Since the launch of the first successful weather satellite, TIROS-1, by the United States in 1960, the observation of weather from space has moved from a novelty to a routine activity. A modern forecast, whether produced by a national meteorological service or a private company, begins with data from an international constellation of satellites whose combined output runs to terabytes every day. The history of how this system was built, and how it affects the forecasts that reach the public, is less well known than the forecasts themselves.

Weather satellites fall into two broad categories. Geostationary satellites orbit at 36,000 kilometres above the equator, matching the earth's rotation so that they remain above a fixed point on the ground. From that altitude they cannot see the polar regions well, but they can produce an image of almost an entire hemisphere every fifteen minutes. Polar-orbiting satellites, by contrast, circle the earth at altitudes of a few hundred kilometres in orbits that pass over the poles. They provide much finer detail, including direct measurements of temperature and humidity in the atmosphere, but they see any particular point only twice a day.

Geostationary satellites are most familiar to the public through the moving cloud imagery shown on television weather bulletins. They detect visible light during the day and infra-red radiation at night, allowing cloud patterns to be tracked continuously. More recent instruments also measure specific infra-red channels that correspond to particular atmospheric gases, enabling the detection of fog, dust and smoke as well as cloud. Meteorologist Dr. Olufemi Adebayo has explained that the value of geostationary data lies largely in its continuity: a forecaster can watch a thunderstorm develop over hours, with no gaps, in a way that simply was not possible before the 1970s.

Polar-orbiting satellites, which are less visible to the public, carry most of the information that actually feeds numerical forecasting models. Their sensors sample many layers of the atmosphere and produce profiles of temperature and moisture that ground-based weather stations cannot match. Without such data, modern medium-range forecasting - the three-to-five-day outlook that is now routinely useful - would not be possible. Weather forecasting made a major step forward in the 1990s when global models began to assimilate polar satellite data in something approaching real time.

The cost of these programmes is considerable, and several international arrangements ensure that data is shared. The European Organisation for the Exploitation of Meteorological Satellites, Eumetsat, co-operates closely with the American NOAA and the Japanese JMA. Data is exchanged free of charge under long-standing agreements, and national forecast agencies use each other's satellite output routinely. Dr. Adebayo has noted that this is one of the clearest examples of large-scale scientific co-operation that survives political disputes among the countries involved.

Improvements have come from more than hardware. The assimilation of satellite data into forecasting models is itself a major scientific task, requiring complex statistical methods to match imperfect measurements with model variables. A 2019 assessment by the European Centre for Medium-Range Weather Forecasts estimated that satellite observations now account for more than ninety percent of the reduction in forecast error achieved over the past three decades, with surface-based observations and ground-launched balloons contributing the remainder.

Not every aspect of the system is in good health. Several critical satellites approach the end of their design lives, and the lead time for replacements is long. A gap in polar-orbiting coverage in the mid-2020s was widely anticipated and, as of recent reports, has largely been avoided, but only through careful extension of older platforms. Meanwhile, private operators have begun to launch small satellites whose data is sold to forecast providers. The value of this commercial data is debated; it adds coverage but raises questions about consistency and long-term availability.

The influence of all this technology on daily life is deeper than most users realise. Flight routing, harvest planning, shipping insurance, disaster preparedness and many municipal services depend on forecasts that would be markedly less reliable without the satellite system. Dr. Adebayo puts it plainly: 'We do not notice weather satellites because we do not notice the forecasts working.' A rare failure of a key platform, when it occurs, is usually the first reminder that a service the public treats as automatic is, in fact, the product of a carefully maintained orbital infrastructure.