imagers.
Subjective bogus, as required.
Research performed in this field
Benefits of the research done in the context of the GDPS
The research reported in section 4.2.1.2 for the subjects listed below is also beneficial for the RDPS.
15

o
Ensemble-Variational (En-Var) data assimilation
o
GPS-RO
o
Conventional observations
Regional LAM Analysis
Research continued in 2012 on the development of a 4D-VAR extension of the regional analysis and has
been finalized. Results from a first version of the system are reported in Tanguay et al (2012). This
system is coupled to a GEM-10 km resolution model. Optimisation of the VAR code on the new Power-7
IBM machines has been completed. The full system runs within its allocated 20 min real time and
assimilates data every 15 min binning rather 45 min over the 6h data assimilation window. The
performance of this new regional system is significantly better than the previous operational 3D-VAR
system in the range 0-48h. This new system has been implemented operationally October 3, 2012. See
also the section “Summary of highlights” at the beginning of this document.
4.3.2 Model
4.3.2.1
In operation
Initialization
Formulation
Domain
Numerical technique
Diabatic Digital Filter (Fillion et al., 1995).
Hydrostatic primitive equations.
LAM domain.
Finite differences: Arakawa C grid in the horizontal and Arakawa A grid in
the vertical.
996X1028 latitude-longitude grid having a uniform 0.09º (~10 km)
resolution covering North America and adjacent oceans
Grid
Note: This LAM is piloted by a Global model as per section 4.2.2.1. This
global model grid is rotated to be aligned with the LAM grid.
80 hybrid levels. Model lid at 0.1 hPa.
Implicit, semi-Lagrangian (3-D), 2 time-level, 300 seconds per time step
(Côté et al., 1998a and 1998b).
x, y,  and time.
E-W and N-S winds, temperature, specific humidity and logarithm of
surface pressure, liquid water content, Turbulent kinetic energy (TKE).
MSL pressure, relative humidity, QPF, precipitation rate, omega, cloud
amount, boundary layer height and many others.
Levels
Time integration
Independent variables
Prognostic variables
Derived variables
Geophysical variables:
Derived from analyses
initial time, predictive
at
Surface and deep soil temperatures, surface and deep soil moisture ISBA
scheme (Noilhan and Planton, 1989; Bélair et al. 2003a and 2003b); snow
depth, snow albedo, snow density.
Derived from climatology at
initial time, fixed in time
Sea ice thickness.
Derived from analyses, fixed
in time
Sea surface temperature, ice cover.
Derived from a variety of
geophysical
recent
data
bases
using
in
house
software, fixed in time
Orography, surface roughness length (except over water), subgrid-scale
orographic parameters for gravity wave drag and low-level blocking,
vegetation characteristics, soil thermal and hydraulic coefficients, glaciers
fraction.
Horizontal diffusion
Del-6 on momentum variables only, except del-2 applied on temperature
and momentum variables at the lid (top 6 levels) of the model.
Parameterized (McFarlane, 1987; McFarlane et al., 1987).
Parameterized (Hines, 1997a,b).
Orographic gravity wave drag
Non-orographic gravity wave
drag
Low level blocking
Radiation
Surface scheme
Surface roughness
over water
length
Parameterized (Lott and Miller, 1997; Zadra et al., 2003).
Solar and infrared using a correlated-k distribution (CKD) (Li and Barker,
2005).
Mosaic approach with 4 types: land, water, sea ice and glacier (Bélair et al.,
2003a and 2003b).
Charnock formulation except constant in the Tropics.
16
Boundary-layer
turbulent
mixing (vertical diffusion) with
wet formulation
Shallow convection
Stable precipitation
Deep convection
4.3.2.2

Based on turbulent kinetic energy (Benoît et al., 1989; Delage, 1988a and
1988b), with statistical representation of subgrid-scale clouds (Mailhot and
Bélair, 2002; Bélair et al., 2005). Mixing length from Blackadar.
Kuo Transient scheme (Bélair et al., 2005).
Sundqvist scheme (Sundqvist et al., 1989; Pudykiewicz et al., 1992).
Kain & Fritsch scheme. (Kain and Fritsch, 1990 and 1993).
Research performed in this field
Benefits of the research done in the context of the GDPS
The research reported in section 4.2.2.2 for the subjects listed below is also beneficial for the RDPS.

o
Vertical discretization
o
Solver
o
All aspects of physical parameterizations
Resolution increased to 10 km
As specified above, in October 2012 we implemented a new version of the RDPS. The research
performed in 2012 to obtain a final version with increased resolution involved readjusting several of the
sub-grid scale parameterizations, including the formulation of roughness length, vertical diffusion and
latent heat fluxes over oceans. Some work on solar radiation and the triggering of deep convection was
also performed. Verification of precipitation against different analyses was performed. A boundary layer
scheme with hysteresis effect has been introduced to maintain a more realistic profile in inversion
situations.

Microphysics
Research has been performed to start adapting the Milbrandt-Yau microphysics scheme (Milbrandt and
Yau, 2005a, 2005b) for lower-resolution applications (~10 to 30 km) by introducing a cloud fraction
formulation to the scheme.

High-resolution Deterministic prediction System
Research work to finalize the operationalization of a 2.5 km grid spacing system over western Canada
has been performed. Twice-daily forecasts up to 42 hours were tested.
The connection between the surface and the atmosphere has been studied through the lowering of the
first dynamical level to just a few meters above the surface.
Research has been performed to provide a Canadian-wide high-resolution (2.5 km grid spacing) shortrange forecasting system to the Canadian Meteorological Centre. Emphasis is put on improving the spinup time scale of the model to allow for relatively rapid data assimilation cycles.
4.3.3 Operationally available NWP products
4.3.3.1
Analyses
A series of upper-air analysis products are available in electronic or chart form (MSLP and fronts, 1000-500
hPa thickness, geopotential height, temperature and winds at different pressure levels).
4.3.3.2
Forecasts
A wide variety of forecast products are available in electronic or chart form. These include the classic charts
such as MSLP and 1000-500 hPa thickness, 500 hPa geopotential height and absolute vorticity, cumulative
precipitation and vertical velocity, 700 hPa geopotential height and relative humidity. Series of special charts
are produced in the context of the summer or winter severe weather (tropopause, stability indices, wind
shear, helicity, wind chill, liquid water content, streamlines, low-level maximum wind, vertical motion, etc.) or
in the specific support for aviation forecasting (icing, freezing level, height of cloud ceiling, momentum flux,
turbulence, etc.). A wide range of bulletins containing spot forecasts are produced for many locations over
North America.
A new production system has been developed over the past few years to standardise and unify the
production of charts and calculations described above. The system called SPOOKI stands for Système de
Production Orienté Object contenant une Kyrilel d’Informations – Object oriented production system
containing a myriad of information. A first version of this system has been implemented on an experimental
17
basis in 2012 to produce some aviation bulletins. SPOOKI will become operational in 2013 with additional
products related to summer and winter severe weather.
4.3.4 Operational techniques for application of NWP products (MOS, PPM, KF, Expert
Systems, etc.)
4.3.4.1
In operation
Perfect
Prog
Model
Output
Statistics
(MOS)
Diagnostic
techniques
on direct
model
output fields
Same as in 4.2.4.1 except based on the regional model and for lead time within 48
or 54 hours.
An Updateable MOS system (Wilson and Vallée, 2001 and 2002) is used for the
statistical post-processing of the direct regional model outputs. This regional postprocessing system currently provides forecasts for :
 2-m surface temperatures and dew point temperatures at spot locations at
three-hour intervals between 0 and 48-hour projection times at 00 and 12 UTC,
and between 0 and 54-hour at 06 and 18 UTC.
 10-m surface wind speed and wind direction at spot locations at three-hour
intervals between 0 and 48-hour projection times at 00 and 12 UTC, and
between 0 and 54-hour at 06 and 18 UTC.
 6h and 12h probability of precipitation at spot locations at the 0.2 mm threshold
between 0 and 48, or 54 hour projection times.
 Total cloud opacity at three-hour intervals in four categories.
 Surface winds at maritime locations (mostly buoys) at six-hour intervals between
0 and 48 or 54-hour projection times. Forecasts are produced for more than 100
locations including part of Pacific and Atlantic oceans but also for some large
Canadian inland water bodies.
 Equations are valid for the four runs (00, 06, 12 and 18 UTC).
Before implementing a new version of the numerical model, the statistics are
updated using R&D final tests.
Aviation Package: Charts of forecast icing (Tremblay et al., 1995), turbulence
(Ellrod, 1989), cloud amounts with bases and tops, freezing levels and tropopause
heights. The charts are produced at 6h intervals out to 24 hours.
Summer Severe Weather Package: Forecast charts of buoyant energy, helicity,
convective storm severity index, low level wind shear, precipitable water, low and
high level wind maximum, surface temperature and dew points, heights and
contours at 250 hPa and tropopause heights. The charts are produced at 6-hour
intervals out to 24 hours.
Winter Severe Weather Package:.Forecast charts of precipitation type (Bourgouin,
2000), 250 hPa contour heights and vorticity, precipitable water, 6-hour precipitation
amounts, wind chill, surface temperature, thickness values and warm or above
freezing layers with bases and tops. The charts are produced at 6h intervals out to
24 hours.
Air Quality Package:.Forecast charts of the mean sea level pressure at 21 UTC
with the forecast precipitation amounts between 12 and 00 UTC; charts of the
streamlines at 21 UTC with the wind mileage (time integration of the wind speed)
between 12 and 00 UTC; charts of the forecast minimum and maximum boundary
layer height and the ventilation coefficient. These charts are valid for today and
tomorrow.
Direct model outputs are used to forecast upper air winds and temperatures for
aviation purposes.
Several parameters are interpolated at stations, formatted and transmitted
operationally to Regional Offices.
Automated
computer
The calculations for these charts are in the process of being converted to the
SPOOKI post-processing tool.
A system, named SCRIBE, is running at all the Regional Weather Centres in
Canada to generate a set of automated plain language forecast products, including
18
worded
forecast:
Scribe
Weather
element
matrices
4.3.4.2
public, agricultural, forestry, snow, air quality and marine forecasts from a set of
weather element matrices for days 1, 2 and 3. (Verret et al., 1993; 1995; 1997).
See the following section Weather element matrices. SCRIBE is the main tool for
operational public forecast preparation. Operational meteorologists use an interface
to add value to the automated forecast as required. Once the meteorologist has
reviewed the weather element, SCRIBE generates the forecast products
automatically
Same as section 4.2.4.1, except the data is valid at projection times between 0 and
48 hours at 00 and 12 UTC, and between 0 and 54 hours at 06 and 18 UTC, and
UMOS guidance is used instead of Perfect Prog. Scribe matrices are produced four
times a day (00, 06, 12 and 18 UTC).
Supplementary weather element matrices are also used. The content of these
matrices include mean sea level pressure, surface pressure, lifted index, highest
freezing level, mean wind direction and speed over the four lowest  levels of the
driving model, boundary layer height and ventilation coefficients at time of minimum
and maximum temperatures, instantaneous and accumulated downward infra-red
and visible radiation fluxes, model temperature and dew-point at 925 and 850 hPa,
wind speed and direction at 925 and 850 hPa, model boundary layer height,
concentration of ozone near surface, as well as PM2.5, PM10, NO2, NO and SO2.
The time resolution of these matrices is 3 hours, with projection times out to 48
hours.
Research performed in this field
Tests are being conducted to implement a 2-dimensional kriging interpolation procedure of UMOS 2-m drybulb and dew-point temperatures innovations using trials computed by the Regional Deterministic Prediction
System (RDPS). This new procedure will improve the forecasts from numerical models at sites without
observations but in vicinity of sites with UMOS forecasts. Work is also planned to insert MOS marine winds
in weather element matrices.
4.3.5 Ensemble Prediction System (EPS) ( Number of members, initial state, perturbation
method, model(s) and number of models used, number of levels, main physics used,
perturbation of physics, post-processing, calculation of indices, clustering)
This system is referred to as the Regional Ensemble Prediction System (REPS).
4.3.5.1
In operation
The 20 members of the Regional Ensemble Prediction System (REPS) consist of limited-area configurations
of the GEM model over a continental, North American domain at 33 km grid spacing. Initial conditions are
provided by the operational Global Ensemble Kalman filter, and the lateral boundary conditions are provided
by the GEPS. All members use the same physics and dynamics. However, physical tendencies of individual
members are perturbed with Markov chains.
4.3.5.2
Research performed in this field
Research has been performed to better represent precipitation processes in the context of stochastic
perturbations. Improved representation of low-level blocking and planetary boundary layer processes were
incorporated into the system. Research work to bring the horizontal grid spacing to 15 km and increase the
number of vertical levels has been performed.
4.3.5.3
Operationally available EPS Products
There is no operational product available in 2012. We will implement something in 2013 (see
section 6.1.1).
19
4.4 Nowcasting and Very Short-range Forecasting Systems (0-6 hrs)
4.4.1 Nowcasting system
4.4.1.1
In operation
The SCRIBE Weather Forecast Product Expert System is capable of ingesting the latest observations and
nowcasting model data to update in real time the Scribe weather elements. This sub-system has been
developed to minimize the necessary manual adjustments done by the forecaster to merge the current
weather conditions with the forecast.
The Scribe Nowcasting uses surface observations (North American radar mosaic data have been temporarily
removed) and lightning data from the Lightning Detection Network. These observations are used to feed
short term forecast models. A statistical model called “PubTools” uses the surface observations to forecast
the probabilities of occurrences of weather elements. The observed radar reflectivities are projected during
the next 6 hours with a vector motion calculated from observed imageries 20 minutes apart. Finally, an
algorithm has been developed at CMC to predict the probabilities of thunderstorm occurrences based on the
forecast position of the lightning clusters. For weather elements not observed or not associated to a short
term forecast system, the regional deterministic prediction system run four times a day is used as a fall back.
All these observed and forecast data are processed by a rules based system to determine the best sequence
of weather elements representing the current observation and short term tendencies. UMOS for dew point
was introduced in 2012 in replacement of direct model values.
4.4.1.2

Research performed in this field
Radar QC/QPE Project
Latest versions (Versions 2.7 and 2.8) of the Unified Radar Processor (URP) development cycle include
a major upgrade to the QC/QPE data analysis software whereby bad and no data regions are explicitly
identified and where improved ground clutter algorithms are included. Masks for blocked areas are
added. A slightly improved version of the precipitation accumulation (PA) software is also included.
Version 2.7 has been implemented and adopted by CMC for QPE products. Version 2.8 was submitted
in early 2012 for software integration, and includes updated QPE configuration for Regional Centres.

Aviation Related Research
The Canadian Aviation Nowcasting System (CAN-Now) has been further improved and, although not
operational, is now running continuously for many airports across Canada with an emphasis on the
HUBs, the Hibernia area, and Defense Weather Services. Two Nowcasting systems are being tested.
The Integrated Weighted Model (INTW) system integrates in-situ observations and several models to
produce 0-8 hour forecasts of significant aviation variables such as visibility, ceiling, precipitation rate,
wind speed and direction, wind gust, cross winds, etc. The Adaptive Blending of Observations and
Models (ABOM) uses in-situ observations and the output of individual models to produce 0-6 hour
forecasts. The INTW system has been tested for possible use in the INCS system. The INTW system is
also being tested as an input into a First Guess TAF product. Both INTW and ABOM have undergone
significant statistical verification tests in 2012 and show considerable promise as operational tools.

Fog Remote Sensing And Modeling Project (FRAM)
A field program for the Fog Remote Sensing And Modeling Project (FRAM) was undertaken in Yellowknife.
This project was focused on investigating the microphysics of ice fog formation, evolution and dissipation in order to
better detect and forecast ice fog conditions. A manual for warm fog forecasting , which included extensive
overviews of all detection and prediction methods was produced for the operational forecaster
community. FRAM ice fog project work will be completed during 2013-14 fiscal year. A BAMS manuscript
is accepted for representing project scientific goals and its effect on operational meteorology.

