Swedish Numerical Weather Prediction: The Forgotten First

Summary

Sweden ran the world’s first operational numerical weather prediction (NWP) forecasts in 1954, a full year before the United States. The effort was led by Carl-Gustaf Rossby at Stockholm University’s Institute of Meteorology (MISU), used the BESK computer (briefly the fastest in the world), ran a barotropic model, and was operationally sponsored by the Royal Swedish Air Force – not SMHI, which was initially skeptical. This fact is surprisingly absent from most popular accounts of NWP history, which tend to credit the US Joint Numerical Weather Prediction Unit (JNWPU), which began operations in May 1955.

1. Carl-Gustaf Rossby’s Return to Sweden (1947)

Why He Left Chicago

In 1947, at the height of his career in the United States – holding the chair of the Department of Meteorology at the University of Chicago, having founded journals, reorganized the AMS, and trained thousands of military meteorologists during WWII – Rossby made the surprising decision to return to his native Sweden. He was invited by the Swedish government, which created a personal professorship for him at Stockholm University (then Stockholms Hogskola).

What He Built

Rossby became the first professor at the newly established Department of Meteorology (MISU) at Stockholm University in 1947. In parallel, he created the International Meteorological Institute (IMI), formally established by a decision of the Swedish Parliament in 1955 (though it had functioned informally under Rossby since at least 1947–1948). The IMI’s stated purpose: “to conduct research in meteorology and associated fields and to promote international scientific co-operation within meteorology.”

With dedicated funding from the Swedish government, IMI brought leading scientists to Stockholm for sabbaticals and shorter visits. Rossby also founded the journal Tellus (first issue 1949), published by the Swedish Geophysical Society, to serve as a counterweight to existing European meteorological publications. In its very first year, Tellus published articles by some of the biggest names in meteorology from around the world. Crucially, the landmark Charney-Fjortoft-von Neumann paper on the first ENIAC weather forecast was published in Tellus.

The Bridge Figure

Rossby was the unique connector between three traditions:

The Bergen School (Vilhelm Bjerknes) – where he trained as a young man from 1919
American meteorology (MIT, Chicago, Woods Hole) – which he largely built
Stockholm – where he returned to create an international research centre

He did not fully abandon the US. He divided his time between Stockholm, the University of Chicago, and Woods Hole throughout the period 1947–1957. As he said: “Communications are the alpha and omega of meteorology.”

2. The NWP Group in Stockholm

Key People

Rossby assembled a remarkably international group at MISU/IMI. The core members involved in the NWP effort included:

Swedes:

Bert Bolin (1925–2007) – Rossby’s most distinguished student. Met Rossby during military service in Stockholm in 1947. Spent 1950–1951 at IAS Princeton working on the ENIAC forecasts with Charney and von Neumann. Returned to Stockholm and became a key contributor to the 1954 operational forecasts. Published “Numerical forecasting with the barotropic model” (Tellus, 1955) extending forecasts to 72 hours. Received his Ph.D. in 1956. Succeeded Rossby as director of IMI in 1957. Later became first chairman of the IPCC (1988–1997). Supervised Paul Crutzen’s doctoral work (Crutzen went on to win the 1995 Nobel Prize in Chemistry).
Bo Doos (1922–2010) – Swedish meteorologist, integral to the 1954 operational forecasts on BESK. Co-authored “Numerical weather map analysis” with Bergthorsson (Tellus, 1955), presenting a method to analyse upper-air charts numerically. Professor at MISU until 1970. Later served as director of the Global Atmospheric Research Programme (GARP) from 1971 to 1982. Member of the Royal Swedish Academy of Sciences.

Icelanders:

Pall Bergthorsson (Iceland) – Key participant in the 1954 Stockholm forecasts. Co-authored the Bergthorsson-Doos method for numerical weather map analysis. Two of the forecasts he worked on, for 23 and 24 March 1954 at 03 GMT, are believed to represent the first forecasts finished in time to be of use in operation. Later served as director of the Icelandic Meteorological Office (1989–1993).
Geirmundur Arnason (1915–?, born Akureyri, Iceland) – Fil.Lic. in Meteorology from Stockholm (1951), Ph.D. from Stockholm (1964). Research Associate and Lecturer at Stockholm 1951–1955. Published on baroclinic models in Tellus (1952, 1953). Later became Senior Research Scientist at the US JNWPU (1955–1958), then Director of Research at US Navy Fleet Numerical Weather Prediction Facility, and finally Professor at SUNY Albany.

