Preparing forecasts of assorted yields for forest crops is fundamental to managing forests. Studies of growth and yield in the form of systems of prediction equations provide managers with information on instantaneous, periodic, and whole stand growth and yield information, which provide a first means of regulating yield from forests. The availability of appropriate compatible stern volume and taper equations provides further quantitative information about resources in terms of merchantable lengths, end diameters and volumes of particular sections of the stern of the tree necessary to determine the mix of products. This thesis describes in particular how a system for forecasting assorted yields of Douglas fir in the South Island of New Zealand was developed.

The system comprises two models, namely DfirTree and DfirStand.

DfirTree is a compatible tree volume and taper prediction system, developed to cater for Douglas fir trees grown throughout the South Island, in Canterbury, Nelson and Southland. The volume - taper prediction system is based on the principle of splines (segmented polynomials) and provides two approaches with which to determine volumes of any part of the stern: (i) volume based and (ii) taper based. DfirStand is a simultaneous growth and yield model for simulating growth of Douglas fir in all four regions of the South Island, namely Canterbury, Nelson, Southland and Westland. DfirStand is developed through a state - space approach, the variables used to describe the state of the system at any time being mean top height, net stand basal area/ha, stocking, thinning history and local environment.

Both components of the overall yield prediction system show how regional attributes can be aggregated and modelled in a more realistic manner through the use of dummy variables to explain locality adaptation(s) where applicable. Rather than having a proliferation of models or an unwieldy quantity of adjustment factors, this system envisages a return to the traditional general volume, taper and yield prediction systems that can be developed with modern technology, ones which utilize the power of user - friendly computer hardware and software to provide the requisite sensitivity for forecasting assorted yields.