Research Support Desk Project (RSD)
Research support desks were active in the Ontario Region and Prairie Region Storm Prediction Centres.
In each location, the close proximity of research and operational staff during high impact weather events
was used to provide a feedback process between research and operations, whereby research staff
learned the needs and gaps in the operational program and operations staff learned improved
forecasting and nowcasting techniques. Significant development in the last year was focused on
improving the weather depiction interface to the forecaster in order to optimize the efficiency of warning
production. These techniques were demonstrated but have not yet been implemented.
20

Interactive Forecasting, Nowcasting and Alerting Systems (0-48 hour)
The interactive Convective Analysis and Storm Tracking (iCAST) prototype for forecasting, nowcasting
and alerting (Sills et al. 2009) has been further developed to address the goals of MSC’s Warning Reengineering Signature Project and Next Generation Prediction System Signature Project. The prototype
pursues an area-based, MetObject-oriented approach that allows for a wide range of forecaster
interaction. Probabilistic verification approaches were further explored. Development of NWP-based firstguess MetObject fields for thunderstorm initiation and occurrence continued to be investigated and their
use with iCAST evaluated. Finally, forecaster workload tests related to interaction with the MetObject
database were undertaken to ensure the feasibility of the MetObject approach.
Much of the testing and evaluation of the iCAST prototype has taken place in a real-time setting at a
Research Support Desk situated in the operational area of MSC’s Ontario Storm Prediction Centre. A
similar desk has also been active at the Prairie Storm Prediction Centre in Edmonton. At each location,
the close proximity of research and operational staff during high-impact weather events was used to
provide a feedback process between research and operations, whereby research staff learned the needs
and gaps in the operational program and operations staff learned improved forecasting and nowcasting
techniques. Many of the techniques have been developed from field research on lake- and sea-breeze
circulations (e.g. Sills et al. 2011) and dryline circulations (Taylor et al. 2011).
4.4.2 Models for Very Short-range Forecasting Systems
4.4.2.1
In operation
No very short range forecasting system in operations at this time.
4.4.2.2

Research performed in this field
LAM urban window
Research has been done on an “urbanized” version of GEM-LAM at 250-m grid spacing over the
Vancouver metropolitan urban area, developed as part of the Urban Meteorology Modeling System. This
system will soon be tested over other cities, like Toronto, Canada in preparation for the 2015 Pan
American Games and Tokyo, Japan as part of a collaboration with the Japan Meteorological
Administration (JMA). It is planned to implement in experimental mode several computational windows
covering a few key urban areas of the country (e.g., Toronto, Montreal, Vancouver), shortly after the
2015 Pan American Games which will be used for demonstration.
21
4.5 Specialized numerical predictions (on sea wave, storm surge, sea ice, marine
pollution transport and weathering, tropical cyclones, air pollution transport
and dispersion, solar ultraviolet (UV) radiation, air quality forecasting, smoke,
sand and dust, etc.)
4.5.1 Assimilation of specific data, analysis and initialization (where applicable)
4.5.1.1

In operation
Surface fields
Fields
Analysis Grid(s)
Method
Trial Field
Frequency
Data Source
Surface air
temperature
a)1080x540 gaussian
b) regional grid
Optimum
interpolation
Surface dew
point
depression
a)1080x540 gaussian
b) regional grid
Optimum
interpolation
a) 6 hours
b) 24 hours at
18 UTC
a) 6 hours
b) 24 hours at
18 UTC
Synop, Metar, SA,
Ship, Buoy, Drifter,
Metar,
Synop, Metar, SA,
Ship, Buoy, Drifter
Sea surface
temperature
anomaly
a)1080x504 gaussian
b)1800x900 gaussian
Optimum
interpolation
Model forecast of
temperature at
hybrid=1.0
Model forecast of
dew point
depression at
hybrid=1.0
Previous analysis
24 hours
(at 00z)
Snow depth
a)1080x540 gaussian
b)Variable resolution
15 km grid;
c)2.5 km grid over
British Columbia
1080x540 gaussian
Optimum
interpolation
Previous analysis
with estimates of
snowfall and
snowmelt
6 hours
Ship, Drifter.
BuoyBuoy,
AVHRR-GHRSST,
A/ATSR , A/ATSR
(Brasnett, 1997);
Plus AAMSR-E for
b) (Brasnett, 2008)
Synop, Metar, Sa
(Brasnett, 1999)
Data averaging with a return to
climatology in areas where data are
not available.
24 hours
Deep soil
temperature
1080x540 gaussian
6 hours
Soil moisture
400 x 200 gaussian
Derived from climatology and a
running mean of the surface air
temperature analysis
Derived from climatology
Albedo
400 x 200 gaussian
Derived from albedo climatology,
vegetation type, the snow depth
analysis and the ice cover analysis
6 hours
Ice cover

SSM/I, plus ocean
and lake Ice data
from the Canadian
Ice Centre
No direct
measurements
available
No measurements
available
No direct
measurements
available
Regional Deterministic Precipitation Analysis (RDPA)
The Regional Deterministic Precipitation Analysis (RDPA) was upgraded in October 2012, following the
operational implementation of a 10 km configuration of the Regional Deterministic Prediction System
(RDPS). This analysis system provides 6-hourly and 24-hourly estimates of precipitation accumulation
on a 10 km grid covering North America in near real-time (a preliminary analysis is available at T+1h,
and a final analysis is available at T+7h). Monthly and seasonal anomaly maps are also generated in
experimental mode by MSC.
This version of the RDPA combines a background field obtained from a short-term forecast of the GEM
model in its regional configuration (at 10km) with rain gauge data. Observations are obtained by
combining the reports from the synoptic observation network with reports from COOP networks
(currently only over the US and over the Province of Quebec).
Objective evaluation of this new version showed that both the categorical bias and the equitable threat
score were slightly improved. The most significant improvements in skill were observed for the 24h
analysis over the United States during summer. A slight degradation of the precipitation mass is however
observed for the 6h-analysis in summer, due to more intense precipitation events being predicted by the
10 km RDPS.
4.5.1.2

Research performed in this field
3D-VAR sea-ice analysis
22
A new sea-ice analysis system has been developed, based on the 3D-Var approach to data assimilation.
The main system has a domain that includes all ice-covered waters surrounding North America
extending up to the North Pole, with a horizontal resolution of ~5 km (Buehner et al. 2013). A persistence
forecast from the analysis 6 hours earlier is used as the background state. Retrievals of ice
concentration from passive microwave data (AMSR-E and SSM/I) and the subjective analyses produced
by the Canadian Ice Service that heavily depend on RadarSAT images, are assimilated. This system,
which is running in experimental mode at CMC, will be upgraded at the beginning of 2013 by additionally
assimilating data from 3 SSM/IS sensors and from the scatterometer instrument ASCAT. A configuration
of the system has also been adapted to produce global sea ice analyses for use in the global and
regional NWP systems. The implementation in an experimental mode of the global system is planned for
the beginning of 2013.

Regional Deterministic Precipitation Analysis (RDPA)
The most important input for hydrological prediction and land data assimilation systems is generally
precipitation. This led to the development of a Canadian Precipitation Analysis (CaPA).
Research currently focuses on including other sources of observation in the analysis, including
observations of clear sky from GOES imagery, ground radar QPE, and lightning observations. Efforts are
also devoted to increasing the number of COOP network stations in the analysis and to correcting bias in
solid precipitation measurements.

High-resolution external land surface prediction system (200 m)
A prototype for this new system was successfully tested for several applications, including the
Vancouver 2010 Winter Olympic Games (Bernier et al. 2012).. A national extension of this system, at
200-m grid spacing, is currently being tested to improve short-range prediction of near-surface
meteorological variables, including air temperature, humidity, and winds. A major innovation with this
system is the representation and resolution of all Canadian cities. A first implementation in experimental
mode is expected for early 2014.

CaLDAS
Development of a first version of the Canadian Land Data Assimilation System (CaLDAS) has been
completed. The new system assimilates a larger amount of data using an Ensemble Kalman Filter
technique. For soil moisture, remote sensing data from ESA’s Soil Moisture and Ocean Salinity (SMOS)
mission and from NASA’s Soil Moisture Active and Passive (SMAP) is being examined. This data will be
assimilated in conjunction with near-surface air temperature and humidity.
For snow, a project has been completed to use space-based high-resolution optical information (e.g.,
from MODIS) to specify snow fractional coverage and microwave information (e.g., AMSR-E or SSM/I) to
retrieve snow water equivalent.
Finally, work is also underway to improve the first guess for the assimilation of leaf area index (LAI). The
Canadian Terrestrial and Ecosystem Model (CTEM), predicting the evolution of ecosystems including
fluxes of water, energy, carbon, and nitrogen, will be used for the evolution of vegetation. Results from
CTEM will be provided to a simple LAI assimilation system developed a few years ago at Environment
Canada. Current projects have proceeded to couple CaLDAS with upper-air assimilation systems for
global and regional analyses. The impact of CaLDAS land surface analyses (for surface temperatures,
soil moisture, and snow) is currently being tested for medium-range NWP systems. It will also soon be
done for short-range and longer-range systems.

Chemical data assimilation
Research is ongoing on the application of 3D-var chemical data assimilation system. This system is
based on the 3D-var meteorological assimilation system (v10.2.2) and uses RTTOVS9.2 for ozone
sensitive channels. There is also added capability for using averaging kernels for retrieval data such as
SBUV/2. The system has been used to assimilate near-real time ozone profiles, ozone partial columns
and total columns from MLS, SBUV/2 and GOME-2. The evaluation has been performed against
independent observations including ACE-FTS, MAESTRO, OSIRIS, MIPAS, Brewer, and ozonesonde
measurements. It also uses a linearized stratospheric photochemical module developed by McLinden et
al. (2000) that is on-line in the GEM NWP model and is also used to evaluate the radiative impact. A
study funded by the Canadian Space Agency to that effect was completed last year. This year, some of
the ozone assimilation experiments conducted in the previous study were repeated following an
improvement in the tropospheric configuration of the simplified ozone model and adjustments of the
ozone error variances using the Desroziers approach (Desroziers and Ivanov, 2001; Chapnik et al.,
2004; Desroziers et al., 2005). As part of the study, it was also shown that the diurnal variation of the
23
total column ozone from short-term forecasts is fairly consistent with that from Brewer measurements.
Limited time was also devoted to the continuation of the ozone-sensitive radiance assimilation study.
These studies are to be pursued next year. As extension of this work a new project was initiated for the
implementation of near-real time global ozone assimilation for improved UV index forecasting and for
providing ozone boundary conditions for the GEM-MACH regional air quality forecast model.
This system had also previously been used in an OSSE study for the proposed PREMIER mission (ESA,
2008) for the assimilation of ozone, water vapour and temperature. The PREMIER OSSE was a sub-task
of the PREMIER Impact Study sponsored by the European Space Agency (ESA/ESTEC) and led by the
Jülich Research Centre (Germany). A minor update to the PREMIER OSSE report was provided in June
2012. Significant improvements in tropospheric ozone and water vapour medium range forecasts were
obtained from the assimilation of PREMIER data. The same OSSE system also served in NWP studies
on the calibration of simulated errors for IASI and AIRS radiances and of atmospheric motion vectors for
the proposed Polar Communications and Weather mission (PCW, 2009)
Initial steps towards the migration of this chemical assimilation system in the Environment Canada NWP
assimilation EnVar/EnsKF modularization and unification project were also undertaken this year but
there is still more work to be done to complete the integration into the new assimilation system. Some
time was also devoted to the adaptation and verification of the forecasting and assimilation system to
modifications of the computing environment.

Objective Analysis for Surface Polluants
The regional chemical analysis system (for O3 and PM2.5) that was developed with the model
CHRONOS has been adapted to the operational air quality forecast model GEM-MACH. In collaboration
with the Canadian Meteorological Centre (CMC), all the steps towards an operational implementation
has been completed. Several issues needed to be addressed and completed, such as having a uniform
format and secure real-time feed for surface observations of pollutants that included AirNow
observations and all Canadian observations. Quality assurance, quality control, and process control of
the objective analysis suite for operations similar, in some ways, to the meteorological assimilation
practice were also completed.

Assimilation for Climate Applications
The main focus of this research is the development of an operational system for the routine estimation of
greenhouse gas fluxes from the surface to the atmosphere, based on atmospheric measurements. Such
a system would be used by Environment Canada for planning and policy purposes but also for
addressing questions of observing system needs, whether ground or space-based. The system is called
EC-CAS (EC Carbon Assimilation System) and it uses the operational air quality model (GEM-MACH)
and will use the GEPS as a basis for an Ensemble Kalman smoother. This year (2012) saw major
progress in the adaptation of the operational weather forecast model for CO2 simulation. Four issues
with the vertical diffusion scheme used by air quality applications were identified and rectified, two
convective mass transport schemes were tested and the deficiency of boundary layer mixing was
revealed. However these boundary layer modeling issues were rectified when a later version of GEM
(v4.5) with vertically staggered levels was adopted in 2013. The lack of mass conservation in the
forecast model due to the semi-Lagrangian advection scheme was studied and found to be significant.
Comparisons against observations and the US (NOAA) CarbonTracker are now greatly improved
compared to last year. Because GEM-MACH was employed without reactive chemistry and assessed
over long time scales (1 year), some bugs were uncovered by this project and these were found to have
potentially large impacts on air quality forecasts.
4.5.2 Specific Models (as appropriate related to 4.5)
4.5.2.1

In operation
Air Quality Model
Since November 2009, the Regional Air Quality Deterministic Prediction System (RAQDPS) has been
based on the model GEM-MACH. GEM-MACH combines the weather forecast model GEM with an inline chemical transport model. The air quality process representations include gas-phase, aqueousphase, and heterogeneous chemistry and aerosol processes. It uses a 2-bin sectional representation of
the PM size distribution, but PM chemical composition is treated in more detail and additional processes
affecting PM concentrations have been included (Anselmo et al., 2010). In October 2012, the version of
GEM-MACH used in the RAQDPS was upgraded to the GEM 3.3.8 and PHY 5.0.4.2 libraries, and the
grid-spacing in the horizontal was reduced from 15 to 10 km. In addition, the number of levels in the
upper part of the domain (between 0.1 and 100 hPa) was increased from 12 to 24, bringing the total
24
number of levels from 58 to 80. This latter change was done to increase the numerical stability of the
model.
The SMOKE emissions processing system is used to produce hourly anthropogenic input emission files
on the GEM-MACH rotated latitude-longitude grid. The emissions files account for hour, day, month and
primary emissions type (on-road mobile, off-road mobile and area, major and minor point sources, and
biogenic). Biogenic emissions are estimated on-line using the BEIS v3.09 algorithms and depend on
near-surface temperature, solar radiation, and Julian day.
The RAQDPS runs twice daily at 00 and 12 UTC to produce 48 hour forecasts. The limited-area domain
covers the bulk of North America and adjacent waters and is initialized and piloted at the boundaries by
meteorological fields from the RDPS (see section 4.3). Chemical species are initialized using the 12-h
forecast of the previous RAQDPS run. Climatological chemical profiles are applied at the lateral
boundaries.

Ozone and UV index forecasting
The Canadian Global model is used to prepare ozone and UV Index forecast at the 18 and 42 hour
projection times based on 00 UTC data and at the 30 hour projection time based on 12 UTC data
(Burrows et al., 1994). A Perfect Prog statistical method is used for forecasting total ozone, which is then
supplemented with an error-feedback procedure. UV Index is calculated from the corrected ozone
forecast. Correction factors have been added to take into account the snow albedo, elevation and
Brewer angle response.

Regional coupled atmosphere-ocean-ice forecasting
A fully-interactive coupled atmosphere-ocean-ice forecasting system for the Gulf of St. Lawrence (GSL)
has been implemented on 9 June 2011 at the Canadian Meteorological Centre (CMC). The strategy
includes three main parts: (1) An oceanic pseudo-analysis cycle, (2) A superimposed sea ice. The
forecast analysis based on direct insertion of Radarsat analysis and (3) a coupled forecast cycle. Note
that the coupled forecast cycle is a 48 hour simulation based on 00 UTC data where both of its models
(GEM-LAM and OCEAN-ICE) run at the same time. It provides 00-48 hour weather, sea ice and ocean
forecasts (Smith et al. 2012). In October 2012, the resolution of the system has been increased to 10 km.

Wave forecasting
The regional deterministic wave prediction system (RDWPS) is based on the dynamical wave model
WAM (WAve Model - version 4.5.1). RDWPS is configured to provide sea-state forecasts over the
following domains: Arctic, Eastern Pacific, and North Atlantic Oceans, the Gulf of St. Lawrence, and four
Great Lakes (Ontario, Erie, Huron and Superior). Each domain runs up to six times a day at (00 UTC, 06
UTC, 12 UTC, and 18 UTC). Four runs are forced by forecast winds from the Regional Deterministic
Prediction System. The other two runs are for long-range forecasts (120h) and are based on forecast
winds from the Global Deterministic Prediction System.
The resolution over the Arctic and Eastern Pacific Oceans is set to 0.5º. The North Atlantic Ocean is at a
resolution of 0.15º and a resolution of 0.05º is used over the Gulf of St. Lawrence and the Great Lakes.

Storm Surge forecasting in the Atlantic region
The Atlantic Storm Prediction Centre (ASPC) located in Halifax produces operational storm surge
forecasts over Eastern Canada using the barotropic version of Dalcoast developed at Dalhousie
University specifically for this region (Bernier et al. 2010). It is run twice daily (00 UTC and 12 UTC).
The storm surge model is driven with surface air pressure and winds from the Regional Deterministic
Prediction System (RDPS). The RDPS runs at a resolution of 0.08° and covers the North West Atlantic
Ocean and Gulf of St. Lawrence.

