High-performance modeling and simulation are playing a driving role in decision making and prediction. For time-critical emergency support applications such as severe weather prediction, flood and wildfire modeling, and influenza modeling, late results can be useless.   With HPC resources distributed in a naturally fault-tolerant way across the nation, the community can build infrastructures and policy frameworks for utilizing high-end computational resources in support of emergency computation.  A specialized infrastructure is needed to provide computing resources quickly, automatically, and reliably. SPRUCE is a system to support urgent or event-driven computing on both traditional supercomputers and distributed Grids.  Currently, SPRUCE is deployed at several large supercomputer centers. The Linked Environments for Atmospheric Discovery (LEAD) project is one of our initial applications. LEAD used the SPRUCE system embedded into their portal for simulating and predicting severe weather events during the tornado season.  We are also beginning to work with projects modeling wildfires, influenza and pandemics.

To fulfill the mission of serving the computational, data management, and research needs of the atmospheric and earth science communities, the NCAR Computational and Information System Laboratory provides stable services utilizing proven technologies in high performance computing, data analysis, and data archival, and at the same time pursues promising new technologies designed to allow scientists to address transformative scientific problems.  This presentation will describe the cyberinfrastructure environment supported at NCAR, outline the challenges that exist in maintaining and enhancing this environment, and discuss the short-term plans and long-term options for augmentation of the computational and related facilities.

When stars like our sun are young they spin much more rapidly than the current solar rate and appear to have much stronger magnetic activity. This may indicate that stellar dynamo action is stronger in these rapidly rotating solar-like stars. We have begun simulating the global-scale stellar convection and resulting dynamo action in these stars with the 3-D compressible MHD anelastic spherical harmonic (ASH) code. As a great surprise, we've found that these stellar dynamos build global-scale magnetic fields in the bulk of the convection zone. Exploring how these complex three dimensional fields interact with the turbulent convection has been greatly facilitated by the development of advanced 3-D analysis tools, including Vapor. Here we explore the nature of these dynamo generated fields and the possibility of similar processes working within stars like our sun.

We describe our experiences with climate modeling applications on Blue Gene. In particular we describe ongoing work with the Parallel Ocean Program (POP) on the 20-rack Blue Gene/L system at IBM Thomas J. Watson Research Center. We also provide an update on the progress of an ultra-high resolution Community Climate System Model (CCSM) configuration capable of utilizing Blue Gene and other massively parallel systems.

The Earth System Modeling Framework is a scalable, portable infrastructure for turning modeling codes into collections of components with standard interfaces. The Earth System Curator is a portal for browsing, archiving, and retrieving components, and for storing information about viable model assemblages along with associated datasets. Together, these pieces open up possibilities for collective model construction and validation that transcend the current paradigm of single-institution ownership of complex, grand-challenge modeling systems.

Mixing in seawater is caused by a combination of shear-driven turbulence and convection. Convection often results from double diffusive instability, in which the different molecular diffusivities of heat and salt generate regions of anomalously high and low density. Shear-driven turbulence and double diffusive convection have been studied individually and are relatively well understood, but their interactions are not, despite the fact that the two mechanisms act in concert in about one half of the ocean interior.

When ambient horizontal currents vary in the vertical, double diffusive convective motions are organized into “salt sheets”, alternating planar regions of rising and sinking fluid which contain within them opposed horizontal flow patterns parallel to the sheets. Early in the flow development, the salt sheets attain large amplitude and three-dimensionality is introduced via two distinct secondary instabilities. The first causes the sheets to buckle into nearly-horizontal strips. The second induces billowing motions at the upper and lower tips of the sheets. Shortly after the appearance of these secondary instabilities, a complex and poorly-understood sequence of events leads the flow to a fully turbulent state which has properties of both shear-driven turbulence and double diffusive convection.

The process described here depends heavily on the vast difference on the molecular diffusivities of heat and salt in water, which renders the simulation a computational grand challenge. The simulations described in this talk were the first to take account of that difference and to evolve to a fully turbulent state. They are therefore the first direct simulations of turbulence in seawater.