“Data Assimilation” (DA) is a fundamental technique to combine large-scale numerical simulations and observational/experimental big data through a Bayesian estimation on simulation model parameters and time-dependent status behind target systems. DA was originally developed in geophysical areas, especially in meteorology and oceanology, and is nowadays applied in wide variety of fields such as weather forecast, space, life and industrial sciences. Our eventual goal is to establish DA methodology, which is capable of generating next-generation simulation models that can predict future statuses from big data. Despite DA nowadays being applied in wide areas of science, opportunities to discuss and exchange information about DA are only limited in each specific community. This special session provides a forum to DA researchers of both natural science and engineering. Dissemination to attendees unfamiliar with DA and seeking new areas within which DA is applicable are also objectives of the proposed special session.
Topics of interest
This special session accepts various topics in wide areas of information science and applied statistics; (1) development of next-generation simulation models that are capable of predicting future status through embedding observational/experimental big data to existing simulation models, (2) development of algorithms that enable us to extract information behind big data, (3) development and improvement of sequential Bayesian filters in an attempt to apply for DA, (4) application of DA to real problems for extraction of structures and essences behind observational/experimental big data, (5) validation and reconstruction of simulation models, (6) implementation of DA algorithms on massively parallel computers, and (7) miscellaneous topics related to DA and/or big data including visualization software and cloud computing services.
Special Session Chairs
- Hiromichi Nagao, Earthquake Research Institute (Japan)
- Masaya M. Saito, The Institute of Statistical Mathematics (Japan)