Oh boy… the BoM has a habit of getting seasonal forecasts of both temperature and precipitation wildly wrong (Warwick Hughes has a habit of juxtaposing forecast and eventual reality like this). Now they want to claim climate forecasts?
HIDDEN BEHIND A nondescript door in the Bureau of Meteorology’s office on Collins Street, Melbourne, is one of the world’s great supercomputers.
In a high-ceilinged room laced with aqua electrical cables, great grey and black machines, more than two metres high, whir noisily tabulating important information. They are not calculating the existence of obscure sub-atomic particles or understanding the speed of light, they are instead pondering the most quotidian of questions: do you need to bring your brolly?
The Sun Microsystems computer has 4,600 central processing units, running at a top speed of 53,912 gigaflops, which to you and me means a brain the equivalent of about 10,000 home PCs.
When the BoM booted up the computer for the first time in 2010, it was the 113th most powerful computer in the world. These days it’s slipped to 314, but is still an impressive machine.
Chris Ryan is the Centre Director at the National Meteorological and Oceanographic Centre (NMOC). According to Ryan, we need all this computing power because “the atmosphere is a chaotic, complex system.”
Ryan says 23 million pieces of information are fed into the machine – observations of temperature, pressure, humidity from all over the globe and far up into the atmosphere.
Crunching the numbers takes about an hour and produces a picture of what the weather will be every 10 minutes for the next 10 days for each of the hundreds of thousands of observation points. Each time the computer runs, the output takes up 200 gigabytes of disk space. The Bureau runs these calculations four times a day.
It’s a far cry from the first use of mathematics and physics to produce a weather forecast. Starting with observations across Europe, physicist Lewis Fry Richardson calculated with a slide rule and Newtonian physics what the weather would be like six hours from his starting point of 7am on 10 May 1910. The calculations took two years. And he was wrong (wildly wrong: he predicted cyclones over Germany, for instance). But the basic idea was sound, and when mechanical computers were invented during World War II, they were quickly put to use calculating weather patterns, with far greater success.
Melbourne is one of three World Meteorological Centres, each with supercomputers collating and disseminating weather data from around the globe. The other two are in Moscow and Washington.
When not predicting if it will rain tomorrow, the computers have another important function: predicting whether it will rain in 20 years’ time.