Probabilities of occurrence for events in the Geysers region have been posted
with other evaluations of predictions for the region. While it is true that events
do occur there every day, these events are generally of Ml<2.5. Earthquakes in
the region between Ml 2.5-4.0 only occur during relatively infrequent active periods
which generally last for 2-3 days per month. Since the beginning of 2001, there
have been 15 5-day periods during which an earthquake has occurred within
40 km of the epicenter identified by Don within the magnitude range 2.4-2.7.
In the same time there have been 66 such periods. The probability of success
of Don's forecast is thus 15/66 =0.23 or odds of about 1 in 4.
Don has been making a series of predictions over the past several monts for
the Geysers region. Since seismicity in this area is sporadic at the Ml>=2.5
level, this is actually a good test of whatever method he is using. The probabilities
of success are relatively high, but multiple successes without intervening failures
suggest his method has some validity.
While some predictors choose to forecast only large events, the statistics on such
forecasts must be collected for years (because of the long intervals between
large events). Most predictors give up long before enough predictions have been
made for any meaningful statistics to be done under these circumstances.
The prediction of small earthquakes, while not something that will headlines
across the nation's newspapers, is, nevertheless extremely important. Any predictive
technique worth it's salt must be shown to be accurate over long time frames and
consistent. For the most part, this can only be done with the prediction relatively small events.