Upscaling NZ-DNDC using a regression based meta-model to estimate direct N2O emissions from New Zealand grazed pastures.

摘要: The availability of detailed input data frequently limits the application process-based models at large scale. In this study, we produced simplified meta-models simulated nitrous oxide (N2O) emission factors (EF) using NZ-DNDC. Monte Carlo simulations were performed and results investigated multiple regression analysis to produce EF. These then used estimate direct N2O emissions from grazed pastures in New Zealand. Zealand EF maps generated with national scale soil maps. Direct pasture calculated by multiplying map a nitrogen (N) map. Three considered. Model 1 included only organic carbon top 30cm (SOC30), 2 also clay content factor, 3 added interaction between SOC30 clay. median annual estimated each model (assuming errors purely random) were: 9.6GgN (Model 1), 13.6GgN 2), 11.9GgN 3). values corresponded an average 0.53%, 0.75% 0.63% respectively, while corresponding inventory was 0.67%. If error can be assumed independent for pixel 95% confidence interval order ±0.4-0.7%, which is much lower than existing methods. However, spatial correlations could invalidate assumption. Under extreme assumption that identical approximately ±100-200%. Therefore further work needed assess degree correlation errors.