iCAS Post Conference Summary

By Staff
10/29/2013 - 12:00am

High-performance computing systems, adapting next-generation weather and climate applications to future computing architectures, current efforts in climate and weather research, and the challenges of big data in the atmospheric sciences were central themes at this year’s International Computing in Atmospheric Sciences (iCAS) conference.

The meeting, held in Annecy, France, brings together international colleagues to discuss information technology advances and developments in scientific computing, allowing scientists to study atmospheric questions as part of the earth system model.

Over 70 participants from 12 countries gathered for this year’s conference, which also included corporate sponsors IBM, Cray, Mellanox, Nvidia, Intel, PGI, Bull, Oracle and NetApp.

Also discussed at the meeting were challenges in designing and constructing facilities capable of accommodating current and future computing systems while reducing power consumption and operating costs.

The conference, which took place during the second week of September, included several keynote presentations, beginning with a talk by Dr. William Mahoney of NCAR on the challenges and opportunities in prediction of wind and solar energy.

In his presentation, Mahoney explained the importance of predicting extreme events, and the need for more detailed and representative data to help inform wind turbine design and describe solar irradiance over a variety of climates.

Dr. Charles Hansen of the University of Utah shared a wealth of imagery to illustrate the importance of scientific visualization to understanding processes and phenomena in the natural world.

In a talk that incorporated recurring topics of the presentations, Dr. James Kinter of COLA described his team’s experience during NCAR’s Advanced Scientific Discovery program, an initiative that allows scientists large allocations on a new supercomputer during the testing and implementation period.

Kinter’s project, which ran on the Yellowstone supercomputer, involved dramatically increasing the resolution of weather and climate models, and observing how improved resolution affected model fidelity and predictive skill.

“We must resolve the weather in order to get answers about climate,” Kinter explained, illustrating the need for continuing to push the limits of computational science, in order to better understand and predict natural processes that affect our everyday lives.

All the conference presentations are available at https://www2.cisl.ucar.edu/cas2k13/program