Enhancing the efficiency of Climate and Weather Simulation in High Performance Computing Environments

Award Information
Agency:
National Aeronautics and Space Administration
Amount:
$124,169.00
Program:
SBIR
Contract:
NNX12CD31P
Solitcitation Year:
2011
Solicitation Number:
N/A
Branch:
N/A
Award Year:
2012
Phase:
Phase I
Agency Tracking Number:
114844
Solicitation Topic Code:
S6.01
Small Business Information
Tellus Applied Sciences, Inc.
1769 Jamestown Road, Williamsburg, VA, 23188-2300
Hubzone Owned:
N
Woman Owned:
N
Socially and Economically Disadvantaged:
N
Duns:
780575218
Principal Investigator
 Atanas Trayanov
 Principal Investigator
 (757) 220-3182
 atrayanov@tellusappliedsciences.com
Business Contact
 Henry Pierce
Title: Business Official
Phone: (757) 272-7851
Email: pierce@tellusappliedsciences.com
Research Institution
 Stub
Abstract
A central focus of NASA's Global Modeling and Assimilation Office (GMAO) atmospheric general circulation modeling effort is the development of an atmospheric model suitable for data assimilation, weather forecasting, and climate simulation. Ongoing developments are focused on highly parallel processing, global simulations of increasing resolutions, and increased coupling of the earth system's process models. While model computation scales very well with number of available processors, a major constraining factor on the efficiency of these simulations is the input processing of Terabyte-size source data files used by the gridded component models. Our objective in this research is to increase the efficient use of CPU time associated with these simulations by paralleling I/O processing operations using an 'I/O Staging Server' which captures and makes available the required source data asynchronously with the simulation run. More efficient I/O for reading model restarts and boundary conditions, and writing model output and checkpoint files will free up processing resources that are currently idling during I/O. As a result, we will realize a significant increase both the number of models that can be involved in the simulation and the achievable resolutions of the grid components. It is estimated that currently up to 25% of a forecast run is consumed by I/O, a factor we think can be reduced by at least 50% or more through the use of State-of-the-art I/O processing approaches and supporting software infrastructure.

* information listed above is at the time of submission.

Agency Micro-sites

US Flag An Official Website of the United States Government