Performance Impact of Streaming Doppler Radar Data on a Geospatial Visualization System
A geospatial visualization tool that is capable of visualizing weather data in conjunction with the terrain and object data typical of GIS visualization tools provides an important advancement in the study of severe weather formation by providing a tool for weather scientists to study of the impact of land topography and human changes. In particular, unlike the plains region, tornados in the Southeast form quickly, are generally small, and are short lived. Weather scientists would like to better understand the impact that high human population density areas, such as metro Atlanta, and topography have on tornado formation. Such an understanding could lead to better and earlier tornado predictions. The visualization of Doppler radar data raises interesting questions for the particular geographic visualization tool we are working with, the Virtual Geographic Information System (VGIS). In particular, Doppler radar data is streaming, that is, it arrives as a continuous stream in near real-time. GIS visualization sytems such as VGIS are often designed under the assumption that the data to be visualized is relatively static. In this paper we assess the performance impact of visualizing Doppler radar with VGIS, and propose a novel data structure and indexing scheme that reduces the performance overhead of retrieving and visualizing streaming data.