Danes:

Aksel Wiin-Nielsen (1924–2010) – Danish meteorologist who moved to Stockholm’s Institute of Meteorology under Rossby. He participated in “the first numerical prediction that completed its computation ahead of the time for which the forecast was made.” Later spent 15 years in the US (JNWPU, NCAR, University of Michigan). Became the first Director of ECMWF (1974–1979), then Secretary-General of WMO (1980–1983). A direct line from Rossby’s Stockholm to the European centre that now leads global weather prediction.

Norwegians:

Arnt Eliassen (1915–2000) – Norwegian meteorologist, a pioneer in numerical analysis for weather forecasting. Had a brief stint at the International Meteorological Institute in Stockholm under Rossby in the late 1940s. Also worked at IAS Princeton with Charney (1948–1949) on the foundational ENIAC forecasts. Won the Carl-Gustaf Rossby Research Medal in 1965. Primarily based at the Norwegian Meteorological Institute and University of Oslo, but a key collaborator.

Americans and Other Visitors:

Norman Phillips (USA) – Rossby invited Phillips to Stockholm in autumn 1953 to help set up a weather model on the BESK computer. Phillips’s role was to rewrite the code that existed for ENIAC for BESK. This was significant: it would take a year and a half before the Americans at JNWPU were able to produce their first operational forecast.
Philip Thompson (USA) – American military meteorologist who considered Rossby’s department a “listening post” for meteorological intelligence. Established a US Air Force research centre in Sweden in association with MISU. He was the first director of the computer project at Princeton (1948–1949).
Chester Newton (USA) – After graduating from the University of Chicago (1946–1951), Newton and his wife Harriet were invited to work at IMI Stockholm (1951–1953) on a two-year visit.
Erik Palmen (Finland) – Distinguished Finnish meteorologist, collaborator of Rossby’s at both Chicago and Stockholm. Member of IMI’s scientific council from 1955.
H. Bedient (USA) – Listed as a participant in the 1954 forecasting work.

3. The SMHI Connection

Initial Skepticism

When Rossby approached the Swedish Meteorological and Hydrological Institute (SMHI) about implementing operational NWP, the agency’s leadership was skeptical. They believed something this experimental should be left to universities, not an operational weather service. This institutional conservatism forced Rossby to find an alternative sponsor.

The Military Weather Central (MVC)

Rossby turned to the Military Weather Central (MVC) under the Royal Swedish Air Force. The head of the MVC, Oskar Herlin, was described as “a practical man” who had observed stagnation in conventional techniques for short-range forecasting. The relatively modest cost (compared to military budgets) made him receptive. Herlin’s advocacy convinced the Swedish Air Force to fund the operational NWP trials.

SMHI’s Gradual Adoption

The spectacular success of the 1954 forecasts during military exercises eventually shifted SMHI’s institutional attitude. New SMHI leadership recognised computing’s meteorological potential. However, the timeline of SMHI’s adoption was slow:

1954: Operational NWP run by Air Force Weather Service + MISU, not SMHI
1961: SMHI finally operationalised NWP on the BESK computer
1973: SMHI assumed full responsibility for all NWP production in Sweden

This 19-year gap between the first Swedish operational forecast and SMHI taking full ownership is itself a remarkable story about institutional inertia.

4. The First Operational Forecasts (1954)

The March 1954 Milestone

The earliest forecasts that can be called “operational” in the sense of being completed in time to be of use were computed for 23 and 24 March 1954 at 03 GMT. These are believed to represent the first NWP forecasts anywhere in the world that were finished ahead of the time for which the forecast was valid.

The September 1954 Military Exercise

The defining moment came in September 1954, when 45000 troops in central Sweden were conducting military manoeuvres that included testing nuclear protection equipment. Real-time forecasts of upper-air movement – strongly determinative of the trajectory of simulated nuclear fallout – would therefore be critical.

In what amounted to a six-week operational trial, Rossby’s group produced real-time NWP forecasts during the exercise. The results were dramatic: the operationalised NWP forecasts turned out to be tremendously successful compared to subjective (human-only) forecasts. This success validated the technology and generated institutional momentum for continuing the programme.

Routine Operations from December 1954

Starting in December 1954, the Royal Swedish Air Force Weather Service in Stockholm made weather forecasts for the North Atlantic region three times a week using the BESK computer running the barotropic model developed at MISU. These were the first routine real-time numerical weather forecasts in the world.