Environmental Emergency Response models
The CMC is able to provide real-time air concentrations and surface deposition estimates of airborne
pollutants. These fields are obtained from 3-D short / long range atmospheric
transport/dispersion/deposition Lagrangian Stochastic Particle Models named MLDP0, MLCD, MLDP1
and Trajectory model. Important applications from these models are the estimation of the concentrations
of radionuclides and volcanic ash. Based on this operational capability, the CMC is designated by the
WMO as a Regional Specialized Meteorological Centre (RSMC) with specialization in Atmospheric
Transport Modelling Products for Environmental Emergency Response. In addition, CMC is designated
by the International Civil Aviation Organization (ICAO) as a Volcanic Ash Advisory Centre (VAAC).
There has been an increased use of these operational atmospheric transport modelling tools to the
25
dispersion of chemical and biological agents in the context of the response to local environmental
emergencies.
The Lagrangian Particle Dispersion Models are "off-line" models. Therefore, fields of wind, moisture,
temperature and geopotential heights must be provided to them. These are obtained either from the
Global Deterministic Prediction System (GDPS) or Regional Deterministic Prediction System (RDPS)
forecasts and analyses. Please refer to the above sections 4.2 and 4.3 for more information on these
NWP products.
All Environmental Emergency Response (EER) models can be launched easily with a flexible Graphical
User Interface (GUI) called SPI. SPI has been under development for many years at CMC and allows
duty officers to respond efficiently to emergencies.
o
MLDP0 (Modèle Lagrangien de Dispersion de Particules d’ordre 0)
MLDP0 is a Lagrangian Particle Dispersion Model described in D’Amours and Malo, 2004, and
D’Amours et al. 2010. In this model, dispersion is estimated by calculating the trajectories of a very
large number of air particles (or parcels). The trajectory calculations are done in two parts: 3-D
displacements due to the transport by the synoptic component of the wind, then 3-D displacements
due to unresolved turbulent motions. Dry deposition is modelled in term of a deposition velocity. Wet
deposition will occur when a particle is presumed to be in a cloud. The tracer removal rate is
proportional to the local cloud fraction.
The source is controlled through a sophisticated emission scenario module which is a function of the
release rate of each radionuclide over time. For volcanic eruptions, a particle size distribution is used
to model the gravitational settling effects in the trajectory calculations according to Stokes’s law. The
total released mass can be estimated from an empirical formula derived by Sparks et al., 1997,
which is a function of particle density, plume height and effective emission duration. In MLDP0,
tracer concentrations at a given time and location are obtained by assuming that particles carry a
certain amount of tracer material. The concentrations are then obtained by calculating the average
residence time of the particles, during a given time period, within a given sampling volume, and
weighting it according to the material amount carried by the particle.
MLDP0 can be executed in forecast mode for up to day 10, using the operational GDPS products,
and up to 48 hours using the operational RDPS products. MLDP0 can also be executed in hind cast
mode, globally. MLDP0 can be executed in inverse (adjoint) mode. The model has been used
extensively in this configuration in the context of the WMO-CTBTO cooperation.
o
MLCD (Modèle Lagrangien à Courte Distance)
MLCD is a Lagrangian Particle Model described in details in Flesch et al. 2002. It is designed to
estimate air concentrations and surface depositions of pollutants for very short range (less than ~10
km from the source) emergency problems. As in MLDP0, this 3-D Lagrangian dispersion model
calculates the trajectories of a very large number of air particles. MLCD is a first order Lagrangian
Particle Dispersion Model because the trajectories of the particles are calculated from the velocities
increments, while MLDP0 is a zeroth order Lagrangian Particle Dispersion Model since the
trajectories of the parcels are updated from the displacements increments.
The Langevin Stochastic Equation is based on the turbulent components of the wind associated to
the turbulent kinetic energy (TKE). These fluctuating components, vertical and horizontal are
generated from a "user provided" set of wind observations (velocity and direction) time dependant
through a "two-layer" model (Flesch et al., 2004). MLCD can take into account the horizontal
diffusion for unresolved scales operating at time scales longer than those associated to the TKE.
The removal processes of radioactive decay, wet scavenging and dry deposition can also be
simulated by the model. MLCD can be run in forward or inverse mode.
o
MLDP1 (Modèle Lagrangien de Dispersion de Particules d’ordre 1)
A full 3-D first order Lagrangian Particle Model called MLDP1 has been implemented for short range
dispersion problems on horizontal domains of 100-200 km, with a time horizon of about 12 hours.
This stochastic dispersion model is well described in Flesch et al. 2004. The fluctuating components
of the turbulent wind are obtained by partitioning the TKE calculated in the driving NWP models.
MLDP1 is parallelized to run on several nodes on the supercomputer at CMC.
o
Trajectory Model
The trajectory model is a simple tool designed to calculate the trajectory of a few air parcels moving
in the 3-D wind field of the atmosphere. The model is described in D’Amours and Pagé, 2001. Only
26
transport by the winds is considered without taking into account any other physical or atmospheric
processes. The advection of an air parcel is computed according to a Runge-Kutta scheme.
The model estimates the trajectories of the parcels, originating from or arriving at the same
geographic location, but for different levels in the vertical. The location and levels are defined by the
user. The model can be run to obtain a quick estimate of the expected trajectory of an air parcel,
whose point of origin or point of arrival (back trajectory) is specified as the input parameter.
4.5.2.2

Research performed in this field
Air Quality Model
The operational GEM-MACH15 AQ forecast was a participant in an experimental ensemble prediction
system that was operated during the CalNex experiment in the summer of 2010. The CalNex
experiment focused on trans-Pacific transport to North America from Asia. The ensemble prediction
setup was lead by NOAA; other members of the ensemble forecast aside from GEM-MACH15 included 8
members (NOAA’s operational CMAQ/NAM 12km model, a research version of CMAQ/NAM utilizing
RAQMS lateral boundary conditions, BAMS 15km MAQSIP, BAMS 45km MAQSIP, BAMS 45km CMAQ,
WRF/CHEM 20km (with 2 real-time variations with and without RAQSM lateral boundary conditions)),
and GO-CART with/without data assimilation was run in retrospective mode. The study suggested that
there is considerable room for improvement in the models; while GEM-MACH had the highest rcoefficient score for ozone (0.69), none of the models beat the r-coefficient score of persistence across
all observation points and over time (0.81). GEM-MACH also had the highest r-coefficient for PM2.5 of
the models (0.72), while persistence r-coefficient was 0.89. The study highlighted several important
differences in the partitioning of emitted volatile organic compounds and speciated particulate matter,
between the models. Retrospective analysis of the study is still underway, and is expected to continue
throughout 2013
During the 2012 fire season an experimental setup for ingesting active wildfire emissions into GEMMACH was tested.. Daily 48h forecasts were prepared through the summer and made available for
internal evaluation. The prototype is one of the outcomes of the collaboration between Environment
Canada and the Canadian Forest Service. Development of the prototype will continue in 2013 and an
improved suite of experimental products will be available to forecasters for the summer of 2013.

2015 Pan-American Games High-Resolution Air Quality Modeling
This multi-disciplinary project has field monitoring and meteorological and air quality modelling
components. The air quality modelling component will develop, implement and evaluate a highresolution air quality model (GEM-MACH 2.5km) and a new objective analysis tool for making predictions
of the Air Quality Health Index (AQHI) at sporting venue locations across the Greater Toronto Area
(GTA). The comparison of the current operational predictions and the high-resolution predictions,
evaluated against an expanded network of AQHI measurements during summer 2014 and 2015 over the
GTA will provide a pilot opportunity to accelerate the rate at which high-resolution air quality forecasting
becomes operational for cities across Canada. Currently, improvements are being made to the pollutant
emissions processing for on-road vehicles using results from a link-based, traffic flow model for the GTA.

High resolution land surface prediction
Several new models and approaches are currently being examined to better predict surface or nearsurface conditions over land. An external land surface modelling system has been developed and is now
integrated at grid sizes much smaller than that of the atmospheric models. This increased resolution
allows better exploitation of geophysical information on orography, land use / land cover, and water
fractional coverage. Adaptation, or downscaling, of atmospheric forcing (precipitation, temperature,
humidity, winds) is used to more realistically drive the surface processes. The success of this approach
has been demonstrated in mountainous regions (as a prototype was prepared for the 2010 Vancouver
Winter Olympic Games). A first implementation of this system at CMC-Operations is expected in the
next year.
Improvement or inclusions of several aspects of land surface modelling are currently being tested. A
multi-budget multi-layer land surface scheme called the Soil and Vegetation Scheme (SVS) has been
designed, coded, and is now being tested for several applications. In addition, a new approach to the
coupling between the land surface canopy (including vegetation and cities) has been developed and is
also being tested. Other aspects of surface modelling that are being examined include the
representation of snow over sea-ice, blowing snow (using the PIEKTUK model), and urban modelling
with the Town Energy Balance (TEB) model. The very high resolution land surface system is being
proposed for several applications, including meteorological predictions in cities and surface hydrological
predictions (soil moisture, snow on ground, runoff and drainage).
27

Regional coupled atmosphere-ocean-ice systems
R&D is being performed on several regional coupled atmosphere-ocean-ice systems:
o
A fully-interactive coupled atmosphere-ocean-ice forecasting system for the Gulf of St. Lawrence
(GSL) has been implemented in 2011 at CMC. The oceanic component is the ocean-ice model of the
GSL at 5-km resolution developed at the Maurice-Lamontagne Institute. Work has been done to
replace that component by an adapted NEMO ocean model version. Investigation is also ongoing to
better understand the effect of the coupling on the low level atmosphere.
o
A similar system to the previous one is currently in development over the Great-Lakes using a 2km
resolution configuration of NEMO. The NEMO ocean model is currently under evaluation and initial
development work on a coupled GEM-NEMO configuration for the Great Lakes has begun.
o
Development and validation of an integrated marine Arctic prediction system. The objective is to
allow the expansion of the end-to-end analysis and forecasting system (with a few days lead time)
for production of marine information products beyond the Canadian Arctic into the international
waters of METNAV Areas XVII and XVIII.
The research efforts will focus on:
o
Development, validation and implementation of marine forecasts with lead times to 1-3 days based
on the EC regional high resolution coupled multi-component modeling (atmosphere, land, snow, ice,
ocean, wave) and data assimilation system
o
Development and validation of a sea ice analysis system
o
Development and validation of automated techniques for near-real-time utilization of various satellite
data types to improve surface wind estimation, extract sea ice features and detect and estimate the
presence of ice hazards (icebergs, ice islands)
The operational implementation of this system will be made in phases beginning with a stand-alone sea
ice prediction system. This will be followed by subsequent additions of coupling to an ocean component
and full coupling to the atmosphere.
The initial stand-alone ice prediction system used for the first phase of the METAREAs system is based
on the 3DVAR sea ice analysis system combined with the Los Alamos CICE v4.1 multi-category
dynamic/thermodynamic sea ice model. CICE also includes a simple mixed-layer ocean model in order
to allow growth/melt during the forecast integration. CICE is initialized with the 3DVAR sea ice
concentration analyses, CMC sea surface temperature analyses and a climatological ice thickness field.
48hr forecasts on a 5km grid are then produced by forcing the sea ice model with 10 km atmospheric
forecasts from Environment Canada’s Regional Deterministic Prediction System and by monthly
climatological ocean currents.
For a one-year period (2010), daily 48hr 5-km resolution ice forecasts have been validated against
Canadian Ice Service (CIS) daily ice charts, sea-ice concentration data derived from RADARSAT
measurements and products from the NSIDC Ice Mapping System. Overall, results show a significant
improvement of the forecasting skill as compared to persistence of the 3DVAR analysis. This system
has been run 4 times/day since August 2012. These forecasts form the basis of a collaborative
evaluation between RPN-E Section and CIS operations.

Global coupled atmosphere-ocean-ice system
Environment Canada (EC), Fisheries and Oceans Canada (DFO), and the Department of National
Defence (DND) are preparing an operational global coupled atmosphere-ocean-ice data assimilation and
prediction system that can ingest in-situ Argo float data and satellite observations such as sea surface
height and temperature.
In order to initialize coupled atmosphere-ocean-ice forecasts an ocean analysis is required. To this end,
the Mercator-Ocean assimilation system has been setup on EC computers. This system has been
running in R&D mode since December 2010 producing weekly analyses and 10 day ice-ocean forecasts
in real-time using the NEMO modeling system and the Mercator assimilation system. The Mercator data
assimilation system is a multi-variate reduced-order extended Kalman filter that assimilates sea level
anomaly, sea surface temperature (SST) and in situ temperature and salinity data. Ice fields are
initialized using Canadian Meteorological Centre (CMC) daily ice analyses. Research activities have
focused on evaluating the SST and sea ice forecasts.
A 2-way interactive coupling of the GEM and NEMO global models has been completed using a fluxcoupling approach exchanging fields via a TCP/IP socket server called GOSSIP. A series of daily 16day
forecasts has been produced over the winter 2011 period and is under evaluation.
28

Wave modeling
Wave modeling research is being conducted primarily under four projects:
o
METAREAS – “Provision of wave forecasts for METAREA XVII/XVIII in support of the Global
Maritime Safety and Distress System (GMDSS)”.
o
PERD (Program for Energy Research and Development) – “Improvements to Canada’s operational
ocean wave forecasting system in support of offshore activities along the East Coast and Northern
waters of Canada”.
o
SAR-NIF (A High-resolution Ensemble Prediction System for SAR)
o
PanAm Games – A forecast demonstration project for the 2015 Toronto Pan and Parapan American
Games
The projects’ objectives are:

o
Improvements to the physics of the wave model with the replacement of WAM by WaveWatch3
o
Development and implementation of a Global Deterministic Wave Prediction System for 0-5 day
forecasts.
o
Development and implementation of a Global Ensemble Wave Prediction System for 0-10 day
probabilistic forecasts. The integration of Canada's 20 members in a North American Ensemble is
being investigated under NAEFS. NOAA and the US NAVY currently each contribute 20 members
to the super ensemble (Alves et al. 2013).
o
Investigation of impacts of coupling with other systems such as the atmospheric, ocean, and ice
models on wave forecast skills and other component forecast skills.
o
Development and implementation of a Great Lakes Ensemble Wave Prediction System. The
combination of 20 members from Canada and 20 members from NCEP into a Great-Lakes super
ensemble via NAEFS is under investigation.
Storm Surge forecasting in Canada
Storm Surge modelling research is currently conducted to implement nationally an operational Storm
Surge model system called the Regional Deterministic Surge Prediction System (RDSPS). The first step
will be to implement an operational Storm Surge based on Dalcoast for the Atlantic coast of Canada.
Other dynamical models and the design of a near global forecast system that would allow for fewer
numerical domain under a full coverage of Canadian coastlines will be explored. The main short term
project objectives are to proceed to the implementation of an operational storm surge forecast system for
the Atlantic coast of Canada with:

o
Improved and increased forecast lead time to 144 hours (6 days)
o
1/30 degree grid spacing
o
Developing and implementing an Ensemble Surge Prediction System for Atlantic Canada to produce
probabilistic storm surge forecast for lead times up to 240 hours
Hydrological and water cycle modeling system
Development of an ensemble hydrological forecasting system is progressing, with efforts being focused
in the Great Lakes and St. Lawrence watershed.
Both short-term (2-3 days) and long-term (up to one month) forecasting systems are being assessed. It
remains difficult to show that higher resolution ensemble forecasts from the Regional Ensemble
Prediction System (REPS) provide better forecasts than the Global Ensemble Prediction System
(GEPS), partly because the period over which outputs from both the REPS and the GEPS are available
is very short. Reforecast experiments, in particular for high impact events, would be required. In all
cases, it is expected that statistical post-processing will be required in order to obtain reliable
probabilistic forecasts. A Bayesian methodology is currently being tested for updating a climatological
prior distribution based on ensemble forecasts of stream flow.

Experimental 2D Hydrodynamic forecasting system
Operational Hydrodynamic Simulation (OPHS) has been successfully transferred to an operational (as
an experimental system) status in May 2013. The OPHS system produces numerical analyses and
simulations of various parameters of interest for the St. Lawrence River and its tributaries. It produced
reliable currents and water levels that are used in numerous applications e.g. water quality modelling in
29
emergency responses. It is a brand new system that will be used as a valuable decision-making tool
regarding the integrated management of the St-Lawrence. In version 1.0.0 of OPHS, the analyses are
produced for the portion of the St. Lawrence between Montréal and Trois-Rivières. These analyses and
simulations help us better understand important details of the ecosystem associated with these
waterways and can assist those who make use of the St. Lawrence River system. In this version the
products that are available include images and ascii datasets. In the future, OPHS will produce an
analysis and a forecast of water flow, water temperature, ice movement and other applications. OPHS
will also be extended in short term to the entire St. Lawrence River and to other waterways as well.

