Exploring sonar data using clustered visual features and coordinated geovisualization

Exploring and analyzing oceanographic sonar data is a difficult task due to the extreme ratios in the dimensions of the data. While sonar data may consist of many hundreds of thousands of sonar pings coving hundreds of kilometres, the ocean depth of the data is at a much lower resolution. As a result, visual representations of the sonar data (echograms) are normally shown as long and narrow ribbons of data. As an analyst zooms in to show the echogram in sufficient detail, much of the contextual information is lost and the analyst must perform horizontal scrolling to explore the data. In this research, we outline an approach that couples a technique for visually clustering slices of the echogram based on visual similarity, with a geovisualization method that shows the spatial location of echogram slices on a virtual globe. Panning and zooming within each of these views of the data results in coordinated filtering, such that data outside of the viewport in one view is dimmed and de-emphasized in the other view. This approach provides data analysts with a powerful geovisual analytics tool for exploring sonar data. In particular, analysts may filter the data based on spatial regions of interest, visually identify important features within the data, and observe the spatial relationships among the locations of the echogram slices