This page contains a Flash digital edition of a book.
inside views Making a prediction


In late 2010 the Australian Bureau of Meteorology predicted that the next tropical cyclone season would most likely be active. Cyclone Yasi, which recently hit Queensland, Australia, showed they were right. But with so many variables needing to be taken into account, how do meteorologists make their predictions accurate?


resolution and computational capacity of supercomputers means that more sophisticated models and simulations are being deployed in the prediction of weather events and climate change. We asked two experts for their opinions, and take a look at how research groups are benefi tting from Numerical Algorithms Group (NAG) numerical routines.


T


Supercomputers are a must for creating dynamic simulations, explains Jane Strachan, a Willis Research Fellow at the Department of Meteorology, University of Reading


‘Simulation of tropical cyclones is a big test for climate models and computational resources. Tropical cyclones are relatively small-scale systems requiring high-spatial resolution; they occur over relatively small timescales, but they are affected by (and affect) large-scale climate activity. Running the models at high enough resolution for a suffi ciently long time, and with suffi cient physical complexity, is a computational challenge that is constrained by computational limitations, including data storage and visualisation. ‘Until relatively recently, climate models or General Circulation Models (GCMs) did not have the resolution to simulate tropical cyclones dynamically. ‘But several modelling centres around the world are running climate models at resolution not only able to simulate general atmospheric and oceanic circulation, but also relatively small-scale weather events, like tropical cyclones, with increasing realism. The resolution still places limitations on the intensity of such simulated events, but this is a huge leap for climate models.’


42 SCIENTIFIC COMPUTING WORLD


he accuracy of weather forecasting is still something of a clouded issue, but it would be impossible without high-performance computing. The


A clearer image


Brian Etherton, senior atmospheric scientist at the University of North Carolina’s Renaissance Computing Institute (RENCI), discusses modelling needs


‘Before joining RENCI I recognised a gap in the information being produced by computer models in the United States; namely the resolution they were capable of. I now have the computing power to go to a higher resolution and have recently been creating models of both one and three kilometres that physically track tropical cyclones as they


move out from Africa up the Carolina coast, and then back out to sea. ‘We are primarily using the weather research and forecasting model (WRF) version known as advanced research WRF (ARW), but one of its drawbacks is its scalability when a lot of processors are working on it. That’s proven to be a software limitation as you get to the point where the I/O dominates, and doubling the number of processors does not halve the run time. To take this to a much larger scale, you would need to solve this problem so that, if you apply more processors, you can reduce the time for completion. ‘Here stateside (and to be frank, Europe is a bit ahead of us in this regard), ensembles are the future of numerical weather prediction. It’s the notion of not just saying it will rain or not, but coming up with that spectrum of possibilities for proper probabilistic forecasting as well as risk assessment. WRF itself is being developed nicely and that’s going to carry on. But if you really want to make an ensemble then the workfl ow management from all things from CPUs to disk storage is where the greatest needs will be. There is going to be a lot of information and data to keep on track.’


By the numbers


Researchers at the National Oceanic and Atmospheric Administration (NOAA), at Oak Ridge National Laboratory (ORNL) doing advanced modelling with the Climate Modelling and Research System (CMRS), will now be able to use numerical routines from the Numerical Algorithms Group (NAG) Library on the new Cray XT6 supercomputer, known as Gaea. The CMRS will provide a dedicated high-


performance computing (HPC) resource for NOAA and its research partners, and provide a signifi cant increase in computing capacity


to address some of the most pressing global climate change questions. Gaea will be the world’s most powerful HPC system dedicated to climate research. NAG routines already help to power the


work of other climate and weather research teams across the globe, including those at the UK Ocean Observing and Climate Research Group in Southampton, at the German Potsdam Institute for Climate Impact Research, at the Italian Institute of Atmospheric Sciences and Climate and at the India National Centre for Medium Range Weather Forecasting.


www.scientific-computing.com


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44