The Model

Type: Barotropic model (single-layer atmosphere)
Equation: Barotropic vorticity equation
Level: 500 hPa (500 mb) pressure surface – geopotential height
Forecast range: Initially 24 hours; Bolin extended to 48 and 72 hours by 1955
Forecast area: North Atlantic region, approximately 9000 x 12000 km
Performance (Bolin 1955): Correlation between computed and observed changes: 0.85 for 24h, 0.82 for 48h, 0.70 for 72h forecasts over Western Europe. The most successful 72-hour forecast achieved a correlation of 0.87.

The Workflow

The operational workflow required close coordination between military and academic institutions:

The Swedish Air Force Weather Service prepared weather observation data and upper-air charts
Weather maps were sent by military couriers to the BESK computer installation on Drottninggatan in Stockholm
The data was entered into the computer and the barotropic model was run
Forecast output was returned to the weather service for operational use
Forecasts were produced three times per week

Key Publications

Staff Members, Institute of Meteorology, University of Stockholm. “Results of forecasting with the barotropic model on an electronic computer (BESK).” Tellus 6 (1954): 139–149.
Bolin, B. “Numerical forecasting with the barotropic model.” Tellus 7 (1955): 27–49.
Bergthorsson, P. and Doos, B. “Numerical weather map analysis.” Tellus 7 (1955): 329–340.
Bolin, B. “An improved barotropic model and some aspects of using the balance equation for three-dimensional flow.” Tellus 8 (1956): 61–75.

5. The BESK Computer

Background: Sweden Builds Its Own Computer

In the late 1940s, the Swedish Board for Computing Machinery (Matematikmaskinnamnden, later Matematikmaskinnamnden or “SBCM”) initially attempted to procure computers from the United States. When the US State Department imposed export controls on computer technology in 1948 (Cold War restrictions), Sweden pivoted to domestic computer development.

Technical Specifications

Full name: Binar Elektronisk SekvensKalkylator (Binary Electronic Sequence Calculator)
Completed: 1953 (operational December 1953)
Operational until: 1966
Architecture: Based on von Neumann’s IAS machine design from Princeton
Word length: 40 bits
Memory: 512 words electrostatic (Williams tube) memory; upgraded with ferrite core memory in 1956
Instruction length: 20 bits (2 instructions per 40-bit word)
Speed: Addition in 56 microseconds; multiplication in 350 microseconds (~18000 additions/second)
Hardware: 2400 vacuum tubes and 400 germanium diodes (partly solid state)
Power consumption: 15 kVA
Developer: Led initially by Conny Palm (died December 1951), then Stig Comet; hardware by Erik Stemme; architecture by Gosta Neovius and Olle Karlqvist

Briefly the World’s Fastest

For a short period after completion in 1953, BESK was the fastest computer in the world. This was a remarkable achievement for a small country.

Shared Use

BESK was not dedicated to weather forecasting. Computer time was expensive and shared among multiple users:

Meteorology: Weather data for Rossby’s group and SMHI
Military aviation: Wing profiles for the Saab Lansen attack aircraft
Telecommunications: Statistics for Televerket (Swedish telecom)
Road engineering: Road profiles for Vagverket (road authority)
Signals intelligence: During nights, the Swedish National Defence Radio Establishment (FRA) used BESK for cracking encryption of radio messages

This shared-time arrangement meant that weather forecasting had to compete for machine time – one reason forecasts were initially limited to three times per week rather than daily.

Comparison to the ENIAC

The Inventing Europe project notes that BESK was “similar to ENIAC, but considerably faster.” Where a 24-hour forecast on ENIAC took approximately 24 hours to compute (making it useless as an operational tool), the BESK could complete the same computation well ahead of the valid time – the crucial threshold that separates a “research experiment” from an “operational forecast.”

Successor: Facit EDB

BESK was eventually succeeded by the Facit EDB computer. Paul Crutzen, who arrived at MISU in 1959 as a computer programmer, later noted that Stockholm University housed among the fastest computers in the world with BESK and its successor.

