Constraining carbon cycle climate interaction with a joint land-atmosphere carbon data assimilation approach

In recent years, observations of the global carbon cycle in the atmosphere and on land have increased dramatically, including FLUXNET network that measures surface fluxes of water, heat and CO2, space-based CO2 observations from SCHIAMACHI, AIRS, GOSAT and OCO2, and other remotely sensed derived ecosystems attributes such as vegetation indices and fluorescence. Yet such data are often used independently and in isolated fashion. We describe a novel joint land-atmosphere data assimilation system that will simultaneously handle multiple modeling components and multiple streams of data. This challenging task is achieved using the Local-Ensemble-Transform-Kalman Filter (LETKF) that has been shown to be a powerful tool in a variety of atmosphere, ocean and carbon data assimilation settings. It includes a number of advanced features such as quantification of transport error using ensemble meteorological analysis, variable localization, and up-scaling flux tower via a footprint observation operator. The system offers an opportunity to combine the complementary attributes of 'top-down' atmosphere approach that gives best large-scale constraint with the 'bottom-up' ground-based approach that provides greater insight for small-scale processes . Some initial results using a combination of atmospheric models SPEEDY, GEOSChem and terrestrial ecosystem model VEGAS will be presented.