Environmental Emergency Response models
Some corrections and improvements were made to the operational atmospheric dispersion models
(MLDP0, MLDP1 and MLCD). Work has started to merge the different models to produce a unified
dispersion model MLDPn. Work is ongoing to develop a transport and dispersion modeling capability to
address the problems of dispersion at the urban scale. The Canadian Urban Dispersion Modeling
(CUDM) system is a multi-scale system that aims to simulate the mean flow, turbulence and
concentration fields in urban areas. The system involves a cascade of meteorological models from the
operational regional model to an urbanized meso-scale model. Outputs from the urbanized
meteorological model are used to drive a Computational Fluid Dynamics (CFD) model running at the
urban scale. In turn, urbanSTREAM provides the high resolution wind and turbulence fields (5-15 m
resolution) to drive urbanLS, a Lagrangian stochastic particle trajectory model.
4.5.3 Specific products operationally available

Air quality
The operational output from the RAQDPS consists of hourly concentrations of surface tropospheric
ozone, PM2.5, PM10, nitrogen dioxide, nitrogen monoxide, and sulphur dioxide, as well as select
meteorological fields. The forecast of PM levels is based on primary PM emissions and the chemical
formation of secondary PM (sulphate, nitrate, ammonium and secondary organics). Post-processing is
performed on this output to provide users with maximum and mean forecasts of ozone, PM 2.5 and PM10
in the boundary layer per 6-hour forecast period. These products are available on the web at:
http://weather.gc.ca/aqfm/index_e.html
In 2012, the national Air Quality Health Index (AQHI) forecast program was expanded from 63 sites to 74
sites distributed across Canada. This program, which first began as a pilot project in 2007, provides a
means to communicate to the public the level of risk associated with exposure to a mixture of O3, PM2.5,
and NO2 pollution. Raw model output from the RAQDPS is corrected with the statistical post-processing
package UMOS-AQ and provided to forecasters along with processed pollutant observations to assist in
the preparation of AQHI forecasts. AQHI observations and forecasts are available for all 74 active sites
across Canada at:
http://www.ec.gc.ca/cas-aqhi/default.asp?lang=En&n=450C1129-1
General information on the AQHI is available at:
http://www.ec.gc.ca/cas-aqhi

Environmental Emergency Response
Upon receiving a request for a nuclear or radiological support from a WMO Member Country Delegated
Authority, the CMC (RSMC Montréal) will provide the following standard set of basic products:
o
Three dimensional trajectories starting at 500, 1500 and 3000 m above the ground, with particle
locations indicated at synoptic hours
o
Time integrated pollutant concentration within the 500 m layer above the ground in Bq/m3
o
Total deposition (wet and dry) in Bq/m 2 from the release time to the end of the third time period
The CMC can also provide charts of air concentration estimates for many levels in the atmosphere as
well as total surface deposition estimates at various time intervals.
Inverse (adjoint) modelling is provided upon request to support the activities of the Comprehensive Test
Ban Treaty Organization, as defined in the WMO Manual on the GDPFS (WMO TD 485).
The CMC is also designated as the Montréal Volcanic Ash Advisory Centre and provides modeling and
guidance over its area of responsibility in accordance with ICAO’s Annex 3 on the provision of
meteorological services.
30

Ozone forecast and UV index
Charts of the total ozone forecast and of the UV Index forecast are prepared and transmitted to the
Regional Offices. Bulletins giving the forecast UV Index at an ensemble of stations across Canada are
also generated and made available to the public.
4.5.4 Operational techniques for application of specialized numerical prediction products
(MOS, PPM, KF, Expert Systems, etc.) (as appropriate related to 4.5)
4.5.4.1
In operation
Model
Output
Statistics
(MOS)
Automated
computer
worded
forecast:
Scribe
Weather
element
matrices
4.5.4.2
An Updateable MOS system (Wilson and Vallée, 2001 and 2002) is used for the
statistical post-processing of the direct air-quality model outputs. This air-quality
post-processing system currently provides forecasts at more than 200 observation
sites for :
 Ground-level ozone concentrations at hourly intervals between 0 and 48-hour
projection times.
 Ground-level particulate matter (PM2.5) concentrations at hourly intervals
between 0 and 48-hour projection times.
 Ground-level nitrogen dioxide (NO2) concentrations at hourly intervals between
0 and 48-hour projection times.
 Equations are valid for the two daily runs (00, 12 UTC).
Before implementing a new version system, the statistics are updated using R&D
final tests.
The SCRIBE system (see section 4.3.4.1) is used to generate a set of automated
plain language air quality forecast products for days 1 and 2. See the following
section Weather element matrices.
Supplementary weather element matrices have been developed and implemented
for concentration of ozone near surface, PM2.5, NO2, as well as PM10, NO and
SO2. The time resolution of these matrices is 3 hours, with projection times out to
48 hours. The matrices for the first three (O3, PM2.5 and NO2) are obtained using a
statistical interpolation of the increments between model output statistics (MOS) at
observation sites and the air quality model predicted concentrations. For the
remaining pollutants (PM10, NO and SO2) , the matrices are obtained using the air
quality model predicted concentrations.
Research performed in this field
An improved linear statistical procedure using model, persistence and antecedent predictors (maxima and
minima observed the previous days) and an improved predictor selection procedure is being developed for
an experimental implementation.
31
4.6 Extended range forecasts (10 days to 30 days) (Models, Ensemble and
Methodology)
4.6.1 In operation
As of December 1st 2011, the monthly and seasonal forecast system has been upgraded. The new system
is based on 10-member ensembles of predictions made with two coupled atmosphere-ocean-land physical
climate models (for a total of 20 members), and produces predictions up to a year. For more information
about this new system see section 4.7.1. For the monthly predictions, surface air temperature forecasts are
issued at the beginning of the months and at mid-months.
The surface air temperature forecasts are made in doing first an average of the daily temperature as
predicted by each of the two model ensembles. The climatology of these model ensembles is then
subtracted from the mean forecast monthly temperatures to derive their respective forecast anomalies.
These anomalies are then normalized and combined using an arithmetic average. The surface air
temperature forecast anomalies are the anomalies of the mean daily temperature measured at the
Stevenson screen height (2 metres). Finally the anomalies are divided in three categories (above, near and
below the normal). Charts are produced, showing above normal, below normal and near normal temperature
categories.
4.6.2 Research performed in this field
Work has been done to develop a new monthly forecasting system. The impact of time-evolving SST on the
global EPS has been evaluated, and a suite for conducting historical forecasts spanning 18 years (19942012) has been developed.
Research work has been conducted to develop the Extreme Forecast Index (EFI) for the global EPS
forecasts. EFI products are produced for surface air temperature, 10-meter wind, and 1-day, 5-day and 10day accumulated precipitation. The EFI has been produced operationally since July 2012.
Research was performed on a multi-variable linear regression model based on the status of the MaddenJulian Oscillation (MJO) and persistence in order to forecast wintertime surface air temperature anomalies
over North America out to 4 pentads. Modest skill is found, largely centered over the eastern United States
and the Great Lakes. The model skill is seen to be highly dependent on the magnitude and phase of the
MJO as well. Theoretical work on the dynamics of intraseasonal variability was conducted in combination
with data analysis. Using global atmospheric instability theory, the mechanism of MJO-NAO connection was
investigated. In another study, a new index was developed to monitor and predict the intraseasonal
variability in the East Asian-Western North Pacific monsoon region, which has a global impact. A few other
process studies were performed on blocking climatology of the GEM model, tropical-extratropical
interactions, trends in seasonal hindcasts, and seasonal prediction of frost-killing frequency in middlesouthern Canada.
4.6.3 Operationally available EPS products
Deterministic forecast of monthly temperature anomaly is available on the Internet:
http://weatheroffice.ec.gc.ca/saisons/index_e.html
32
4.7 Long range forecasts (30 days up to 2 years) (Models, Ensemble and
Methodology)
4.7.1 In operation
On Thursday December 1 2011, the Canadian Meteorological Centre (CMC) has started using its newly
developed global coupled seasonal prediction system for forecasting monthly to multi-seasonal climate
conditions. This new system named CanSIPS for Canadian Seasonal to Interannual Prediction System
replaces both the uncoupled (2-tier) prediction system previously used for producing seasonal forecasts with
zero and one month lead times and the CCA statistical Prediction system previously used for forecasts of
lead times longer than four months. With CanSIPS, Environment Canada is now able to issue on monthly
basis predictions of seasonal climate conditions covering a full year. This represents substantive progress
with respect to the previous system. CanSIPS can also skilfully predict the ENSO phenomenon and its
influence on the climate up to a year in advance. The development and the implementation of this multiseasonal forecast system is the result of a close collaboration between CMC and the Canadian Centre for
Climate Modelling and Analysis (CCCma).
CanSIPS is a multi-model ensemble (MME) system based on two climate models developed by CCCma. It is
a fully coupled atmosphere-ocean-ice-land prediction system, integrated into the CMC operational prediction
suite and relying on the CMC data assimilation infrastructure for the atmospheric, sea surface temperature
(SST) and sea ice initial states. The two models used by CanSIPS are:

CanCM3
This model uses the atmospheric model CanAM3 (also known as AGCM3) with horizontal resolution of
about 315 km (t63) and 31 vertical levels, together with the ocean model CanOM4 with horizontal
resolution of about 100 km and 40 vertical levels and the CLASS land model. Sea ice dynamics and
thermodynamics are explicitly modelled.

CanCM4
This model uses the atmospheric model CanAM4 (also known as AGCM4) also with a horizontal
resolution of about 315 km (t63) but with 35 vertical levels. The CanOM4 ocean, CLASS land and sea
ice components are essentially the same as in CanCM3.
A detailed description of these models is provided in Merryfield et al. (2013a).
CanSIPS has two modes of operation:

Assimilation mode
CanSIPS uses a continuous assimilation cycle for 3D atmospheric temperatures, winds and specific
humidities as well as sea surface temperatures and sea ice. The assimilated data comes from the six
hour CMC 4D-VAR global atmospheric final analyses and the daily CMC SST and sea-ice analyses.
Additionally, just before launching the production of the forecasts, an NCEP 3D ocean analysis is
assimilated into the CanSIPS ocean model background state. All the 20 members’ initial conditions are
independent but statistically equivalent in the sense that their differences are of the same order than the
observation uncertainties.

Forecast mode
CanSIPS forecasts are based on a 10-member ensemble of forecasts produced with each of two
CCCma climate models for a total ensemble size of 20. Monthly to multi-seasonal forecasts extending to
12 months are issued the first day of each month. Additionally, a one-month forecast is issued at midmonth (15th). Deterministic and probabilistic forecasts for surface temperature and precipitation are
produced for each category (above/near/below normal) for seasons made of months 1-3, 2-4, 4-6, 7-9,
10-12. The probabilistic forecasts are done by counting members in each of the three possible forecast
categories: below normal, near normal and above normal. The probabilistic forecasts are not calibrated
yet but a reliability diagram with error bars is provided with each forecast.
CanSIPS climatology is based on a hindcast period covering 1981-2010 and was produced during phase 2
of the Coupled Historical Forecast Project (CHFP2) research effort. The ensemble size (20) is the same for
the forecast and the hindcasts.
More technical information on CanSIPS is available in Merryfield et al., 2011.
33
4.7.2 Research performed in this field

Impact of model biases on seasonal forecast skill
The influence of the quality of a model’s representation of climate on its ability to predict seasonal
climate anomalies was investigated by Kharin and Scinocca (2012) through application of a novel
procedure for correcting a model’s seasonally-varying biases. They found that, in addition to reducing
these biases, such a procedure improves the model’s representation of seasonal climate variability, as
well as its skill in predicting seasonal climate anomalies.

Stratospheric contribution to skill
The seminal work of Baldwin and Dunkerton (2001) established that large anomalies in stratospheric
circulation, occurring primarily in winter, gradually propagate downward, suggesting the existence of a
potentially predictable influence on surface climate up to two months or more following such an event.
That this source of predictability can be realized in model-based predictions was demonstrated by
Sigmond et al. (2013b) through a series of retrospective ensemble forecasts using the Canadian Middle
Atmosphere Model. Specifically, forecasts initialized immediately following sudden stratospheric warming
(SSW) events exhibited statistically significant skill in predicting the Northern Annular Mode, as well as
surface temperatures in northeastern North America and Northern Eurasia, over the 16-60 days
following these events, whereas a control set of forecasts initialized in the absence of SSWs did not.

Seasonal predictions of Arctic sea ice
While coupled models that include a sea ice component are increasingly being applied to seasonal
prediction, two Environment Canada studies are among the first to examine the ability of such models to
predict future sea ice conditions. Sigmond et al. (2013a) looked at historical forecasts of Arctic sea ice
area by CanSIPS, and found that the skill of such predictions in most seasons attributable mainly to long
term trends except in the first 2-3 months of the forecasts. Merryfield et al. (2013b) considered multimodel predictions obtained by combining forecasts from CanSIPS and the NCEP CFSv2 systems, and
found the combined predictions to be more skilful that those of either system according to a broad range
of measures.

Development of a 1-tier GEM version
Work continues on a coupled version of the GEM NWP forecast system, which will become part of the
coupled seasonal forecast system when finalized.

MJO impact on North American temperature
A multi-variable linear regression model was developed based on the status of the MJO and persistence
in order to forecast wintertime surface air temperature anomalies over North America out to 4 pentads
(Rodney et al. 2013). Modest skill is found, largely centred over the eastern United States and the Great
Lakes. The model skill is seen to be highly dependent on the magnitude and phase of the MJO as well.

Blocking climatology in GEM
The performance of the GEM model in reproducing atmospheric low-frequency variability was evaluated
in the context of Northern Hemisphere blocking climatology. The validation was conducted by applying a
comprehensive but relatively simple blocking detection algorithm to a 20-year integration of GEM in
climate mode (Dunn-Sigouin et al. 2013).
4.7.3 Operationally available products
Deterministic and probabilistic products of seasonal forecast are available on the Internet:
http://weatheroffice.ec.gc.ca/saisons/index_e.html
Probabilistic products of seasonal forecasts are available with zero and 1 month lead time. Deterministic
products of seasonal forecasts are available with 0, 1, 3, 6 and 9 months lead time. These forecasts are for
three months periods and are issued on the first day of each month.
Charts and model output grids for the season 1 are available in real time on Internet at the following site:
http://meteo.gc.ca/grib/grib2_cansips_e.html
The forecast digital data are on a 2.5 degrees grid in GRIB2 format. Monthly means of surface air
temperature, precipitation, 500 hPb heights, 850 hPa temperature and mean sea level pressure are available
for each of the 20 models runs used to prepare the official forecast. Also, hindcast data as well as their
climatological averages are available for each model.
34
5. Verification of prognostic products
5.1 Annual verification summary
Objective verification of the operational numerical models is carried out continuously at the CMC. CMC
participates in a monthly exchange of NWP verification data following WMO/CBS recommendations originally
implemented in 1987. The table on the following page is a summary of the CMC verification scores for 2012
according to the recommended format. This is a subset of scores exchanged with the other participating
NWP centres.
The current standards for the exchange were established in 1998. The WMO/CBS Coordination Group on
Forecast Verification has updated the standards to include mandatory use of a standardized list of
radiosonde observations for verification, a standardized interpolation method, an updated climatology for the
calculation of anomaly correlation and the addition of new scores (mean absolute error, root mean square
error of the forecast and analysis anomalies, standard deviation of the forecast and analysis fields).
Verification data based on the new standard is already available for some of the participating NWP producing
Centres on the web-site of the Lead Centre-Deterministic NWP Verification (ECMWF):
http://apps.ecmwf.int/wmolcdnv. Work toward implementing the new standards is progressing at CMC, and
the data should be available from the LC-DNV web-site in 2013.
Verification summary - 2012
Canadian Meteorological Centre
Global Environmental Multi-scale (GEM) Model (33 km, L80)
Verification against analysis
Area
N. Hemisphere
Tropics
S. Hemisphere
Network
N. America
Europe
Asia
Australia - N.Z.
Tropics
N. Hemisphere
S. Hemisphere
Parameters
T+24h
T+72h
T+120h
00UTC
12UTC
00UTC
12UTC
00UTC
12UTC
8.2
3.8
2.6
4.2
10.4
4.0
8.6
3.8
2.5
4.2
10.3
4.0
23.6
8.3
3.8
6.6
30.0
8.9
23.8
8.3
3.7
6.7
30.6
9.0
47.0
13.5
4.6
8.3
58.1
14.5
47.2
13.5
4.5
8.4
59.1
14.6
Verification against radiosondes
T+24h
T+72h
00UTC
12UTC
00UTC
RMSE (m), GZ 500 hPa
9.4
9.8
23.4
RMSVE (m/s), Wind, 250 hPa
6.0
5.9
10.3
RMSE (m), GZ 500 hPa
12.3
11.6
26.2
RMSVE (m/s), Wind, 250 hPa
5.5
5.2
9.8
RMSE (m), GZ 500 hPa
13.4
13.4
23.5
RMSVE (m/s), Wind, 250 hPa
5.9
5.9
9.3
RMSE (m), GZ 500 hPa
10.6
12.4
17.8
RMSVE (m/s), Wind, 250 hPa
5.6
5.8
8.5
RMSVE (m/s), Wind, 850 hPa
4.2
4.1
4.9
RMSVE (m/s), Wind, 250 hPa
5.8
5.8
7.3
RMSE (m), GZ 500 hPa
12.8
12.6
25.9
RMSVE (m/s), Wind, 250 hPa
5.6
5.6
9.6
RMSE (m), GZ 500 hPa
12.8
14.3
24.2
RMSVE (m/s), Wind, 250 hPa
5.8
6.0
9.3
12UTC
24.3
10.5
24.6
9.6
23.1
9.3
21.2
8.8
4.7
7.5
25.7
9.6
27.6
9.5
T+120 h
00UTC
44.5
15.9
51.8
16.0
40.2
13.1
32.3
12.7
5.5
8.6
48.4
14.7
43.2
13.9
12UTC
46.5
16.0
50.1
15.8
39.3
13.0
40.0
12.9
5.3
8.8
48.5
14.7
49.8
14.5
RMSE (m), GZ, 500 hPa
RMSVE (m/s), Wind, 250 hPa
RMSVE (m/s), Wind, 850 hPa
RMSVE (m/s), Wind, 250 hPa
RMSE (m), GZ 500 hPa
RMSVE (m/s), Wind, 250 hPa
Parameters
5.2 Research performed in this field

There has been on ongoing focus on verification of surface and weather element parameters, including
temperature, dew-points, winds and cloud cover

There was also ongoing work in the area of verification of ensemble prediction and high-resolution
models

Exploration of the use of a relational data base for verification purposes as a tool to manipulate
observation and forecast data efficiently and to make conditional verification results

Statistical tool for precipitation verification against GPCP analyses

Algorithm for detection, tracking and verification of tropical cyclones
35
6. Plans for the future (next 4 years)
6.1 Development of the GDPFS
6.1.1 Major changes in the operational DPFS which are expected in the next year

Computer systems
The existing high performance, networked, parallel file system, as described in section 2, will have the
server portion replaced by 14 new IBM blades. The storage capacity available will also increase by
800TB.
The mass storage system, as described in section 2, will have the server portion replaced by Quantum
meta data controllers.




