PhD PROJECT, fully funded (EPSRC)
Earthquake forecasting: Seismic source zonation and spatio-temporal modelling of Earthquakes
Supervisors: Mark Naylor1 , Ian Main1 , Janine Illian2 , David Bolin3 & Roger Musson4
1. School of GeoSciences, U. of Edinburgh ([log in to unmask], [log in to unmask])
2. School of Math. & Stats., U. of St. Andrews ([log in to unmask])
3.Mathematical Statistics, Chalmers University of Technology, Sweden
4. British Geological Survey, Edinburgh
Summary: This project combines recent advances in the modelling of spatial-temporal systems with earthquake data to improve seismic hazard maps. Related fundamental questions include: How stationary are patterns of seismicity? How reliable are methods to characterize sequence properties e.g. swarms or mainshockaftershock sequences? How do we integrate expert judgement?
Background: Earthquake forecasts quantify where and when future events are likely to occur. Seismic source zonation is the process by which observed earthquake data are combined with interpreted geological and geophysical information to define regions within which the spatial distribution and rates of earthquakes are likely to be similar.
However, this process is somewhat subjective and there is the potential to improve current practice using recent statistical advances. Spatial-temporal point processes are mathematical models used to describe and analyse the evolving geometrical structure of patterns formed by objects that are irregularly distributed in space – such as the location of earthquakes and tectonic faults. However, fitting these models using standard methodology such as Markov chain Monte Carlo methods (MCMC) tends to be computationally prohibitive, especially in the context of realistically complex models relevant to earthquake forecasting. Recent developments in computationally efficient model fitting involving integrated nested Laplace approximation (INLA) make fitting more complex models feasible. They have been applied with some success in modelling spatio-temporal population dynamics in ecology, but have not so far been applied to address earthquake hazard assessment.
This project will develop and apply statistical methodology within the area of computationally extensive spatial statistics. The methodological development undertaken in this project will include the translation of the existing methodology into the context of spatial point processes as suitable for the geological application in quantitative seismic hazard analysis, as well as the computational implementation of these using appropriate computational methods.
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|