Spatio-temporal modelling of forest monitoring data

This thesis is about developing models to investigate spatio-temporal trends in defoliation levels in European forests. The dataset used is provided by the International Co-operative Programme on Assessment and Monitoring of Air Pollution Effects on Forests (ICP), which has been measuring crown condition on several thousand survey plots across Europe since the 1980s. Initially, a pre-existing generalised additive mixed model (GAMM) for defoliation aggregated at the survey plot level is adapted for use with the ICP data, and an improved method of displaying spatial patterns of change over time is developed. We then consider modelling spatio-temporal trends at the level of individual trees within the GAMM framework with suitable response distributions, before moving on to set up a cumulative logistic regression model (incorporating temporal autocorrelation) for defoliation as an ordinal response. Finally, we begin to develop an MCMC algorithm to estimate this model, and offer suggestions for future improvements.