The national forecasting centres (e.g. NCEP in the USA, Meteo France, UK Met, DWD) generally make their model data available in what is called grib or grib2 format. For more info see:
https://en.wikipedia.org/wiki/GRIB
This comes straight off the supercomputers (NCEP's is known as WCOSSS) and is basically dumped onto file servers. It is then up to the end user (whether that be one of the big agencies or people like me) to take that and do something with it.
In some cases further processing is done by the national forecasting centre, for example cluster analysis on the ensemble output. This may or may not be made available to other users in grib2 format. NCEP pushes everything out freely as far as I am aware. In the UK the Met Office have a product called Decider which I don't currently have access to but which is available to the "public" to licence. (See https://www.metoffice.gov.uk/services/business-industry/energy/decider).
Therefore, I think the idea that organisation X, Y or Z has a "secret stash of data" is very misleading. If they can leverage advantage it will mostly be through having the best meteorologists to produce the forecasts and the best analysts to process the data sets and produce tools to make it easier to identify trends and the most probable scenarios.
Originally Posted by: Brian Gaze