Modern technology has the ability to generate large datasets over space and time. Such data typically exhibit high autocorrelations over all dimensions. The field trial data motivating the methods discussed in this talk were collected to examine the behaviour of traditional cropping and to determine a cropping system which could maximise water use for grain production while minimizing leakage below the crop root zone. They consist of moisture measurements made at 15 depths across 3 rows and 18 columns, in the lattice framework of an agricultural field.

I describe the initial modeling of a single day's data which led to the ‘layered-CAR’ model, a simple extension of the (Besag et al, 1991) conditional autoregressive model, and included an errors-in-variable model, all fitted in WinBUGS. In moving to four dimensions, we again used the layered CAR model  (which accounts for differences in scale between horizontal and vertical measurements), and developed a  block updating Gibbs sampler in pyMCMC (Strickland, 2010),  to analyse the 59 days' of data collected over five years. The Bayesian framework permits transparent specification and easy comparison of the various complex models compared.

Besag, J., York, J. and Mollie, A. (1991). A Bayesian image restoration with applications in spatial statistics (with discussion). Annals of the Institute of Mathematical Statistics, 43, 1-59

Strickland, C. (2010). pyMCMC: A Python Package for Bayesian estimation using Markov chain Monte Carlo, Queensland University of Technology


Dr Margaret Donald

Research Area

Clements & Associates, Sydney and Prince of Wales Clinical School, UNSW


Fri, 10/08/2012 - 4:00pm to 5:00pm


OMB-145, Old Main Building, UNSW Kensington Campus