Advances in probabilistic earthquake forecasting; A quiet revolution

Global Change Seminar

14/02/2018

Dr Maximilian Werner, University of Bristol

Pictures shows after effect of an earthquake; Image credit goes to National Geographic

 In this talk, the research is based onComparison with the quiet revolution of weather forecasting where it all link to how good it is in prediction to southern and northern of atmosphere.Review of in recent earthquakes forecasting and avenue for future research.

Methods

·      CSEP: platform for developing and evaluating forecast models

·      Recent insights from CSEP’s results and plan for future. Predicting earthquakes without relying to the past data hence eliminating all the biases. Testing centre and model’s location of CSEP.

·      Geophysics based forecast; the evolution of strain field (stress changes can help clamp and unclamp faults)

·      Fault response stress changes; friction model of nucleation

·      How strong is the aftershock patterns?

·      Predictive power of coulomb hypothesis?

·      Stress changes associated with the ruptures

·      Physics based model: too deterministic, not integrating all complicate ddata, the complicated geometry of faults etc

·      Improving the model

·      Aftershocks; damages

·      Newest model – physics but with more ability to predict strain changes rat

 RESULTS

o  Small earthquakes help the prediction of the future earthquake

o  First experiment: in California 2007-2011

o  Few models of combination of small-larger earthquakes, only large magnitude earthquake, (geodesy) based on gps data changing strain rate to seismological rate on par with usgs models, geological; fault predictions

o  Using probability mass function which only have one parameter; the rate of global strain or Geodesy

o  Focusing on the plate boundary and also some continental rifting 

o  So, if the combination with past seismicity (smoothed seismicity) would provide the better prediction?

o  This model has become the reference in global scale

o  Conclusion: combining the two is useful and much more reliable

o  Plate boundaries and where the crust is deforming, the experiment in California is encouraging that the data are useful, and the model can be used

Recent highlights

·      Importance of small earthquakes

·      Predictive skill of strain-rate based methods

·      Improved forecasting and the eolution of earthquakes cascades

·      2^ndgeneration coulomb predictions with encouraging results

·      New ensemble modelling and evaluation methods

further readings
The Silent Revolution of Probabilistic Voting
Earthquake prediction by hydrogeochemical record

Next Post: Exploring Zealandia; New continent of the earth