Development of marginal emission factors for N losses from agricultural soils with the DNDC-CAPRI meta-model

The article discusses marginal emission factors for N losses from agricultural soils, with rape and wheat as examples, and presents results for EU15 as high-resolution maps and aggregated to Member State level. The results are generated by linking the economic model for the agricultural sector CAPRI (Common Agricultural Policy Regional Impact) with spatial down-scaling, and a statistical meta-model for the bio-physical model DNDC (DeNitrification-DeComposition). For a given agro-economic scenario, CAPRI supplies for each crop the crop share, yield and fertilizer application rate spatially downscaled to clusters of I km x 1 km grid cells. The results from CAPRI are processed by a meta-model of DNDC to estimate the local greenhouse gas emissions from the soil. DNDC is a dynamic process-oriented model, which estimates trace gas fluxes and nutrient turnover in agricultural soils. The fit of the regressions is typically very good (similar to 0.95R(2) for the majority of the regressions), and all coefficients are significant at 99% probability. The meta-model allows a seamless integration between the economic and the bio-physical models, offering additional benefit such as the site-specific calibration of the bio-physical model ensuring the match between simulated and observed yield at the grid-level. The meta-model is used to calculate marginal emission factors for a 1 kg ha(-1) increase of mineral N and manure fertilizer rates for rape and wheat, at different levels of fertilization. They show that for Western European farming practice, only a small fraction of extra nitrogen fertilizer would go into increased yields: most of it would be emitted to the environment. The largest spatial variability is observed for N2O emissions. The derivation of marginal emission factors is just one of the many possible uses for the linked regionalized agro-economic and soil chemistry model, which exploits to a large extent both geo-referenced and regionally available statistical information at European scale. (C) 2009 Elsevier B.V. All rights reserved.