Project Abstract
|
We will carry out formal likelihood tests of earthquake forecasts developed under the RELM project, an dpublish technical descriptions of those models constructed at UCLA. RELM investigators have been developign forecasts (defined as statements of probability per unit area, depth, magnitude, time, and focal mechanism parameters) for testing against each other using future earthquake occurrence. Based on recent consensus decisions, RELM forecasts should be submitted this year as numerical tables of predicted earthquake rates in cells of specified latitude, longitude, magnitude, and focal mechanism parameters. Two main menu formats have been encouraged: long-term forecasts of magnitude 5 and larger earthquakes to be evaluated over a 5 year period, and short-term forecasts of magnitude 4 and larger events to be tested and updated daily. Other formats are welcome if two or more forecast models use them.
Testing the various models will help to reselve major questions in earthquake science (as manifested in the various forecast models), such as the validity of fault segmentation and quasi-periodic recurrence, the effect of Coulomb and dynamic stress on earthquake triggering, and relationship of surface deformation to later earthquake offurrence. Testing will also help to identify those methods appropriate for consensus models, such as the planned RELM-ERF (earthquake rupture forecast) models for seismic hazard and loss estimation. |
