A Methodological Study of Big Data and Atmospheric Science

Indiana University, Bloomington IN

Investigators

Abstract

This award supports a two-year project that investigates fundamental methodological aspects of Big Data in atmospheric science. The PI, in collaboration with scientists at the National Center for Atmospheric Research and several research assistants (a post-doctoral fellow and some graduate students), plans to establish some basic foundations, understandings and analysis of the relative merits of a variety of methods in the analysis and application of Big Data within atmospheric science and modeling. The main questions to be addressed by the researcher and her collaborators include the following. How do practices and technologies for data collection, dissemination and use affect the production of scientific knowledge? What is the role of theories and hypotheses within research practices and data analysis? If data-driven research constitutes a distinctive mode of knowledge production, how is that knowledge best delivered? The researcher intends to establish a philosophy of Big Data in atmospheric science; she plans to disseminate the results of her research to different audiences by producing several papers for diverse professional journals. The researcher also intends to formulate and communicate any knowledge for policy that might result from her Big Data research. She also plans to train a post-doc and some grad students so that they may serve as resources for policy makers to facilitate effective application of atmospheric science to public policy. The PI and her collaborators will examine what is involved in moving from the Big Data context of the outputs of multiple atmospheric models involving terabytes of data, to the applications and reduction of that data to a particular city's request for specific temperature forecasts, and how this analysis might become more automated through analysis of Big Data in a way not being done at present. This fundamental problem facing the regional modelers in modeling groups around the world is that there are tens of thousands of city and regional planners who need information from the regional weather models, but the information these users and impact-personnel need is not available to read off of the model without the help of the scientists who created it. They need translators between the models and the impact-personnel and users. One modeling group the PI would be working with at the National Center for Atmospheric Research is attempting to develop automation of various kinds to answer a range of questions from users and impact-personnel, automation that reduces Big Data into small and specific answers to specific questions that avoids various pitfalls and peculiarities of the models. That is, they are trying to build Big Data software systems that could act as translators. These problems are significantly exacerbated by the data being Big in one way or another, and they find that the available Big Data analytics are not helping them in the way they ideally could. The PI and collaborators propose to highlight, clarify, and define more precisely what exactly this group and others could use in their applications to social contexts. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.