moinsen / Software historyDE ↗
← 1950

Event

ENIAC berechnet Wettermodelle

Context and research status
  • Kennung: 1950-eniac-wettervorhersage
  • Zeitraum: April 1950 / April 1950
  • Datumsgenauigkeit: Monat nach späterem Rückblick / month according to later account
  • Art: Forschungsanwendung / research application
  • Kategorien: Technik, Akteure, Ereignisse, Zahlen
  • Beteiligte: Jule Charney, Ragnar Fjørtoft und John von Neumann: Forschungsbericht; weiteres Team: Berechnung / research paper; wider team: computation
  • Region: USA / United States

In 1950 a research team calculated weather models on ENIAC. A paper published that year by Jule Charney, Ragnar Fjørtoft and John von Neumann describes four 24-hour forecasts based on actual data. NOAA later dated the first one-day nonlinear calculation to April 1950. These were research experiments, not an established daily forecasting service. The British experiments of 1952 followed later.

Meteorologist Peter Lynch later explained the method using a 19-by-16 grid covering much of North America: eight three-hour steps produced a one-day forecast. A NOAA archive account reports that one calculation took about 24 hours; the experiments were still far from a fast daily service. Context

Homé McAllister later recalled the practical work: after a run, the team tabulated punched cards and sorted them by new coordinates before feeding them into the next run. A Ballistics Research Laboratory group had previously prepared for the project and its coding in Princeton. Firsthand recollection

Sources