Nature 546, 485–491 (2017); doi:10.1038/nature22974

Owing to a production error, the area means in Fig. 3a appeared incorrectly as −0.676 μm instead of −0.68 μm, and in Fig. 3c as –0.745 g m–2, instead of +0.75 g m−2. We also note a mistake in our estimate of the effective radiative forcing (ERF) for the experiment that considers a fissure eruption in June–July. Instead of −0.29 W m−2 (+40%), the revised ERF is −0.61 W m−2 (+194%). The incorrect estimate of the June–July ERF appears in the main text (page 6) and in Supplementary Discussion section S12 and Supplementary Fig. S12.1b. The ERF was calculated as the difference in the two-month average (June–July) top-of-the-atmosphere net radiation between the simulation including the volcanic emissions (experiment) and the simulation excluding the volcanic emission (control). The error in the calculation of the June–July ERF comes from loading the same model output for the month of July for both the experiment and the control simulations. In effect, our published estimate (−0.29 W m−2) accounts only for the perturbation simulated for June (but halved) because of the cancellation error for July. These errors have been corrected in the online version of the paper.