Sections

Services

Information

UK United Kingdom

How your computer could reveal what’s driving record rain and heat in Australia and NZ

Australians and New Zealanders can now use their computers to help scientists discover if climate change has contributed…

Golden Bay in New Zealand after damaging flooding in December 2011. Gerry Draper, via NIWA, CC BY-NC-ND

Australians and New Zealanders can now use their computers to help scientists discover if climate change has contributed to record heatwaves, droughts and flooding across both countries.

The Weather@home project, launched in Australia and New Zealand today, is the latest stage of what has been dubbed “the world’s largest climate modelling experiment”, started in the UK a decade ago.

Anyone with a computer and access to the internet can take part by volunteering their computer’s spare processing power to run climate and weather modelling simulations, even while continuing to use their computer normally.

There are 20,000 people worldwide currently helping with similar climate prediction experiments for Europe, USA and southern Africa. Over the past decade, people in 138 countries with nearly 100,000 different computers have been involved.

In the UK, that has enabled the equivalent of 20,000 years of simulations to be run in just three weeks, testing the likely contributing factors to this year’s devastating floods. Live results from the UK testing can be seen here.

Where in the world people have been volunteering their computers over the past decade to help run experiments on extreme weather. Climateprediction.net, CC BY-NC-ND

Now scientists from the University of Oxford, the UK Met Office, the University of Melbourne, the University of Tasmania and New Zealand’s National Institute of Water and Atmospheric Research (NIWA) will examine the record-breaking heatwave in Australia and extreme drought in New Zealand in early 2013.

They also plan to assess the possible role of climate change in Australia’s 2009 Black Saturday bushfires, and record rain and flooding across eastern Australia in 2010 and 2011, as well as in Golden Bay and Nelson in NZ three years ago.

How it works

Speaking at an Australian Science Media Centre briefing on the new project, the researchers told reporters there was no risk to people’s computers – including from hacking – if they took part.

“Yes, [it] is safe,” said Dr Friederike Otto from Oxford University’s Environmental Change Institute. “We use software called BOINC, developed at the University of California, Berkeley, that has been specifically developed for these sort of citizen science projects – and it has never been used to hack participants' computers.”

Beyond studying weather and climate change, the BOINC software has been widely used for years to help with new science discoveries, including trying to cure diseases, discover pulsars and even search the universe for alien life.

It allows people to use idle time on their computer (with Windows, Mac, Linux, or Android) to run experiments, creating a virtual “supercomputer” across the world, which automatically shares the results with researchers.

NIWA scientist Suzanne Rosier explains Weather@home on Vimeo.

Testing extremes in Australia and NZ

2013 was a record-breaking year for extreme heat in Australia and New Zealand.

Highest daily maximum temperatures across Australia during the first two weeks of January 2013. Bureau of Meteorology, CC BY-NC-ND

More than 70% of Australia recorded temperatures above 42°C, with temperatures exceeding 48°C at a number of locations. On 7 January 2013, Australia experienced its hottest day on record with a national average maximum temperature of 40.3°C.

Meanwhile, in New Zealand the news at the time was dominated by drought. From January to March 2013, the North Island experienced an average of almost 80 days without rain, far more than ever recorded previously.

For the Weather@home project, researchers need to run two very large “ensembles”, or groups, of weather simulations.

Soil moisture deficit across New Zealand on 17 March 2013, probably the point in the drought when soils were at their driest. Circled areas indicate sample locations analysed in a 2013 NIWA report, from which this figure is taken. NIWA, CC BY-NC-ND

One ensemble will represent “2013 as observed”. This will use both human-caused greenhouse gas emissions and natural emissions from things such as volcanoes, to help simulate weather events that would be have been possible given those climate conditions.

The other ensemble will represent the “2013 world that might have been”, using only the natural emissions, to see what climate conditions might be like without any human-caused emissions.

The data from those two parallel worlds will then be collected from computers all around the world and sent back to the Tasmanian Partnership for Advanced Computing, at the University of Tasmania, to be analysed.

“Basically the more people participate, the more science we can do,” said Dr Sam Dean, from NIWA in New Zealand.

Clearing up a controversial debate

Professor David Karoly from the University of Melbourne initiated the Weather@home project in Australia and New Zealand, and said it could help clear up the ongoing debate about the connection between climate change and extreme weather events.

“There is uncertainty in the public about how much climate change has contributed to individual extreme events. People like the Prime Minister and the Minister for the Environment have commented that there is no link, while climate scientists say there is a connection,” said Professor Karoly.

