A system for large-scale visualization of streaming Doppler data

The NEXRAD Level II super resolution Doppler radars continuously scan the atmosphere above the continental USA, providing a stream of temporally and spatially misaligned large volumetric data about cloud reflectivity, wind velocity, and spectrum width. This data is used for immediate and long term weather predictions. However, because this large amount of sparse streaming data is not temporally aligned, the existing approaches rely either on a 2D projection of the 3D data, or the display of the 3D data only for a single radar. We present a framework that enables users to interactively access, analyze, and visualize the Doppler reflectivity data directly in 3D for multiple radars. Our approach extends the existing body of work on large-scale storage of global weather data and out-of-core volume rendering using CUDA ray-casting. The asynchronously streamed reflectivity data from multiple radars are first temporally aligned and then processed to a hierarchical format that is suitable for a large-scale volumetric visualization in near-real time with a minimal run-time processing. This approach also allows for varying precision and level of detail.