UCERF2

The 2008 Uniform California Earthquake Rupture Forecast, Version 2, or UCERF2, is one of a series of earthquake forecasts prepared for the state California by the Working Group on California Earthquake Probabilities (WGCEP), collaboration of the U.S. Geological Survey, the California Geological Survey, and the Southern California Earthquake Center, with funding from the California Earthquake Authority. UCERF2 was superseded by UCERF3 in 2015.

Of the hundreds of seismogenic (earthquake causing) geologic faults in California, UCERF classifies only six faults as Type A sources, meaning there is sufficient information to both estimate and model the probability of a Magnitude (M) 6.7 or greater earthquake within 30 years. These six faults (summarized in Table A, below) are the: (1) San Andreas (split into northern and southern sections, (2) San Jacinto, (3) Elsinore, (4) Garlock, (5) Calaveras, and (6) Hayward-Rodgers Creek. Faults which are known to be slipping (and therefore seismogenic) but lack sufficient information to fully model how close they might be to rupture are classified as Type B. About twenty of these faults (see Table B) are estimated to have a 5% or greater chance of an M ≥ 6.7 earthquake within 30 years. An additional six areas where strain is accumulating but where knowledge is insufficient to apportion slip onto specific faults are classified as Type C sources.

There is additional chance of earthquakes on faults that were not modeled, and of lesser earthquakes. Northern California has an estimated 12% chance over the same 30 years of an M ≥ 8 megathrust earthquake on the Cascadia subduction zone. UCERF has also prepared "participation probability maps" of the chance that any area will experience an earthquake above a certain magnitude from any source in the next 30 years (see figure).

UCERF probabilities of an earthquake on a given fault are based on four layers of modeling:

These are used to produce both time-independent and time-dependent forecasts of earthquake probabilities. The former are based on "stress-renewal" models of seismic stress being released by an earthquake, then renewing (or rebounding; see Elastic-rebound theory) until it triggers another earthquake. In time-dependent models the probability of a fault rupturing thus depends on how long stress has been accumulating since the last rupture. However, the details of how this happens are not adequately known, so time-dependent methods estimate the periodicity and currently accumulated strain based on observed seismicity. Out of this a time-independent earthquake rate model (ERM) is produced, from which the time-dependent probability model (UCERF) is derived. (For more information see "Can earthquakes be predicted?" in the External links.)

...
Wikipedia