Fig. 7

Schematic of the mechanism for volcanic El Niño development following a Pinatubo-like eruption in June. a September–October–November of year 0 mean tropical precipitation (mm day⁻¹, shading) and surface wind (m s⁻¹, vectors) anomalies; b September–October–November of year 0 relative sea surface temperature (°C, shading) and 0–100 m mean zonal ocean current (cm s⁻¹, red contours); c June–July–August of year 1 mean tropical precipitation (mm day⁻¹, shading) and surface wind (m s⁻¹, vectors) anomalies; and d June–July–August of year 1 relative sea surface temperature (°C, shading) and 0–100 m mean zonal ocean current (cm s⁻¹, red contours) anomalies after the Pinatubo eruption in the IPSL-CM5B coupled model ensemble for ocean initial conditions leading to neutral ENSO state at the end of the eruption year in the unforced control run based on the warm water volume content. Surface winds are shown when at least two thirds of the individual members display consistent sign anomalies. The white contours on precipitation and relative sea surface temperature anomalies indicate anomalies that are significantly different from zero at the 90% significance level according to a two-tailed Welch’s t-test. Reduced rainfall from August to October of the eruption year induces a tropospheric heating anomaly over equatorial Africa, which forces a Matsuno–Gill characteristic response, with easterly wind anomaly to the west and westerly wind anomaly to east of the heat source, as a result of atmospheric Rossby and Kelvin wave propagation. The downwelling Kelvin wave reduces deep convection along its path, in particular over the western Pacific, which further strengthens the surface westerly wind signal there. The easterly and westerly wind anomalies are represented by the thick black arrows in (a, b, d). The red block arrows in (b, d) represent the Pacific Ocean downwelling Kelvin waves response to the westerly wind. The year after the eruption, from August to October, the weaker but persistent tropical land cooling and the Bjerknes feedback both contribute to weaken the Walker circulation (c) in the Pacific Ocean, thereby allowing the full development of the El Niño-like event (d)