“We won’t be able to say climate change is the sole cause of extreme weather. What we want to do is look at the contribution of climate change to increasing the frequency and intensity of those extremes, particularly as we see heatwaves, record high temperature, drought and bushfires.”

With the help of volunteers at home, the researchers will be able to conduct far more experiments than they could hope to on their own.

“We need to run the simulations a lot of times because extremes are rare events and we might not get many of them if we just run the simulations once,” Professor Karoly said.

“If we run repeated simulations many thousands of times, we can really have a look at how likely are, both in the world as it was in 2013 and how it would have been without human caused climate change."

Nathalie Schaller explains the science behind the UK flooding project.

The rise of citizen science

Dr Philip Roetman from the Citizen Science Program at the University of South Australia said that while citizen science projects have been around for a long time, they were becoming more popular and more important.

“Departments are having their budgets cut and they are thinking, how am I going to do this research? One way is to get the community involved,” said Dr Roetman.

He said the Weather@home project would have been impossible 20 years ago because not everyone had internet-connected computers at home. And beyond helping scientists, he said citizen science could also help public interest in and understanding of science.

“It’s not a magic bullet; just because people get involved in citizen science doesn’t mean they’re going to change their views towards everything. But it’s a great way to get people involved, and then have that discussion.”

Join the conversation

24 Comments sorted by

  1. Comment removed by moderator.

    1. Liz Minchin

      Queensland Editor at The Conversation

      In reply to Michael Eagleton

      You're welcome to comment Michael – and thanks. However, if you put a link to your business other readers will report it as possible spam and we don't have a choice but to delete it as advertising.

      report
  2. Andrew Nichols

    Digital Drudge

    I signed up straight away!

    I'm approaching 5% of my first work units.

    report
    1. William Raper

      Retired

      In reply to Andrew Nichols

      I'm keen, but as a retired scientist, I need to know what ISP charges (in gigs, not$) are likely to take place?

      report
    2. Liz Minchin

      Queensland Editor at The Conversation

      In reply to William Raper

      Good question. If we can get one of the project's researchers to reply, hopefully they can answer that for you William.

      report
    3. David Karoly

      Professor of Atmospheric Science at University of Melbourne

      In reply to William Raper

      The data download to start a run is modest and the output that is uploaded at the end to the data server is not huge, as we have tried to select key daily data to analyse, including minimum and maximum temperatures, rainfall, windspeed, sea level pressure and relative humidity at each grid point over land.

      Weather At Home download and upload file sizes

      The size of download files for a Weather At Home task (including workunit, restart files and ancillary files): 131MB

      Currently the size of the uploaded files and restart files are:
      Weather At Home ANZ: (five daily variables output)
      Each upload file (every model month, 12 in total): 14MB
      Restart files (end of every run): 47MB
      Total size of files produced: 215MB

      Average computation time: 4.5 days

      report
    4. David Karoly

      Professor of Atmospheric Science at University of Melbourne

      In reply to David Karoly

      Oops, the above is for an earlier version, so here is the correct data download/upload sizes, still modest.

      Weather At Home download and upload file sizes

      The size of download files for a Weather At Home task (including workunit, restart files and ancillary files): 131MB

      Currently the size of the uploaded files and restart files are:
      Weather At Home ANZ: (six daily variables output)
      Each upload file (every model month, 12 in total): 26MB
      Restart files (end of every run): 47MB
      Total size of files produced: 359MB

      Average computation time: 4.5 days

      report
    5. Blair Donaldson
      Blair Donaldson is a Friend of The Conversation.

      Researcher & Skeptic

      In reply to William Raper

      Hi William, up until about 12 months ago I had been signed up to the SETI project. Data segments were around .5 MB as I recall. There was an option to download as much or as little as you like on the BOINC application preferences.

      The number of projects available now as increased quite a bit so there is bound to be something there that interests you.

      report
  3. John Rimmer

    logged in via Facebook

    As a long term Stanford folding@home volunteer I am conflicted. I want to join in here in addition but the electricity costs are not insignificant. Perhaps you should negotiate a rebate system with Origin Energy?

    report
  4. Henry Verberne

    Once in the fossil fuel industry but now free to speak up

    I have downloaded the software and will give it a go.

    report
    1. Ross Barrell

      Aikido Student

      In reply to Henry Verberne

      I'm about to give it a go. I actually have a (pre loved) machine in surplus) that can be used. I'm just in the process of figuring out if the Linux software will run in Kubuntu 12.10. We shall see. :)

      report
    2. Ross Barrell

      Aikido Student

      In reply to Ross Barrell

      No problems. There's actually a distro specific version of BOINC for Kubuntu 13.10, so it's pretty much a no brainer. very good.

      report
  5. Craig Somerton

    IT Professional

    Thanks for the article. I signed-up too and am happy to add my spare processing time toward the task.

    report
  6. Comment removed by moderator.

