of the Pacific might, by telegraphic communication, receive eighteen to twenty-five hours’ warning of the coming of destructive sea waves resulting from earthquakes in South America.
Based on Milne’s observation that seismic waves propagate much more slowly than electromagnetic waves, single-seismometer warning systems were first deployed in Japan in 1950 to stop trains after earthquakes. New communication and computing technologies have advanced these warning systems over the past 50 years (97). Real-time seismic systems have also been deployed in Taiwan, Mexico, and the United States, involving dedicated digital seismic networks that are automatically analyzed by computers and broadcast to interested users (e.g., emergency service organizations, utilities, train operators). In Mexico, a seismic alert system, operating since 1991, warns of subduction-zone earthquakes several hundred kilometers west of Mexico City. During the September 14, 1995, Guerrero earthquake (M 7.3), its warning arrived 72 seconds before the strong ground motion (98). An early warning system set up by the USGS after the 1989 Loma Prieta earthquake gave workers on the collapsed freeways in Oakland (about 100 kilometers from the rupture zone) as much as 20 seconds’ notification before the shaking from strong aftershocks, allowing them to evacuate hazardous areas (99).
The requirements for a real-time monitoring system are: reliable components that will perform during an earthquake, broad bandwidth communications facilities, and seismic instrumentation with high dynamic range and broad band-recording capabilities. With this approach, real-time systems in the United States are being developed in southern (Caltech USGS Broadcast of Earthquakes) and northern (Rapid Earthquake Data Integration Project) California (100). With partnerships between university researchers, private corporations, and government (federal, state, and local), these systems have largely evolved by upgrading the communications and data-processing infrastructure of existing seismic networks. Significant improvements in the southern California seismic network are currently under way (101).
The recent deployment of high-performance systems in California has motivated an active dialogue regarding operational goals and practices. This has focused attention on the differing data delivery needs before, during, and after the start of ground shaking, the type of information required for different purposes (e.g., simple warning, magnitude, hypocenter, extent and location of strong ground shaking), the tolerance for false alarms, and the policies for broadcasting data. Consideration of these questions has revealed a much broader range of applications for real-time data than originally anticipated. At first, the systems were expected to deliver warnings before strong ground shaking began so that evasive