En français

ECCC logo

TOC > MSC data > CMC Operational System's changelog

Changes to the Canadian Meteorological Centre (CMC) Operational System

2020

April 1, 2020

January 21, 2020

2019

July 23, 2019

July 3, 2019

April 15, 2019

March 4, 2019

January 30, 2019

2018

November 28, 2018

September 18, 2018

June 28, 2018

June 7, 2018

April 4, 2018

April 3, 2018

March 2, 2018

March 1, 2018

January 24, 2018

2017

December 14, 2017

November 1, 2017

September 6, 2017

2016

November 9, 2016

September 7, 2016

June 21, 2016

April 14, 2016

April 7, 2016

March 16, 2016

April 14, 2016

2015

December 15, 2015

September 17, 2015

August 20, 2015

July 9, 2015

June 23, 2015

June 11, 2015

June 3, 2015

April 16, 2015

2014

November 18, 2014

May 6, 2014

2013

December 4, 2013

November 20, 2013

November 07, 2013

August 20, 2013

August 12, 2013

*Adjustments made so that scribe global matrices now available about 30 minutes earlier

July 23, 2013

July 12, 2013

July 3, 2013

June 18, 2013

  • Inclusion of Td from UMOS into the Scribe matrices.
  • Update to the production of the bulletins FPCN80, FPCN48-49, FPCN46-47.

May 17, 2013

May 13, 2013

May 7, 2013

April 15, 2013

February 28, 2013

February 26, 2013

February 13, 2013

2012

October 19, 2012

October 3, 2012

2011

December 22, 2011

December 1, 2011

November 16, 2011

Wednesday October 20, 2010

A new limited area version of the regional forecast system becomes operational at the Canadian Meteorological Centre.

Starting on October 20 2010 with the 12 UTC run the Canadian GEM regional forecast system will be modified to incorporate a limited area version of the gem model referred to as REG-LAM3D. Changes to the analyses will also be provided to the model.

One of the changes to the forecast model includes a western and southern extension of the 15 km high resolution grid. The use of the Limited Area Model (LAM) brings significant changes in the way the analsysis is performed. Details of those changes are in the notes provided below.

Some adjustments were made to ensure all products, including those products using data near our outside the edges of the LAM grid are still available. Even though the new forecast system involves additional components, the computational cost of running a driving global model and a lam is the same as with the previous system. No changes to the time of availability of the products are expected.

Objective and subjective verifications of the reg-lam3d over a three month parallel run revealed no difference in the quality of forecasts. Special attention was given to weather systems over the atlantic ocean, especially ones that would track along the eastern boundary of the domain (coastal development), or ones moving from an area outside the domain (tropical storms) and no deterioration of the forecast was noted.

The REG-LAM3D will now pilot the air quality forecast model GEM-MACH15.

Additional information is available in the official announcement available here.

Users wishing more technical information can access a technical note by clicking here.

Tuesday March 2, 2010

Update to the emissions data set for the GEM-MACH15 air quality model at the Canadian Meteorological Centre

On March 02 2010 starting with the 1200 UTC run, the Canadian Meteorological Centre (CMC) implemented a new emissions data set for the operational air quality forecast model GEM-MACH15 (Global Environmental Multi-scale model - Modelling Air quality and CHemistry).

The objective of this update was to address the occasional severe overprediction of fine particulate matter (pm2.5) in some major Canadian urban centres, primarily during events of stable/shallow boundary layer conditions and light winds.

Additional information is available in the official announcement available here.

Users wishing more technical information can access a technical note by clicking here.

Thursday December 3, 2009

CMC Updates Data Assimilation Systems with ATOVS Data

On December 3 2009, the Canadian Meteorological Centre (CMC) will update its data assimilation systems with additional ATOVS data. These are the AMSUA and MHS data from the METOP and NOAA-19 polar orbiting satellites. These new data will now be made available to the analysis of the GEM-15km and GEM-global models, as well as the ensemble prediction models. This update is needed in order to maintain or improve the current level of assimilated data as new or replacement satellites are launched by the operational satellite programmes of NOAA and EUMETSAT.

AMSUA and MHS data from METOP and NOAA-19

Presently AMSUA data from NOAA-15, NOAA-16, NOAA-18 and AQUA are assimilated. AMSUB or MHS data from NOAA-15, NOAA-16, NOAA-17 and NOAA-18 are assimilated. All these satellites are either secondary (backup) or research satellites. The operational satellites are presently METOP and NOAA-19. The incorporation of data from METOP and NOAA-19 will result in better data coverage (up to 30% more data), improved timeliness of data and a more robust data assimilation system in case of instrument failure.

Differences in improved data coverage resulting from this implementation can be seen on this image for AMSUA data, and on this image for AMSUB data.

Impact on forecasts

Rigorous tests have shown the net impact on model performances as positive but small. The largest improvement was noted in the global forecasts in the southern hemispere. In the ensemble prediction system there is a minor improvement beyond day 3. This slight impact is attributed to the added coverage from the additional METOP and NOAA-19 ATOVS data that benefit the assimilation system. All the differences observed are judged acceptable within the context of our ongoing effort to maintain the robustness of our forecasting systems.

This change was announced in the official announcement available here.

For specific questions concerning this implementation, please write to production-info(at)cmc.ec.gc.ca

Wednesday November 18, 2009

GEM-MACH15 Air Quality Forecast Model to replace CHRONOS at the Canadian Meteorological Centre

On November 18 2009 starting with the 1200 UTC run, the Canadian Meteorological Centre (CMC) will implement GEM-MACH15, a 15-km configuration of the new on-line operational air quality forecast model called GEM-MACH (Global Environmental Multi-scale model - Modelling Air quality and CHemistry). GEM-MACH15 replaces CHRONOS which has been in service since 1999. the new model, which features a more sophisticated representation of chemical processes and an improved set of emissions, will immediately provide improvements to the air quality forecast system, while enabling more significant advancements in the years to come.

