Hi Don, thanks for posting that, I agree with your diagnosis, but there is another factor,(perhaps more than 30% probable) an important consideration, where the recurrence period for some faults is so far apart that we haven't yet had a chance to see an event occur in recorded time. There have been several large earthquakes which occured on faults that were either thought to be inactive or were not even known to be faults. Many new faults are identified each year by concussion mapping such as in the Seattle area and Los Angelas areas and others. Of the faults we know of, Rodgers Creek poses a serious menace. Unfortunately there are so many unknown faults such as the one that produced the 5.2 at Yountville, that added together, there may be more risk from unknown faults than any one specific known fault. Hence the accelerated progrem of fault mapping that is happening of late. Forecasting by deterministic prediction techniques (officially) can only be done on well understood faults such as the SAF. Northridge and Kobe Japan were poorly understood faults, as seismology (arguably) is still in it's infancy. Then there's Parkfield just to throw a wrench in the gears! The 1991 Loma Prieta (World Series) quake was identified as one of the most likely areas to break and had two 5+ foreshocks, 2 and 15 months before the mainshock occured. The quakes at Topok lately may be part of a similar foreshock series with a 5.7 Aug 1 99 and now, further tremors. The official stance of the worlds top geophysists is that we should give up trying to predict earthquakes and spend the money used for prediction research on mitigation. The Joint Association of Geophysists was told in a meeting in London that "despite a huge international effort spanning many decades, not a single reliable sign for an impending earthquake has yet been identified." It was "highly unlikely" that quake precursors exist claimed Geller who accused some researchers of deliberately exaggerating the success of quake forecasting. "Deterministic prediction is just absolutely not not possible" Stuart Crampin of the University of Edinburgh told the meeting.
Let's compare weather forecasting with quake forecasting- If a forecast for a bad, (but not SEVERE), storm meant that an evacuation or heightened emergency level was undertaken every time, they would have to say that weather could not be accuurately forecast either. Because there is less at stake from an inaccurate weather forecast, it doesn't need to be 98% accurate to be called successful. With the cost of earthquake warnings being acted upon, the accuracy needed to be called successful must be higher. One, it's money, and two, the warning may cause panic, looting or deaths by stress related heart attack during evacuation. They knew Loma Prieta was about to occur at least three hours before by instrument readings (I'm being brief here)but said nothing because people were mostly at home on the holiday, and safer than if warned. It was an accurate diagnosis, when if a warning was given, the freeway that collapsed could have been bumper to bumper, and police and emergency services may have been tied up by looting or panic. Narrowing the window of earthquake occurence to a few hours or days would be extremely useful but is the hardest part of earthquake prediction. I could go on and on and probably will after I get my daughter a snack...mb. Take care all thanks for the dialogue...