Spatial Estimates for CH4 and N2O Emissions at the Continental Scale Using a Direct Inversion Technique With Recursive Source Area Aggregation

The representativity of mixing ratio observations in the boundary layer is an important issue and depends for example on the location of the site and its climatology, sources and sink strength in the immediate surroundings, sampling strategy and the observed component. Depending on the component, area and processes one wishes to study the sampling strategy will therefore vary strongly. For long lived greenhouse gases with high background mixing ratio values and relatively small sources (and sinks, if any) this will work out completely different as for short lived reactive tracers like ozone or terpenes. In all cases the variability through emissions or reactions at time scales of seconds to hours will be observed together with variability due to atmospheric transport and mixing processes on these same time scales. On longer time scales of days to months the mixing ratio signal observed is always a composed signal due to processes on connected larger spatial scales from continental to hemispheric scale. On the annual to climatic time scales global processes will start to dominate the variations. Disentangling the different processes at different timescales requires the use of atmospheric transport models that are sufficiently adequate at the required scales in time and space. A combination of coupled transport models each working at different characteristic scales in time and space seems to be a feasible way to carry out analysis of observed time series at discrete observation sites. In this paper we present a combination of the offline coupled eulerian global transport model TM5, providing the global boundary conditions, and a set of Eulerian and Lagrangian regional transport models. From this combination we derive by direct inversion a spatial explicit emission map for methane and nitrous oxide in western Europe, based on continuous mixing ratio observations of these gases at a set of background and tall tower sites in Europe. The regional models used are the eulerian WRF model and the lagrangian models Flexpart and COMET. The inverted emission maps vary in resolution from 10*10 km close to the observation points to 1000*1000 km at large distances from the observations. Total emissions for methane on a country by country comparison with bottom up estimates compare generally well, although significant differences larger than the uncertainty estimates are found. The differences between inversion results and bottom-up inventory nitrous oxide, where the largest uncertainty in emissions is expected, are smaller than for methane.