Two time scales
Weather prediction works on different time scales. Forecasts for the next several days depend heavily on computer models. The next few hours, often called nowcasting, depend more on watching what is happening right now and extending it forward. The word is generally applied to periods from zero to about six hours, though definitions vary. Satellites contributed to both, but their history is largely the story of images arriving faster and more often. Our article on TIROS covers the first cloud pictures of 1960; this one picks up from there.
Pictures once a day
In the 1960s, the satellites that gave the most useful regular coverage were in polar orbits. Because a polar-orbiting satellite passes over a given place only a few times a day, forecasters had a picture of a particular storm perhaps once or twice a day. That was sufficient to recognize the large-scale organization of clouds but too slow to follow fast changes. Pictures were often transmitted to ground stations as they were taken, and later systems used scanning radiometers that built up an image line by line rather than relying on television cameras.
Some operational programs began in the middle of the decade, including the ESSA series and later the NOAA polar satellites, which carried more instruments. Satellite pictures also began to be used in broadcast weather reports, making the view from space familiar to the public.
Staying over one place
The step that made frequent imagery possible was the geostationary satellite, which stays above the same point on the equator and can therefore view a region repeatedly. ATS-1, launched in 1966, took early images of the Earth’s disk from this orbit. Its images were built up using a spinning scanner, and animated sequences showed clouds moving over the day. The first operational geostationary weather satellites in the United States, SMS-1 in 1974 and GOES-1 in 1975, brought regular coverage, with images typically at intervals of about half an hour. Europe’s Meteosat began in 1977 and Japan’s GMS followed in the same period. Looping images allowed forecasters to see how storms were moving and growing.
Geostationary satellites have a drawback: from 35,786 kilometers they are far away, and they see the polar regions only at a glancing angle. That is why polar and geostationary satellites are used together.
From images to measurements
Alongside faster pictures, instruments improved. Infrared channels showed clouds at night and let analysts estimate cloud-top temperatures. A technique devised in the 1970s allowed forecasters to estimate the strength of tropical cyclones from the appearance of their clouds in satellite pictures. It was valuable in ocean areas where aircraft rarely flew, although it requires judgment and has limits. Measurements of winds, found by tracking clouds between successive images, were also used in computer models. More recent instruments measure the temperature and moisture of the atmosphere in layers, and these data are fed into the models that make longer forecasts.
Faster and sharper
The most visible change in recent years has been the speed of imaging. The American GOES-R series, whose first satellite was launched in 2016, can scan the full disk of the Earth about every ten minutes in its usual mode, with a faster option of about five minutes, and it can view smaller regions at much shorter intervals, as often as every minute or so. Japan’s Himawari-8, launched in 2014, and Europe’s newer Meteosat generation similarly offer more channels and more frequent scans than earlier satellites. The exact figures depend on the satellite and the mode being used, so check an agency’s current documentation for specifics.
Lightning mappers on some of these satellites detect flashes continuously. Because lightning often rises sharply in storms that are intensifying, such data can help forecasters decide when to issue warnings. This is an example of how imagery has moved from describing the weather as it was to supporting warnings about what may happen within the hour.
Working with radar and ground data
Nowcasting rarely depends on one source. Forecasters combine satellite images with weather radar, surface observations and lightning data. Radar shows rain intensity and motion beneath the clouds, while satellites show the cloud tops and the environment in which storms develop, including areas outside radar range. Computer techniques extrapolate the motion of storms forward for the next hour or two. These results are useful but uncertain, particularly when storms develop or decay quickly.
Remaining limits
Better images have not removed the underlying uncertainty of the atmosphere. Forecast skill has improved over decades, in part because of satellite data, but that remains a general statement rather than a promise for any particular day. Satellites may fail or lose data, and the instruments must be calibrated. Space agencies keep spare satellites for this reason.
The path from a few blurry pictures in 1960 to rapid-scan views shows how a single technology can change its role. Early weather satellites showed that clouds could be seen from above. Later ones turned the view into a continuous tool for short-term forecasting, and satellites now contribute in some way to nearly every weather forecast.