The Visualization Group is currently working with the Space Sciences Laboratory on an information visualization project for supernova spectra called SpectraVis. SpecrtaVis is an interactive visual interface for browsing and displaying supernova data. Implemented in Java using Piccolo, a toolkit providing support for 2D graphics and the development of zoomable user interfaces, the tool will be used by the astrophysicists that are researching the mysterious dark energy, one of the latest most important scientific discoveries. Understanding and exploring the nature of the dark energy - which acts as an accelerator for the expansion of the universe - comes mainly from the study of type Ia supernovae. The Nearby Supernova Factory (SNFactory) - an international astrophysics collaboration - is the largest data volume supernova search currently in operation. The large amount of data (80GB per night) has to be rapidly and accurately analyzed in order to find supernovae. Studying them in detail so that they can be used as cosmological distance indicators is a challenge and here is where we proposed a tool: a visual interface to display and browse supernova data. The interface will use both scientific and information visualization. My work was on the information visualization part of the project. One of the main info vis techniques we used was the "fisheye" visualization. The fisheye view provides a detailed view by magnifying the area of interest, while preserving the context with less important information.
The browser will be part of a comprehensive supernova data catalog, workflow visualization and scientific analysis application called Warehouse.
Interface Description The fisheye browser uses a grid layout to preserve the time-series concept behind the study of SNe. Each dot of the grid corresponds to a supernova spectrum. The vertical axis contains the target names of the supernovae and on the horizontal has the age in days of the supernova or the phase. The period of scientific interest in the study of type Ia SNe ranges between 15 days before peak brightness and 40 days after peak brightness. In order to see a more detailed spectrum, the user clicks on one of the spec nodes and the fisheye technique makes the clicked dot and its neighbors bigger. The user can click one more time on the spectrum of interest and the zooming is taken at a higher level.

The semantic zooming is taking place at four levels: the initial dots, the two small spectra of focus and neighbour size, and the enlarged graph:

The fisheye lens approach to the visual browser was combined with standard information visualization techniques such as highlighting and filtering. While the fisheye changes the display size relative to focus, the highlighting changes the display type relative to focus.

Rolling over the phase or target names, the corresponding column or row is highlighted in red. Same happens when rolling over the grid.

Filtering is one of the features in progress. Momentarily, a double range slider was created to remove or select data for a certain phase interval. The slider can be operated by either adjusting its length or by entering values in the text fields and pressing return.

A similar double range slider will be implemented for the number of targets, allowing the user to add or remove as many supernovae as he wants.

A summer student's Presentation and Talk on the project.

Download tarball.