|
Titel |
On the statistical analysis of maximal magnitude |
VerfasserIn |
M. Holschneider, G. Zöller, S. Hainzl |
Konferenz |
EGU General Assembly 2012
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 14 (2012) |
Datensatznummer |
250062914
|
|
|
|
Zusammenfassung |
We show how the maximum expected magnitude within a time horizon [0,T] may be
estimated from earthquake catalog data within the context of truncated Gutenberg-Richter
statistics. We present the results in a frequentist and in a Bayesian setting. Instead of deriving
point estimations of this parameter and reporting its performance in terms of expectation
value and variance, we focus on the calculation of confidence intervals based on an imposed
level of confidence α. We present an estimate of the maximum magnitude within an
observational time interval T in the future, given a complete earthquake catalog for a time
period Tc in the past and optionally some paleoseismic events. We argue that from a
statistical point of view the maximum magnitude in a time window is a reasonable parameter
for probabilistic seismic hazard assessment, while the commonly used maximum
possible magnitude for all times does almost certainly not allow the calculation
of useful (i.e. non-trivial) confidence intervals. In the context of an unbounded
GR law we show, that Jeffreys invariant prior distribtution yields normalizable
posteriors. The predictive distribution based on this prior is explicitely computed. |
|
|
|
|
|