Decision-support systems for establishing radiata pine plantations in the central North Island of New Zealand

摘要

A framework for decision-making relating to establishment of radiata pine plantations was defined, with provision for both numerical models and non-numerical representations of knowledge. Data from Nelder-design experiments were used to investigate the amount of between-tree competition occurring in young radiata pine plantations. Dbhob was found to be unrelated to initial stocking prior to year five. Modelling of basal area/ha growth and yield in a Nelder-design experiment showed that functions used in traditional basal area models under-estimated basal area growth during the two years following the time when mean height was 1.40 m. An adjustment was made to these functions, allowing for allometric assumptions on which growth models are based, which improved models estimates of early basal area/ha growth. Models of young radiata pine survival and size class distribution models were built for crops aged 0 to 5 years in the Central North island region of New Zealand. Data came from site preparation experiments, and the models are sensitive to variations in altitude and site preparation practices. Off-site preparation practices studied, weed control was found to have the largest effect on both initial survival and growth. Mounding improved growth to a lesser extent, and cultivation improved survival of young trees. Fertilisation with nitrogen and phosphorous was found to have a negligible effect on growth and no effect on tree survival. The basal area/ha function incorporated the allometric adjustment developed during the analysis of Nelder-design experiments in a way which resulted in compatible mean height and basal area/ha models. As an illustration of the potential for non-numerical decision-support tools, a knowledge-based computer program was developed to assist forest managers in selecting herbicidal treatments prior to, or during the years following plantation establishment. The system was built using techniques developed for artificial intelligence applications, in a form which allows updating of knowledge relating to weeds, herbicides, surfactants, application methods and treatments, by experts unfamiliar with computer programming. Opportunities for incorporation of these tools into a comprehensive decision-making and control system are discussed.