Dear all!
I have been asked to forward this seminar invitation to the list. Sorry
for cross-posting.
Claus
-----------------------------------------------------------------------------------------------------------------------
You are invited to attend this seminar talk
TITLE: "Spatio-temporal models for red pine decline"
SPEAKER: Professor Jesper Møller (Aalborg University, Denmark)
DATE: Monday 27th of June 2005, 4:00pm.
VENUE: SIMBIOS Centre, Kydd Building (5th floor),
University of Abertay, Bell Street, Dundee, DD1 1HG
Please see below for the abstract. If you have any queries please
contact Janine Illian ([log in to unmask]).
-----------------------------------------------------------------------------------------
Professor Jesper Møller, Aalborg University, Denmark
"Spatio-temporal models for red pine decline"
Abstract:
Red pine decline is characterized by expanding pockets of dead trees in
plantations throughout the Great Lakes Region. Elucidation of exact
mechanisms of pocket development and expansion remain elusive since a
single site has never been observed over more than two years. In the
present study, we fit space-time models to a seven-year data set of
annual investigation of all trees in a plantation. Each year, each of
the 2,715 trees was examined for presence/absence of Ips spp., tree
condition and the number of pitch tubes, each of which signifies
colonization by a turpentine beetle. We attempt to answer the following
questions: How do we model the space-time progression of turpentine
beetles, Ips colonized trees and the expanding pockets of dead trees? Do
turpentine beetles predispose trees to attack by Ips spp.? Do Ips spp.
kill the trees?
This is an ongoing research project together with Kenneth F. Raffa and
Brian Aukema, Department of Entomology, University of Wisconsin, Jun
Zhu, Department of Statistics, University of Wisconsin, and my
Ph.D.-student Jakob G. Rasmussen, Department of Mathematical Sciences,
University of Aalborg. In the talk I'll compare two different approaches
based on a discrete-time model and a continuous-time model. In the
discrete-time model the likelihood function depends on unknown
normalizing constants, which need to be estimated when finding maximum
likelihood estimates. Moreover, an unknown ratio of normalizing
constants appears in the Hastings ratio when doing straightforward MCMC
posterior simulations, and I'll show how a new auxiliary variable
technique eliminates this problem. In the continuous-time model the full
likelihood is tractable, but for the marginal likelihood we need to
account for missing data, using again MCMC methods.
***********************************************************************************
Claus-D. Mayer | http://www.bioss.ac.uk
Biomathematics & Statistics Scotland | email: [log in to unmask]
Rowett Research Institute | Telephone: +44 (0) 1224 716652
Aberdeen AB21 9SB, Scotland, UK. | Fax: +44 (0) 1224 715349
***********************************************************************************
|