Title: Data Assimilation: Leveraging Big Data for Scientific Determination
Speaker: Mark D. Butala, Technical Staff, Ionospheric and Atmospheric Remote Sensing Group, NASA Jet Propulsion Laboratory, California Institute of Technology
Abstract: Big data, the accelerating volume of information resultant from ever less expensive sensing and processing technology, is proving itself transformative in many domains, from how businesses, financial institutions, and governments make decisions, to the forecasting of Earth’s meteorological and overall climate systems. The opportunities of big data are now available to the space physics community with its multibillion dollar fleet of spacecraft dedicated to observing the Sun, near-Earth space, and intermediate regions, remotely measuring this connected system with an increasing number of viewpoints, spectral diversity, resolution, and cadence. For example, the Solar Dynamics Observatory, launched in 2010, down-links 1.5 terabytes of scientific data per day. This significant investment of scientific resources will continue to grow as our increasingly technological society must better understand and forecast the impacts of extreme space weather phenomena, i.e., solar flares and coronal mass ejections, that could, in an extreme case, devastate the Global Positioning System, power grids, and dependent technology.
In this seminar, I present my contributions to data assimilation, an emerging discipline for utilizing Bayesian inference methodology to formulate scientific conclusions based on big data. The computational solution to the resultant inference problem is challenging both for the high-dimensionality and the a priori constraint given by a first principles physics model of system dynamics. My research, throughout graduate school and now professionally at the Jet Propulsion Laboratory, has focused on leveraging the increasing space physics data volume to improve data assimilative models of the solar corona and Earth’s ionosphere, space weather source and sink regions, respectively. Specific contributions discussed include a proof of ensemble Kalman filter convergence, a simplified mathematical framework for tomographic coronal temperature structure estimation, and the first global reconstruction of coronal structure based on data from multiple sensors in space.
Speaker Bio: Mark D. Butala received an Honors Bachelor of Electrical Engineering degree with Distinction from the University of Delaware in 2002, graduating summa cum laude. At the University of Illinois at Urbana-Champaign, he received the M.S. and Ph.D. in Electrical and Computer Engineering degrees in 2004 and 2010, respectively, funded in part by competitive fellowships awarded on the basis of research proposals, including an NSF Graduate Research Fellowship, and by funding secured from research laboratories/industry in the form of an MIT Lincoln Laboratories Graduate Fellowship. In 2010, he joined the technical staff at the Jet Propulsion Laboratory, California Institute of Technology as a member of the Ionospheric and Atmospheric Remote Sensing group within the Tracking Systems and Applications section. His contributions have been recognized with NASA Group Achievement Awards “for outstanding development of real-time techniques to detect ionospheric perturbations due to tsunami using the Global Positioning System” and “for outstanding achievement in the operation and successful execution of the Curiosity rover’s mission of exploration to the surface of Gale Crater on Mars.” His research interests include the theory of remotely sensed image formation, Monte Carlo and statistical signal processing theory and practice, and the application of rigorous Bayesian methodology to big data problems in science.