Fortelling our climate future
NCI is pleased to announce increased support for climate change and earth system science research, benefiting both the Centre of Excellence for Climate System Science, and the major agency partners of NCI, namely BoM and CSIRO.
Through significant external funding initiatives involving a formal collaboration between NCI and its peak system vendor, Fujitsu, together with funds made available through the NCRIS 2013 initiative, a three year program of development to enhance the computational and I/O performance of the ACCESS modelling suite on Raijin and its successor will be developed. These activities, which involve the partners of NCI as well as Fujitsu Limited of Japan, will also seek to prepare the component codes for the next generation of many- and multi-core processor architectures.
Formally replacing the Bureau of Meteorology’s suite of prediction models in August 2010, ACCESS (the Australian Community Climate and Earth-System Simulator) is the most powerful climate model ever created and run in this country and one, potentially, capable of predicting the global climate of 2100 or the outlook for rainfall trends round Narrandera, NSW, or Katanning, WA, through 2030.
ACCESS has been assembled and tested on the NCI supercomputer by a research consortium including CSIRO, the Bureau of Meteorology and several Australian universities precisely to perform these prodigious feats of climate prediction and local weather forecasting from models that replicate, far more faithfully than any yet, events great and small in the Earth’s churning atmosphere and oceans as global warming takes hold.
ACCESS represents the fusion of no less than six of these enormous mathematical constructs and is so byte-ravenous it can run effectively on no other computer in the country, says project leader Dr Tony Hirst of CSIRO. It fuses the power and precision of the UK Hadley Centre’s atmospheric chemistry model with the US’s Geophysics and Fluid Dynamics Lab’s unsurpassed oceans model, the French-developed OASIS ‘coupler’ to link them at the right resolutions, an advanced sea-ice model which incorporates this vital element into the earth’s heat budget and ocean circulation, and CABLE, an Australian land-surface model, which describes soil water retention and runoff, seepage and carbon uptake and release in local detail.
The output from these phenomenal calculations will arrive in every living room and farm ute in Australia in the form of more ‘skilful’ local weather forecasts and seasonal predictions – and helped shape vital policy decisions when the Intergovernmental Panel on Climate Change (IPCC) delivered its fifth assessment in 2013.
When implemented in 2010, says Tony, ACCESS ran for many hundreds of years of test simulations, to ensure the model performed consistently. The team then run it in direct comparison with the actual 20th Century, to see how accurately it reproduces what happened historically with climate. They are now sending it surging forward in time to raise the veil on what the future holds for Australia in the remainder of the 21st century.
Dr Hirst says the Australian community is already seeing the benefits of the model in nightly weather forecasts, made with greater confidence and precision that possible before.
“The southern hemisphere and Southern Ocean in particular are really vital in terms of what happens to the world’s climate and energy budget, but most of the northern hemisphere models reproduce their effects quite poorly. Our job is to help refine the global models by contributing improved modelling of southern processes. The Hadley Centre, for example, is most appreciative of our input to its atmospheric model, which is widely acknowledged as one of the world’s best,” Dr Hirst says.
Now fully operational, ACCESS has begun to explore vital but so far under-tested facets of climate change, such as how the earth’s biological systems will respond to warming and elevated CO2 — will they absorb or release more carbon, damping the greenhouse process or pushing it into overdrive? At a continental scale it will explore how the Australian landmass itself will ‘breathe’ carbon and water, taking them in and releasing them and the subsequent net gains or losses.
The answer to questions such as these affect everyone on the planet, determining how quickly society must move to zero carbon emissions and how steep will be the adjustments forced on people and the economy by the risk of dangerous change.
To run such vast models and support the memory they demand will absorb a massive 3 terabytes – three trillion bytes – of computational power a month, a feat that nothing but the NCI’s supercomputer can support.