Combining Data and Models for Robust Estimates of Global Air-Sea CO2 Fluxes

Observations of partial pressure of carbon dioxide (pCO2) are needed to calculate air-sea fluxes of CO2, but these are limited due to sparse sampling by a limited number of field campaigns. These observations are especially limited in the Southern Hemisphere. Gap filling is a way to obtain a global gridded dataset. The uncertainties in pCO2 produced by these methods are not known. Therefore, the project will use Earth system model (ESM) large ensembles as an observation system simulation experiment (OSSE). The OSSE involves extracting the actual ship track samples from the model simulated pCO2 along with other quantities needed to train the neural network or drive the mixed layer model to produce a reconstructed pCO2 that will be compared with the model simulated 'truth.' The size of the large ensembles will produce the uncertainties in pCO2 generated by the two different methods. In addition, the group will also develop a new gap filling method with reduced uncertainty.