    1. Ross Barrell

      Aikido Student

      In reply to Andy George

      a) What a bleak outlook.
      b) Are you suggesting the whole thing be abandoned?
      c) have you a better suggestion to solve the issues you have raised?
      d) I am retired and have no career to pad out. I would bet that a significant number of others who read and contribute to TC also have no vested (and therefore no conflict of) interest in being part of this project.

      report
    2. Liz Minchin

      Queensland Editor at The Conversation

      In reply to Andy George

      Hi Andy,

      In reply to your initial comment: "Soooo... we're going to burn coal to power calculations to help predictions on what the effect of burning the coal will be...?"

      That's a fair question and the researchers were asked about electricity use in the Australian Science Media Centre briefing yesterday; we didn't include it as the story was long as it was. But for you or anyone else wondering, they essentially said "yes that's true if you rely on mains power, but there are options to reduce…

      Read more
    3. Lynne Black

      Latte Sipper

      In reply to Liz Minchin

      Thanks, Liz, for bringing attention to a few of the fundamental rules that guide our written contributions to these great articles. It has bothered me for some time that, whilst most of the comments are thoughtful and intelligent, some comments can be snide, insulting personal attacks on the other respondents. It's usually the same people that are guilty of this. I'm sure these hurtful comments could dissuade some readers from contributing their thoughts. Thank you for the timely reminder that courtesy and good manners should be prerequisites for making comments online.

      report
    4. Liz Minchin

      Queensland Editor at The Conversation

      In reply to Lynne Black

      Thanks Lynne

      There's also a small "Report" button under comments – if you ever see something you think Cory our dedicated moderator or one of us editors needs to be alerted to, do hit it and mark whether you think it could be considered "abuse", "spam" (such as someone posting what appears to be an ad), off-topic etc. We'll then take a look as soon as we can.

      We're trying really hard to keep these comments/conversations polite, respectful (*especially* when you disagree with someone), on-topic and useful for all readers. It's impossible to do that perfectly, but we're doing our best. And we really appreciate that most Conversationalists do do the right thing by others.

      report
  7. Gary Luke
    Gary Luke is a Friend of The Conversation.

    thoroughly disgusted

    This is a general question about comparing statistical models to real world occurrences, and as a flow on, the utility of this project. Suppose the average of the results of one of the simulations shows an 85% likelihood of an occurrence, and the average of the results of the second simulation shows that its set of parameters produced a 12% likelihood. And suppose whatever it is that's being modeled actually did occur. Is there any way to determine which set of circumstances and parameters used in either simulation or not even considered in either actually caused the real world conditions?

    report
    1. Liam J

      logged in via email @gmail.com

      In reply to Gary Luke

      " Is there any way to determine which set of circumstances and parameters used in either simulation or not even considered in either actually caused the real world conditions?"
      As i understand it, only in a probabilistic sense. Which might not satisfy those used to working in easy cause-effect fields, but is not uncommon in complex fields like climate, medicine, materials engineering, ecology, etc.

      -

      Re BOINC, i've been running it for some time, no security issues or noticeable impact on concurrent processing.. good job! I wish they'd licence their IP to miscrosoft.

      report
  8. Patrick Maher

    Retired Doctor of Psychology and Academic

    I would happily volunteer two computers I have running - but, until the download/upload freight is costed and the issue settled with clarity, I am not prepared to pay for others to use my limited ISP bandwidth and have my ISP charge me to do it.

    report
    1. David Karoly

      Professor of Atmospheric Science at University of Melbourne

      In reply to Patrick Maher

      Patrick,

      That is a good question and we should have told users that information in advance. I answered the question above and here is the response again.

      The data download to start a run is modest and the output that is uploaded at the end to the data server is not huge, as we have tried to select key daily data to analyse, including minimum and maximum temperatures, rainfall, windspeed, sea level pressure and relative humidity at each grid point over land.

      Weather At Home download and upload file sizes

      The size of download files for a Weather At Home task (including workunit, restart files and ancillary files): 131MB

      Currently the size of the uploaded files and restart files are:
      Weather At Home ANZ: (six daily variables output)
      Each upload file (every model month, 12 in total): 26MB
      Restart files (end of every run): 47MB
      Total size of files produced: 359MB

      Average computation time: 4.5 days

      report