An Integrated System for Near Real-Time 3 D Visualization of NEXRAD Level II Data on TeraGrid

The occurrences of severe weather have caused many injuries, fatalities, and severe damage to personal and public properties. Weather forecasting and severe weather prediction help reduce such damages by providing opportune warnings so that people can protect their lives and properties, change economic operations, and plan daily activities ahead of time. Traditionally, the evaluation and analysis of actual weather conditions are done using 2D satellite images. In the past few years, the National Weather Surveillance 1988 Doppler Radar Network (WSR-88D) [1] also known as the Next Generation Weather Radar system (NEXRAD) – became available, providing high spatial and temporal resolution 3D data on a continuous basis. Despite the availability of 3D information in the new generation of radar data, this data is most commonly displayed today as 2D images, simple 3D point clouds, or iso-surfaces. The resulting images provide limited information about the movement of the frontal systems, as viewed on web-sites such as www.weather.com, www.wunderground.com or when viewed on TV. The data is typically displayed only for single radar and the data from multiple sites are just overlapped using transparency. These results are usually enough for a simplified view but cannot accurately represent details. A better visualization of the radar data captured by the Doppler network can significantly help weather forecasters and researchers to gain fresh insights on weather conditions and could greatly improve our knowledge and help educate future scientists. Major computational challenges exist in providing visual displays that fully utilize the 3D information in the radar data in real-time, and we have addressed some of the challenges by using TeraGrid resources. In this paper, we present an integrated solution for near real-time data delivery and 3D visualization that can be deployed as a service gateway to engage experts and non-experts alike. This system utilizes the NEXRAD data distribution already available on the TeraGrid and the TeraGrid Condor resource to speed up the processing of data from multiple radar sites, coupled with hardware accelerated graphics processing for interactive 3D visual analytics. The paper begins with an introduction where we discuss the scientific motivation, challenges, and our contributions. We then review previous approaches, related works and existing methods in Section 3. Section 4 describes the design of the integrated system. Sections 5 to 7 describe the schemes and methods of our approach in detail. Experimental results are presented in Section 8. Section 9 concludes our work and addresses some future improvements and extensions of this work.