Dr. YanChing Zhang in the EPA has produced great visualization comparing model predictions with satellite images using Vis5D. We try to visualize the same dataset with feature-based techniques with  FVis5D. We believe the enhanced visualization as a result of feature tracking, coupled with some other standard visualization techniques can provide better insight of the dataset. 

In the above image, the features are first segmented with all the small objects(volume < 5) removed, and then colored with tracking history information. The GOES satellite image is overlaid over the map outline on the bottom of the 3D box. Color is used to depict the object history information. Click to see the mpeg movie. Meanwhile, the number of objects changes can be plotted as below.


 

As one may notice that the dataset is still quite complicated. It would be necessary to remove the visual clutter to reveal the object of interest. In the image below, we isolate the largest cloud (gray) starting from t=3. The evolution of the isolated object can be seen here. Furthermore the integrated content of the object is computed over time.


 
 

Simon Xin Wang <xswang@vizlab.rutgers.edu>