The Infrasound Network (ISNet):as a 88D Adjunct Tornado Detection Tool: A Perspective After Operations Since the Spring of 2003

After operation through three years (2003-2005) the purpose of this paper is to summarize the status of our efforts to evaluate the potential of ISNet for tornado detection and warning. The network implementation has involved parallel efforts at hardware design, detection algorithm development, designing displays, application of a fully compressible numerical model to understand infrasound generation processes, and modeling propagation using a 3-D acoustic ray tracing program. In combination, these efforts are placing us in a position to realistically assess the potential of infrasound to help with tornado detection and warning. One of our initial expectations, (that a 3-station network separated by about 200Km would robustly triangulate on acoustic tornadic sources) was shattered. Only at longer ranges (typically > 200Km because of an upper atmospheric wave guide) did the 3 stations (at Boulder, CO, Pueblo, CO, and Goodland, KS) monitor the same source with clear detections. On the other hand, individual stations show regional tornado detection skill with different stations at times simultaneously detecting different regional tornadoes. Ray trace simulations have explained these regional detection differences, indicating that there is a need for closer network station spacing. Examples of both good and failed detections of tornadic storms are presented and recommendations made for optimum network spacing. The needs for further evaluation are reviewed; including implementation of a denser network, improved verification (network stations can contribute guidance to field programs), improved display options and timeliness. An intriguing aspect of our observations is that storms often produce infrasound about 30 minutes prior to a first tornado report. Numerical simulations in progress should help illuminate the infrasonic source processes. Although there are other potential uses for such networks (e.g. detection of turbulence aloft and avalanches at long ranges), we continue to focus on the potential for tornado detection and warning.