A Prototype Global Carbon Cycle Data Assimilation System (CCDAS)

Wolfgang Knorr (MPI-BGC, Jena), Marko Scholze (MPI-Met, Hamburg), Peter Rayner (CSIRO, Melbourne), Thomas Kaminski (FastOpt, Hamburg) and Ralf Giering (FastOpt, Hamburg)

Atmospheric inversion studies have become an important tool for identifying terrestrial sources and sinks of CO2 at the interannual time scale. Such traditional top-down studies have so far delivered important insights into the large-scale patterns of atmosphere-biosphere and atmosphere-ocean CO2 exchanges. For determining more detailed patterns, they suffer from the inverse problem being seriously underconstrained. Such methods are usually contrasted with bottom-up approaches mostly using process-based terrestrial or oceanic models. Such models, however, cannot take the rich information into account that is contained in CO2 measurements from the extensive flask sampling network. Here, we present results of a two-stage assimilation study of satellite radiances (identifying vegetation activity) and atmospheric concentration data into a terrestrial biosphere model. The assimilation optimizes the tunable parameters in the model (some spatially explicit, some global), and thus we predict the model evolution using these optimized parameters. We fit CO2 data from 1979-99 and, via the optimized model, identify the processes responsible for the mean terrestrial fluxes and their variability. For example, we find a highly significant correlation between El Nino/Southern Oscillation and terrestrial CO2 fluxes, with CO2 lagging by a few months. Net CO2 outgasing during El Nino events is caused mainly by reduction of photosynthesis in large parts of the tropics, notably the Amazon basin and Central Africa. The most important deviation from the good correlation between global net terrestrial CO2 fluxes and ENSO is found after the eruption of Mount Pinatubo in 1991. The system also predicts average sources and sinks of CO2 in the terrestrial biosphere, where a similar latitudinal trend is found as for example in the Euroflux eddy covariance measurements. The uncertainties on the parameters, generated as part of the assimilation procedure, also allows us to identify those processes which are well constrained by the data. Finally, the system allows the computation of uncertainties in net CO2 fluxes, which is discussed for the example of some major regions. The system thus provides important quantitative information on our ability to constrain regional CO2 sources and sinks by giving explicit uncertainty bounds. It is also flexible enough to accommodate global versions with variable resolution that allow focussing on particular regions of interest, and can be extended to handle new streams of data, such as flux measurements, and to predict further prognostic quantities plus their uncertainties.

back to conference programme