Global Deterministic Prediction System
o
Grid spacing from 33 to 25 km
o
Introduction of a hysteresis effect in the planetary boundary layer scheme
o
Flow-dependent bulk coefficient for low-level orographic blocking scheme
o
Use of Charney-Phillips vertical staggering
o
Increased resolution of 4D-Var inner loop (240x120 to 400x200 grid points)
o
Increased number of time bins (9 to 21) in 4D-Var cycles
Global Ensemble Prediction System
o
Localization length scale in the EnKF is height-dependent
o
EnKF horizontal grid from 400x200 to a 600x300
o
EnKF number of vertical levels from 58 to 74
o
Reduced satellite data thinning (2.7 times more observations)
o
Improved bias correction of observations
o
Refinement to stochastic physics perturbations in the case of strong vertical motion (convective and
orographically-induced)
o
Hysteresis effect in planetary boundary layer scheme
o
Flow-dependent bulk coefficient for low-level orographic blocking
o
Multi-parameter approach for bulk drag coefficient in the orographic blocking scheme
o
A single surface scheme is used instead of two
o
Forecast extended to 32 days once a week
o
A reforecast procedure will be used to provide model climate statistics
o
Implementation of Extreme Forecast Index (EFI) based on the Global EPS
Regional Ensemble Prediction System
o
Horizontal grid spacing from 33 to 15 km
o
Refinement of stochastic physics perturbations in the case of strong vertical motion (convective and
orographically-induced)
o
Hysteresis effect in planetary boundary layer scheme
o
Flow-dependent bulk coefficient for low-level orographic blocking
High Resolution Deterministic Prediction System
The domain covering Western Canada will become officially operational. A 0-42 hours forecast will be
produced twice a day

Regional Deterministic Air Quality Prediction System
36



o
Inclusion of wet deposition due to precipitation, in-cloud and below-cloud scavenging from the KFC
algorithm for convection
o
Modifications in the treatment of minimal and maximal vertical diffusion within and above the
boundary layer, respectively
Nowcasting system
o
Making use of composite radar Precip-ET (product of the Unified Radar Processing) at 4 km
resolution applying 2 additional cleaning algorithms to radar products
o
Introduce new version at 4 km resolution of McGill Algorithm Precipitation Lagrangian Extrapolation
(MAPLE). The observed radar reflectivity and lightning are projected during the next 6 hours with a
vector motion calculated from observed imageries 20 minutes apart
o
New sampling of quantity of precipitation forecast from RDPS to generate probability of precipitation
o
Improve rules for probability of precipitation nowcasting
o
Update Nowcasting dictionary
Wave forecast system
o
Implementation of Canada’s first Global Deterministic Wave Prediction System
o
First system implemented at EC based on WaveWatchIII (WW3)
o
Implementation of Canada’s first Global Ensemble Wave Prediction System, also based on WW3
Post-processing system (SPOOKI)
With the rapid evolution of models, grid changes, scales changes, etc., a highly adaptable postprocessing tool is needed in order to provide products to forecasters in an efficient and timely manner.
SPOOKI (Object oriented production system containing a myriad of information) is based on a modular
“plug-in” approach where each plug-in component is specialized, reusable and autonomous. These
object oriented programming characteristics simplify the maintenance of the system. A particular
attention is also given to create a user-friendly system for novice users.
There are 2 main components to the system:
o
the heart, spooki_run, which contains 3 libraries and an executable program
o
the plug-ins, which is a series of dynamic modules
The general concepts behind the system are:
o
decoupling and encapsulation (tasks separated in specialised re-usable modules and autonomous
operations)
o
multi-directional communications (capacity of the modules to share information through a common
interface)
o
efficiency (allow previous operations and read data to be used by new calculations – avoiding
unnecessary recalculations or data reloading)
The development of plug-ins under SPOOKI allows for collaborative development by various groups
working in both traditional weather forecasting and specialised forecasts. The advantages of the system
are:
o
reduce code generation
o
early detection of bugs
o
rapid prototyping
o
faster transition from development to operations
In the next year we will start producing a few bulletins using SPOOKI.

Monthly and seasonal forecast system
o
http://weather.gc.ca/saisons/index_e.html will be updated to display ‘all-in-one’ calibrated maps
o
CanSIPS assimilation cycle will ingest the new CMC GDPS 25 km analysis
o
CanSIPS assimilation cycle will use the new CMC sea-surface temperature analysis for nudging
37
o
CanSIPS assimilation cycle will use the new CMC sea-ice analysis for nudging
o
A bias correction for the soil moisture will be implemented as part of the assimilation cycle of
CanSIPS. This will correct the too low precipitations and too high temperatures especially seen in
summer.
o
Monthly forecast will be based on the extended GEPS (0-32 days) instead of CanSIPS
38
6.1.2 Major changes in the operational DPFS which are envisaged in the next 4 years



Computer Systems
o
A complete overhaul of the archiving system is forecasted for 2014-2015
o
Replacement of the Power7 supercomputer in 2014-2015
Data assimilated for both the GDPS and the RDPS
o
Increased vertical resolution of data from radiosondes and aircraft, combined with the 4D balloon
position available from radiosonde reports in BUFR code will be incorporated into the operational
analyses
o
An improved bias correction scheme for temperature and humidity observations for radiosondes will
be developed.
o
Modification to the assimilation of radiosonde humidity observations and modification to the vertical
interpolator used for RTTOV
o
Bias correction scheme for aircraft data
o
The assimilation of near-surface observations from land stations, ships and buoys will be revised
o
The variational assimilation system will be adapted to the Charney-Phillips vertical staggered
coordinate
o
The assimilation of ozone sensitive channels from AIRS and IASI is envisioned, coupled with an
ozone analysis replacing climatology
o
Assimilation of CrIS and ATMS radiances from NPP. Implement revision of observation error for all
radiances. Revise horizontal thinning of radiance data
o
Develop assimilation of surface sensitive channels over land from infrared and/or microwave
radiances
o
Implementation of the Canadian Land Data Assimilation System (CaLDAS)
o
Data from new satellites will be assimilated: NPP will provide data from CrIS and ATMS, and MetopB data will also be assimilated. Data from AMSR-2 on GCOM-W will be used (SST, Ice) as work will
be initiated to assimilate data from ADM-Aeolus once available
o
Implement additional CSR data for GOES E/W and MTSAT-2 from F17 and F18 additional AIRS and
IASI channels, and AVHRR polar winds; modify the bias correction predictors in the infrared as
originally planned
4D-Var to En-Var for both the GDPS and the RDPS
An En-Var approach to data assimilation is under investigation and should replace the current 4D-Var
approach. The forecast error covariance matrix will be computed from the ensemble of short range
forecasts of the EnKF system. This will be used for both the Global deterministic system and the
Regional deterministic system.



Dynamical cores
o
The vertical discretization for all GEM models will use the Charney-Phillips vertical staggering
o
Implementation of the Semi-Lagrangian Inherently Conserving and Efficient (SLICE) approach is
being studied for tracer transport (see Mahidjiba et al., 2008)
Physical parameterizations
o
Reduction of cold bias near the tropopause
o
Addition of thermal tendencies to the gravity-wave drag scheme
o
Improved treatment of cloud/radiation feedbacks
Changes specific to the Global Deterministic Prediction System
o
Decrease the horizontal grid spacing to 10 km
o
Implementation of the Yin-Yang horizontal discretization to eliminate the pole problems
39





o
A high-resolution external land surface prediction system will be coupled with the atmospheric
component
o
Coupling of atmospheric component with ice-ocean forecasting model
o
The deterministic forecasts will be considered as a high-resolution control member of an ensemble
system
o
Increased vertical resolution in the lower troposphere
o
Surface fields will come from the CaLDAS system
Changes specific to the Regional Deterministic Prediction System
o
New vertical discretization: Charney-Phillips
o
Model top piloting
o
Vertical resolution increase in the lower troposphere
o
Inclusion of an adapted version of the Milbrandt-Yau 2-moment microphysic scheme
o
Reformulated boundary layer parameterization
o
Minor changes to physics package
o
Eventually this system will be replaced by the National LAM 2.5 km resolution over Canada
o
Coupling with ice-ocean model
Global EPS assimilation
o
The data thinning will be reduced
o
The sequential algorithm will be modified to permit the assimilation of more data
o
The horizontal resolution will increase from 100 km to 45 km
o
The number of level will increase from 58 to 74 (top at 2 hPa)
Global EPS forecast
o
Implementation of the Yin-Yang horizontal discretization to eliminate the pole problems
o
Grid spacing decreased to 30-40 km
o
Model lid from 2 to 0.1 hPa
o
Stochastic physics will likely replace the multi-parameterization approach.
o
Land surface fields will come from the CaLDAS system
o
Coupling of atmospheric component with ice-ocean forecasting model
o
Stochastic representation of surface model errors
o
In general, improvements to the deterministic system will be adapted to the global EPS
Regional EPS
o
The horizontal grid spacing at 10 km
o
Increased vertical resolution, especially in the PBL
o
A multiple En-Var approach for initial conditions
o
Stochastic representation of surface model errors
High Resolution Deterministic Prediction System
o
Increase the number of vertical levels from 58 to 72, with emphasis on the PBL
o
Replace multi-grid system with a single grid (2.5-km horizontal grid spacing) covering all of Canada
and the Arctic (with 2-4 runs per day, 36/42 h). Will have its own assimilation system in 2015
o
Implement upper-boundary nesting from the piloting model (the RDPS)
o
Changes to PBL scheme (consistent with changes to RDPS)
40




o
Modified microphysics scheme (prognostic graupel density, improved aerosol treatment for droplet
nucleation, prognostic snow-liquid ratio)
o
Initial land-surface fields from high-resolution land-surface scheme (CaLDAS)
o
Evolving ice and SST from marine forecasting system
Seasonal to multi-seasonal forecast system
o
http://weather.gc.ca/saisons/index_e.html will be significantly upgraded
o
From 20 to 40 members in the current CanSIPS (CanGMC3 & CanGCM4) configuration
o
Implement a land surface data assimilation based on CalDAS.
o
Implementation of a coupled ocean-atmosphere ensemble forecast system for multi-seasonal
forecasts based on the GEM-NEMO modeling system. It is planned to add 10 members of this GEMNEMO configuration to CANSIPS
o
New forecast products for SST, Niño indices, sea ice, etc
Nowcasting System
o
Replace PubTools by Integrated Weighted Model (INTW) for some of the weather elements
o
Make use of Limited Area Model West Window at 2.5km
o
Migrate from a point based system to area based in the Next Generation Forecast System
o
Making a meso-analysis of precipitation types, convection and cloud cover from metar data and
refining this analysis with radar, satellite and lightning observations
Air Quality system
o
Migration of the RAQDPS to GEMv4 which includes the Charney-Phillips vertical staggering
coordinate.
o
Introduction of a global chemical deterministic prediction system based on the GDPS code for
producing ozone analyses and forecasts as well as improved global UV index forecasts
o
Use of a global air quality deterministic prediction system for piloting the RAQDPS meteorological
and chemical variables to extend model lead time from 2 to 3 days
o
Addition of UV index forecasting as part of the RAQDPS
o
Making use of surface ozone analyses as initial conditions within the RAQDPS
o
Making use of daily remote sensing information on wildfires to include their emissions in the
operational RAQDPS
Wave forecast system
o
Migration of all wave forecast systems from WAM to WW3
o
Improvements of the Wave model physics
- Investigation of improved communication for massively parallel processing
- Introduction of wave-ice interaction
- Additional source terms and new constraints on some wave and ice parameters
- Investigation of wave-current interaction
- Investigation of data assimilation
o
Development and implementation of a Global Deterministic Wave Prediction System for 0-10 day
forecasts.
o
Development and implementation of a Global Ensemble Wave Prediction System for 0-10 day
probabilistic forecasts. The integration of Canada's 20 members in a North American Ensemble is
being investigated under NAEFS. NOAA and the US NAVY currently contribute 20 members each
to the super ensemble.
o
Investigation of impacts and development where appropriate of coupling with other systems such as
the atmospheric, ocean, and ice models on wave forecast skills and other component forecast skills.
41
o

Development and implementation of a Great Lakes Ensemble Wave Prediction System. The
combination of 20 members from Canada and 20 members from NCEP into a Great-Lakes super
ensemble via NAEFS is under investigation.
Storm surge forecast system
o
Improvements of the Surge model core
- Introduction of non-linear terms
- Increase in resolution
- Introduction of tide-wave-surge interactions


o
Development and Implementation of a Global Ensemble Prediction System for 0-10 days
o
Extension of the deterministic forecast lead time to 144h (6 days)
Changes or new implementation of other operational systems
o
Implementation of North Atlantic regional analysis and forecast (sea ice)
o
Implementation of a global ocean-ice data assimilation system (based on Mercator system)
o
Implementation of a fully global atmos-ocean-ice forecast system
o
Implementation of an external high-resolution (100 m) land surface prediction system
o
Implementation of a hydrological forecasting system for the Great-Lakes and St. Lawrence basins
o
Implementation of a WAM EPS
Post-processing system (SPOOKI)
All the post processing production converted to SPOOKI (see section 6.1.1 for a description of the
system).


o
All the post production from the Regional Deterministic Prediction System (summer sever weather,
winter sever weather and aviation packages)
o
All of the post production of the Global Deterministic Prediction System
o
New products from the High Resolution Deterministic Prediction System and Ensemble forecasting
developed with the SPOOKI framework
Radar QC/QPE Project
o
The technology development prototype for the antenna mounted radar receiver and data transfer to
the signal processor which are required for dual polarization upgrade of the radar network has been
implemented on the S & T Branch research radar at King City. The adaptation of this new technology
to the Environment Canada radar modernization project will be done
o
The upgrade of 10 of the radars in the EC network to add dual polarization capability will be
completed
o
Technical transfer of radar processing software cycle 2 for severe weather detection and
extrapolation
o
Technical transfer of new data quality control algorithms for radar data (software cycle 3) which
include dual-polarization applications
Canadian HUB Airport Nowcasting System (CAN-Now)
o
Implementation of an automated first guess TAF production system which has a forecaster in the
loop methodology and which is based on climatology, conditional climatology, nowcasting output and
model output
o
Technical transfer of new algorithms for ceiling and visibility prediction
o
Technical transfer of algorithms for wind gusts and runway visual range
o
Development of enhanced applications for Pearson nowcasting products to TAF production
o
Specification of TAF production requirements for the NinJo consortium
o
Technical transfer of latest lightning forecasting algorithm to CMC (see Burrows et al., 2005)
42

Fog Remote Sensing And Modeling Project (FRAM)
o
Technical transfer of new fog and icing detection algorithms for GOES satellites to the operational
workstation
o
Technical transfer to CMC of new visibility algorithm for application in NWP models
o
Technical transfer to CMC for ice fog and visibility predictions over northern regions
o
Innovations on instruments development
43
6.2 Planned research Activities in NWP, Nowcasting, Long-range Forecasting and
Specialized Numerical Predictions
6.2.1 Planned Research Activities in NWP
6.2.1.1


Data assimilation
Global ensemble Kalman filter
o
Radiance observations from satellites can have significant spatial and temporal error correlations.
These correlations can be simulated in the context of the ensemble Kalman filter.
o
Initialization is currently performed using a digital filter finalization. The use of a more selective filter
will be investigated.
o
It will be attempted to reduce model spin-up in the data-assimilation cycle. This will likely involve a
closer link between the model and the assimilation components.
Regional (continental) EnVar assimilation (RENVAR)
Since Fall 2012, a Regional Ensemble Variational Analysis method has been developed for the RDPS
10 km system. The source of ensemble forecasts comes from the global EnKF system as normally used
by the global EnVar system except the innovations comes from a short term GEM-10 km forecast. This
analysis procedure is a first step towards fully merging an upcoming Regional GEM-2.5 km forecasting
system (different forecast domain then the RDPS) with a truly regional high-resolution EnKF analysis at
10 km (referred to as RENKF-10). RENKF-15 km is currently under development which embed the
GEM-2.5 km domain and more suited to improve also tropical storm tracking and mid-latitude transitions.
Results with the RENVAR system based on extensive objective verifications are as good as the
operational 4D-Var system except slightly better performance during summer over the United States but
which comes however with a significant deterioration of 0-24h accumulated precipitation bias. Once
finalized, plans are to transfer the first-stage RENVAR system to operations during summer 2014.

