Simulation ensembles such as the ones simulating deep water asteroid impacts have many facets. Their analysis in terms of detecting spatio-temporal patterns, comparing multiple runs, and analyzing the influence of simulation parameters requires aggregation at multiple levels. We propose respective visual encodings embedded in an interactive visual analysis tool.

Physical simulations aim at modeling and computing spatio-temporal phenomena. As the simulations depend on initial conditions and/or parameter settings whose impact is to be investigated, a larger number of simulation runs is commonly executed. Analyzing all facets of such multi-run multi-field spatio-temporal simulation data poses a challenge for visualization. It requires the design of different visual encodings that aggregate information in multiple ways and at multiple abstraction levels. We present a top-down interactive visual analysis tool of multi-run data from physical simulations named MultiVisA that is based on plots at different aggregation levels. The most aggregated visual representation is a histogram-based plot that allows for the investigation of the distribution of function values within all simulation runs. When expanding over time, a density-based time-series plot allows for the detection of temporal patterns and outliers within the ensemble of multiple runs for single and multiple fields. Finally, not aggregating over runs in a similarity-based plot allows for the comparison of multiple or individual runs and their behavior over time. Coordinated views allow for linking the plots of the three aggregation levels to spatial visualizations in physical space. We apply MultiVisA to physical simulations from the field of climate research and astrophysics. We document the analysis process, demonstrate its effectiveness, and provide evaluations involving domain experts.