Understanding the behavior of the natural carbon cycle requires knowing the fluxes of carbon into and out of land and ocean sinks, but the terrestrial biosphere’s behavior is hard to observe directly. To overcome this limitation, researchers have turned to atmospheric inversion modeling, which uses the distribution of CO2 in the atmosphere to back-calculate the location and size of sources and sinks that would produce that spatial pattern.

The Sarmiento and Pacala groups are tackling a variety of challenges in inversion modeling, improving model components and sampling strategies as well as applying models to important questions about the distribution of carbon sources and sinks. This work is laying the groundwork for a carbon observing system that will incorporate forest inventory data, data on atmospheric composition, satellite observations, and eddy flux information to monitor both short and long-timescale changes in the carbon cycle, as well as provide predictions for the future.


Model evaluation and data requirements

A critical question for the inversion exercise is whether northern hemisphere mid-latitude land carbon sources and sinks can be accurately estimated from atmospheric data, and if so, how high a data density would be needed to obtain estimates to a given precision. A previous study by the group indicated that signals above the mixed layer are so tiny that, with the currently achievable measurement precision and accuracy, they are likely too small to be helpful on their own. Manuel Gloor and colleagues’ work suggested that detecting even large changes in terrestrial sinks will require frequent high-density sampling in the mixed portion of the planetary boundary layer, a consideration that should influence future sampling campaigns.

Another important step in developing inverse models of sources and sinks is evaluating the performance of the underlying transport model. A recent comparison of predictions from Mozart, the group’s atmospheric transport model, with observations of SF6, a long-lived anthropogenic tracer, showed generally good agreement in the annual mean inter-hemispheric gradient (Figure 12). Some discrepancies in the seasonal distribution in SF6 are leading the team to retune parameterizations in the model.

Fig. 12. Latitudinal distribution of annual mean atmospheric SF6 concentration from observations and simulations with the Mozart model.

The team is now evaluating the use of satellite data to compensate for the lack of terrestrially-based CO2 observations and better constrain surface carbon fluxes on land. Comparison of satellite data with atmospheric transport model predictions shows good overall agreement between all estimates of seasonal cycles and North-South gradients of CO2 from 30S to 30N, but indicates further improvements in retrievals, and better understanding/validation of lower-to-upper troposphere transport and its modeling, are required to constrain surface fluxes with this method.

Finally, the team has designed a direct carbon budgeting approach using model simulations to infer carbon sources and sinks and improve the North American carbon budget. Direct budgeting puts a control volume on top of North America, balances air mass in- and outflows into the volume and solves for the surface fluxes. The use of CO2 vertical profiles simulated at the planned 19 stations of North American Carbon Plan network has given an estimation of the error of 0.39 GtC/y within the model world. The researchers suggest this error may be achieved through a better estimation of mass fluxes associated with convective processes affecting North America, and through adding a few stations in the North-West and the North-East of the continent.


Pinning down fluxes and variability

One major finding of the inversion modeling team has been that incorporating data from the ocean’s interior significantly improves estimates of regional ocean fluxes. When combined with atmospheric observational constraints, their results suggest that there is no need for a large CO2 fertilization sink in the tropics to balance the deforestation source. This has significant implications for future predictions by terrestrial biosphere models, which up to now have included a major CO2 fertilization sink.

The researchers have also made considerable progress in understanding the interannual variability of air-sea and air-land vegetation carbon fluxes, an important step toward making predictions of carbon pool behavior under a changing climate. One issue of recent debate has been the role played by the North Atlantic Ocean. Predictions of ocean biosphere models embedded in ocean circulation models suggested that North Atlantic variability was small, while time-series data from one ocean station and a previous inversion study of atmospheric data seem to indicate the opposite.

Using a novel inverse method, Manuel Gloor and colleagues showed last year that the primary source of global CO2 air-sea flux variability is in the Pacific Ocean, a finding that is consistent with biogeochemical modeling. Both methods also indicate that the Southern Ocean is the second-largest source of air-sea CO2 flux variability, while the variability throughout the Atlantic, including the North Atlantic, is diagnosed and predicted to be small.

This year, Keith Rodgers has used ocean models to study air-sea fluxes of CO2 over the North Pacific Ocean. The ocean modeling work has shown that there is a “window” of uptake of anthropogenic CO2 in the Kuroshio Extension region of the North Pacific, and that this window exhibits a very strong seasonality. His findings should reduce a large source of uncertainty in estimating variability in the CO2 budget over the North American continent.