Visual Analytics for Large Scale Scientific Simulations, Fiscal Year 2019

Scientific ensemble data sets have played increasingly more important roles for uncertainty quantification in various scientific and engineering domains, such as climate, weather, aerodynamics, and computational fluid dynamics. Ensembles are collections of data produced by simulations or experiments conducted with different initial conditions, parameterizations, or phenomenological models. They are usually used to describe complex systems, study sensitivities to initial conditions and parameters, and mitigate uncertainty. The goal of this proposal is to develop visual analytic techniques for large scale scientific ensemble data sets. Using ensemble simulations as an example, for a single run of such a simulation, there can be data generated in the range of several hundred gigabytes to tens of terabytes. A large scale ensemble dataset can consist of hundreds or thousands of such instances, with many variables in the form of scalar, vector, or tensor, and has a large number of samples in the high-dimensional input parameter space.