Japan launched its first early warning system in October 2007. When an earthquake is detected, warnings are broadcast over TV and radio that give at least a few seconds to respond. That can save people's lives and a great deal of money. If only 1 percent of the costs of a major earthquake can be saved, the system will pay for itself many tmes over. (from http://geology.about.com/od/quake_preparedness/a/quakewarnings.htm )
Japan to Get Quake Warning System 23/8/2007
Toward Real-Time Earthquake Warnings
Seconds before earthquakes, an early-warning system ... May 02, 2003
Research Center for Earthquake Prediction
This group at Kyoto University shows what real scientists are studying these days.
Quake Warnings Before the Shaking Starts
Earthquakes and Schoolhouses
Saving Lives with Earthquake Engineering
Sleeping Well in Seattle
The large earthquake that struck the area on February 28, 2001, was just a reminder of that state of affairs. This time, not a single person was killed. Seattleans should sleep better now, knowing that earthquake preparation really pays.
Strengthening houses, evaluating school buildings, compiling long-term landslide maps, and placing strong-motion seismometers in structures are all elements of this effort.
Starting to Save Shangri-La Earthquake Safety Day is an annual step forward for a precarious city.
"Earthquakes Cannot Be Predicted"
This classic 1996 article from Science hit the seismological community like a thunderclap.
Prominent seismologist Robert Geller reported on the RAS-JAG meeting in the pages of Eos:
The overwhelming consensus of the meeting was that earthquake prediction, in the popular sense of deterministic short-term prediction, is not possible at present. Most of the participants also agreed that the chaotic, highly nonlinear nature of the earthquake source process makes prediction an inherently unrealizable goal. . . . The mass and trade media's highly optimistic reports on prediction differ greatly from the extremely pessimistic consensus of the meeting. Participants agreed that efforts should be made to correct the media's misconceptions, but there was pessimism about the prospects for real improvement.
The word "chaotic" is significant, because the mathematics of chaos is a step forward. The pattern of earthquakes along a fault resembles other chaotic phenomena like the avalanching of a sand pile under a random rain of sand grains—while each individual occurrence is unpredictable, the bulk result can be modeled quite precisely. In the case of the sand pile, the bulk result is a cone with sides at the angle of repose. For earthquakes it is an overall level of energy release that matches the tectonic movement of the earth's plates. From that information we can confidently map the expected long-term hazards for a region—that is, we can construct long-term forecasts. This is vitally useful for planners, emergency agencies, and designers of buildings and other structures.
A related development was a 1996 paper from Science, by the all-star cast of Robert Geller, David Jackson, Yan Kagan, and Francesco Mulargia, was about as strong a statement as scientists can make on the subject, starting with its title—"Earthquakes Cannot Be Predicted."
(from http://geology.about.com/od/eq_prediction/a/aa_EQprediction.htm )
1906 San Francisco earthquake
Tangshan - The Deadliest Earthquake (唐山大地震)
Tangshan Earthquake: Politics & Response