A supercomputer for all seasons
GroundCover™ Issue: 110 | Author: Rebecca Jennings
Australian grain growers are tapping into a supercomputer with the processing power of 10,000 home desktop computers when they access seasonal forecasting information from the Bureau of Meteorology (BoM).
The latest climate tool – the Predictive Ocean Atmosphere Model for Australia (POAMA) or, as it is more commonly known, BoM’s new seasonal climate outlook, is a state-of-the-art system to predict climate up to a whole season ahead.
These new outlooks are based on a dynamical climate model. Unlike the previous statistical climate model, which used past history to guide the outlook, the dynamical model combines actual physics of the atmosphere, oceans, land and ice to calculate climate conditions over Australia for the coming three months.
If you like big numbers, than POAMA has them: to provide a requested outlook the system crunches the calculations on more than 40 million pieces of data, sourced from observation systems around the globe. It uses a supercomputer rated at 104 teraflops (a teraflop is one trillion operations per second) with more than 36,000 gigabytes of memory and 214 terabytes of data storage.
But what does all this mean for the farm?
Dr Andrew Watkins, who manages BoM’s climate prediction services, says the seasonal climate outlooks are general statements about the likelihood of wetter or drier-than-average weather over a three-month period.
As these are probabilistic outlooks (not categorical ‘yes/no’ forecasts), they are best used as part of a suite of farm risk-management tools.
In particular, Dr Watkins says POAMA provides better forecasts in autumn than the old statistical model: “Autumn has traditionally been called a ‘predictability barrier’ by the forecasters because the Pacific Ocean, particularly those regions where El Niño and La Niña tend to form or decay, has less influence on our climate at this time,” he says.
He says grain growers can come up with a reasonable degree of probability for a season by combining POAMA outlooks with other information – such as soil-moisture levels – to make decisions.
The POAMA model is a work in progress as contributing technologies develop. BoM started using the newest version, POAMA-2, in May 2013. Since then, the seasonal climate outlooks (issued monthly) have all been based on forecasts from this new system.
Looking ahead, BoM plans to develop POAMA further, including increasing its resolution from 250 kilometres to 150km, to provide greater climate detail.
“The skill is also expected to increase as science, observations and computing power all develop even further in the coming years,” Dr Watkins says. “Using a dynamical model also means the door opens to potentially more agriculturally focused climate outlooks, such as for wind, humidity or sunshine-hours.”
Regional three-month seasonal climate outlooks are available at: www.bom.gov.au/climate/ahead
GRDC Project Code MCV00034
Region North, South, West