GEM-MACH is an on-line model in which the chemistry is integrated in-step with the dynamical and physical weather processes along with many other improvements. It is a is a limited area model (LAM) configuration with 15-km horizontal grid spacing that provides a better coverage of the Canadian landmass.

GEM-MACH15 has been tested extensively over summer and winter seasons since 2007 and most recently, the model has been run in parallel with CHRONOS during the summer of 2009. During this period, a coordinated evaluation of the model by operational forecasters at regional weather offices and at CMC, by model developers, and by Environment Canada air quality science units was undertaken. Subjective and objective evaluation results are available in the documents referred to below.

Products that were previously produced from CHRONOS data will now be produced using GEM-MACH15 data and those products should be available 45 to 80 minutes sooner than equivalent products produced previously CHRONOS.

Additional information is available in the official announcement available here.

Users wishing more technical information can access a technical note by clicking here.

Monday June 22, 2009

Upgrades to CMC's Ensemble Assimilation and Ensemble Forecast Systems

On Monday June 22, 2009 the Canadian Meteorological Centre (CMC) will implement major changes to its ensemble assimilation (EnKF) system and minor changes to its ensemble prediction system (EPS). EnKF changes will begin with its 0600 UTC run and EPS changes will begin with its 1200 UTC run.

Changes to the EnKF system are (1) An increase in number of vertical levels to 58 while keping the model top at 10 hPa. Trial fields produced as a result are better from 400 hPa up to the model top. (2) The GPS-RO data are now assimilated. This additional data has a significant positive impact on the analyses particularly in the southern hemisphere. (3) The size of the observation groups in the sequential assimilation algorithm has been increased leading to a lesser number of groups with more observations each and a reduction in the randomness associated with having multiple analysis passes.

The corresponding implementation of the global deterministic model extended into the middle atmosphere (cf. meso-strato) also leads to a better bias correction of data particularly in the stratosphere. The EnKF is using the bias corrected observations coming from this deterministic system and as a result a clear positive effect on the analyses has been seen particularly at high vertical levels.

Minor changes to the EPS model members are (1) Horizontal diffusion is now done on real winds components instead of the calculated image winds. This does not affect the forecast quality but leads to greater numerical stability of the model. (2) The number of processors has also been doubled leading to a reduction of the integration time by 15 minutes.

Extensive EPS testing was performed in development on summer (july 2006) and winter (january 2007) data. A parallel run in spring 2009 was also conducted.

Probabilistic verification of the forecast against radiosondes has shown improvement for the first days of integrations for temperature, geopotential height, wind and humidity particularly at 250 hpa. Significant forecast improvements have been found to extend to days 1 to 3 in the northern hemisphere, and to lead times covering days 1 to 7 in the southern hemisphere.

Outputs of the parallel runs were also evaluated by operational meteorologists and no clear changes to system behavior or quality have been noticed by them.

More details of subjective and objective evaluations are available in the official announcement referred to below.

It is expected products from the EPS should now be available about 10 minutes sooner than their present time of availability.

Additional information is available in the official announcement available here.

Users wishing more technical information can access a technical note by clicking here.

Extension into the Middle Atmosphere of the global and regional forecast systems of the Canadian Meteorological Centre

On Monday June 22nd starting with the 0600 UTC surface analysis run, the Canadian Meteorological Centre (CMC) will make important changes to the global and regional systems leading to improvements in the forecasts, especially at lead times of 3 to 6 days in the global forecast system.

Changes to the global and regional data assimilation systems include assimilation of higher level atmospheric data from ATOVS, AMSU-A channels, some GPS radio occultation (GPSRO) data. New observation and background error statistics are also added.

Changes affecting both global and regional models include raising the model lid to 0.1 hPa (64 km), replacing the sigma-like (or eta) vertical coordinate with a new hybrid coordinate and using a non-orographic gravity wave drag scheme.

An additional change in the global model is the use of a new radiative transfer scheme previously implemented in the regional model and leaing to a general improvement in tropospheric temperature forecasts.

New global and regional forecast systems were tested and evaluated extensively during their development phase and were run in parallel alongside the operational systems during the spring of 2009. Outputs of the parallel systems were verified objectively against analyses, against observations and evaluated subjectively on a daily basis by operational meteorologists.

Objective evaluations of the global model against radiosondes show very significant improvement in the stratosphere at all forecast lead-times, with consistent improvement throughout the troposphere. Temperature bias is generally improved. Large biases in zonal wind components in the lower stratosphere are greatly reduced. Verifications against analyses show improved anomaly correlations for temperature and geopotential and generally reduced wind rms errors in the troposphere at all forecast ranges. QPF verification shows the change to be neutral over north america.

Objective verification of the regional system against radiosonde indicate large improvements in the stratosphere. Impacts were neutral or slightly positive in the troposphere. Surface fields were not significantly affected as the new radiative scheme had already been implemented in the regional model in march 2009.

Subjective evaluation of the global system showed a general improvement and was recognizably better for the geopotential at 500 hPa than for sea level pressure, and some gains were noted for the QPF. Improvements were seen in the 3 to 5 day lead-time forecast lows affecting Atlantic Canada and the Bering sea. Evaluations also suggest the new system will not overforecast tropical storms as much as the previous global system.

Subjective verification of the regional system noted gains in the quality of the mass fields except in the Arctic where the effect was neutral.

Additional details of subjective and objective evaluations are described in the official announcement referred to below.

Forecast systems using outputs of regional or global models, (UMOS, SCRIBE, wave models (WAM)) are expected to have improvements proportional to those seen in the regional and global systems. UMOS system applied to the global model temperature output have been retrained. No retraining was required for the regional UMOS since training had been done before.

