Growth and yield modelling of Pinus radiata in Canterbury, New Zealand
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
During this study two growth and yield models were developed for Pinus radiata D.Don plantations in Canterbury, New Zealand, namely CanSPBL and CanJuv. CanSPBL is a model for the whole rotation of stands owned by the Selwyn Plantation Board Limited in Canterbury. CantJuv is a model for juvenile growth from after planting to before thinning in Canterbury. An existing stand growth and yield model CANTY was examined using a newly established relational database of PSP measurements. Projection bias was shown for mean top height, basal area per hectare and volume per hectare. Height-diameter relationships were modelled for estimating individual tree height and mean top height. Of the sixteen functional forms evaluated the Petterson equation with exponent -5 and the two-parameter Richards' equation led to the smallest mean square error at stand level. Incorporating stand age, site index and altitude into the selected Petterson equation reduced the mean square error by 72% for a pooled regional data set. The new model, CanSPBL, was achieved with precision at both stand and tree level. The components of the stand model include mean top height, basal area per hectare, stems per hectare, volume per hectare and diameter distribution. The inclusion of altitude into a chosen polymorphic Schumacher difference equation significantly reduced mean square error by 17% for mean top height and 41% for basal area. An examination of this stand model using two sources of data at plot and stand levels showed little apparent bias. Two main model components of tree diameter and tree mortality were developed to complete an individual-tree projection system. For projection of diameter, an approach based on relative basal area was found best, compared with many sigmoid difference equations. A logistic regression procedure showed that relative diameter, altitude, the interaction between initial stocking and projection interval length, and site index were significantly correlated to tree mortality. Nine sets of experimental data and thirty-one temporary plots were used to refine the juvenile growth model of stand level, CanJuv. The components of the model were mean top height, basal area, sectional area at ground level, and survival. Regression analyses identified age, annual rainfall and weeding as the significant factors in determining growth and mortality.