6. Comparison with the United States

Timeline

Event	Sweden	United States
First forecasts completed ahead of valid time	March 1954 (BESK)	Not until 1955
First operational trial during real-world exercise	September 1954	–
Routine operational forecasts begin	December 1954	May 1955 (JNWPU)
Operational agency fully owns NWP production	1973 (SMHI)	1955 (JNWPU)

The JNWPU

The US Joint Numerical Weather Prediction Unit (JNWPU) – a joint project of the US Air Force, Navy, and Weather Bureau – opened in Suitland, Maryland in 1954 and began routine real-time forecasting in May 1955, using a newly purchased IBM 701. However, the initial US results were disappointing: “when the model began running operationally later in 1955, the results were very disappointing and not usable by forecasters.”

By contrast, the Swedish system had been validated during the September 1954 military exercise and was producing useful forecasts from December 1954.

Why Is This Not Better Known?

Several factors explain why Sweden’s priority in operational NWP is often overlooked:

US dominance of the narrative. The history of computing and NWP is often told as an American story: ENIAC, von Neumann, Charney, JNWPU. The Swedish work was published in Tellus rather than American journals and attracted less attention in the Anglophone world.
Institutional ambiguity. The Swedish forecasts were produced by an ad hoc collaboration between the Air Force and a university department, not a permanent national forecasting agency. SMHI did not take over until 1961/1973. The JNWPU, by contrast, was a clearly defined operational unit.
Scale. The US built a permanent, well-funded organisation that ran forecasts twice daily on a dedicated computer. Sweden ran three forecasts a week on shared time on a computer that also calculated wing profiles and cracked codes.
Rossby’s death. Rossby died suddenly in August 1957, removing the charismatic advocate who might have championed the Swedish achievement more forcefully on the international stage.
Swedish modesty. Anders Persson of SMHI has been one of the few historians to document the Swedish NWP story in detail (see his 2005 paper series in Meteorological Applications). His work is the primary English-language source for much of this history.

Other Countries

For comparison, the global timeline of first operational NWP:

Country	Year of First Operational NWP
Sweden	1954
United States	1955
Japan	1959
Canada	1963
United Kingdom	1965 (Met Office, using KDF9 computer)
Australia	1969

UK note: Britain conducted early experimental NWP work in 1952 (Fred Bushby and Mavis Hinds, under guidance of John Sawyer, using the EDSAC computer at Cambridge and the LEO computer of Lyons Co., with a 12 x 8 grid and 260 km spacing). However, the Met Office did not produce its first operational NWP forecast until 2 November 1965, using the English Electric KDF9 computer. A Ferranti Mercury (“Meteor”) had been installed at Dunstable in 1959 for NWP research, and an operational trial ran in early 1960.

7. Rossby’s Role: The Connector

Rossby was the indispensable catalyst for Swedish NWP. Without him:

There would have been no MISU or IMI at Stockholm University
There would have been no international talent pipeline bringing Bolin, Doos, Bergthorsson, Arnason, Phillips, Newton, Thompson, Wiin-Nielsen, and others to Stockholm
There would have been no Tellus to publish the results
There would have been no connection to von Neumann’s computer project at Princeton (Rossby had been contacted by von Neumann in 1946 about using ENIAC for weather prediction)
There would have been no advocate who could bypass SMHI’s conservatism and secure military funding through Oskar Herlin

Rossby’s career traced an extraordinary arc: from Bergen (the birthplace of modern meteorology) to MIT and Chicago (where he built American meteorology) to Stockholm (where he brought NWP to operational reality). He was arguably the greatest institution-builder in the history of atmospheric science.

8. Quotes and Anecdotes

Rossby

“Communications are the alpha and omega of meteorology.” – Rossby’s guiding principle, reflected in his founding of journals, international institutes, and visiting scientist programmes.
“He died while making great plans for the future.” – On 19 August 1957, Rossby collapsed from a heart attack during a conference in Stockholm. He was 58. Close friends knew he had suffered from rheumatic fever as a boy and had a weak heart.

Bolin on His Meeting with Rossby

During military service in Stockholm in 1947, the young Bert Bolin came into contact with Professor Rossby, who had just returned from the United States. This chance encounter shaped the rest of Bolin’s career – from NWP to the carbon cycle to the IPCC.

Norman Phillips on Rossby (from earlier research)

“Yes Norman, and it should be that!” – Rossby’s reported exclamation upon seeing that Phillips’ general circulation model reproduced realistic atmospheric features. (Note: this quote’s exact provenance needs verification.)

Chester Newton on Rossby’s Atmosphere

“When Rossby was in town, the department was in tumult… it was exciting but exhausting.”

Aksel Wiin-Nielsen

Wiin-Nielsen participated at Stockholm in “the first numerical prediction that completed its computation ahead of the time for which the forecast was made” – the essential threshold between research and operations.