The CHRONOS air quality model was run in parallel using outputs from the new regional model. Because regional outputs now used by CHRONOS are the result of an interpolation from the new hybrid levels onto eta levels, a slight deterioration in objective scores was noted with the parallel CHRONOS but no significant divergence occured throughout the evaluation period. This is a temporary situation in view of improvements to the air quality prediction system planned for later this summer.

Details of impacts on the Ensemble Prediction System (EPS) from changes to the global forecast system are described in a seperate GENOT and technical note that were published for the EPS system.

Products from the new global and regional models will now be available about ten to fifteen minutes later than previously delivered. CMC is working to reduce these delays.

Additional information is available in the official announcement available here.

More detailed technical information on changes to the global system is available by clicking here.

More detailed technical information on changes to the regional system is available by clicking here.

Wednesday June 10, 2009

Implementation of the 0600 UTC and 1800 UTC Regional Prediction System

The transmission between 20h15 and 20h30 UTC on June 10th 2009, of scribe matrices files produced from outputs of the 1800 UTC run of the Gem-Regional model, makes official the implementation of the 0600 UTC and 1800 UTC runs of the Regional Prediction System. Doubling the frequency of the production of short-term forecasts from the Gem-Regional model seeks to address the needs expressed by customers of the Meteorological Service of Canada, among others those of civil aviation.

These runs have been executed for nearly a year in experimental mode. GRIB and BUFR format digital data have been transmitted for a few months already from the new intermediates runs. In addition, a full suite of graphical or alphanumeric forecast products from these runs will be activated gradually during the next few months.

It should be noted that these runs are integrated to 54 hours to ensure their compatibility with the 0000 UTC and 1200 UTC runs which are integrated to 48 hours.

Additional information is available in the official announcement available here.

Tuesday March 31, 2009

Improvements to the assimilation system of the Canadian Meteorological Centre

Starting on March 31 2009 the Canadian Meteorological Centre will update its data assimilation systems with additional satellite data. These are the GPSRO from the METOP and GRACE satellites, MODIS direct broadcast winds and the ocean surface winds from the ASCAT instrument on-board the METOP satellite. These new data will now be made available to the analysis of the GEM-15km and GEM-global models, as well as to the ensemble Kalman-filter system. This update is needed in order to maintain or improve the current level of assimilated data as new or replacement satellites are launched by the operational satellite programs of NOAA and EUMETSAT.

Preliminary tests have shown the net impact on model performances as small for forecasts up to day 3. For day 4 onward a slight improvement was noted in the global forecasts over both hemisperes. This slight impact is attributed to the added coverage from the additional satellite data that benefit the assimilation systems. All differences observed are judged acceptable within the context of our ongoing effort to maintain the robustness of our forecasting systems.

Additional information on these changes is available in the official announcement available here.

Wednesday March 12, 2009

Improvements to the global and regional data assimilation systems and to the regional forecast system of the Canadian Meteorological Centre. (Originally scheduled for March 11 2009).

Starting on March 12 at 0600 UTC, the Canadian Meteorological Centre will implement changes to incorporate new data into the global and regional data assimilation systems. Furthermore, the GEM model of regional data assimilation and forecast system will be improved by using a new radiative transfer scheme and by extending its uniform-resolution core grid to better cover the arctic regions.

Data assimilation changes consist of assimilating GPSRO data from six COSMIC satellites into the global and regional assimilation systems. These data provide refractivity profiles from satellite gps measurements and refractivity can be directly linked to atmospheric temperature and moisture conditions. Changes will be ported on the global 4d-var variational assimilation system and the regional 3d-var assimilation system. It is planned in the near future to incorporate these profiles into the ensemble assimilation and prediction system and to also obtain additional profiles of this type from other satellites including GRACE, CHAMP and METOP.

The gem model itself used in the global system remains unchanged.

This is a significant increase in the amount of temperature profile data and its impact is multiple. These data have very little biases and excel at detecting the location of the tropopause. During tests we noted improvements particularly above 300 hpa, both in terms of the first guess fit and analyses fit to radiosonde data. This is particularly true in the southern hemisphere but was noted in all areas. Another significant impact is in reduction of biases in the stratosphere, which has an impact on the assimilation of satellite radiance data such as AMSU and AIRS. We have noted a readjustment of the bias correction needed for these radiance data.

Changes to the regional data assimilation and forecast systems also include the use of a new radiative transfer scheme. The previous infrared and solar radiative transfer parametrization is now replaced by an accurate and more detailed new scheme that makes use of a correlated k-distribution (ckd) method described in Li and Barker (2005). The new scheme's main impact for short term forecasts is the reduction of a known overall cold bias at the surface. The implied general warming will produce a decrease of the frequency of strong cold errors. This effect will be greater than the increase in frequency of strong warm errors. In winter, warming is expected both at night and during the day. In summer, warming is expected at nighttime.

Another change in the regional model is the extension into the polar regions of the uniform-resolution core grid. This portion of the variable-resolution grid has been expanded over the arctic regions to include the north pole as well as most of the Arctic ocean. The horizontal resolution of the core grid has not been changed and still uses a 15-km grid spacing. This means weather systems impacting the Arctic regions and adjacent regions will be better analysed and forecast. It should be noted that new and old grid points are not collocated. As a consequence, users who extract data from direct model output may have to revise their procedure, such as the selection of a representative grid point for a given latitude/longitude location.

Additional information on the performance evaluations and impacts on other components and products resulting from these changes are described in the official announcement available here.

Users wanting more technical information on changes relating to the addition of new new satellite data can access the associated technical note available here.

Users wanting more technical information on changes that apply to the regional model can click here to access the technical document.

Wednesday March 4th, 2009

Public forecasts extending to include days 6 and 7

Starting on Wednesday March 4 2009 at 1700 UTC the public forecasts which currently cover days 1 to 5 will be extended to include days 6 and 7. This improvement will entail modifications to the production of the scribe matrices and to scribe itself.

