Re-evaluating the 1940s CO2 plateau

The high-resolution CO2 record from Law Dome ice core reveals that atmospheric CO2 concentration stalled during the 1940s (so-called CO2 plateau). Since the fossil-fuel emissions did not decrease during the period, this stalling implies the persistence of a strong sink, perhaps sustained for as long as a decade or more. Double-deconvolution analyses have attributed this sink to the ocean, conceivably as a response to the very strong El Nino event in 1940-1942. However, this explanation is questionable, as recent ocean CO2 data indicate that the range of variability in the ocean sink has been rather modest in recent decades, and El Nino events have generally led to higher growth rates of atmospheric CO2 due to the offsetting terrestrial response. Here, we use the most up-to-date information on the different terms of the carbon budget: fossil-fuel emissions, four estimates of land-use change (LUC) emissions, ocean uptake from two different reconstructions, and the terrestrial sink modelled by the TRENDY project to identify the most likely causes of the 1940s plateau. We find that they greatly overestimate atmospheric CO2 growth rate during the plateau period, as well as in the 1960s, in spite of giving a plausible explanation for most of the 20th century carbon budget, especially from 1970 onwards. The mismatch between reconstructions and observations during the CO2 plateau epoch of 1940-1950 ranges between 0.9 and 2.0 Pg Cyr(-1), depending on the LUC dataset considered. This mismatch may be explained by (i) decadal variability in the ocean carbon sink not accounted for in the reconstructions we used, (ii) a further terrestrial sink currently missing in the estimates by land-surface models, or (iii) LUC processes not included in the current datasets. Ocean carbon models from CMIP5 indicate that natural variability in the ocean carbon sink could explain an additional 0.5 PgC yr(-1) uptake, but it is unlikely to be higher. The impact of the 1940-1942 El Nico on the observed stabilization of atmospheric CO2 cannot be confirmed nor discarded, as TRENDY models do not reproduce the expected concurrent strong decrease in terrestrial uptake. Nevertheless, this would further increase the mismatch between observed and modelled CO2 growth rate during the CO2 plateau epoch. Tests performed using the OSCAR (v2.2) model indicate that changes in land use not correctly accounted for during the period (coinciding with drastic socioeconomic changes during the Second World War) could contribute to the additional sink required. Thus, the previously proposed ocean hypothesis for the 1940s plateau cannot be confirmed by independent data. Further efforts are required to reduce uncertainty in the different terms of the carbon budget during the first half of the 20th century and to better understand the long-term variability of the ocean and terrestrial CO2 sinks.