Norman Phillips’s Defence of Bolin’s Thesis (1956)

Faculty opponent Norman Phillips wrote: “Mr Bolin’s defence of his thesis must be considered as extremely competent.”

The Military Exercise

In September 1954, NWP forecasts for 45000 troops conducting nuclear-protection exercises proved “tremendously successful compared to subjective forecasts.” This was the validation moment.

9. After Rossby’s Death (1957)

Immediate Succession

Bert Bolin became the natural successor as director of IMI and, in 1961, the first holder of a permanent professorship in meteorology at Stockholm University. He maintained and expanded the international character of the department.

Continuation of NWP

Swedish NWP continued after Rossby’s death:

1961: SMHI finally operationalised NWP on the BESK
1966: Paul Crutzen, who had arrived at MISU as a computer programmer in 1959, was involved in the conception, development and running of some of the very first numerical models for weather prediction
1973: SMHI assumed full responsibility for all NWP production in Sweden

The ECMWF Connection

Sweden’s NWP tradition had a direct lineage to the European Centre for Medium-Range Weather Forecasts (ECMWF), founded in 1975 with eighteen member states including all Scandinavian countries.

Key connections:

Aksel Wiin-Nielsen (Denmark) – trained under Rossby at Stockholm, participated in the 1954 BESK forecasts – became ECMWF’s first Director (1974–1979).
Three senior figures starting ECMWF in the 1970s began their careers in Rossby’s NWP project 25 years earlier. Only one was Swedish, reflecting Rossby’s international recruitment.
By the 1970s, Sweden diverted NWP resources toward ECMWF’s collective European effort.
Bo Doos directed the Global Atmospheric Research Programme (GARP, 1971–1982), a major international programme that helped pave the way for ECMWF’s medium-range forecasting mission.
Bert Bolin chaired GARP’s planning (from 1967) and later the IPCC.

The Broader Stockholm Legacy

MISU’s influence extended far beyond NWP:

Paul Crutzen arrived at MISU in 1959 as a computer programmer, did his doctoral work under Bolin, and won the 1995 Nobel Prize in Chemistry for his work on stratospheric ozone
Bolin’s carbon cycle research, inspired by Rossby, ultimately led to the creation of the IPCC and the global climate change framework
The Bolin Centre for Climate Research was established at Stockholm University in 2006, with over 400 researchers

10. The Swedish Computer Industry Connection

Why Sweden Had a Computer at All

The story of Swedish NWP is inseparable from the story of the Swedish computer industry. When US export controls blocked access to American computers in 1948, the Swedish Board for Computing Machinery (SBCM) built their own. BESK, their second machine, was directly influenced by the design of von Neumann’s IAS machine at Princeton. The irony: the same American restrictions that were meant to protect Western technological superiority actually stimulated Sweden to build a machine that enabled the world’s first operational weather forecasts.

The Decline

Sweden’s nascent computer industry ultimately failed when US export controls were liberalised in the 1960s and IBM competition intensified – a pattern mirrored in the UK and Japan. BESK was succeeded by the Facit EDB, but Sweden never became a major computer manufacturer.

Key Dates Summary

Date	Event
1919	Rossby joins Bergen School under Vilhelm Bjerknes
1928–1939	Rossby builds American meteorology at MIT
1940–1947	Rossby chairs Department of Meteorology at University of Chicago
1946	von Neumann contacts Rossby about NWP on ENIAC
1947	Rossby returns to Sweden; founds MISU at Stockholm University
1947	Bolin meets Rossby during military service in Stockholm
1948	US export controls on computers; Sweden begins building BESK
1949	Rossby founds Tellus journal
1950	Charney-Fjortoft-von Neumann ENIAC weather forecast (published in Tellus)
1950–1951	Bolin at IAS Princeton, works on ENIAC forecasts
Autumn 1953	Norman Phillips visits Stockholm to help port code from ENIAC to BESK
December 1953	BESK becomes operational (briefly world’s fastest computer)
March 1954	First forecasts computed on BESK that are completed ahead of valid time
September 1954	Six-week operational NWP trial during military exercise (45000 troops)
December 1954	Routine operational NWP begins – 3x/week for North Atlantic (world’s first)
1955	Bolin publishes barotropic forecasting results extending to 72 hours
May 1955	US JNWPU begins operational NWP (one year after Sweden)
1955	IMI formally established by Swedish Parliament
1956	Bolin receives Ph.D.; Rossby on cover of Time magazine
19 August 1957	Rossby dies of heart attack in Stockholm
1957	Bolin succeeds Rossby as director of IMI
1959	Paul Crutzen arrives at MISU as a computer programmer
1961	SMHI operationalises NWP on BESK; Bolin becomes first permanent professor
1973	SMHI assumes full NWP production responsibility
1974–1979	Wiin-Nielsen (trained in Stockholm under Rossby) serves as first ECMWF Director
1988–1997	Bolin serves as first IPCC Chairman
1995	Paul Crutzen (Bolin’s student at MISU) wins Nobel Prize in Chemistry

