Simulator to evaluate tsunami warning performance for coastal HF radars

The four-decade old prediction that coastal HF radars intended for current mapping could detect and warn of tsunamis via their orbital velocities has been confirmed by dozens of offline observations over the past four years. A detection method for SeaSonde systems (of which 600 now operate worldwide) has shown how the unique (but often weak) tsunami signal can be pulled out of background currents and noise by a “q-factor” pattern recognition algorithm. As with any event detector, one must trade off probability of detection with false alarm rate. We developed and describe here a simulator, that allows one to optimize tsunami performance based on local bathymetry and background conditions. Running on real-time systems in the field, it correlates false-alarm and detection probabilities with the background, injecting a simulated model-derived tsunami signal into noise and echo normally used for current mapping. This allows optimizing algorithm performance and detection thresholds masked by background. Minimal tsunami-intensity is predictable at a given site. Filters that correlate alert spikes with other information (e.g., seismic events) further reduce false alarms, so that robust warnings are passed along to our TWCs (tsunami warning centers). Our company has begun an official partnership with NOAA's NWS and NOS to evolve this capability into TWC operations.