Updating urban forest inventories: An example of the DISMUT model

Canberra is a unique city in Australia where the trees on public land that dominate the urban forest were planned for at the city's inception. In the mid-1990s, a 100% census of street and park trees was completed, and together with simple health, growth and yield models, this database formed the basis of a decision information system to support the management of the urban trees - DISMUT. The accuracy of the models was evaluated in a study in 2005 where models to predict total tree height were found to be unbiased and precise, tree crown dimension were under-estimated for small trees, and tree health was over-estimated. The over-estimate of health may be due to the relatively poor rainfall conditions over the past 10 years while the biases in crown dimension predictions are more likely due to a too simple model form. However, the existence of DISMUT predictions over all streets and parks in Canberra means that statistically efficient two-phase sampling approaches can be used to correct for any bias in the mean estimates of tree numbers and size, and also to predict the mean value of other environmental, economic or social parameters of interest that are correlated to tree size.