Project Abstract
|
We are constructing both long-term (years) and short term (days) parametric models of earthquake probability as a function of location, magnitude, time and focal mechanisms. We used them to forecast earthquakes over amgnitude 5 for yearly periods, and magnitude 4 for daily periods in Southern California. We have begun testing the models against other published ones. In the process we learned that small earthquakes are collectively as important as large ones in triggering future events; clustering is very strong (about 70% of earthquakes are triggered); and earthquake catalogs miss many earthquakes just after a moderate to large one. In the coming year we plan to (1) apply the same techniques to forecast earthquakes in Northern California, (2) apply the short-term clustering techniques at longer and longer periods, hoping the use the same techniques for long-period forecasting, and (3) develop and improved long-term model based fault slip rates, geodetic deformation rates, and an improved relationship between fault length and maximum magnitude.
Models will be based on magnitude 2 and larger quakes, used to predict magnitude 4 (short term) and 5 (long term). Tests will be based on the larger magnitudes appropriate to the time scale. Our triggered seismicity model will be part of the RELM effort to develop and evaluate a variety of viable models.
An important part of our work is also to complete publication of two manuscripts. One is abstracted in our anual report fo rlast year, and it is ready for submissiona after very minor revisions. The other is the precise methodology by which we have constructed short and long term forecasts specifically for the RELM forecast test. |
