How Do I Rate This?
The blue stars show the average user rating for this item. To add your own rating, move your cursor over the stars to highlight them in gold, and click to show your rating. One star highlighted is the lowest rating, all five is the highest. Once you have rated an item, your rating is added to the average.
Utilization of High Performance Computing
High-capability computing research is performed to support and improve interaction between science observations and science models, along with creating state-of-the-art engineering simulations. Computer scientists and modelers make use of JPL's hardware, middleware, and application resources to help investigate science and engineering questions that could support future NASA missions.
Computational science and engineering facilitates analysis and discovery are difficult to achieve by theory and/or experiment alone. Modeling, simulation, visualization, and analysis help to retire risk in engineering designs through trade-space exploration, and help to gain physical understanding within the Earth and space sciences. Computation has had a significant role in NASA mission success across engineering design, navigation, science data processing, and physics-based models in Earth and astrophysical sciences.
Accomplishing NASA’s future missions would require computational support for model-based prediction that would enable new insights for forming scientific questions and informing science investigation proposals. Moreover, the missions and spacecraft that would pursue these new and difficult investigations must, to be successful, be the product of focused analyses and design trades and the careful characterization of relevant physical phenomena in the space environment.
Development Plan
Selected Hardware Resources
JPL Supercomputing and Visualization Facility (SVF)
The SVF provides high-performance computing and visualization environment to support all JPL engineers and scientists, and their collaborators at other institutions, in solving computationally intensive problems.
Parallelizing MOPEX software enabled optimization of mosaics from large-area Spitzer imaging surveys.
This supercomputing facility provides a variety of supercomputers, such as a Dell 1024-Xeon CPU distributed system and three SGI Altix 3000 machines. Additionally, the facility can provide terabytes of storage space and employs visualization resources that allow scientists and engineers to view their high-resolution images at animation speeds.
One example of successful use of this facility is a project (“Efficient Mosaicking of Large Area Spitzer Surveys”) in which a parallelized implementation of existing image-mosaicking software (MOPEX) was developed to run on JPL’s 1024-processor cluster computer. The improved performance was used to optimize mosaics from large-area Spitzer Space Telescope imaging surveys. With the ability to complete a MOPEX run in two hours optimization of MOPEX parameters to improve cosmic ray detection has become feasible. This has enabled optimization of cosmic-ray detection and production of improved IRAC shallow-survey mosaics to facilitate searches for brown dwarfs and other rare objects. A sample of other success stories can be found at the SVF website.
Selected Middleware Projects and Resources
Grid Computing
Science missions and instruments continue to produce volumes of useful data, and scientists depend on the data systems and tools that archive this data as a means to access and analyze it. These existing legacy systems do not interoperate well, and scientists must access each data system and its corresponding science data independently through tools that have been custom-built for the particular science data system or mission. The Object Oriented Data Technology (OODT) task is working on the distributed resource location service, which will allow location and exchange of geographically distributed data. Advances in Internet and distributed object technologies provide an excellent framework for sharing data across multiple data systems.
GEOS-Chem simulation of ozone outflow from northeastern United States.
Parallelization
Parallelization is employed at JPL to enhance the speed and capabilities of a number of computing tasks. Here are three examples: (1) It has been applied to GEOS-Chem, a state-of-the-science global 3-D atmospheric chemical transport model, in an effort to increase performance and to remove limitations on global spatial and time scales over which simulations can be run. (2) Parallelizing a coupled earthquake-ocean tsunami model has enabled faster prediction of tsunami generation and demonstrated how JPL’s satellite capabilities can improve tsunami prediction and coastal management. (3) Parallelization has also been applied to JPL radar section’s point target simulator in an effort to enable distributed target simulation, in which tens of thousands of point targets are used to simulate the effects of natural terrain, allowing the study of phenomenological effects.
Selected Applications
The following are three examples of how supercomputing is employed at JPL:
High-Performance-Computing Soap Development
In one study, the computational performance of the Satellite Orbit Analysis Program (SOAP) was enhanced by utilizing the distributed-computing environment provided by JPL’s Dell cluster computer. The parallel parametric study routine was tested with the parametric study, which determined the best combination of right-ascending node and solar-panel orientation for a Mars orbiter to maximize the intensity of sunlight on the panels for a given 3-day period. The test example for the parallel contour calculation routine was to calculate the fraction of time at least one satellite of a two-satellite molniya constellation is visible from Earth for one day over a 120 x 120 grid.
Modeling Charged-Dust-Particle Interactions with Spacecraft for Lunar and Mars Missions
In another study, JPL researchers developed the first physics-based modeling tools for understanding the behavior of dust in the lunar environment in the presence of a spacecraft, a capability that is critical to assessing effects on spacecraft hardware, astronauts and science, including remote sensing (astronomical observations). Follow-on work would allow simulation of dusty plasmas, where the charged dust density influences the local plasma characteristics. This advance in the model would allow simulation of dust behavior on Mars and other objects in the solar system to increase understanding of dust transport, adhesion and removal mechanisms. This could be critical to Mars Sample Return canisters/seals, and to understanding dust dynamics for small bodies (comets, asteroids) and the potential effects on landers, impact probes, etc. Dust dynamics on the surfaces of small bodies (asteroids, moons) could be modeled using the enhanced simulation tool.
Cluster-Based Large Scale Surface Rover Simulations
In a third study, JPL researchers developed the capability to evaluate the performance of a rover technology over a range of environmental (or other) parameters by embedding the technology in the ROAMS simulator and running many simulations in parallel on a JPL supercomputer, each evaluating one point in the parameter space. The task demonstrated the applicability to testing the GESTALT rover-navigation algorithm, which is slated to be used on the MSL rover and to receive limited hardware testing time. In 2.5 hours, the testing framework performed tests that would have required approximately 5 days of continuous usage on a single hardware platform, assuming no downtime, no operator error, no time to reset to initial conditions, and assuming that the test conditions, such as varying slope and soil conditions, even could be provided for a hardware test.
ROAMS-simulated MER rover on a flat plane, a grid of rocks, and a sloped plane with a single rock.
ECCO
Estimating the Circulation and Climate of the Ocean is a joint cooperative enterprise of JPL, MIT, and the Scripps Institute of Oceanography. It was established in 1998 as part of the World Ocean Circulation Experiment (WOCE) with the goal of combining a general circulation model with diverse observations in order to produce a quantitative depiction of the time-evolving global ocean state.
RSS
Send
Bookmark
