Mesoscale simulations and inversions of atmospheric CO2 using airborne and ground-based data
Estimating the spatiotemporal structure of carbon fluxes to the atmosphere to a high degree of certainty is indispensable in the context of ongoing climate change. The quantification of these fluxes is crucial to assess potential feedbacks between the climate and the carbon cycle, and to manage changes in the carbon cycle. This is also essential in the context of climate change mitigation, in terms of monitoring the success of current greenhouse gas emission reduction policies (e.g. Kyoto Protocol) as well as implementing adequate emission reduction and sequestration strategies in the near future. Monitoring the state of atmospheric CO2 concentrations provides valuable information on CO2 exchanges between the surface and the atmosphere. This information can be quantitatively deduced using Bayesian inverse transport modeling together with global networks of observations. This thesis presents high-resolution simulations of atmospheric CO2 over Europe as well as over a region with complex topography (Ochsenkopf in Germany). An inverse technique is applied at high spatial and temporal resolution, taking into account the strong variability of CO2, to derive a regional estimate of biosphere-atmosphere exchange of carbon, together with its associated uncertainty. Overall, the thesis shows that a mesoscale model-data fusion system consisting of a weather prediction model, a Lagrangian adjoint transport model, and a diagnostic biosphere flux model is capable of simultaneously utilizing information from tower-based mixing ratio observations, eddy covariance flux measurements, and remote sensing of the biosphere to estimate surface-atmosphere fluxes at an unprecedented spatial resolution of 2 km. The high-resolution allows for a much better representation of the near-field around atmospheric observing sites, significantly reducing model representation errors, specifically for complex terrain sites such as mountains, compared to previous inverse transport models.