This is a reminder of next week's meeting on space-time modelling, to be
held next Wednesday afternoon at the Royal Statistical Society headquarters
(see http://www.rss.org.uk/about/direction.html for directions). Full
details can be found below,
Richard
==================================================================
ENVIRONMENTAL STATISTICS SECTION
Wednesday 1st December, 1.30-5.30 at the RSS (Tea at 3.20)
Spatial-temporal statistics in environmental science
PETER DIGGLE (Lancaster University)
Spatio-temporal Point Processes, Partial Likelihood, Foot-and-mouth
In this talk I will first make some general comments on the
analysis of spatio-temporal point process data, distinguishing
in particular between data in the form of a discrete-time sequence
of spatial point patterns or a continuous-time record of individual
events, and using several epidemiological data-sets for motivation.
I will then discuss work in progress on a retrospective
analysis of the recent UK foot-and-mouth epidemic in which I
use a partial likelihood method to fit a model which can
then be used to investigate the impact of different control
strategies on the spread of the disease.
RICHARD LAW (University of York)
Plant population dynamics in space and time
Spatial structure in plant communities both determines and is
determined by neighbourhood-dependent birth, death and growth processes.
This talk discusses some recent research on the coupling of spatial
structure and temporal dynamics. The work involves stochastic
birth-death processes in time and space, and deterministic
approximations based on dynamical systems of first- and second-order
spatial moments. These neighbourhood-dependent models have some
properties quite different from those of mean-field models
traditionally used in ecology.
PATRICK BROWN (Lancaster University)
Space-time modelling of animal diseases
Farm animals in Tanzania were tested for 5 tick-born diseases, with each
animal being sampled repeatedly during the course of the study. It is
believed the diseases have spatial dependence as they are transmitted by
insects. Temporal dependence may exist within animals or be part of a
general trend common to all animals. Covariate information such as age
of the animals is also believed to be a factor. In order to assess how
the disease spreads in space and over time, a model with a latent
Gaussian spatio-temporal process and binary observations is
constructed. Inference on the model parameters provides information
about the nature of the disease's epidemiology.
VALERIE ISHAM (University College London)
Space-time modelling of rainfall for continuous simulation.
Spatial-temporal models of precipitation fields have been developed by an
interdisciplinary team of statisticians from UCL and hydrologists from
Imperial College. These are being used to support the continuous simulation
of rainfall-runoff models for use in hydrological design. Our approach to
rainfall modelling combines the strengths of both point-process-based
stochastic models and statistical (generalised linear) models. In
particular, the former models are able to represent high space-time
resolution, while the latter more easily enable spatial and temporal
nonstationarities, including possible climate-change scenarios, to be
incorporated. In this talk, we describe the construction and fitting of a
stochastic-mechanistic model that is stationary in space and time, and
discuss the assessment of results from the continuous simulation of the
fitted model. We then consider extensions of this model to allow for
spatial and temporal nonstationarities (using a GLM to drive the
nonstationarity).
=====================================================================
Richard E. Chandler
^^^^^^^^^^^^^^^^^^^
Room 135, Dept of Statistical Science, University College London,
1-19 Torrington Place, London WC1E 6BT, UK
Tel: +44 (0)20 7679 1880 Fax: +44 (0)20 7383 4703
Internet: http://www.ucl.ac.uk/Stats (department)
http://www.homepages.ucl.ac.uk/~ucakarc (personal)
email: [log in to unmask]
---------------------------------------------------------------------
|