Biogeochemical “sinks” in the oceans and terrestrial biosphere currently absorb nearly half of the co2 emitted by fossil fuel combustion, but the sink processes are not yet well understood. Changes in biogeochemical processes (e.g., sink saturation or reversal) are currently one of the leading sources of uncertainty in projections of changing climate in the 21st century. Diagnostic modeling and data assimilation is an important way to test quantitative hypotheses about global biogeochemical responses to climate variability and change, and has emerged as a major research focus in carbon cycle science. This effort is analogous to the interaction between meteorological analysis and climate modeling, in which operational experience with forecasting leads to improved physical representations for longer-term simulations.

Models of ocean carbon cycle processes include air-sea gas exchange, ecosystem dynamics, and transport of nutrients and other biogeochemical constituents in the oceans. Terrestrial carbon cycle models include photosynthesis and respiration; seasonal phenology; allocation, nutrient cycling, and decomposition; disturbance and succession; land use and land cover change. Atmospheric carbon cycle processes included in models are advection, convection, turbulence, and reactions of carbon gases. A rich suite of observations from both in-situ and spaceborne platforms is available for comparison to the predictions of such models. Estimation theory and methods have been developed over decades in meteorological and other geophysical data assimilation research and are now being applied to state and parameter estimation in carbon cycle science.

As examples of recent results, we present estimates of air-sea exchange of CO2 derived by assimilation of satellite sea-surface temperature, topography, and winds into a coupled physical-biogeochemical ocean model. The system describes spatial, seasonal, and interannual variations in CO2 fluxes over the multidecadal record, highlighting effects of El Nino and other climate variations. We also present gridded estimates of terrestrial sources and sinks derived from remotely-sensed vegetation imagery and in-situ measurements of atmospheric CO2 using both global and mesoscale transport models.