Planted forest carbon monitoring system - forest carbon model validation study for Pinus radiata

A plot based inventory system in conjunction with models is being used to facilitate predictions of carbon stocks and changes in New Zealand's planted forests. The models include the 300 Index Growth Model for Pinus radiata D.Don to predict stem gross and net volume under bark over a rotation using plot data, linked with a wood density model to convert stem volume to carbon, and C_Change to calculate carbon stocks annually in four pools - above-ground biomass, belowground biomass, dead wood and litter. This linked suite of models is called the Forest Carbon Predictor version 3. This model validation paper aims to empirically determine the accuracy and precision of carbon stock and change estimates and predictions from the Forest Carbon Predictor, using independent above-ground biomass measurements acquired at permanent plots located in 39 stands throughout New Zealand and dead organic matter measurements from 14 stands. Model error was assessed using plot inventory data acquired in the same year that biomass measurements were made (model estimation error), and using plot measurements made nominally 5 years before or 5 years after the biomass measurement (model prediction error). Model bias and 95% confidence interval of the bias averaged -1.2% ± 2.6 m3 ha-1 for stem volume, -0.8% ± 1.9 kg m-3 for wood density, 3.7% ± 7.9 t ha-1 for total carbon (excluding mineral soil carbon), -0.9% ± 5.6 t ha-1 for above-ground biomass carbon, and 4.7% ± 12.6 t ha-1 for dead organic matter. The model prediction error was similar to the model estimation error over growth projection intervals of ± 5 years. Total carbon stock estimates at the inventory date and stock change projections over a 5 year interval are expected to average within 5% of actual values. © 2011 New Zealand Forest Research Institute Limited.