Baisden, WT (2006). Agricultural and forest productivity for modelling policy scenarios: evaluating approaches for New Zealand greenhouse gas mitigation. JOURNAL OF THE ROYAL SOCIETY OF NEW ZEALAND, 36(1), 1-15.
The New Zealand Government is currently reviewing policies for the Kyoto Protocol's first commitment period, and entering international negotiations that will define the rules for the second commitment period. Policy makers and stakeholders require robust tools to evaluate national scale policy scenarios. Models for robust scenario analysis must combine the best available biophysical and economic information, and include understanding of uncertainties. I evaluate three different datasets as candidates for estimating the biological net primary production (NPP) of pastoral and forest land, as well as C sequestration potential under exotic or indigenous forest. For scenario analysis of land-use change, estimating NPP is critical to represent the flow of agricultural and forest production into the economy, while estimating C sequestration potential in forests is required to estimate the quantity of C available for credits and liabilities. First, a 4-year average of NASXs 1 km annual NPP product from the MODIS satellite sensor was verified using published New Zealand data, and used to test and calibrate the following indices. The MODIS NPP data were compared to average stocking indices for pasture and site indices for Pinus radiata forests derived from 10- to 30-year-old mapping in the Land Resources Inventory (LRI). Third, a Storie Index approach was developed using factors representing climate and soil properties that co-limit potential productivity, and calibrated to MODIS NPP. Each approach represents the apparent variation in NPP at the national scale, but may risk serious errors at a more local scale. The calibrated Storie Index approach offers the best predictions for pastoral and tussock land (R-2= 0.71) and indigenous forest (R-2 = 0.38), while the LRI-derived site index is most predictive for exotic forests (R-2 = 0.16). While both the LRI-derived indices and the Storic Index are recommended for modelling policy scenarios, the Storie Index approach shows the greatest versatility because it does not depend on current or past land use for estimation, provides the highest level of spatial detail, and can easily be improved.