Overview and State-of-the-Art of Uncertainty Visualization - Scientific Visualization: Uncertainty, Multifield, Biomedical, and Scalable Visualization - page 19

Fig. 1.5 The EnsembleVis tool [ 84 ] for exploring short-range weather forecast data
1.6.2 Weather and Climate
Uncertainties are prolific in weather and climate applications and arise not only from
insufficient models, but also from our inability to accurately measure current weather
conditions and obtain precise knowledge on parameter settings. The typical approach
for mitigating uncertainties in weather and climate applications is to perform multi-
run simulations, often using a collection of models, parameter perturbations, and
initial conditions to generate outcome results for multiple variables and time steps.
While the variables contained in the output of both weather and climate simulations
are similar, the main differences between the two domains are the spatial region of
interest and the duration of time covered. Weather applications are typically only
interested in a small subsection of the planet, such as North America, and run to
cover time steps within the near future. In contrast, climate modeling has a spatial
interest of the whole planet and is run over hundreds of years.
The uncertainty resulting from these multi-run simulations are typically captured
in what is know as “Ensemble data sets”. These ensembles combine the multiple runs
such that notions of probability of outcome can be explored. An ensemble consists
of multiple simulation realizations and are often generated by pre-defined parameter
perturbations. The visualization and analysis of these data sets aims to understand
variations between models and effects of parameters and initial conditions, culminat-
ing in an understanding of the phenomenon leading to weather and climate events.
An example of a visualization and analysis tool can be seen in Fig. 1.5 , which
shows a screen shot of the EnsembleVis framework for the exploration of short-range
weather forecasting data [ 84 ]. This tool uses a multiwindow approach to provide a
collection of views for the end user, an approach used by other tools [ 92 ]. This