Interannual variability in the Australian carbon cycle over 2015–2019, based on assimilation of OCO-2 satellite data

Abstract. In this study, we employ a regional inverse modelling approach to estimate monthly carbon fluxes over the Australian continent for 2015–2019 using the assimilation of the total column-averaged mole fractions of carbon dioxide from the Orbiting Carbon Observatory-2 (OCO-2, version 9). Subsequently, we study the carbon cycle variations and relate their fluctuations to anomalies in vegetation productivity and climate drivers. Our five-year regional carbon flux inversion suggests that Australia was a carbon sink averaging −0.46 ± 0.08 PgC yr−1 (excluding fossil fuel emissions), largely influenced by a strong carbon uptake (−1.04 PgC yr−1) recorded in 2016. Australia semi-arid ecosystems, such as sparsely vegetated regions (in central Australia) and savanna (in northern Australia), were the main contributors to the carbon uptake in 2016. These regions showed relatively high vegetation productivity, high rainfall and low temperature in 2016. In contrast to the large carbon sink found in 2016, the large carbon outgassing recorded in 2019 coincides with an unprecedented deficit of rainfall and higher than average temperature across Australia. Comparison of the posterior column average CO2 concentration against the Total Carbon Column Observing Networks (TCCON) and in situ measurements offers limited insight into the fluxes assimilated with OCO-2. However, the lack of these monitoring stations across Australia, mainly over ecosystems such as the savanna and areas with sparse vegetation, impedes us from providing strong conclusions. Comparison of our flux inversion to the ensemble mean carbon flux of the OCO-2 Multi-model Intercomparison Project (MIP) (2015–2018) agrees with our findings, and their results also suggest that Australia was a strong carbon sink in 2016 (−0.73 ± 0.41 PgC yr−1). The analysis of the variability of the nine models that participate in the OCO-2 MIP also aligns with our findings, and it gives us the confidence to say that changes in rainfall and temperature drive most of the carbon flux variability across Australia.