Breadcrumbs

2018 NZMS Forder Lectureship - Applied stochastic modelling for structured physical processes

Date Friday 13 April 2018
Time 4:10pm - 5pm
Where S.G.01, University of Waikato
Presenter Prof. Valerie Isham, University College London
Contact Steven Miller
Contact email steven.miller@waikato.ac.nz
Admission Cost Free

Mechanistic stochastic models aim to represent an underlying physical process (albeit in highly idealised form, and using stochastic components to reflect uncertainty) via an analytically tractable model, in which interpretable parameters relate directly to physical phenomena. They contrast with highly detailed deterministic models that seek exactly to describe the evolution of a process from fixed initial conditions with no allowance for uncertainty. Both classes of models contribute understanding of the process dynamics and can be used to develop control strategies.

In this talk, I will consider some general approaches to modelling physical processes, in both time and space-time, beginning with the purposes of modelling, some questions to be addressed and choices to be made. I will describe some useful stochastic model components and their properties, and ways of modelling population structure, including the use of random networks.

I will illustrate these ideas by showing how showing how model components can be combined to build temporal and spatial-temporal models of rainfall, and of soil moisture, for hydrological applications.