Sources

Primary Scholarly Sources

Persson, Anders. “Early operational Numerical Weather Prediction outside the USA: an historical Introduction. Part 1: Internationalism and engineering NWP in Sweden, 1952–69.” Meteorological Applications 12 (2005): 135–159. Cambridge Core – Accessed: 2026-04-03
Persson, Anders. “Early operational Numerical Weather Prediction outside the USA: an historical introduction. Part III: Endurance and mathematics – British NWP, 1948–1965.” Meteorological Applications 12 (2005). Wiley – Accessed: 2026-04-03
Persson, Anders. “Fifty Years Later.” History of Meteorology 6 (2014): 7–13. PDF – Accessed: 2026-04-03
Staff Members, Institute of Meteorology, University of Stockholm. “Results of forecasting with the barotropic model on an electronic computer (BESK).” Tellus 6 (1954): 139–149.
Bolin, B. “Numerical forecasting with the barotropic model.” Tellus 7 (1955): 27–49. Semantic Scholar – Accessed: 2026-04-03
Bergthorsson, P. and Doos, B. “Numerical weather map analysis.” Tellus 7 (1955): 329–340. PDF – Accessed: 2026-04-03
Bauer, P., Thorpe, A., and Brunet, G. “The quiet revolution of numerical weather prediction.” Nature 525 (2015): 47–55. Nature – Accessed: 2026-04-03

Biographical Sources

“Bert Bolin (1925–2007) – a world leading climate scientist and science organiser.” Tellus B 65 (2013). Tellus B – Accessed: 2026-04-03
“Bert Bolin – founder of the climate panel and world-leading scientist.” Stockholm University. Link – Accessed: 2026-04-03
“Aksel Wiin-Nielsen passed away.” ECMWF. Link – Accessed: 2026-04-03
Aksel C. Wiin-Nielsen. Wikipedia. Link – Accessed: 2026-04-03
Arnt Eliassen. Wikipedia. Link – Accessed: 2026-04-03
“Geirmundur Arnason.” University at Albany, SUNY. Link – Accessed: 2026-04-03
“Carl-Gustaf Rossby: Theorist, institution builder, bon vivant.” Physics Today. Link – Accessed: 2026-04-03
“Carl-Gustaf Rossby The Stockholm period 1947–1957.” Tellus B 51 (1999). Tellus B – Accessed: 2026-04-03
Paul J. Crutzen – Biographical. NobelPrize.org. Link – Accessed: 2026-04-03

Institutional Sources

“About us.” Department of Meteorology (MISU), Stockholm University. Link – Accessed: 2026-04-03
“About the centre.” International Meteorological Institute in Stockholm (IMI). Link – Accessed: 2026-04-03
“History.” ECMWF. Link – Accessed: 2026-04-03
“Numerical Weather Prediction at 60: A journey of innovation at the Met Office.” Met Office. Link – Accessed: 2026-04-03

Computing History Sources

BESK. Wikipedia. Link – Accessed: 2026-04-03
“The Royal Swedish Air Force Weather Service Computes the First Routine Real-Time Numerical Weather Forecasting.” History of Information. Link – Accessed: 2026-04-03
Yang, Charles. “The First Compute Arms Race: the Early History of Numerical Weather Prediction.” arXiv:2506.21816. arXiv – Accessed: 2026-04-03
“New calculations in the Cold War.” Inventing Europe. Link – Accessed: 2026-04-03

General NWP History

History of numerical weather prediction. Wikipedia. Link – Accessed: 2026-04-03
“The History of Numerical Weather Prediction.” NOAA. Link – Accessed: 2026-04-03
“50th Anniversary of Operational Numerical Weather Prediction.” ResearchGate. Link – Accessed: 2026-04-03

Michał Brennek