Investigating a Potential Role for the AMO on Modulating the Link between ENSO and the NAO.

A recent study (Zhang et al., 2019) found that the connection between ENSO and the NAO during NH winter can be modulated by the phase of the Atlantic Multidecadal Oscillation (AMO). When ENSO and the AMO are in phase, the effect of ENSO on North Atlantic region circulation is amplified such that +ENSO/+AMO results in a more robust -NAO signal and a -ENSO/-AMO results in a more robust +NAO signal. A suggested mechanism that tropical Atlantic SSTAs may provoke convective anomalies significant enough to initiate a Rossby wave response, impacting surface pressure in the North Atlantic region.

I calculate the AMO as described by Zhang et al. (2019) and create a probability density function that illustrates the effect of +ENSO/+AMO conditions on the NAO during JFM, seen in Fig. 1. This shows that in both CESM-LE and AMIP-TOGA, there is a negative shift in the NAO when +ENSO conditions are in phase with the AMO are in phase.

Fig. 1: Probability density function of NAO amplitude during JFM for +ENSO/+AMO conditions and +ENSO/-AMO conditions.

Next, I investigate one of two possibilities that can contribute to a weakened +ENSO/-NAO linkage after volcanic eruptions.
(1) The AMO was in a cold phase prior to the eruptions.
(2) The eruption contributed to a cold phase of the AMO.

Possibility 1: The AMO was in a cold phase prior the eruptions.

We examine the phase of the AMO in CESM-LE and AMIP-TOGA across all three of the eruptions to explore these two possibilities. Fig. 2 shows the evolution of Tropical North Atlantic (TNA) SSTAs (5-25N, 55W-15W), a metric chosen by Zhang et al. (2019) to understand the evolution of SSTAs in the region of the tropical Atlantic most important for generating convective anomalies. There is no consistent cool AMO phase before the eruptions. A consistent signal in the CESM-LE (42 ensembles averaged) is for cooling in the TNA region, which is to be expected with volcanic aerosol cooling. According to a mechanism proposed by Zhang et al. (2019), warm convective anomalies can initiate a northeastward propagating Rossby wave train across the Atlantic, favoring a -NAO like anomaly in the North Atlantic. This is consistent with a higher likelihood of -NAO conditions when +ENSO/+AMO conditions are occurring simultaneously. The cooling caused by volcanic aerosol over the tropical Atlantic can favor a +NAO in the North Atlantic. However, explaining differences between individual eruptions is far more difficult based on this information. All three eruptions exhibit anomalous cooling in CESM-LE while the differences between the three eruptions in the AMIP-TOGA runs do not explain the differences in circulation. Although there is a minor warm bump in Agung during the first JFM after the eruption, it seems unlikely that such a small anomaly could be consequential. However, only after Agung are TNA SSTs not anomalously cool, unlike during El Chichon and Pinatubo. It’s possible that the lag between TNA SSTAs and a response in the North Atlantic could actually explain why Agung has a more -NAO response (warmer TNA SSTs) while El Chichon and Pinatubo have a more +NAO Response (cooler TNA SSTs). This idea of a lag is supported by Zhang et al. (2019).

The differences between CESM-LE and AMIP-TOGA indicate that cooling is too strong in model generated SSTs compared to the observed SSTs. When the model generates SSTs, it produces a colder Tropical North Atlantic, which has the effect of creating a more +NAO pattern. The differences between CESM-LE and AMIP-TOGA also suggest the AMO mechanism is not responsible for the +NAO pattern in AMIP-TOGA despite the +ENSO. Otherwise, the CESM-LE would have a stronger +NAO pattern compared to AMIP-TOGA. It is unlikely that the +ENSO in AMIP-TOGA is enabling the +NAO response because it goes against what the model typically does during +ENSO winters.

The differences between CESM-LE and AMIP-TOGA suggest that the AMO mechanism is not responsible for a the +NAO/+ENSO pattern that occurs in AMIP-TOGA. AMIP-TOGA does not have a more robust -AMO signal compared to CESM-LE, but AMIP-TOGA does have a more robust +NAO signal. In CESM-LE, the TNA region SSTs are consistently colder during JFM which should favor a +NAO pattern compared to AMIP-TOGA. However, the exact opposite result occurs. The +NAO signal is more robust in AMIP-TOGA compared to CESM-LE. This mechanism is likely a dead end for explaining why a +ENSO appears to be necessary for a +NAO pattern.

Fig. 2: Tropical North Atlantic SSTAs following the three major volcanic eruptions of the 20^th century.

A possible way to explain the differences in circulation between the eruptions in AMIP-TOGA (ignoring CESM-LE for a moment and assuming that the difference is because of the forced signal) is that the Central Pacific El Niño pattern after Agung may enhance the probability of a -NAO signal in opposition to the forcing from the stratospheric temperature gradient. Graf and Zanchettin (2012) suggest that a CP El Niño is more closely associated with a -NAO pattern compared to an EP El Niño. However, in AMIP-TOGA, despite a CP El Niño being present after Agung, there is very little difference in the +NAO response. If anything, Agung has more of a +NAO response in AMIP-TOGA because the strength of the El Niños after El Chichon and Pinatubo are far stronger than the weak CP El Niño after Agung. Including the observed tropical ocean temperatures leads to a +NAO response – but I still am not able to understand why this is the case.

I am still unable to explain the differences between CESM-LE and AMIP-TOGA. In summary:

CESM-LE has a weaker +ENSO response and a weaker +NAO response but a stronger +AMO.

CESM-LE should have a stronger +NAO if one just took into account the weak ENSO and strong +AMO response.

Next steps

Below are a few ideas for exploring this further and explaining the discrepancies between AMIP-TOGA and CESM-LE.

(1) Examine Eliassen-Palm flux to determine if there is a difference in wave flux between CESM-LE and AMIP-TOGA. There likely will be, because a stronger polar vortex refracts wave, but it’s worth calculating the wave flux out of the equatorial regions in both models. If it’s lower in AMIP-TOGA despite warmer tropical SSTs, then that is the fundamental reason why AMIP-TOGA has a +ENSO/+NAO. The next step is figuring out WHY wave activity isn’t getting to the poles in AMIP-TOGA.