Verifications have shown that the ensemble forecast system gave better results than using the global deterministic forecast sytem for days 6 and 7. The ensemble system also showed better stability from forecast jumpiness.

A new series of scribe matrices already in production from the Canadian ensemble forecast system and have been transmitted since november 12 2008. These matrices will now be used to produce the public forecasts of days 6 and 7. These matrices will be based on 0000 UTC and 1200 UTC integrations and will be available by 0530 UTC and 1730 UTC respectively.

For more information, please consult the official announcement.

Users wishing more technical information can access this technical document.

Wednesday May 28th, 2008

Improvements to the global, regional and ensemble assimilation systems of the canadian meteorological centre

Starting on May 28 at 12 UTC, the Canadian Meteorological Centre's global, regional and ensemble assimilation systems will have the following changes applied to their data assimilation systems:

First, radiance data from 87 infra-red AIRS channels from the AQUA satellite and 7 microwave channels from the SSM/I of the F-13 and F-14 satellites will be assimilated by the global and regional assimilation systems only.

Further more ocean surface winds from quickscat's scaterometer, GOES satellite winds from the 3.9 microns channel and supplemental data from AMSUA, AMSUB or MHS near swath edges from the NOAA and AQUA satellites will also be assimilated in the global, regional and the ensemble forecast system.

Also dynamical bias correction of satellite radiances will be accompanying this new package including the usage of version 8 of the fast radiative transfer model RTTOV and for the global and regional models the use of a new vertical interpolator from the RTTOV levels to the gem levels for the assimilation of radiances.

Finally AMSUA channel 3 data will not be assimilated anymore (due to contamination by contribution from the surface) and a new cloud filtering will be introduced for the AMSUB and MHS data.

The ensemble system will make use of the wind profiler data. This system will also be using the observations coming from the first quality control process (background check) instead of waiting for the final variational quality control. This results in a gain of 40 minutes in product issuance time with no significant decrease in the forecasts quality.

The gem model itself used in the global and regional forecast system remain unchanged except for a correction on the diffusion of the winds near the top of the model. Because of the additional data, the delivery of products may arrive 15 minutes later. For more information, please consult the official announcement.

Users wishing more technical information can access this technical document.

Tuesday March 11th, 2008

Improvements to the AQHI forecast program

On Tuesday March 11, 2008 several improvements were made to the processing system that supports the national Air Quality Health Index (AQHI) pilot project.

  1. The processing of observations was moved forward 10 minutes from 50 mins past the hour to 40 mins past the hour in order to improve the timeliness of the delivery of AQHI observations to the public.

  2. In preparation for the distribution of SCRIBE v3.10 (the first version of SCRIBE to include an air quality component) to the regions, model output from the operational air quality forecast model CHRONOS was added to the SCRIBE matrices.

  3. Code was put in place for the addition of the city of Montreal as an official pilot forecast site. However, public dissemination of the product for Montreal has not yet begun.

2008

CMC reduces from 24h to 12h the lag time in initial conditions of its automated seasonal forecast system

Description of the system: 2-tier system Persisted SST anomaly with no ocean model Sea-ice relaxed toward climatology 4 models GEMCLIM + SEF+ GCM2 + GCM3 40 members; 10 members each model 12 h lags initial conditions 12 seasons; 0 and 1 lead time Hindcasts 1969-2004

Thursday November 22nd, 2007

Extension of the National AQHI Pilot Project

On Thursday November 22, 2007 the Air Quality Health Index (AQHI) Pilot project was officially extended to 14 communities in British Columbia. These communities include 4 sites in the Metro Vancouver region, Central Fraser Valley, Eastern Fraser Valley, South, Central, and North Okanagan, Kamloops, Prince George, Quesnel, Nanaimo/Parksville and Victoria/Saanich.

The national AQHI Pilot project was officially launched July 9, 2007 as a joint effort involving Environment Canada, Health Canada, as well as various regional, municipal, and local health authorities. The 14 new sites join the city of Toronto which was the first community to participate in the program. More communities will join the program in the coming months.

The AQHI is the first health-based index that combines the quality of air with known health effects. The AQHI takes into account how the level of exposure to multiple pollutants, even at low levels of exposure, can affect your health. Observations of AQHI for a particular community are available every hour, with 2-day forecasts being issued twice daily.

For more information, please consult the following web-site:

Tuesday July 10th, 2007

Major improvements to the Ensemble Prediction System (EPS)

On Tuesday July 10 2007, starting with the 12 UTC run the Canadian Meteorological Centre (CMC) will implement an upgraded ensemble prediction system (EPS).

Changes to the ensemble analysis component of the system include a better simulation of the model error and the capacity to use observations valid at at different times in the data assimilation window.

Changes to the ensemble prediction system include the removal of the SEF model, the use of the GEM model as the main ensemble model, an increase in the number of members from 16 to 20, an increase in the horizontal resolution to 400x200 (0.9 degrees or 100 km at the equator) as well as a modernization of the parameterization schemes.

Due to the increased resolution and greater number of members of the system, the forecasts will be available about 30 minutes later than before. Work is underway to improve the delivery time of the EPS products.

Further details on this implementation are available in the official announcement.

Monday July 9th, 2007

Launch of the AQHI National Pilot Project

On Monday July 9, 2007 the national Air Quality Health Index (AQHI) pilot project was officially launched as a joint effort involving Environment Canada, Health Canada, as well as various regional, municipal, and local health authorities. The first community to participate in the program is the city of Toronto. Other communities will join the program in the coming months.

The new AQHI is the first health-based index that combines the quality of air with known health effects. The AQHI takes into account how the level of exposure to multiple pollutants, even at low levels of exposure, can affect your health. Observations of AQHI for a particular community are available every hour, with 2-day forecasts being issued twice daily.

