Thread: SSA session on forecasting

Started: 2018-12-21 21:50:41
Last activity: 2018-12-21 21:50:41
Topics: SSA Meetings
Michael, Andrew
2018-12-21 21:50:41
On behalf of my co-conveners, let me bring the following SSA session on
earthquake forecasts, from short-term to PSHA to your attention and from
seismological to communication methods.

See you in Seattle,

Better Earthquake Forecasts

Earthquake forecasts have a wide range of applications from short-term
guidance during earthquake sequences and swarms to being an ingredient in
long-term Probabilistic Seismic Hazards Assessments (PSHA). In this
session, we will discuss what makes an earthquake forecast useful and how
to improve them. For short-term forecasts of swarms and earthquake
sequences, most current, official forecasts are based on statistical models
of earthquake clustering such as the Reasenberg & Jones model or the ETAS
(Epidemic Type Aftershock Sequences) model. Can we improve these by
including physics-based models of stress transfer or results from numerical
simulators of earthquake occurrence on fault networks? For long-term
forecasts, PSHA often relies on seismicity rates obtained by smoothing
declustered earthquake catalogs. Would other declustering methods improve
the forecasts or should we abandon declustering altogether and include
aftershocks in hazards assessments and building codes? Some PSHA now also
includes deformation information from plate motions or geodetic monitoring.
How do we best combine that information with the seismicity rates? For all
forecasts, how do we include fault-based information and do we need better
ways to address earthquake catalog incompleteness and uncertainty? A
critical step is testing these forecasting methods and the forecasts
themselves, for example using approaches from the Collaboratory for the
Study of Earthquake Predictability (CSEP). As we develop tests we need to
consider the role of local versus global tests, prospective versus
retrospective tests and tests of forecast ingredients versus complete
forecasts. Questions about testing are particularly timely as CSEP develops
its second phase of operations. Finally, we need to communicate these
forecasts with different users to help inform a variety of decisions. These
communications methods range from hazard curves for engineers to simplified
text or graphics for the people impacted by the earthquake, broadcast
media, emergency managers and first responders. Working alongside our
social science colleagues is an important step to understanding more about
our users, the channels they prefer and what information they need most to
inform their decisions. We seek contributions that address any of the
questions posed above or other ideas on how to improve earthquake forecasts.


Andrew J. Michael, U.S. Geological Survey (ajmichael<at>

Camilla Cattania, Stanford University (camcat<at>

David D. Jackson, University of California, Los Angeles (djackson<at>

Sara K. McBride, U.S. Geological Survey (skmcbride<at>

Warner Marzocchi, Istituto Nazionale di Geofisica e Vulcanologia (

Maximilian J. Werner, University of Bristol (max.werner<at>

Note: my email changed to ajmichael<at> on May 7, 2018 but
michael<at> should continue to get to me.

19:14:05 v.f0c1234e