ALBERT

Description: This study investigates some of the principal errors arising in atmospheric inversion estimates of N2O surface fluxes. Using a synthetic data set of model-generated atmospheric N2O mixing ratio data, representative of the current observation network, we investigate the influence of errors in the stratospheric N2O sink and in vertical transport. Our inversion framework uses a variational formulation of the Bayesian problem, and atmospheric transport is modeled using the global circulation model LMDz. When only optimizing the surface fluxes (with a prescribed sink), bias errors in the sink magnitude translate into substantial bias errors in the retrieved global total surface fluxes. Conversely, we find that errors only in the temporal and horizontal distribution of the N2O sink (nonbiased magnitude) have a very small impact on tropospheric mixing ratios and thus on the retrieved surface fluxes. Bias errors in the modeled vertical transport, however, lead to notable changes in tropospheric N2O and, in particular, in the phase of the seasonal cycle. This also leads to bias errors in the spatial distribution of the derived surface fluxes, although not in the global total. Last, the simultaneous optimization of the surface fluxes and the sink magnitude was tested as a means to avoid biasing the fluxes by incorrect prior assumptions about the N2O lifetime. With this approach, a significant reduction in the error of the sink magnitude was achieved and biases in the surface fluxes were largely avoided, and this result was further enhanced when aircraft data were included in the inversion.