Workshop Report: Petascale Computing in the Geosciences in View of the National Science Foundation's Recent Announcement Entitled: Leadership- Class System Acquisition -creating a Petascale Computing Environment for Science and Engineering, Which Calls for Deployment of a Petascale Computational Fac

15 floating point operations per second) to solve scientific questions of strategic importance by or around the year 2011, and because the time between now and then is not more than ample for scientists to prepare to use such a computer, a series of workshop around Petascale Computing in the Geosciences were organized to wit GARPA, GARPA2, Petascale Computing and the Geosciences I and II. With the permission of NSF, the workshops GARPA2 and Geosciences II were combined as they have largely overlapping goals. The objectives of these workshops was to examine the opportunities for progress in the geosciences that could be enabled by the petascale computational capability and to determine the steps necessary to ensure that this community is prepared to take advantage of such resources when they come on line. The joint workshop report is below, with a concise bullet list of recommendations provided first for ease of reference, and the more detailed analysis and findings that led to these recommendations follows. 1. Recommendations The potential benefits of petascale computing to advance scientific discovery in the geosciences, and to improve economic competitiveness of the U.S and, and the climate for all world citizens, and to better understand the world we live on, as identified in the analysis and findings below, are manifold. However, to achieve these benefits, several specific steps are required, both of the community of research scientists in the geosciences and computer science that should advance relevant science investigations, and by funding agencies that should provide resources to carry out these science investigations. The recommendations of this workshop, developed from the analysis and findings section are: • A portfolio of candidate petascale applications should be established, and development funding should be provided, for collaborative teams of geoscientsits and computer scientists to prepare calculations that can both advance scientific discovery and run at petascale. As a community we need to assemble a portfolio of application classes on the path to petascale. We should not strive to make this a large, comprehensive portfolio at first, but rather seek to obtain high success by enabling a few strategic geoscience computations at petascale in the first few years of system availability, particularly to solve science problems of impact. This set of workshop participants identify the following set of broad geoscience application areas as being inherently suitable for development to go to petascale, and have science impact, both in the near …