Production suite overview

Main Page

Analysing and forecasting the weather, using numerical models, requires an impressive amount of computing and data processing power. This is why a powerful supercomputer is needed to ingest and process all this data. The process which leads to the production of a forecast can be seen as an endless cycle of computer processes, which are called "operational runs". A complete operational run can be summarized as follows:

Data ingest:

decoding, re-formatting and quality control of large amounts of meteorological data from many sources worldwide.

Data assimilation (analysis):

derivation of an up-to-date snapshot of the state of the atmosphere. This must be an accurate three-dimensional depiction of atmospheric winds, pressure, moisture, and temperature.

As a first step in the assimilation of the observational data into the forecasting system, the analysis begins with a best guess, called a trial field, which is a 6-hour model forecast representing our accumulated knowledge of the state and trends of the real atmosphere. The new analysis is then obtained by using sophisticated mathematical methods to strike a judicious balance between the trial field and the newly-acquired observational data. The end-result is called an objective analysis.

Forecasting:

prediction of the behaviour of the atmosphere for up to ten days into the future.

The updated objective analysis is used as the starting point in running the GEM (Global Environmental Multiscale) atmospheric model.

Weather models, such as GEM, use a three-dimensional mathematical grid to represent space. Generally speaking, the tighter the grid, the more accurate the model's representation of the atmosphere becomes. But a tight model grid comes at a steep price in execution costs. A balance must therefore be reached to run the model at the highest possible resolution, while attaining the best results in terms of economy and timeliness. For this reason, the GEM model is run on two types of operational grids: an evenly-spaced global grid, well suited for medium-term forecasts (3-10 days), and a so-called "regional" grid, featuring a variable-mesh grid which is tighter over North America and adjacent waters. The regional grid's higher resolution is well suited for short-term forecasts (up to 2 days) over the North-American region and the adjacent Oceans

Post-Processing:

raw numerical results from the GEM model are fed into post-processing systems to generate finished and semi-finished forecast products: gridded numerical data, charts, bulletins, etc.

The GEM model output is also used as input to other models, for the prediction of

• ocean waves
• volcanic ash/radionucleide atmospheric transport
• air quality (smog)
• stratospheric ozone and solar UV radiation

Other refinements include the use of statistical methods to improve on the raw numerical forecast of temperature, wind, and probability of precipitation, for specific locations.

Dissemination:

large amounts of data produced by the analysis and forecasting system are sent to offices of the Meteorological Service of Canada which, in turn, use them as input into their local forecasts or disseminate them to the Canadian public and other clients. In addition, data from the operational runs are sent directly to a wide variety of outside users including federal and provincial agencies, the Departement of National Defense, Nav Canada, and to some major commercial clients. Data are sent out via satellite or land telecommunication networks, as well as more and more being made available on the world-wide web.

Recent changes to the operational runs

Technical outline of the global forecast system (data assimilation and operational model)

Technical outline of the regional forecast system (data assimilation and operational model)

Technical outline of the surface analyses

Technical outline and timetable of the main operational runs

Timelines of Numerical Weather Prediction in Canada

The Green Lane^TM,
Environment Canada's World Wide Web Site.