Local LAMs
In the context of the recently launched Canadian experts research-team project called “Marine
Environmental Observation Prediction and Response Network” (meopar.ca), a relocatable, mesoscale,
limited-area Ensemble Kalman Filter Data Assimilation (EnKF) and Forecasting system is currently
under development for research purposes. The first-stage atmospheric component of this system
(referred to as HREnKF), itself driven by a Regional EnKF 15-km system, will be running at 2.5 km
horizontal resolution over two designated domains basically corresponding to the soon to be operational
2.5 km forecasting regions over West Coast and East Coast. The first research version planned to be
running by the end of 2014, will be assimilating more conventional data (as compared to current lower
resolution operational analysis systems at CMC) and will also have the ability to assimilate radar data
such as radial winds and reflectivities. Short-term 12-h ensemble forecasts at 2.5 km will be produced
twice a day. Atmosphere-Ocean coupling with that system should take place during year 2015. Full
relocatability over Canadian regions is planed during a 5-year development plan.

GPS (radio occultation reflections)
A project on the analysis of sea surface reflections observed during GPS radio occultation was
continued. The objective is to extract geophysical information relevant to the low troposphere, with
potential applications to meteorological data assimilation. An algorithm for the extraction of
supplementary information on the refractivity in the lower troposphere was proposed (Boniface et al,
2011), and implemented through an extended bending angle operator. Quantification of the benefit
obtained from this information (signals have suffered refraction and reflection), above that obtained from
standard processing (signals have only refracted) is underway.

Microwave and infrared vertical sounders surface channels radiance assimilation
New surface emissivity databases are now available with RTTOV-10 in the microwave and infrared
domain as well as revised emissivity modeling over ocean for microwave radiances. In the infrared
region of the spectrum, the work on the assimilation of surface sensitive channels over land has been
initiated with the arrival of a postdoc in May 2013. The selection of channels for CrIS is done, and
formats were tested with simulated data. NPOES ATMS (microwave) will also be assimilated
operationally. Work on error correlation associated with AMSU data was carried out in the context of
ENKF. The resource associated with this work is leaving, but the project will be pursued.
44

Cloudy infrared radiance
A methodology to assimilate cloudy infrared radiances was demonstrated in a 1D-var context and is well
adapted to hyperspectral sensors such as AIRS and IASI (Heilliette and Garand, 2007). This
methodology was implemented in the 4D-var assimilation system, with cloud parameters estimated
within the global minimization problem. That work will be pursued in the context of EnVar.

Data assimilation for climate applications
In 2013, EC-CAS (Carbon Assimilation System) development will continue with migration to the latest
GEM version (GEM-MACH v2.0), implementation of convective transport of tracers, and a conservative
semi-Lagrangian advection scheme. Adaptation of the ensemble Kalman Filter (GEPS) for flux
estimation will also begin. Extension of the EC-CAS for global methane simulation with validation
against surface observations and CarbonTracker-methane will be done. Development of CO 2 prior fluxes
for EC-CAS will include implementation of CTEM (used in CRD’s Earth System Model) terrestrial
biospheric CO2 fluxes and the improvement of these fluxes by the assimilation of measurements of
vegetation properties. Updates and refinements to anthropogenic CO 2 fluxes will also be implemented.
A detailed analysis of the biases between GOSAT CO 2 data and the model simulations will be carried
out in preparation for assimilating the satellite data.

Chemical data assimilation
Past efforts in chemical data analysis and assimilation are moving towards an operational
implementation.
The chemical analysis system of surface pollutants (ozone and PM2.5) has been approved to become
operational and this should be done soon. The chemical analysis has a relatively high spatio-temporal
resolution (10 km) and is issued every hour. It is based on a modified optimal interpolation algorithm and
combines on an hourly basis, chemical trial fields from GEM-MACH and surface observations from
AirNow US/EPA database ( ~ 1100 sites for ozone and ~ 600 for PM2.5) and most of land based
Canadian surface stations including CAPMON and NAPS networks (~ 200 sites). Research and
development continues on how to improve the analyses (higher resolution, e.g. 2.5 km resolution,
adaptive bias correction scheme and error statistics) and extend to other pollutants (NO2, SO2, PM10)
the existing analyses. Recent experiments which assessed the impact of surface ozone analyses on
forecast show a positive impact up to 24 hours. An investigation on the use of satellite observations of
AOD will also be undertaken for the next two years, and is likely to become operational sometime
afterwards. On the other hand, multi-year objective analyses for ozone (period 2002-2012) and for
PM2.5 (period 2004-2012) have been produced hourly and are freely available to the scientific
community with a paper being submitted to ACP (Robichaud and Menard, 2013). Application of the
multi-year analysis is intended to support epidemiological studies, to produce a climatology of surface
pollutants in North America and to compute long-term trends. In the short term, we plan to produce an
objective analysis of NO2 and of the AQHI, and to produce a high resolution (2.5 km) analysis for the
PanAm games.
The stratospheric ozone assimilation and UV index forecasting effort will proceed on two fronts. One is
the application of the current global system for near-real time (but not operational) assimilation and
forecasting as of the Fall 2013. The other is migration of the existing global chemical data assimilation
capability (3D-Var Chem) to the new EnVar/EnsKF assimilation framework developed for NWP.
Operational ozone data for assimilation includes SBUV/2 partial column and GOME-2 total column
retrievals, and IASI/AIRS ozone-sensitive radiance channels. The chemical data assimilation system will
eventually move to include other radiative constituents such as methane and aerosols optical depth
measurements. The global system will deliver global ozone and UV index forecasts, and will be used for
providing upper and lateral boundary condition to the regional GEM-MACH forecasts.
The development plan for air quality chemical data assimilation at Environment Canada for the next five
to ten years is to develop an integrated regional-global chemical data assimilation using GEM-MACH
and the EnVar assimilation environment. In parallel, the EnsKF regional-global chemical data
assimilation capability will also be developed by the climate and air quality research divisions with initial
applications for carbon assimilation and source estimation by the climate research division. For air
quality applications, the ensemble Kalman filter capability will be minimal at first for the regional system,
whereas the global assimilation system shall be fully developed using the simplified linear chemistry of
LINOZ (McLinden et al. 2000). Research will continue on the development of a regional chemical data
assimilation combining surface and satellite observations, and developing computationally feasible
ensemble chemical forecasts.

PCW mission
45
CSA is proposing, with EC and DND as main partners, the Polar Communications and Weather mission
in 2018 time frame. This mission will provide seamless communications (Ka and X band, possibly UHF)
and satellite imagery over the entire region 50-90 N from two satellites in HEO (High Elliptical Orbit)
orbits. Phase A was completed in 2012, and provides a comprehensive description of the various
components of the mission. Orbit related trade-offs were studied (Trishchenko and Garand, 2011,
Trishchenko et al, 2011). Another study evaluated the advantage of the PCW satellite system in
comparison to existing constellation of low orbiting satellites (Trishchenko and Garand, 2012). The
Science Team is now proposing a 16-h orbit characterized by three apogees, which minimizes radiation
hazards substantially in comparison to the 12-h orbit Molniya originally proposed. The main
meteorological payload is an advanced 20-channel imager similar to those planned for the next
generation of GEO satellites (MTG, GOES-R). Phase-A studies to evaluate other potential payloads
(referred to as PHEOS) such as a UV-VIS-NIR imager and a hyperpectral infrared sounder were
completed. Research activities at EC focus on demonstrating the technical feasibility and merit of the
mission for meteorology. In particular an observing system simulation experiment (OSSE) was
developed and allowed evaluating the impact on forecasts of polar winds from PCW. Predictability at
day 3 is expected to be improved by 3 hours in the Arctic region and 6 hours in the Antarctic region. That
work is soon to be published (Garand et al, 2013). CSA is studying means to realize the mission in
Public Private Partnership (PPP). The US military could contribute financially. The target for approval
by Treasury Board is 2014 budget.

Sea-ice analysis
Research will be conducted to improve the variational data assimilation system for producing analyses of
sea ice conditions and for initializing coupled models that include sea ice. The focus of this work is on
the use of a sea-ice model to provide the background state within the data assimilation cycle (instead of
persistence) and the assimilation of new observation types. The new observation types include AVHRR
visible/infra-red data and synthetic aperture radar data from the RadarSat-2 satellite. Research is also
being conducted on approaches to automatically compute the characteristics values (also referred to as
“tiepoints”) required by sea-ice retrieval algorithms and simple observation forward operators for
assimilating satellite observations.

Land data assimilation
Several aspects of CaLDAS will be improved in the next few years. First, significant progress is
expected on the inclusion of space-based data in CaLDAS. Data from SMOS, SMAP, and SSM/I or
AMSR are expected to provide much more information on soil moisture and snow. The use of data from
VIIRS will also be examined for the specification of vegetation characteristics. An incremental version of
CaLDAS will be coded and tested to provide land surface initial conditions for the new sub-km prediction
systems that are expected to become experimental in the next several years. Finally, a strategy will be
developed and tested to jointly assimilate both surface-based and space-based observations, and
space-based information on the vegetation canopy (e.g., LAI will be assimilated in CaLDAS with a first
guess based on the Canadian Terrestrial Ecosystem Model (CTEM).
46
6.2.1.2


Models
Dynamical cores
o
A 3D iterative solver for more stable implicit time stepping over steep orography will be studied
o
More efficient time stepping schemes will be studied (of the type predictor-corrector)
o
Options to use tracer mass fixing or inherently conserving transport schemes should become
available
o
Optimization of advection scheme following the use of the Yin-Yang grid
o
Icosahedral grid with finite volume
Physical parameterizations
o
Changes in PBL scheme: study the turbulent total energy approach
o
A reformulation of the boundary layer parameterization and its connection with the surface will be
achieved
o
Cloud and precipitation processes will be improved
o
Improved treatment of cloud/radiation feedbacks
o
General shortening of model spin-up time scales
o
General reduction of main model biases
o
More sophisticated treatment of stochastic physics
o
Research is planned to update the radiative transfer scheme through:
- The use of global maps of band dependent surface emissivity and albedo
- Improved parameterization of effective radius for ice crystals and
- Vertically varying trace gases climatology


o
Research is planned on the production of an ozone analysis as well as on the impact of its use in the
NWP model
o
Research to improve the grid-scale precipitation scheme; in particular, plan to develop and test a
multi-moment microphysics (prognostic precipitation) scheme suitable for mesoscale configurations
(10-30 km horizontal grid spacing)
o
Major research efforts are towards the vertical diffusion and the shallow convection
o
Research will be performed to improve the physics of the continental GEM model at ~10 km grid
spacing, as well as on the tuning of the physics necessary when the resolution of a model is
increased
o
Research will be performed to further improve the land surface models, especially the hydrological
component of the new M-ISBA scheme. Research will also be done to improve the coupling
between the surface and the low-level atmosphere
GEPS
o
Additional perturbations will be added near the surface. This may include the use of different fields
for different members for variables like sea surface temperature, soil moisture and roughness length.
o
To enable the migration of the global EPS to the Yin-Yang horizontal grid, we need to investigate the
behavior of the parameterization for Stochastic Kinetic Energy Backscatter in the context of this grid.
Urban meteorology modelling system
In the context of the CRTI (CBRN - Chemical, Biological, Radiological, and Nuclear - Research and
Technology Initiative) project, the Meteorological Service of Canada (MSC) has developed an urban
version of the LAM model with an urban surface scheme (the Town Energy Balance – TEB) to better
represent the effects of large cities for reliable prediction of flows and dispersion in the complex
urbanized environments of populated North American cities (Mailhot et al., 2006). Further development
has been done by taking advantage of the Environmental Prediction in Canadian Cities (EPiCC)
Network. This network has provided urban measurements in the Montreal and Vancouver areas on a
continuous basis for a period of two years, together with remotely-sensed data for land cover and urban
47
structures, which allowed the validation of model aspects which have not been tested extensively so far.
Emphasis in the next few years will be on evaluating the impact of built surfaces on km-scale and subkm-scale NWP of high-impact weather, including wind events, strong precipitation, low-visibility, and
precipitation phase. Most of this work will be done in preparation to the 2015 Pan American Games in
Toronto, Canada, and as part of a collaboration with the Japan Meteorological Agency (JMA) featuring
numerical integrations over Tokyo, Japan. It is also planned to proceed to a reconstruction of TEB, in
order to make it more modular and to allow for the inclusion of new sub-models for buildings, roads, and
vegetation.

The regional coupled atmosphere-ocean-ice system
Following the recent implementation at CMC of the coupled Gulf of St. Lawrence (GSL) system, future
developments focus on coupled GEM-NEMO configurations. Current efforts focus on two GEM-NEMO
coupled models: for the Great Lakes, the GSL and the Arctic/North Atlantic. It is planned that these
projects will dovetail into a single regional coupled atmosphere-ice-ocean system for METAREAs
covering the Arctic and North Atlantic and Canadian inland waters. Within close coordination with
CONCEPTS and the French operational oceanographic group Mercator-Ocean, regional ocean data
assimilation will be developed.
Another area of focus is the tuning of the sea ice model parameters as well as sea ice forecasting case
studies in order to improve the forecasts. Current developments also include:
o
Specific validation tools for the ice edge.
o
Calculation of the ice internal pressure (important information for navigation).
o
Bidirectional coupling with a METAREA pan-Arctic GEM-LAM.
The METAREA project will also benefit from a collaboration with McGill University (as part of a BREA
project) to improve the formulation of the model sea ice rheology (by using a different yield curve and/or
flow rule). This should allow better simulations of landfast ice, leads and pressure ridge formation.
Furthermore, numerical aspects will also be improved as we have a project in collaboration with Los
Alamos National Lab to implement a Jacobian-free Newton-Krylov solver using IMplicit-EXplicit (IMEX)
techniques.

