Sim Vis The Simulation Visualization Tool

SimVis is a research software framework that was developed for the interactive visualization and analysis of complex (4D, multi-variate) data resulting from computational fluid dynamics (CFD) simulations, e.g., the injection and combustion processes in car engines (Doleisch et al. 2005; Schmidt et al. 2005). The key elements of this software framework such as interactive feature specification and visualization, brushing, linked views, focus+context visualization, derived data, or different types of views, are described in the following.

3.1.1 Feature-Based Visualization and Brushing

SimVis uses a feature-based visualization approach. This type of visualization is characterized by the possibility to focus on especially interesting subsets of the data, the so-called features. To give to the user the opportunity to interactively select features becomes increasingly important when dealing with larger data sets. In SimVis the user can specify features by a brushing mechanism in different views such as scatterplots and histograms. Brushing simply means to select data points directly on the screen (e.g., using the mouse), thus assigning a degree of interest (DOI) attribute to each data point. The DOI can be either 0 or 1 in the case of a discrete feature classification. In SimVis also fractional values of the DOI are possible, representing a "fuzzy" selection (Doleisch and Hauser 2002).

In the different views the user can select data attributes (variables) of interest (e.g., the temperature), which either stem directly from the imported data set or are derived attributes (e.g., the temperature trend).

3.1.2 Linked Views

Another important concept is the use of multiple, linked views. Different aspects of the data set are shown in several views side by side. Brushing in one view is immediately propagated to all other views (Doleisch 2004). Thus changing the selection in one of the views is immediately reflected in all other views, using a different coloring scheme for the selected data points. Visualizing different attributes (e.g., climate state variables or diagnostic variables derived from them) in the views easily reveals correlations as well as other distinctive features between the data attributes.

3.1.3 Focus+Context Visualization, 3D View

The 3D view is a core component of the visualization system and allows the user to orientate both in space and time (if time-dependent data is available). SimVis uses a focus+context (F+C) visualization (Hauser 2005): according to the DOI value, each data item is either drawn in an emphasized way (being in focus) or in a reduced style (e.g., transparent gray in the 3D view). This distinction is consistent through the linking in all views, but plays an especially important role in the 3D view (Fig. 1). The user can easily distinguish between the relevant data with high degree of interest (colored) and the context (in gray) in which it is located.

3.1.4 Derived Data

In order not to limit the user only to the data attributes (variables) already available in the imported data set itself, SimVis provides the possibility to perform certain mathematical operations on the existing attributes. This concept is very flexible

Fig. 1 SimVis 3D view. Shown is the trend in refractivity in the ECHAM5 data set for the year 2044 (seasonal summer means), for all pressure levels ("sheets") from 1000-20 hPa; each "sheet" itself is a latitude-vs.-longitude slice. Data points with high values of the signal-to-noise ratio (see Sect. 3.2) are selected ("in focus") and colored

Fig. 1 SimVis 3D view. Shown is the trend in refractivity in the ECHAM5 data set for the year 2044 (seasonal summer means), for all pressure levels ("sheets") from 1000-20 hPa; each "sheet" itself is a latitude-vs.-longitude slice. Data points with high values of the signal-to-noise ratio (see Sect. 3.2) are selected ("in focus") and colored

and easily extendable in the SimVis system. Currently available formulae include, for example, temporal and spatial gradients, data smoothing, elementary algebraic expressions, and normalizations.

3.1.5 Types of Views

There are several types of views available in SimVis, the most important ones being scatterplot, histogram, curve, and 3D view. In scatterplot and histogram views selections can be made via the brushing mechanism explained above. The curve view (Kehrer 2007; Muigg et al. 2008) shows the variation in time of each data point, drawn on top of each other. In this view, advanced brushing mechanisms can be applied, such as selecting all curves going through a certain volume in time and parameter space. The 3D view is a passive view (i.e., no selections can be made here) displaying the specified features using F+C visualization. Here, interaction in space and time as well as viewing perspective changes are available.

0 0

Post a comment