I am currently
interested in statistical models for the size and rate of occurence
of peaks over threshold river flow events. So far work has
concentrated on data at a single-site only, but we are hoping to
expand this to a multiple site analysis. I am particularly
interested in modelling the over-dispersion in the rate of events
through the use of random effects models.
I am also interested in modelling extremal dependence, especially for
variables with margins which are non-stationary and/or time
dependent variables. This includes many environmental data sets.
A statistical comparison of survival and
replacement analyses for the use of censored data in a contaminant air
database: A case study from the Canadian Arctic. Eastoe, E.F.,
Crispin, C.J., Heffernan, J.E. and Hung, H. (2006) Atmospheric
Environment, 40, 6528-6540.
Modelling non-stationary extremes
with application to surface level ozone. Eastoe, E.F.
and Tawn, J.A. (2009) Applied Statistics, 58,22-45.
A hierarchical model for
non-stationary multivariate extremes: a case study of surface-level
ozone and NOX data in the UK Eastoe, E.F. (2009) accepted
The distribution for the cluster
maxima of exceedances of sub-asymptotic thresholds. Eastoe, E.F.
and Tawn, J.A. (2007).
Nonparametric estimation of the
spectral measure, and associated dependence measures, for multivariate
extreme values using a limiting conditional representation.
Eastoe, E.F., Heffernan, J.E. and Tawn, J.A. (2008).
Statistical models for
over-dispersion in the frequency of peaks over threshold data from
UK flow series. Eastoe, E.F. and Tawn, J.A. (2009).
Risk, rare events and
Faculty of Science and
Technology Christmas Conference,
FREE Annual Science
UK Extremes Meeting,
UK Extremes Meeting,
The title of my
PhD thesis is `Statistical models for dependent and non-stationary
extreme events'. The work was supervised by Jan Heffernan, Jonathan
Tawn and Crispin Halsall. The motivation for most of the work came
from attempting to analysis the extremes of several surface level air
pollutants - NO2, NO and Ozone. The thesis has four chapters
(less the introduction). In the first we consider the distribution of
local threshold exceedances (cluster maxima) under certain conditions
on the sub-asymptotic extremal index (specifically a lack of stability
within the range of observed data). In the second, we suggest a
pre-processing approach for modelling threshold exceedances of a
non-stationary series; this is extended to the multivariate case in
the third chapter. Finally in the fourth chapter, and moving into more
traditional multivariate extremes territory. we derive a new
non-parameteric estimator for the spectral measure and Pickand's
dependence function. A copy of the thesis can be downloaded here [pdf].