The global coupled atmosphere-ocean-ice modeling system
Upcoming steps for the global ocean-ice forecast system include:
o
Use of 3DVAR global ice analyses in place of CMC operational ice analyses
o
Improved blending of ocean and ice analyses
o
Improved SST assimilation in marginal ice zone
o
Produce daily analyses
For global GEM-NEMO coupling, upcoming steps are:

o
Verification of medium range forecasts with 33km GEM coupled to 1/4deg NEMO
o
Evaluation of 1deg GEM coupled to 1deg NEMO under development for monthly forecasts
Air Quality: Regional GEM-MACH
Efforts in the coming years will focus on the following improvements to the forecast of the Regional
GEM-MACH:
o
Aqueous phase chemistry improvements, including a new solver, process treatment for the cold
season, size-dependant treatment of particulate matter and precipitation scavenging
o
Improved inorganic heterogeneous chemistry
o
Improved volatile organic compound speciation and secondary organic aerosol formation processes
(with a focus on emissions from “mobile” sources such as automobiles)
o
Better emissions using an updated inventory, including sea salts and fugitive dust
o
Improvement of species vertical diffusion and PBL mixing schemes
o
Development and implementation of species shallow and deep convective mixing scheme
o
Implementation and evaluation of species mass conservation schemes
o
Improve the lateral chemical boundary conditions
48

o
Better aerosols formation processes
o
New gas-phase chemistry numerical solver
o
Improvement to the gas phase mechanism
o
Introduction of feedback mechanisms between chemistry and radiative and condensation processes
Air quality: Global GEM-MACH
Efforts in the coming years will focus on the following improvements to Global GEM-MACH:

o
Merging of stratospheric (LINOZ) and tropospheric chemistry
o
Incorporation of the GRAHM mercury model
o
Include the production of SO2 from DMS
o
The capability to include elevated area-source emissions as inputs will be added (to allow forest fire /
wildfire inputs to the model)
o
Post-processing: Aerosol Optical Depth calculations from aerosol mass will be added and compared
to observations
o
Tests of an improved inorganic heterogeneous chemistry solver will be completed
o
Parameterizations for lightning NOx generation will be added, with transfer to regional air quality
prediction system if successful
o
Tests of the generation of initial and boundary conditions for GEM-MACH10 will be carried out and
evaluated
o
Global emissions will be updated to 2010 emissions (HTAP)
o
Participate in HTAP model-model and model-observation intercomparison studies
o
Pilot a 7-day AQ forecasting system combining GEM-MACH Global and Regional simulations
o
Implementation of heterogeneous chemistry in LINOZ
Air quality: Local GEM-MACH (within 2 years)
Efforts in the coming years will focus on the following improvements to urban forecasts with the High
Resolution GEM-MACH:

o
On-road mobile emissions for cities based on available results from link-based, traffic flow models
o
Improvements to emissions from airports, trains, food cooking and lawn mower sources
o
Urban gas-phase chemistry based on SAPRC-07 toxics mechanism
o
New parameterizations for secondary aerosol formation and PAHs
o
Ultra-Fine Particle (UFP) predictions implemented, tested and evaluated
o
Nesting down to 500-m grid spacing
Air quality: Longer range (> 2 years) outlook for Environment Canada’s AQ forecasting research
program
The broad direction for GEM-MACH is to create an operational cascade of forecasts analogous to the
operational weather forecast, using the same modelling platform to create global AQ forecasts, which in
turn drive regional, and in turn local/high resolution forecasts. A topic of particular interest and a
direction for future research are investigations into feedbacks between AQ and weather. Environment
Canada is participating in the Air Quality Model Evaluation International Initiative, Phase 2 (AQMEII-2)
which is investigating the relationships between air quality and weather. GEM-MACH has been modified
to include aerosol indirect and direct feedbacks, as well as ozone feedbacks to radiative transfer. These
code modifications include making additional connections between meteorological and chemical portions
of the code, evaluating different methodologies for estimating cloud condensation nucleus formation from
chemical aerosol and speciation information, evaluating different methods for carrying out aerosol
radiative scattering calculations, and installing and testing these new modules within the code. The
testing of this new version of the model is currently underway. Another topic of interest is the
lengthening of the GEM-MACH forecasting period. Current operational products provide forecasters with
a 48-hour lead time. The uses of better initial conditions and, in the case of the limited area models,
better boundary conditions, are considered essential to achieve this goal. Chemical data assimilation
49
work will continue, with the aim of assimilation systems being constructed for aerosols, tropospheric CO
and ozone, and other air-quality constituents. The extension to assimilating radiances from remote
sounding instruments is envisaged. Chemical data assimilation will play an important role in model
evaluation in support of model development. Finally, improvement of the methane oxidation scheme in
GEM using methane analyses and its impact on water vapour will be investigated.
50
6.2.2 Planned Research Activities in Nowcasting



Radar QC/QPE Project
o
Ongoing research to tune hydrometeor classification algorithms to different climate regimes across
Canada
o
Ongoing research into QPE methods that encompass dual polarization information including
hydrometeor classification
o
Participation in the international radar quality control intercomparison project
Aviation Related Research
o
The INTW and ABOM techniques will undergo further testing as part of the First Guess TAF project
o
Algorithms will be further developed to better predict ceiling, visibility, wind gusts, precipitation rate
and type, lightning, blowing snow, etc
o
Improved performance metrics will be developed for aviation related variables
o
Improved methods to allow forecasters to rapidly access model based forecasts and to assess their
accuracy in specific situations
o
More work on probabilistic forecasts for aviation related variables
Fog Remote Sensing And Modeling Project (FRAM)
Visibility reduction in cold climates is important for aviation, transport, and shipping industries. Ongoing
research into the relationship between visibility and weather elements such as blowing snow, rain, snow,
drizzle, and light snow will play an important role for operational weather applications.
To improve the parameterization related to visibility in northern latitudes we need a better research into
instrument systems which could be used to measure light precipitation in Arctic regions, including snow,
drizzle and fog. This project results will be used in coming years to validate model based predictions and
improve sensors used in the field.

SAAWSO (Satellite Applications for Arctic Weather and SAR Operations) Project
This project focuses on fog, cloud and precipitation predictions using satellite observations and surface
in situ measurements. Results will be improved using newly developed satellite based algorithms. First
phase of the project is performed at St. John’s, Newfoundland, and phase 2 will be performed over
Goose Bay, Labrador. Overall goal will be to improve forecasting in northern latitudes based on satellite
data and surface in-situ measurements, and improve forecast models.


Research Support Desk Project (RSD)
o
Ongoing research to test new techniques for nowcasting summer and winter high impact weather
events
o
Significant focus will be on the further development and implementation of meteorological objects
and the paradigm change to operationally forecasting events and areas versus points
o
Ongoing research to optimize the human-machine mix as applied to operational forecasting.
Interactive Forecasting, Nowcasting and Alerting Systems (iCAST) (0-48 hours)
o
Ongoing research to test new techniques for forecasting and nowcasting summer and winter highimpact weather events via iCAST and the Research Support Desk
o
Ongoing research to optimize the human-machine mix as applied to operational forecasting,
nowcasting and alerting
o
Significant focus will be on the further development and implementation of MetObjects and the
operational paradigm change from point-based forecasting/nowcasting to area-based
forecasting/nowcasting
o
Demonstration of the utility of very high-resolution meteorological monitoring and numerical
modelling for thunderstorm nowcasting during Pan Am 2015
o
Demonstration of the effectiveness of the area-based, MetObject-oriented approach to forecasting,
nowcasting and alerting via the Next Generation Warning and Forecast Process Demonstration
during Pan Am 2015
51
o
Technical transfer of requirements for the area-based, MetObject approach to developers of the
MSC forecaster workstation (NinJo)
o
Refinement of probabilistic verification methods for the area-based, MetObject-oriented approach to
summer thunderstorm forecasting and nowcasting using the iCAST prototype / Research Support
Desk
o
Development and testing of new MetObject-based storm tracking, intensity trending and first-guess
warning techniques for summer thunderstorms and related severe weather using the iCAST
prototype / Research Support Desk
o
Development and implementation of an area-based, MetObject approach for winter convective storm
(lake-effect snow) forecasting and nowcasting using the iCAST prototype / Research Support Desk
o
Development of infrastructure for supporting mesoscale meteorological monitoring requirements of
Pan Am 2015
o
Development of infrastructure for supporting the Next Generation Warning and Forecast Process
Demonstration during Pan Am 2015
52
6.2.3 Planned Research Activities in Long-range Forecasting

Improvement of monthly forecasting
Research will continue to be performed toward a seamless long range forecasting system from day to
weeks.

Improvement of seasonal forecasting
The year 2012 marked the first full year of operation for the one-tier, two-model CanSIPS system
following its replacement of the previous two-tier system in late 2011. In addition to develop an array of
new forecast products enabled by this coupled model-based system, efforts have begun to further
improve the forecasting system itself. Avenues for improvement currently under consideration include:

o
increasing ensemble size from 20 (10 per model) to 40 (20 per model)
o
calibration of probabilistic forecasts as in Kharin et al (2009)
o
land initialization based on real-time analysis
o
improved initialization for sea ice thickness
o
addition of a third coupled model (currently under development) based on EC’s GEM NWP
atmospheric model and the NEMO ocean model
Tropical-extratropical interactions
The tropical convective activities associated with the MJO excite Rossby waves that propagate to the
extratropical latitudes and influence the Canadian weather. The objective of this study is to identify
possible links between extreme weather conditions in Canada and tropical organized convections. For
example, heavy precipitation in British Columbia can be related to an intensified low pressure system in
the northeast Pacific which can be a result of deepening of the Aleutian low by a tropical forcing
associated with the MJO. What is the mechanism and how it is represented in a dynamical model is
crucial to improve extended range forecasts of those extreme weather systems.

Stratosphere influence
Besides the tropics, another important source of skill for extended weather forecasts may come from the
stratosphere. An interesting aspect is the downward propagation of the AO (Arctic Oscillation) signal
from the stratosphere, which may influence the AO variability and weather conditions on the ground up
to two months after. It is planned that an assessment is done for GEM-strato and its impact on weather
predictions beyond 10 days. More details on the stratosphere influence on the seasonal forecast skill is
available in Section 4.7.2 under ‘Stratospheric contribution to skill’. Research on this topic will continue
with CanSIPS.

GEM for climate applications
The infrastructure of the GEM model will be adapted to allow for its use in freely running mode, as well
as for climate-type simulations. The same model library will be used for weather applications and long
simulations.
53
6.2.4 Planned Research Activities in Specialized Numerical Predictions

Statistical guidance
There is a plan to expend the UMOS set of guidance to include dew point temperature, blowing snow,
probability of precipitation amounts, and type.

