NASA’s Jet Propulsion Laboratory is chartered to conceive and execute robotic spacecraft that explore other worlds. These craft are sent to the planets to orbit and sense their atmospheres, surfaces and interiors, and to asteroids, comets and moons to image and discover their composition. Other spacecraft land on the surface of Mars and rove the planet acting as field geologists. The sensors and spacecraft used for planetary exploration are also applied to remote sensing of the Earth, producing datasets that not only characterize our planet but also allow predictive modeling. Similarly, a range of observational systems for astronomy produces datasets of the heavens that are categorized and then drive astrophysical modeling.

Large-scale computation is required in a number of areas to achieve the goals described above. The design and simulation of engineered systems requires integrating high-fidelity physics for accurate model-based design. For example, planetary landing, like the recent Curiosity mission, requires simulations of the spacecraft in the presence of a range of uncertainties such as the knowledge of the Mars atmosphere. Similarly, a long-term goal will be the real-time ingestion and analysis of multiple data sets that are used to autonomously guide a landing craft to a scientifically valuable location. In the area of remote sensing, models that can extract the thickness and composition of the icy shell and subsurface ocean on icy moons, such as Jupiter’s moon Europa, requires the development of electromagnetic models and large-scale computation to retrieve the information from radar signals. Finally, spacecraft that capture Earth and astrophysical data sets with many 10’s of terabytes of data per day, down-linked from the observatories, will require state-of-the-art technologies in big data and analytics.

This talk will describe a range of problems requiring large-scale computation and data science as described above, intermingled with recent results from JPL missions.