High-Resolution Upscaling of Closed Chamber Fluxes for N2o Emissions from China’s Agricultural Soils

Abstract:

Moving from local toward global N2O emissions brings up numerous issues related to data processing, aggregation, tradeoffs between model quality and data quality, and prioritization of data collection and/or compilation efforts. We studied these issues in the context of modelling China’s N₂O emissions from agricultural soils. We developed a spatially-explicit model (PKU-N₂O-Agr model) for high-resolution mapping of N₂O emissions based on the idea of Hole-in-the-Pipe Model. We collected 709 site-year records (504 for upland and 205 for paddy) at 106 experimental sites across China from 1994 to 2013 and calibrated the observed N₂O flux by using the Bayesian Recursive Regression Tree algorithm. The calibrated PKU-N₂O-Agr model is applied to simulate China’s N₂O emissions from upland and paddy cropland at 1-km spatial resolution and to examine the variable importance and sensitivity for N₂O emissions as well as scaling dependence of the effect-response relationships. The N₂O emissions in 2008 are 615 GgN₂O/yr and ~25% lower than PKU-N₂O and EDGAR v4.2 global product sampled over China. The average coefficients of determination between observed and simulated results were 0.91 for upland and 0.92 for paddy cropland, which indicate the using a simplified data-driven approach with data of high resolution could produce accurate and reliable results. Emission factors (considering background emissions) for paddy and upland soils are 0.6% and 0.8% of N inputs, which are 2 times of IPCC default but half of the mean of observations, respectively. SOC is the most important for capturing the variability of N₂O emissions from upland, whereas N inflow is the critical factor for paddy cropland. Different with previous works, the marginal sensitivities of environmental factors on agricultural N₂O emissions are calculated, which is of great use for verifying process-based simulation model when being applied in China (e.g., DNDC). Both critical factors and the effect-response relationships are significantly different under diverse spatial scales, which indicate that the variables and structure of process-based models need to be updated for adapting agricultural N₂O simulation at different scales.