Wind energy forecasting
There is an externally-funded research program that focuses on improving low-level wind forecasting at
high resolution. Different aspects are being studied: momentum and heat fluxes, vertical resolution, the
use of small horizontal domain and ensemble approaches. Feasibility study on predicting power lose due
to icing was carried out with simulations of icing episodes.
54
7. References
Alves, J.-H.G.M, P. Wittmann, M. Sestak, J. Schauer, S. Stripling, N.B. Bernier, J. McLean, Y. Chao, A.
Chawla, H. Tolman, G. Nelson, and S. Klotz, 2013, The NCEP/FNMOC Combined Wave Ensemble
Product: Expanding Benefits of Inter-Agency Probabilistic Forecasts to the Oceanic Environment.
Bulletin of the American Meteorological Society, doi: 10.1175/BAMS-D-12-00032.1
Anselmo, D., M.D. Moran, S. Ménard, V. Bouchet, P. Makar, W. Gong, A. Kallaur, P.-A. Beaulieu, H. Landry,
C. Stroud, P. Huang, S. Gong, and D. Talbot, 2010. A new Canadian air quality forecast model: GEMMACH15. Proc. 12th AMS Conf. on Atmos. Chem., Jan. 17-21, Atlanta, GA, American Meteorological
Society, Boston, MA, 6 pp. [see http://ams.confex.com/ams/pdfpapers/165388.pdf]
Aparicio, J., and G. Deblonde, 2008: Impact of the assimilation of CHAMP refractivity profiles on
Environment Canada global forecasts. Monthly Weather Review, 136(1), 257-275.
Aparicio, J., G. Deblonde, L. Garand and S. Laroche, 2009: The signature of the atmospheric compressibility
Aparicio, J., and S. Laroche, 2011: An evaluation of the expression of the atmospheric refractivity for GPS
signals. J. Geophys. Res.,doi:10.1029/2010JD015214.
Baldwin, M. P., and T. J. Dunkerton, 2001: Stratospheric harbingers of anomalous weather regimes. Science
294, 581-584.
Bélair, S., J. Mailhot, J.W. Strapp, J.I. MacPherson, 1999: An examination of local versus nonlocal aspects of
a TKE-based boundary layer scheme in clear convective conditions. J. Appl. Met., 38, 1499-1518.
Bélair, S., L.-P. Crevier, J. Mailhot, B. Bilodeau, and Y. Delage, 2003a: Operational implementation of the
ISBA land surface scheme in the Canadian regional weather forecast model. Part I: Warm season
results. J. Hydromet., 4, 352-370.
Bélair, S., R. Brown, J. Mailhot, B. Bilodeau, and L.-P. Crevier, 2003b: Operational implementation of the
ISBA land surface scheme in the Canadian regional weather forecast model. Part II: Cold season results.
J. Hydromet., 4, 371-386.
Bélair, S., J. Mailhot, C. Girard, and P. Vaillancourt, 2005: Boundary-layer and shallow cumulus clouds in a
medium-range forecast of a large-scale weather system. Mon.Wea. Rev., 133, 1938-1960.
Bélair, S., M. Roch, A-M. Leduc, P.A. Vaillancourt, S. Laroche, and J. Mailhot, 2009: Medium-range
quantitative precipitation forecasts from Canada’s new 33-km deterministic global operational system.
Wea. Forecasting, 24, 690-708.
Beljaars, A.C.M., 2004: A new parametrization of turbulent orographic form drag. Quarterly Journal of the
Royal Meteorological Society, 130, 1327-1347.
Benoît, R., J. Côté and J. Mailhot, 1989: Inclusion of a TKE boundary layer parameterization in the Canadian
regional finite-element model. Mon. Wea. Rev., 117, 1726-1750.
Bernier, N.B. and K.R. Thompson 2010. Seasonal and Annual Tide and Surge Energy budgets off the East
Coast of Canada and Northeastern United States. Continental Shelf Research, 20 (3-4), 353-364
Bernier, N.B., S. Bélair, B. Bilodeau, and L. Tong. 2012. Assimilation and High Resolution Forecasts of
Surface and Near Surface Conditions for the 2010 Vancouver Winter Olympic and Paralympic Games,
Pure and Applied Geophysics, doi : 10.1007/s00024-012-0542-0, pp1-14.
Boniface, K., J. M. Aparicio and E. Cardellach, 2011: Meteorological information in GPS-RO reflected signals,
Atmos. Meas. Tech., 4, 1199-1231, doi:10.5194/amtd-4-1199-2011.
Bougeault, P., and P. Lacarrère, 1989: Parameterization of orography-induced turbulence in a mesobetascale model. Mon. Wea. Rev., 117-1872-1890.
Bourgouin, P., 2000: A Method to Determine Precipitation Types. Wea. Forecasting, 15, 583-592.
Brasnett, B. 1997: A global analysis of sea surface temperature for numerical weather prediction. J. Atmos.
Oceanic Technol., 14, 925-937.
Brasnett, B. 1999: A global analysis of Snow Depth for Numerical Weather Prediction. J. Appl. Meteor., 38,
726-740.
Brasnett, B. 2008: The impact of satellite retrievals in a global sea-surface-temperature analysis. Quart. J.
Roy. Meteor. Soc., 134, 1745-1760.
Brunet, N., 1987 : Development of a perfect prog system for spot time temperature forecasts. CMC Technical
Document, No. 30.
Buehner M, Houtekamer PL, Charette C, Mitchell HL, He B (2010a) Intercomparison of Variational Data
Assimilation and the Ensemble Kalman Filter for Global Deterministic NWP. Part I: Description and
Single-Observation Experiments. Mon. Wea. Rev., 138, 1550–1566.
Buehner M, Houtekamer PL, Charette C, Mitchell HL, He B (2010b) Intercomparison of Variational Data
Assimilation and the Ensemble Kalman Filter for Global Deterministic NWP. Part II: One-Month
Experiments with Real Observations. Mon. Wea. Rev., 138, 1567–1586.
Buehner, M., 2011: Evaluation of a Spatial/Spectral Covariance Localization Approach for Atmospheric Data
Assimilation. Mon. Wea. Rev., 140, 617–636.
Buehner, M., A. Caya, L. Pogson, T. Carrieres and P. Pestieau, 2013: "A New Environment
Canada Regional Ice Analysis System." Atmosphere-Ocean, 51, 18-34.
Buizza, R., M. Miller, and T. N. Palmer, 1999: Stochastic representation of model uncertainties in the
ECMWF ensemble prediction system. Quart. J. Roy. Meteor. Soc., 125, 2887–2908.
Burrows, R. B., M. Vallée, D. I. Wardle, J. B. Kerr, L. J. Wilson and D. W. Tarasick, 1994 : The Canadian
operational procedure for forecasting total ozone and UV radiation. Met. Apps., 1, 247-265.
Burrows, W.R., C. Price and L.J. Wilson. 2005: Warm season lightning probability prediction for Canada and
the northern United States. Weather and Forecasting, 20, 971-988.
55
Chapnik, B., G. Desroziers, F. Rabier, and O. Talagrand, 2004: Properties and first application of an errorstatistics tuning method in variational assimilation. Q. J. R. Meteorol. Soc., 130, 2253-2275.
Charron, M., M. Buehner, P. A. Vaillancourt, C. Charrette, M. Roch, J. Morneau, L. Garand, S. Polavarapu, J.
Aparacio and S. Pellerin, 2011: Impact of the stratosphere on the troposphere in an operational medium
range weather forecasting system. Part I: Results. To be submitted to Mon. Weather Rev.
Côté, J., S. Gravel, A. Méthot, A. Patoine, M. Roch and A. Staniforth, 1998a: The Operational CMC-MRB
Global Environmental Multiscale (GEM) Model: Part I - Design Considerations and Formulation, Mon.
Wea. Rev., 126, 1373-1395.
Côté, J., J.-G. Desmarais, S. Gravel, A. Méthot, A. Patoine, M. Roch and A. Staniforth, 1998b: The
Operational CMC-MRB Global Environmental Multiscale (GEM) Model: Part II - Results, Mon. Wea. Rev.,
126, 1397-1418.
D’Amours, R. and P. Pagé, 2001: Modèles pour les éco-urgences atmosphériques, Internal report, Canadian
Meteorological Centre, Environmental Emergency Response Section, 9 pp.
D’Amours, R. and A. Malo, 2004: A Zeroth Order Lagrangian Particle Dispersion Model: MLDP0, Internal
Report, Canadian Meteorological Centre, Environmental Emergency Response Section, 19 pp.
D'Amours, R., A. Malo, R. Servranckx, D. Bensimon, S. Trudel, and J.-P. Gauthier, 2010: Application of the
atmospheric Lagrangian particle dispersion model MLDP0 to the 2008 eruptions of Okmok and
Kasatochi
volcanoes,
Journal
of
Geophysical
Research,
115
(D00L11),
1–11,
doi:10.1029/2009JD013602.
Delage, Y., 1988a: The position of the lowest levels in the boundary layer of atmospheric circulation models.
Atmos.-Ocean, 26, 329-340.
Delage, Y., 1988b: A parameterization of the stable atmospheric boundary layer. Boundary-Layer Meteor.,
43, 365-381.
Desroziers, G., and S. Ivanov, 2001: Diagnosis and adaptive tuning of observation-error parameters in a
variational assimilation. Q. J. R. Meteorol. Soc., 127, 1433-1452.
Desroziers, G., L. Berre, B. Chapnik, and P. Poli, 2005: Diagnosis of observation, background, and analysiserror statistics in observation space. Q. J. R. Meteorol. Soc., 131, 3385-3396.
Dunn-Sigouin, E., S.-W. Son, and H. Lin, 2013: Evaluation of Northern Hemisphere blocking climatology in
the Global Environment Multiscale (GEM) model. Mon. Wea. Rev., 141, 707-727.
Ellrod, G. P., 1989: An index for clear air turbulence based on horizontal deformation and vertical wind shear.
Preprints of the Third International Conference on the Aviation Weather System, Anaheim, California.
ESA, 2008: Candidate Earth Explorer Core Missions – Report for Assessment: PREMIER – Process
Exploitation through Measurements of Infrared and millimetre-wave Emitted Radiation, SP-20 1313/5,
ESA Publications Division, ESTEC, Keplerlaan 1, 2200 AG Noordwijk, The Netherlands.
Fillion, L., H. L. Mitchell, H. Ritchie and A. Staniforth, 1995: The impact of a digital filter finalization technique
in a global data assimilation system, Tellus, 47A, 304-323
Fillion, L., Tanguay, M., E. Lapalme, B. Denis, M. Desgagne, V. Lee, N. Ek, Z. Liu, M. Lajoie, J.-F. Caron, C.
Page, 2010 : The Canadian Meteorological Center Limited-Area Regional Data Assimilation and
Forecasting
System.
Wea.
Forecasting,
25,
1645–1669.
doi:
http://dx.doi.org/10.1175/2010WAF2222401.1.
Flesch, T. K., J. D. Wilson and B. P. Crenna, , 2002: MLCD: A Short-Range Atmospheric Dispersion Model
for Emergency Response, Contract Report to the Canadian Meteorological Centre by Department of
Earth and Atmospheric Sciences, University of Alberta.
Flesch, T. K., R. D'Amours, C. J. Mooney and J. D. Wilson, 2004: MLDP: A Long-Range Lagrangian
Stochastic Dispersion Model, Internal report in collaboration with the Canadian Meteorological Centre
and the Department of Earth and Atmospheric Sciences from University of Alberta.
Garand, L., J. Feng, S. Heilliette, Y. Rochon, and A. P. Trishchenko, 2013: Assimilation of circumpolar wind
vectors derived from highly elliptical orbit imagery: impact assessment based on observing system
simulation experiments. J. Appl. Meteor. Climatol., to appear.
Gauthier, P., C. Charette, L. Fillion, P. Koclas and S. Laroche 1999: Implementation of a 3D Variational Data
Assimilation System at the Canadian Meteorological Centre. Part I: The Global Analysis, Atmos.-Ocean,
37, 103-156.
Heilliette, S., and L. Garand, 2007, A practical approach for the assimilation of cloudy infrared radiances and
its evaluation using AIRS simulated obervations, Atmosphere-Ocean, 45 (4), pp 211-225.
Heilliette,. S., Y. Rochon, L. Garand, and J. Kaminski, 2013: Assimilation of infrared radiances in the context
of observing system simulation experiments. J. Appl. Meteor. Climatol., 52, 1031-1045.
Hines, C. O., 1997a:: Doppler-spread parameterization of gravity-wave momentum deposition in the middle
atmosphere. Part 1: Basic formulation. J. Atmos. Sol. Terr. Phys., 59, 371–386.
Hines, C. O., 1997b: Doppler-spread parameterization of gravity-wave momentum deposition in the middle
atmosphere. Part 2: Broad and quasi monochromatic spectra, and implementation. J. Atmos. Sol. Terr.
Phys., 59, 387–400.
Houtekamer, P.L., Lefaivre L., Derome J., Ritchie H. and Mitchell, H.L., 1996: A system simulation approach
to ensemble prediction, Mon. Wea. Rev., 124, 1225-1242.
Houtekamer, P. L., Herschel L. Mitchell, Gerard Pellerin, Mark Buehner, Martin Charron, Lubos Spacek, and
Bjarne Hansen, 2005: Atmospheric Data Assimilation with an Ensemble Kalman Filter: Results with Real
Observations, Mon. Wea. Rev., 133, pages 604-620.
Houtekamer, P.L., and H.L. Mitchell, 2005, Ensemble Kalman filtering, Q.J.R. Meteorol. Soc., 131, pp 32693289.
56
Kain, J. S. and J. M. Fritsch, 1990: A one-dimensional entraining / detraining plume model and its application
in convective parameterization. J. Atmos. Sci., 47, 2784-2802.
Kain, J. S. and J. M. Fritsch, 1993: Convective parameterization for mesoscale models: The Kain-Fritsch
scheme. The representation of cumulus convection in numerical models. Meteor. Monogr., 27, Amer.
Meteor. Soc., 165-170.
Kharin, V. V., Q. Teng, F. W. Zwiers, G. J. Boer, J. Derome, J. S. Fontecilla, 2009: Skill assessment of
seasonal hindcasts from the Canadian Historical Forecast Project. Atmos.—Ocn., 47, 204-223.
Kharin, V. V. and J. F. Scinocca, 2012: The impact of model fidelity on seasonal predictive skill, Geophys.
Res. Lett., 39, L18803 doi:10.1029/2012GL052815,
Laroche, S. and R. Sarrazin, 2013 : Impact of radiosonde balloon drift on Numerical Weather Prediction and
Verification. Wea. and Forecasting, in print.
Li, J. and H. W. Barker, 2005: A radiation algorithm with correlated k-distribution. Part I: local thermal
equilibrium. J. Atmos. Sci., 62, 286-309.
Li, X., M. Charron, L. Spacek, and G. Candille, 2008: A regional ensemble prediction system based on moist
targeted singular vectors and stochastic parameter perturbations. Mon. Wea. Rev., 136, 443–462
Lott, F., and M. Miller, 1997: A new sub-grid scale orographic drag parameterization; its testing in the
ECMWF model. Quart. J. Roy. Meteor. Soc., 123, 101-127.
Mahidjiba, A., A. Qaddouri, and J. Côté, 2008: Application of the Semi-Lagrangian Inherently Conserving and
Efficient (SLICE) Transport Method to Divergent Flows on a C Grid. Mon. Wea. Rev., 136, 4850–4866.
Mailhot, J., and S. Bélair, 2002: An examination of a unified cloudiness-turbulence scheme with various
types of cloudy boundary layers. Preprints, 15th Symposium on Boundary Layer and Turbulence, 15-19
July, 2002, Wageningen, Netherlands, 215-218.
Mailhot, J., S. Bélair, A. Lemonsu, L. Tong, A. Leroux, N. Benbouta, and R. Hogue, 2006: Urban modeling at
the Meteorological Service of Canada. IAUC (International Association for Urban Climate) Newsletter,
17, 13-16. (Available online at http://www.urban-climate.org)
McFarlane, N.A., 1987 : The effect of orographically excited gravity wave drag on the general circulation of
the lower stratosphere and troposhere. J. Atmos. Sci., 44, 1775-1800.
McFarlane, N.A., C. Girard and D.W. Shantz, 1987 : Reduction of systematic errors in NWP and General
Circulation models by parameterized gravity wave drag. Short and Medium-Range Numerical Weather
Prediction, Collection of Papers Presented at the WMO/IUGG NWP Symposium, 4-8 August 1986,
Tokyo, 713-728.
McLinden, C.A., S.C. Olson, B. Hannegan, O. Wild, M.J. Prather, and J. Sundet, 2000: Stratospheric ozone
in 3D models: A simple chemistry and cross-tropopause flux, J. Geophys. Res., 105, 14653-14665.
Merryfield, W. J., B. Denis, J.-S. Fontecilla, W.-S. Lee, S. Kharin, J. Hodgson and B, Archambault 2011: The
Canadian Seasonal to Interannual Prediction System (CanSIPS) : An overview of its design and
operational
implementation.
CMC
technical
note
available
here:
http://collaboration.cmc.ec.gc.ca/cmc/cmoi/product_guide/docs/lib/op_systems/doc_opchanges/technote
_cansips_20111124_e.pdf
Merryfield, W. J., W.-S. Lee, G. J. Boer, V. V. Kharin, J. F. Scinocca, G. M. Flato, R. S. Ajayamohan, J. C.
Fyfe, Y. Tang, and S. Polavarapu, 2013a. The Canadian Seasonal to Interannual Prediction System.
Part I: Models and Initialization, Mon. Wea. Rev., doi:10.1175/MWR-D-12-00216.1.
Merryfield, W. J., W.-S. Lee, W. Wang, M. Chen, and A. Kumar, 2013b: Multi-system seasonal predictions of
Arctic sea ice, Geophys. Res. Lett., 40, 1551-1556, doi:10.1002/grl.50317.
Milbrandt, J.A. and M.K. Yau, 2005a: A multimoment bulk microphysics parameterization. Part I: Analysis of
the role of toperation.forecasts.regseri.2013061900_000he spectral shape parameter. J. Atmos. Sci.,
62,3051-3064.
Milbrandt, J.A. and M.K. Yau, 2005b: A multimoment bulk microphysics parameterization. Part II: A proposed
three-moment closure and scheme description. J. Atmos. Sci., 62,3051-3064.
Noilhan, J. and S. Planton, 1989: A simple parameterization of land surface processes for meteorological
models. Mon. Wea. Rev., 117, 536-549
Polar Communications and Weather mission (PCW), 2009: Canadian Space Agency internet site
http://www.asc-csa.gc.ca/eng/satellites/pcw
Pudykiewicz, J., R. Benoit, and J. Mailhot, 1992: Inclusion and verification of a predictive cloud water
scheme in a regional weather prediction model. Mon. Wea. Rev., 120, 612-626.
Robichaud, A. and Ménard, R., 2013. Multi-year objective analyses of warm season ground-level ozone and
PM2.5 over North America using real-time observations and Canadian operational air quality models.
Submitted to ACPD, Feb, 2013.
Rochon, Y., J. W. Kaminski, S. Heilliette, L. Garand, J. de Grandpré, and R. Ménard, 2012 : observation
system simulation experiments for the PREMIER mossion. Proc. 5th WMO Workshop on the impact of
various observing systems on NWP, Sedona, AZ, May 2012.
Rodney, M., and H. Lin, 2013: Subseasonal prediction of wintertime North American surface air temperature
during strong MJO events, Mon. Wea. Rev., in press.
Shutts, G. 2005: A kinetic energy backscatter algorithm for use in ensemble prediction systems,
Q.J.R.Meteorol. Soc. 131, 3079-310
Sigmond, M., J.C. Fyfe, G.M. Flato, V.V. Kharin and W. J. Merryfield, 2013a: Seasonal forecast skill of Arctic
sea ice area in a dynamical forecast system, Geophys. Res. Lett., 40, 529-534, doi:10.1002/grl.50129.
Sigmond, M., J. F. Scinocca, V.V. Kharin, T.G. Shepherd, 2013b: Enhanced seasonal forecast skill following
stratospheric sudden warmings, Nature Geosci., 6, 98-102, doi:10.1038/ngeo1698.
57
Sills, D. M. L., 2009: On the MSC Forecasters Forums and the Future Role of the Human Forecaster. Bull.
Amer. Meteorol. Soc., 90, 619-627.
Sills, D. M. L., J. R. Brook, I. Levy, P. A. Makar, J. Zhang, and P. A. Taylor, 2011: Lake breezes in the
southern Great Lakes region and their influence during BAQS-Met 2007. Atmos. Chem. Phys., 11, 79557973.
Smith, G.C., F. Roy and B. Brasnett (2012) Evaluation of an Operational Ice-Ocean Analysis and
Forecasting System for the Gulf of St. Lawrence, QJRMS, doi: 10.1002/qj.1982
Sparks, R. S. J., M. I. Bursik, S. N. Carey, J. S. Gilbert, L. S. Glaze, H. Sigurdsson, A. W. Woods, 1997:
Volcanic Plumes, John Wiley and Sons, New York, 574 pp.
Sundqvist, H., E. Berge and J. E. Kristjansson, 1989: Condensation and cloud parameterization studies with
a mesoscale numerical weather prediction model. Mon. Wea. Rev., 117, 1641-1657.
Taylor, N. M., D. M. L. Sills, J. M. Hanesiak, J. A. Milbrandt, C. D. Smith, G. S. Strong, S. H. Skone, P. J.
McCarthy, and J. C. Brimelow, 2011: The Understanding Severe Thunderstorms and Alberta Boundary
Layers Experiment (UNSTABLE) 2008. Bull. Amer. Meteorol. Soc., 92, 739-763.
Tanguay, M., L. Fillion, E. Lapalme, M. Lajoie, 2012: Four-Dimensional Variational Data Assimilation for the
Canadian Regional Deterministic Prediction System. Mon. Wea. Rev., 140, 5, 1517-1538.
Trishchenko, A. P., and L. Garand, 2011: Spatial and temporal sampling of polar regions from two-satellite
system pn Molniya orbit. J. Oceanic Atmos. Tech., J. Atmos. Oceanic Tech., Vol 28(8), 977-992..
Trishchenko, A. P., L. Garand, and L. D. Trichtchenko, 2011: Three apogee 16-h highly elliptical orbit as
optimal choice for continuous meteorological imaging of polar regions. J. Oceanic Atmos. Tech., Vol.
28(11), 1407-1422., .
Trishchenko, A. P., and L. Garand, 2012. Observing polar regions from space: Advantages of a satellite
system on a highly elliptical orbit versus a constellation of low Earth polar orbiters. To appear in
Canadian J. Of Remote Sensing.
Tremblay, A., A. Glazer, W. Szyrmer, G. Isaac and I. Zawadzki, 1995: Forecasting of supercooled clouds.
Mon. Wea. Rev., 123, 2098-2113.
Verret, R., 1987: Development of a perfect prog system for forecast of probability of precipitation and sky
cover. CMC Technical Document, 29, 28 pp.
Verret R., 1989: A statistical forecasting system with auto-correction error feedback. Preprints, 11th
Conference on Probability and Statistics, Oct. 1989, Monterey, California, 88-92.
Verret, R., G. Babin, D. Vigneux, R. Parent and J. Marcoux, 1993: SCRIBE: An Interactive System for
Composition of Meteorological Forecasts. Preprints, 13th AMS Conference on Weather Analysis and
Forecasting, August 2-6, 1993, Vienna, Virginia, 213-216.
Verret, R., G. Babin, D. Vigneux, J. Marcoux, J. Boulais, R. Parent, S. Payer and F. Petrucci, 1995: SCRIBE
an interactive system for composition of meteorological forecasts. Preprints 11th International
Conference on Interactive Information and Processing Systems for Meteorology, Oceanography and
Hydrology, January 15-20, 1995, Dallas, Texas, 56-61.
Verret, R., D. Vigneux, J. Marcoux, R. Parent, F. Petrucci, C. Landry, L. Pelletier and G. Hardy, 1997:
SCRIBE 3.0 a product generator. Preprints 13th International Conference on Interactive Information and
Processing Systems for Meteorology, Oceanography and Hydrology, February 2-7, 1997, Long Beach,
California, 392-395.
Vosper, S. B., H. Wells, A. R. Brown, 2009: Accounting for non-uniform static stability in orographic drag
parametrization. Q. J. Roy. Meteorol. Soc., 135, 815–822.
Wells, H.; S. B. Vosper; A. N. Ross, A. R. Brown, S. Webster, 2008: Wind direction effects on orographic
drag. Q. J. Roy. Meteorol. Soc., 134, 689-701.
Wilson, L. J. and M. Vallée, 2001: The Canadian Updateable Model Output Statistics (UMOS) System:
Design and Development Tests, Wea. Forecasting, 17, 206–222.
Wilson, L. J. and M. Vallée, 2002: The Canadian Updateable Model Output Statistics (UMOS) System:
Validation against Perfect Prog, Wea. Forecasting, 18, 288–302.
Zadra, A., M. Roch, S. Laroche and M. Charron, 2003: The Subgrid scale Orographic Blocking
Parameterization of the GEM Model, Atmos. Ocean, 41, 151-170.
58