Modern ways to map out our global climate trends ; Computers and the internet have changed our lives. In the last of his week-long series, Post... [South Wales Evening Post (Wales)]
(South Wales Evening Post (Wales) Via Acquire Media NewsEdge) Modern ways to map out our global climate trends ; Computers and the internet have changed our lives. In the last of his week-long series, Post reporter Richard Youle examines how super computers help weather forecasters and glaciologists in Swansea. @YoulePost email@example.com 01792 514620
WHEN Swansea glaciologist Tavi Murray was an undergraduate she used computer punched cards -- stiff bits of paper with holes in predefined positions.
Now she and Swansea University colleagues have been using IBM supercomputer Blue Ice, which can handle billions of calculations per second, to process large sets of data.
Blue Ice is overshadowed by what looks like a dozen or so vending machines at the Met Office's HQ in Exeter. This particular IBM behemoth was upgraded last year to handle 1,200 trillion calculations per second.
Climate modelling and weather forecasting relies considerably on computer processing and storage capability.
The oceans and atmosphere are involved in a complex and ever- changing tango, while land also absorbs and radiates heat.
Met Office spokesman Charles Powell told the Post: The atmosphere and oceans are constantly trying to get themselves in balance.
There is so much intertwining and fluctuation. Some last 60 or 90 days -- some last one hour. As soon as one fluctuation comes into play, the more it throws others out of play.
The Met Office receives some 10 million bits of data every day from all over the world. Weather stations on land, ocean buoys, radar and satellite all feed information through.
Mr Powell said it resulted in a comprehensive picture of the breadth and depth of the atmosphere.
Short-range forecasts of 6 to 48 hours are mainly provided by the North Atlantic and European forecast model. By running the model over only the North Atlantic region, rather than globally, the level of detail increases.
When forecasting more than two days ahead, a global model is needed because the weather we experience several days ahead can be influenced by the weather happening right now at the other side of the world.
For forecasts as far as two weeks ahead, the Met Office uses ensemble forecasts -- many separate forecasts run for the same period -- each with different starting conditions.
A study published this week by the National Oceanography Centre said forecasters will have a tougher job predicting winter conditions over Europe in some years over others. This was down to the strength of the airflow steaming in from the North Atlantic. A weaker airflow meanders, leading to more complicated weather patterns. You get big differences over Europe, switching between cold Arctic and Siberian air, and milder subtropical air.
Professor Murray's glaciology colleagues Ian Rutt and Suzanne Bevan both have backgrounds in meteorology.
Dr Rutt modelled the atmosphere for his PhD. I took the skills I learned from that to glaciology, he said.
Computer power, he said, had risen considerably over the last decade.
It has become possible to make a super computer out of some of the hardware that you have on a home computer, he said. And the costs of large-scale computing have come down.
Climate models use similar information to forecasting models but over a longer period, enabling scientists to spot trends.
Dr Rutt said: Any kind of model is a representation of the state of the atmosphere -- or the shape of a glacier. The computer model takes this state or shape and predicts what it might be some time later.
We are really interested what happens in the future, but we can't check the future because it hasn't happened! As well as helping us overcome the fact that we are stuck in the present, models enable scientists to overcome another restriction -- we have one Earth. We can only observe changes to our planet. There isn't a control (planet) we can observe in parallel, whose atmosphere for example is not experiencing rising carbon dioxide levels.
There is still plenty of room for human fieldwork and observation.
Professor Murray said: Climate modelling takes three things: observation -- we are still discovering things like new ocean currents; secondly, you need the computer models; third, you need large-scale remote sensing over the whole planet.
Remote sensing uses instruments mounted on satellites or in planes to produce images or scenes of the Earth's surface. These kinds of data sets can check if the climate models are doing anything like (what is happening) on real Earth, said Professor Murray.
In an ideal world, glaciologists would have time series data stretching back hundreds of years.
Our time series is relatively short, said Professor Murray. We have started learning how much the ocean interacts with ice sheets. This goes for the Arctic and Antarctic. Changes in the ocean are driving changes in the ice, and vice-versa.
Asked about striking findings she had observed, Dr Bevan said: Probably the biggest thing was detecting big changes in the flow rates of glaciers in Greenland -- then a slowing down again.
Models must also take into account naturally-occurring feedbacks, like Arctic sea revealed by melting ice absorbing more heat due to its darker colour, compounding the warming effect.
Dr Rutt said: It is thought that changes between glacial and warmer periods are to a great extent (caused) by feedbacks.
Professor Murray added: Humans are doing something pretty large- scale.
How have computers changed your life Are you obsessed with technology www.southwales-eveningpost.co.uk
(c) 2013 ProQuest Information and Learning Company; All Rights Reserved.
[ Back To IBM News 's Homepage ]