Key points are not available for this paper at this time.
In the last century, the global N (nitrogen) cycle has been profoundly disturbed by human influences. One of the most detrimental consequences is the release of large quantities of N2O (nitrous oxide) into the atmosphere, which significantly contributes to global warming (Gulev et al., 2021). Most of the anthropogenic contribution to atmospheric N2O originates from the transformation of excessive reactive N inputs in agricultural food production systems (Tian et al., 2019). Mitigation strategies propose the use of EEFs (Enhanced Efficiency Fertilizers), which have shown large potential in decreasing N2O emissions from various types of agricultural systems (Akiyama et al., 2009; Fan et al., 2022). Two types of EEFs are generally considered: NIs (Nitrification Inhibitors) and CRNFs (Controlled Release Nitrogen Fertilizers). However, the effectiveness of EEFs is yet to be estimated at large spatial and temporal scales. The use of process-based biogeochemical models allows for the estimation of N2O emissions at various spatial and temporal scales and with greater accuracy than widely applied emission factors from the IPCC methodology. Within this thesis, a new routine to model EEFs is implemented in the LandscapeDNDC model framework (Haas et al., 2013). The routine largely follows the recent implementation in the DAYCENT model described by Gurung et al. (2021). For accurate results, biochemical models require their parameters to be calibrated on field data. Therefore, the new LandscapeDNDC routine was calibrated on measurement data from three corn cropping systems in the US. Contrary to DAYCENT model calibration in Gurung et al. (2021), it is the pretense of LandscapeDNDC to not only quantify cumulative emissions but to predict N2O emissions dynamics in higher temporal, e.g., daily time resolution. Thus, the calibration was performed over the entirety of available measurements instead of only on cumulative emissions. Moreover, it was investigated whether calibrating the model over every site simultaneously instead of separately for every site significantly contributes to overall uncertainty in the final results. Our resultsdemonstrate how LandscapeDNDC is able to recreate site and year-specific differences in EEF mitigation potentials. The RRMSE for NIs during the growing season ranges between 1.42 and 2.42. For CRNFs, the range is between 1.05 and 3.52. When reduction factors based on cumulative emissions are concerned, for NIs, the posterior reduction factor proposed by LandscapeDNDC is -12% (- 36% to 12%) (mean and 95% confidence interval), which is lower than the reduction factor suggested by the DAYCENT model -12% (-61.8% to 3.1%) and large observational datasets -38% (-44% to -31%) (Akiyama et al., 2009). For CRNFs, LandscapeDNDC returns a reduction factor of -2% (-28% to +25%), which is again lower than the DAYCENT reduction factor of -12% (52% to +1%) and the reduction factor suggested by large global datasets -35% (-58% to -14%) but compares with a larger observational dataset of multiple US corn cropping systems of -5% (-18% to +7%) (Eagle et al., 2017). However, considering the simulated magnitude and relative EEF reduction potential, large uncertainties remain, which are attributed to site-specific edaphic characteristics and growing season variability.
Weber et al. (Fri,) studied this question.