Wednesday July 4, 2007

Major changes to regional forecast system

On Wednesday July 4, 2007, at 12 UTC, the Canadian Meteorological Centre will implement major changes to the operational regional forecast system.

The length of the regional data assimilation system (RDAS) cycle will be shortened from 12 to 6 hours. The RDAS leading to 12 UTC and 00 UTC regional analysis will be initiated from analysis of the global data assimilation system valid at 06 UTC and 18 UTC respectively.

The RDAS background error statistics will be updated allowing the vertical resolution of the regional analysis to increase from 28 to 58 levels. Furthermore the 3D assimilation scheme will use the FGAT procedure which stands for First Guess at Appropriate Time. That procedure allows the differences between the model trial field and the observations to be computed and taken into account at the time of the observation.

No change is done to the spatial and temporal data selection so the number of assimilated data in the RDAS remains similar. Finally the regional model will benefit from a new regional snow depth analysis using the regional model snowfall forecast as a trial field.

Further details on this implementation are available in the official announcement.

Users wishing more technical information can access this technical document.

2007

CMC's automated seasonal forecast system now includes four model types and additional members

The CMC automated seasonal forecast system now uses a total of 40 members from 4 different types of models: 10 members-GEMCLIM model, 10 members-GCM3 model, 10 members-SEF model and 10 members-GCM2 model. The system produces a forecast with zero month lead time and a forecast with one month lead time

Tuesday October 31, 2006

Major changes to global forecast system

On Tuesday, October 31, 2006, at 12 UTC, the Canadian Meteorological Centre will implement major changes to the operational global forecast system using the Global Environmental Multiscale (GEM) model.

Major changes include increasing significantly the vertical and horizontal resolutions, improving its physical parameterization, improving the condensation and precipitation package, introducing the more sophisticated ISBA surface scheme and improving the model physics. The surface parameterization scheme (ISBA) is now also used by the global model.

A number of improvements have also been introduced into the analysis used by the model a new set of background error statistics and an improvement of 40% in the efficiency of the 4d-var data assimilation system. The cut-off times for availability of observations have been slightly modified to deliver the operational analyses at about the same time as in the previous global forecast system.

Users will see changes in the reception time of global operational products with this implementation. In general, products from the global suite will be received about 40 minutes later than the current reception time. Grib data from this global system will be received 40-55 minutes later than the current times. We will be in a position to reduce this delay with a scheduled increase in computing power early in 2007.

Further details on this implementation are available in the official announcement.

Users wishing more technical information can access this technical document.

Tuesday December 13th, 2005

Major changes to the Ensemble Prediction System (EPS)

Some new types of observations will start being used by the Ensemble Kalman filter (EnKF) data-assimilation system which itself will also benefit from some structural improvements.

The ensemble of forecast models has been revised to arrive at a better simulation of the diurnal cycle and to have a reduction in model biases. The forecasts will be performed twice daily and the forecast period will be extended from 10 days to 16 days.

Further details about this implementation are available in the official announcement note.

Wednesday December 7th, 2005

Assimilation of new satellite data

On Wednesday December 7th 2005, the Canadian Meteorological Centre will update its Assimilation System by including the data from the NOAA-18 and METEODAT-8 satellites recently put into orbit. For the moment only the global and regional GEM models will benefit from this update. The impact on forecasts has been evaluated as being neutral, as expected for such updates.

Further details about this implementation are available in the official announcement note.

Thursday July 21st, 2005

Modifications to the Canadian Wave model

On Thursday July 21st 2005, the Canadian Meteorological Centre will implement modifications to the Wave Forecast System over the Northern domains of the Atlantic and Pacific Oceans. This implementation will also introduce 4 new windows of wave forecast to cover the following domains in the Great Lakes : Ontario, Erie, Huron and Superior.

Further details about this implementation are available in the official announcement note.

Wednesday July 6th, 2005

Improvements to Chronos air quality forecast

On July 6th 2005 the Canadian Meteorological Centre will implement new version of the Chronos air quality forecasting model. Anthrogenic emissions of 2000-2001 will replace the 1995 emissions. The processing of the emission inventory was done using SMOKE (Sparse Matrix Operator Kernel Emissions) instead of CEPS (Canadian Emissions Processing System). The biogenic emissions have also undergone significant modifications using BELD3 vegetation categories instead of USGS (U.S. Geological Survey) and upgrading the associated emissions rates to BEIS3 instead of BEIS2. In brief more vegetation types are now used with updated emissions rates.

Further details about this implementation are available in the official announcement note.

Improvements to the quantity of precipitation forecast in summer situations for the Regional GEM model

On July 6th 2005 the Canadian Meteorological Centre will implement a minor change to the regional forecasting system to improve the quantity of precipitation forecast in summer situations. In addition to this change, a new analysis of sea surface temperature will be introduced featuring higher resolution and fed with more data.

Further details about this implementation are available in the official announcement note.

Tuesday March 15th, 2005

4D-Var: a major improvement to the Canadian global data assimilation system

On Tuesday, March 15th 2005, the 3D-Var (3-Dimensional Variational) data assimilation system for the global forecasting suite will be extended to a 4D-Var scheme. This extension to include time as the fourth dimension of the analysis system implies that the analysis is no longer a simple snapshot of the atmospheric conditions at the time of the analysis (like in the 3D-Var which does this for the primary synoptic hours). Instead, the 4D-Var system provides a time evolution of the atmospheric conditions during the assimilation time window (for the current implementation, this window was chosen to maintain coherence with the 3D-Var, that is 6 hours centered on the main synoptic hours). The 4D-Var analysis is done by using the GEM forecast model itself as part of the assimilation process, and by using observations that are well distributed during the analysis time window. The 4D-Var then computes the best fit between these observations and the model forecasts. The resulting analyses show a much improved dynamical consistency and are accompanied by a significant improvement in the global system performance.

