Werner, Maximilian

Dr. Maximilian Werner joined the group as a postdoctoral fellow in January 2008. He is currently working within the CCES funded project EXTREMES. His research interests lie broadly within the statistical modeling of seismicity in order to test physical ideas about how, where and when earthquakes occur and how large they can be. To this end, he employs advanced statistical methods, including stochastic point processes, extreme value theory and data assimilation.

During his doctoral research at UCLA with Didier Sornette and Dave Jackson, he used the Epidemic-Type Aftershock Sequences (ETAS) model to test claims of universal scaling laws in seismicity, inspired by an analogy to universality in critical phenomena in statistical mechanics. He studied how much of the observed earthquakes might be triggered by small, perhaps undetected, earthquakes, and found that this amount is very significant. He characterized the uncertainties of magnitude measurements in modern earthquake catalogs and studied their adverse impact on model forecasts and on the tests that evaluate the forecasts. He also developed a novel methodology that recognizes such measurement uncertainties and optimally combines measurement errors with model forecasts to generate earthquake forecasts. Inspired by the idea of data assimilation in meteorology, he adapted sequential Monte Carlo methods (particle filters) for earthquake forecasts based on renewal point processes.

Lately, Max has been working on two projects. Firstly, he has generated short term and long term earthquake forecasts for California and Italy, as part of the international Collaboratory for the Study of Earthquake Predictability (CSEP). Secondly, he is estimating probabilities of extremely large earthquakes, for which little or no data exists. The estimates are based on extreme value theory, a statistical method based on mathematical theorems that show that the maxima of many randomly distributed variables are themselves distributed according to one of three so-called extreme value distributions. He is working on combing this purely statistical approach with geological and geophysical knowledge.