The 1954 start of operational Numerical Weather Prediction

The 1954 start of operational Numerical Weather Prediction in Sweden NWP Symposium University of Maryland June 2004

Why Sweden? -C. G. Rossby back from the US 1947 (home-longing, management and politics) -Swedish state-of-art computers (BARK 1950, BESK 1953) -International support (from the USA and Belgium) NWP Symposium University of Maryland June 2004

1953 BESK= Binary Electronic Sequence Calculator NWP Symposium University of Maryland June 2004

BESK In 1953 the world’s “best” computer Arithmetic unit Control desk Electrostatic drum memory “Williams tube” memory NWP Symposium University of Maryland June 2004

The International Meteorological Institute Steyer Hubert Vuorela Eady Rossby V. Mieghem Rossby wanted, like V. Bjerknes after WWI, to play an international role in the political reconstruction after WWII NWP Symposium University of Maryland June 2004

Baroclinic or barotropic models? …yes, but the large scale motion can kinematically be described by a barotropic model The atmospheric motions are driven by thermal processes as reflected in baroclinic developments C. G. Rossby R. C. Sutcliffe NWP Symposium University of Maryland June 2004

Common misinterpretations of Rossby’s barotropic concept -Only valid for stationary waves -Can only perform linear extrapolations -Only valid for barotropic features -Group velocity…. NWP Symposium University of Maryland June 2004

The concept of group velocity tells us that any boundaries have to be placed far enough away from the verification region 24 -hour computational areas for - MISU - UKMO NWP Symposium University of Maryland June 2004

A +24 hour forecast from the UKMO system winter 1953 -54 Boundary errors and numerical instability NWP Symposium University of Maryland June 2004

Only in 1965 did the UKMO go operational with NWP after a personal intervention by the new Director General John Mason Dear members of the press, radio and TV. . . Oh dear. . . NWP Symposium University of Maryland June 2004

In Stockholm tendency calculations were carried out, first by hand in 1952, later on BESK in 1953 The Swedish tabloid “Expressen” published the results almost a year before “Tellus” (forecast left, analysis right) NWP Symposium University of Maryland June 2004

Barotropic re-run of the “tree-feller” storm the 2 -3 January 1954 2 January 1954 00 UTC +24 h barotropic forecast 3 January 1954 00 UTC NWP Symposium University of Maryland June 2004

First real time NWP in late September 1954 Not quite the 72 hour computational area 72 hour verification area NWP Symposium University of Maryland June 2004

From the operational period Dec 1954 -May 1955 -Professor Dahlqvist, when is Spring coming? NWP Symposium University of Maryland June 2004

-Tomorrow at 2 pm! +72 hours NWP Symposium University of Maryland June 2004

…and Spring came! NWP Symposium University of Maryland June 2004

Why Sweden succeeded: -C. G. Rossby back from the USA 1947 -Swedish state-of-art computers -International support -Choice of barotropic model (skilful and providing operational experience) -Efficient automatic analysis system (Bergthorsson-Döös, 1955) -Not too small NWP area (but not too big either, avoiding retrogression of the planetary waves!) NWP Symposium University of Maryland June 2004

From wheat to bread! Bert Bolin shows automated 500 mb forecasts for Ragnar Fjørtoft (Norway) and George Corby (UK) around 1956 NWP Symposium University of Maryland June 2004