Further details about this implementation are available in the official announcement note.

Two published articles are also available.

(1) Click here for the June 2007 MWR article: Extension of 3DVAR to 4DVAR: Implementation of 4DVAR at the MSC.

(2) Click here for the June 2007 MWR article: Impact of Different Components of 4DVAR on the Global Forecast System of the MSC.

The original technical document that was made available at the time of the 2005 implementation is still available by clicking here. These documents on objective and subjective verification are also still available.

Thursday January 13th, 2005

Improvement to the surface temperature forecasts from the GEM regional model

On Thursday January 13, 2005, at 12 UTC, the Canadian Meteorological Centre will implement a minor change to the regional model forecast system in order to improve nocturnal surface temperature forecasts. It has been noted that the operational GEM regional model forecasts excessively cold nocturnal surface temperatures, especially during winter, under clear skies and in weak low-level circulation regimes. This problem has been noted more systematically in mountainous regions as well as in the Arctic. The change aims to correct this problem.

Further details about this implementation are available in the official announcement note.

Wednesday January 12th, 2005

Implementation of the Ensemble Kalman Filter data assimilation method in CMC's Ensemble Prediction System

As of Wednesday January 12 2005 at 00 UTC, the Canadian Meteorological Centre implemented a major change to the data assimilation of the Ensemble Prediction System. The Optimal Interpolation technique was replaced by an analysis cycle based on the Ensemble Kalman Filter (EnKF). The model configuration of the 16 members used to produce the 10 day forecasts remains unchanged. The first forecasts using the new assimilation technique were produced in the 00Z run of 13 January 2005.

You will find details about this implementation inside the official announcement note. A more detailed technical document on the Kalman Filter is available by clicking here.

Tuesday September 21st 2004

New observations incorporated in the regional and global data assimilation systems at CMC

As of Tuesday September 21 2004 at 1200 UTC new observations will be incorporated in the regional and global 3D-VAR (3-Dimensional VARiational) data assimilation system. First, IR radiance data from the 6.7 micron channel from the GOES-12 (East) satellite will be incorporated, in addition to the GOES-10 (West) data which is already used. Also, data from the AMSUA instrument onboard the NASA AQUA satellite will also be assimilated. This will complement the AMSUA data already used for the NOAA-15 and NOAA-16 satellites and will partly compensate for the loss of the AMSUA instrument for NOAA-17 which occurred in late October 2003. There are also some changes to the ATOVS (AMSUA and AMSUB) thinning process and the variational quality control will also be applied to ATOVS data. A third change is the incorporation of satellite winds in the Polar Regions obtained from the MODIS instrument onboard the NASA AQUA and TERRA satellites. Modifications to the data selection aiming at increasing the number of satellite winds from the GOES-P satellite are also included. Finally, data from selected USA wind profilers will also be assimilated.

You will find details about this implementation inside the official announcement note. A more detailed technical document is also available.

Tuesday May 18 2004

Implementation of the 15km GEM regional model and of improvements to the regional data assimilation system

CMC will implement major changes to the operational regional forecast system on Tuesday May 18 2004, at 12 UTC. The horizontal resolution of the GEM model will increase from .22 degree (about 24 km) to 0.1375 degree (about 15 km) in the uniform resolution portion of the grid. The number of vertical levels will increase from 28 to 58, the top of the model remaining at 10 HPa. The model's physics package will be improved. The Kain-Fritsch scheme will replace the Fritsch-Chappell scheme for deep convection while a new shallow convection scheme, based on a Kuo scheme closure, will complement the more selective deep convection scheme. As well, the package features a new sub grid scale orographic parameterization (gravity wave drag and a blocking term).

Changes will also be made to the regional data assimilation system. It will use of satellite radiance data from the AMSU/B instrument, in addition to the data from the AMSU/A instrument that are currently assimilated. Irradiance data from the 6.7 micron channel from the GOES-W satellite (GOES-IR) will also be assimilated. The addition of AMSU/B and GOES-IR data, which are sensitive to both atmospheric temperature and moisture, has the most impact over oceans where more channels are assimilated. These changes are the same that were successfully implemented in June 2003 in the global system. This upgrade to the regional system will bring the two systems to the same level.

You will find details about this implementation inside the official announcement note. Also a more detailed technical document is available.

2004

CMC replaces the SEF model with the GEMCLIM model in its automated seasonal forecast system

The CMC has replaced its SEF model with the GEMCLIM model. GEMCLIM (RPN/CMC) + GCM2(CCCma) From 4 to 12 seasons; still 0 lead time Hindcast extended: 1969-2004

Thrusday June 19 2003

Improvements to the global data assimilation system

As of Thursday June 19, 2003 at 1200 UTC the 3D-VAR (3-Dimensional VARiational) data assimilation system will now make use of satellite radiance data from the AMSU/B satellite instrument in addition to the data from the AMSU/A instrument which are currently assimilated. The AMSU instruments are part of the ATOVS sounder on-board the NOAA-15, NOAA-16 and NOAA-17 satellites. As well, the assimilation of IR radiance data from the 6.7 micron channel on the GOES-W will be introduced. Since AMSU/B and GOES-IR data are sensitive to both temperature and moisture, the addition of the new data has a large impact on the moisture analyses that are used by the numerical models. It is expected that the use of these new data will result in significant improvements in moisture analyses and some modest improvement in the model performance. You will find details about this implementation inside the official announcement note.

Due to time constraints imposed by the conversion to the new IBM supercomputer, the regional data assimilation and forecast system will not make use of these new datasets. The regional system will continue to use the same dataset as it currently does. There will however be some impact on the regional system due to the use of the global analyses at the beginning of the regional spin-up cycle. Work is underway to incorporate the AMSU/B-GOES data in the regional system, but this modification will not be implemented until the new supercomputer is operational.

