Estimating annual N 2 O emissions from agricultural soils in temperate climates CAROLINE ROELANDT, BAS VAN WESEMAEL and MARK ROUNSEVELL Geography Department, Universite ´ catholique de Louvain, Place Louis Pasteur 3, B-1348 Louvain-la-Neuve, Belgium Abstract The Kyoto protocol requires countries to provide national inventories for a list of greenhouse gases including N 2 O. A standard methodology proposed by the Inter- governmental Panel on Climate Change (IPCC) estimates direct N 2 O emissions from soils as a constant fraction (1.25%) of the nitrogen input. This approach is insensitive to environmental variability. A more dynamic approach is needed to establish reliable N 2 O emission inventories and to propose efficient mitigation strategies. The objective of this paper is to develop a model that allows the spatial and temporal variation in environmental conditions to be taken into account in national inventories of direct N 2 O emissions. Observed annual N 2 O emission rates are used to establish statistical relationships between N 2 O emissions, seasonal climate and nitrogen-fertilization rate. Two empirical models, MCROPS and MGRASS, were developed for croplands and grasslands. Validated with an independent data set, MCROPS shows that spring temperature and summer precipitation explain 35% of the variance in annual N 2 O emissions from croplands. In MGRASS, nitrogen-fertilization rate and winter temperature explain 48% of the variance in annual N 2 O emissions from grasslands. Using long-term climate observations (1900–2000), the sensitivity of the models with climate variability is estimated by comparing the year-to-year prediction of the model to the precision obtained during the validation process. MCROPS is able to capture interannual variability of N 2 O emissions from croplands. However, grassland emissions show very small interannual variations, which are too small to be detectable by MGRASS. MCROPS and MGRASS improve the statistical reliability of direct N 2 O emissions compared with the IPCC default methodology. Furthermore, the models can be used to estimate the effects of interannual variation in climate, climate change on direct N 2 O emissions from soils at the regional scale. Keywords: croplands, emission factors, empirical model, grasslands, greenhouse gas, nitrous oxide Received 28 August 2004; revised version received 18 March 2005; accepted 5 May 2005 Introduction The increasing use of nitrogen-based fertilizers, N 2 fixation and atmospheric nitrogen deposition on agri- cultural lands is believed to be largely responsible for the current annual increase in N 2 O atmospheric concentration (314 ppb in 1998; the annual increment between 1990 and 1999 was 0.8 ppb yr 1 ; IPCC, 2001). The Kyoto protocol requires countries to provide national inventories for a list of greenhouse gases including N 2 O. The default methodology proposed by the IPCC estimates direct N 2 O emissions from soils as a constant fraction of the nitrogen input (Bouwman, 1996). This emission factor approach is based only on a limited number of data points and is applied worldwide for agricultural soils regardless of variations in soil char- acteristics, land management or climate. N 2 O is naturally produced in soils as a by-product of microbial processes. Nitrous oxide emissions from agricultural soils not only depend on the addition of nitrogen but also on other factors, e.g. climate, soil and management practices (Firestone & Davidson, 1989; Granli & Bckman, 1994; Davidson & Verchot, 2000). N 2 O production also varies in response to the variability of these factors in both space and time (Flessa et al., 1995; Wagner-Riddle et al., 1997; Kaiser et al., 1998a; Kammann et al., 1998; Smith et al., 1998; Dobbie et al., 1999; Ruser et al., 2001; Dobbie & Correspondence: Caroline Roelandt, fax 1 32 10 472877, e-mail: roelandt@geog.ucl.ac.be Global Change Biology (2005) 11, 1701–1711, doi: 10.1111/j.1365-2486.2005.01025.x r 2005 Blackwell Publishing Ltd 1701