Investigating Northern Hemisphere Warming Responses to the Mt. Pinatubo Eruption through Limited Variability Ensembles

The Mt. Pinatubo eruption on 15 June 1991 is often associated with surface warming in the northern hemisphere during the subsequent winter. The prevailing theory is that increased low latitude stratospheric temperatures caused by the volcanic aerosol injection amplified the meridional temperature gradient and strengthened the northern hemisphere stratospheric polar vortex (SPV). Through the existing teleconnection between the high latitude stratosphere and troposphere the strengthened SPV induces a positive North Atlantic Oscillation (NAO) phase which leads to surface warming over the continental landmasses.

To isolate this localized impact of the eruption we created an ensemble of simulations using the Energy Exascale Earth System Model version 2 (E3SMv2) with prognostic aerosol modifications initialized on 1 June 1991 to limit the intra-ensemble variability at the time of the eruption. The initialization of these simulations was selected to have major climate modes, specifically the quasi-biennial oscillation (QBO) and the El Nino southern oscillation (ENSO), in similar states to those preceding the real-world eruption. We also generated a more traditional ensemble representing the full range of intra-ensemble variability to compare to our limited variability ensemble. For each ensemble member in both the limited variability and full variability experiment we generated a paired counterfactual simulation with the Pinatubo forcing removed to allow for isolation of the Pinatubo impact. In general, the limited variability ensemble has greater coherence in the Pinatubo impact across ensemble members which leads to more statistically robust signal compared to the full variability ensemble.

Investigating the northern hemisphere winter warming post-Pinatubo, we find significant warm anomalies in our limited variability ensemble consistent with observation. However, similarly significant warm anomalies are also present in our limited variability counterfactual simulation, and the difference between the limited variability ensemble members with Pinatubo versus the counterfactuals do not show any significant mid-winter surface warming. Additionally, the full variability ensemble members do not show significant mid-winter surface warming anomalies. All of this suggests that the northern hemisphere mid-winter surface warming present in observations is more likely a result of the state of the major modes in the atmosphere at the time of the eruption, specifically ENSO and the QBO, than a response to the volcanic aerosol injection.

Although we did not find evidence of the traditionally proposed northern hemisphere warming, we did find a statistically robust high latitude surface warming response in fall months immediately following Pinatubo. There was a significant mid-latitude westerly signature in the stratospheric zonal wind that starts less than a month after the eruption and continues through October 1991. During October 1991, as the SPV begins to spin-up, these increased mid-latitude westerlies are associated with a significant weakening of the high latitude westerlies. This weakening of the SPV core is associated with a positive geopotential height impact which penetrates to the surface driving a high-latitude warm temperature impact. This alternative warming pathway is especially coherent in the limited variability ensemble where it is present in more than 90% of individual ensemble members.