Febuary, March and April 2002

Minor changes to the operational system in February, March and April 2002:

19 Febuary 2002: a correction was made in the regional and global assimilation system to the thinning of SATWINDS data in order to ensure a better treatment of the data near the edges of the selection boxes. As well, a correction to the Humsat system was implemented. This corrected a bug which in certain cases was causing a erroneous evalution of the value of the cloud fraction during daytime.

27 March 2002: the thinning process of aircraft data was improved in the data assimilation systems. Also, the rejection criterea applied to "dropsonde" data within the background check process was modified to reduce the number of wind data rejections over the Pacific Ocean. As well, a correction was made to the interpolation process near the computing pole of the grid in order to eliminate problems when processing data at that point.

24 avril 2002: a modification was made to the calculation of the geopotential in pressure coordinates in the GEM regional model. Since 11 January 2001 the geopotential was computed using a cubic interpolation of the model fields in eta coordinates. Recently, it was clearly demonstrated that this method caused a degradation of the geopotential near and above 250 hPa level. Consequently, the previous more precise method was re-introduced which is a linear interpolation done after the recomputing of the geopotential. The meteorological impact of this problem is fairly limited. However, the problem had a negative effect on the monthly verification scores of the geopotential near and above 250 hPa level.

Tuesday, December 11, 2001, 12Z

Implementation of a major change to the analysis and of an improved global GEM model

CMC Operations will implement a major change to the global and regional assimilation systems Tuesday 11 December 2001 at 12UTC. As well modifications will be implemented to the GEM global model in order to improve the treatment of winds over the mountains. The most important change to the assimilation scheme is that temperature and surface pressure will now be directly assimilated thus replacing geopotential height which will now be a diagnosed variable. Also, assimilated radiosonde levels will pass from 16 to 27 levels. Additional datasets will be assimilated. These are dropsondes, temperature data from AMDAR and ACARS, five additional TOVS channels below 500 hPa and satwinds from GOES 8 and 10. The new assimilation scheme will also include a full 3-D VAR quality control module in addition to a background check for the detection of gross errors, thus replacing the OI quality control package. All the above changes are common to both global and regional assimilation systems.

Changes in the GEM global model include a numerical correction to the existing gravity-wave drag scheme and the addition of a low level blocking term to the sub-grid scale orographic scheme. Practically this will result in a better handling of the wind over high mountains like the Rockies and the Himalayas. The regional model remains unchanged.

You will find the official announcement on this change here. A detailed document (PDF Version) describes all the changes and their significant positive impacts on the global and regional analysis and forecast systems.

Tuesday, September 11, 2001, 12Z

Implementation of the ISBA surface scheme in the GEM regional model

CMC Operations implemented a change to the regional forecast system Tuesday 11 September 2001 at 12UTC. A new surface modeling scheme known as ISBA (Interactions, Surface, Biosphere, Atmosphere) replaced the so-called "force-restore" module. The ISBA scheme improves the treatment of surface processes by taking into account the surface fluxes of heat, moisture, and momentum over the various surfaces (land, water, sea ice, and glaciers). The fluxes calculated over these four land types are combined over each grid tile according to their respective weights. These fluxes are then used as a lower boundary for the vertical diffusion. The "force-restore" scheme considered only one land type for each model grid title.

You will find detailed information on this change and its impact in the official note.

Thursday, June 7, 2001, 06Z

Implementation of ATOVS level 1B data in the of global and regional analysis system

CMC Operations implemented a change to the global and regional data assimilation and forecast systems Thursday June 7 2001 at 0600 UTC. The ATOVS-1B data have replaced the TOVS-1D data which were received via the GTS (Global Telecommunication System). The new ATOVS data are now collected directly from a NESDIS server via the internet. As well note that the ATOVS data are obtained from the NOAA-15 and NOAA-16 satellites. The new raw ATOVS-1B data are calibrated at CMC. The new dataset is significantly larger and amounts to approximately 670,000 observations per 24 hours. The impact of the introduction of the ATOVS-1B data was studied in the last few months. The results suggest a general improvement in forecast quality especially in the Southern Hemisphere.

Thursday, January 11, 2001, 12Z

Improvements to the regional analysis: new "eta" coordinates and addition of TOVS and ACARS/AMDAR data

CMC Operations will implement changes to the regional data assimilation and forecast systems Thursday January 11 2000 at 1200 UTC. We will introduce in the regional analysis system the modifications which were made to the global system last June 2000 (analysis on eta model coordinates) and in last September 2000 (addition of TOVS and ACARS/AMDAR data). The 3D-VAR (3-Dimensional VARiational) data assimilation will then be performed on the model coordinates and, as well, will make use of automated aircraft observations, that is ACARS and AMDAR data, in addition to the AIREP data which are currently assimilated. We are also incorporating satellite radiances from NOAA-14 TOVS and NOAA-15 ATOVS instead of the NOAA-14 SATEM data (temperature profiles derived from radiances). You will find detailed information on this change in the official note. As well, technical notes are available on the assimation of TOVS and ACARS/AMDARS data.

Wednesday, September 27, 2000, 12Z

Improvements to the global analysis: addition of TOVS and ACARS/AMDAR data

CMC Operations will implement changes to the global data assimilation and forecast systems Wednesday September 27 2000 at 1200 UTC. The 3D-VAR (3-Dimensional VARiational) data assimilation will now make use of automated aircraft observations, that is ACARS and AMDAR data, in addition to the AIREP data which are currently assimilated. We are also incorporating satellite radiances from NOAA-14 TOVS and NOAA-15 ATOVS instead of the NOAA-14 SATEM data (temperature profiles derived from radiances). You will find detailed information on this change in the official note. As well, technical notes are available on the assimation of TOVS and ACARS/AMDARS data. Note that the regional assimilation and forecast system remains unchanged.

Wednesday, June 14 2000, 12Z

Implementation of a new global analysis on "eta" coordinates

CMC Operations will implement changes to the global data assimilation and forecast systems Wednesday 14 June 2000 at the 12Z run. The new global analysis on "eta" coordinates will be implemented. The 3D-Var data assimilation will be performed directly on the GEM model 28 eta vertical levels rather than on the current 16 standard pressure levels. At the same time, a new version of the GEM model in the global configuration will also be implemented. The main impact of this last change is to increase the level of activity in the model. You will find more information in the official note. The regional assimilation and forecast system remains unchanged.

Wednesday, October 14 1998, 12Z

Implementation of a new global forecast model

CMC Operations will implement the global uniform grid version of the GEM model Wednesday October 14th at the 12Z run. The GEM model in global configuration will replace the T199 version of the global SEF spectral model which is currently used to produce the medium and long range forecasts at CMC. The new model, which has a latitude-longitude resolution of 0.9°, will be also used in the global data assimilation system to generate the trial fields for the global analyses. The official note (GENOT) describes the main differences between GEM global and the spectral SEF model and summarizes the results of the objective and subjective evaluations of the parallel run which has been in place during the past 5 months. The article from Côté et al. (1997) (PDF Version) gives additional details on the GEM model.

With this implementation, CMC now has a unified forecasting system in which a single model (GEM) is used in different configurations (global, regional, local). After running a barotropic model since the 1960s CMC implemented the first version of the spectral hemispheric T20L5 model in 1976. Fifteen years later, in 1991, the global version of the spectral SEF model (T79L21) was implemented. The model's horizontal resolution was increased to T119 in 1993, then to T199 (100km) in 1995. Note that the SEF model will continue to be used for the monthly and seasonal forecasts as well as in the ensemble forecasting system.

Tuesday, September 15 1998, 12Z

Implementation of the GEM, regional 24 km model.

CMC Operations will implement a new version of the GEM regional model on Tuesday, September 15 1998 at the 12Z run. The main changes affect the horizontal resolution of the model grid, which is increased from 35 km to 24 km, and the parametrization of the physical processes. For more information, please consult the official note (GENOT).

Wednesday, March 11, 1998, 12Z

Implementation of the new snow depth analysis.

CMC Operations will implement a new snow depth analysis on Wednesday, March 11, 1998, 12Z. This new analysis will have, among other things, an improved temporal and spatial resolution and will take into acount intermediate synoptic observations. In addition, we have significantly improved the analysis algorithm. For more information, please consult the official note (GENOT) or the detailed documentation (PDF Version) about impacts of this new analysis.

Thursday, January 22, 1998, 00Z

Improvement to the «force-restore» scheme of GEM.

CMC Operations implemented a modification to the GEM model on Thursday January 22nd 1998 at 00Z. This modification is an improvement to the «force-restore» scheme which allows to calculate the model surface temperature and humidity. These modifications will improve the surface temperature and dew point forecast during the winter season. An official note (GENOT) announcing this modification summarizes the results of the objective and subjective evaluations of the parallel run.

Wednesday, September 24, 1997, 12Z

Reintroduction of the regional data assimilation system at CMC

CMC Operations will reintroduce the regional data assimilation system (12 hour spin-up) on Wednesday September 24 1997 at 12Z. This system had been turned off when the GEM model replaced the RFE model last February. The 3D-VAR (3-dimensional variational) data assimilation system was adapted to the regional system which uses the GEM model to produce the 6 hour trial fields needed by the analysis. The official note announcing this modification summarizes also the results of the parallel run.

Wednesday, June 18th 1997 at 12Z

3D-VAR, the new data assimilation system at CMC was implemented

CMC Operations has implemented a new data assimilation system on Wednesday June 18th 1997 at 12Z. The 3D-VAR (3-dimensional variational) data assimilation system has replaced the operational OI (optimum interpolation) scheme that has been used at CMC since the last 20 years. The official note announcing this modification summarizes also the results of the parallel run. A detailed documentation on this new analysis is in preparation. In the meantime, you can consult the images from a Powerpoint presentation prepared by Pierre Koclas for the CMOS conference which was held in Saskaton in early June.

Tuesday June 10th 1997 at 12Z

Modification to the analysis feeding GEM regional implemented

CMC Operations implemented a modification of the regional forecast system on Tuesday June 10th 1997 at 12Z. The GEM regional model is now fed by the 21 level global analysis in sigma coordinates. Since the introduction of GEM, last February 24th, the model was being fed by the 16 level pressure global analysis. An official note announcing this modification summarizes the results of the objective and subjective evaluations of the parallel run.

Monday 24 February 1997 at 12Z

The new GEM regional model was implemented

The new GEM regional model was implemented On Monday 24 February 1997 at 12Z CMC Operations implemented a new regional forecast model: the GEM (Global Environmental Multiscale) model. This new model replaces the operational RFE model that has been used to produce short range forecasts at CMC since April 22 1986. An official note announcing the implementation of GEM describes the differences between GEM and RFE and summarizes the results of the objective and subjective evaluations of the parallel run which has run during the last few months. The article by Côté et al. (1997) (PDF Version) describes the development and the characteristics of the GEM model and offers, as well, a more detailed discussion of the parallel run evaluation.

1996

The Statistical Canonical Correlation Analysis (CCA) is now used for forecasts of month 4-12 over Canada

1995

CMC introduces its first automated seasonal forecast based on a ensemble atmospheric dynamical prediction system

This new system includes - A 2-tier approach - Persisted Sea Surface Temperature (SST) anomaly with no ocean model - Sea-ice relaxed toward climatology - Two atmospheric models: SEF (RPN/CMC) + GCM2(CCCma) - 12 members; 6 members SEF and 6 members GCM2 - Initial conditions: 24 hour lags - 4 seasons; 0-month lead time; 3 months long - Hindcast: 1969-1994