ProjV

Goal: Understand whether or not winter warming pattern after three largest volcanic eruptions during the 20th century was caused by the eruptions or due to natural climate variability.
Scientific question: Do volcanic eruptions produce El Ninos, and is that key to the winter atmospheric circulation response?
Additional work:
1. Calculate the statistical significance of the patterns found in Fig. 1.
2. Further examine LE simulations to see if NCAR model produces an El Niño.
3. See if model produced winter warming in the second winter after volcanic eruption.
4. Are results robust in other climate models? If the other models have a QBO or simulate El Niño better, does that make a difference?
Documents/main/projV/PS/b.e11.B1850C5CN.f09_g16.005.cam.h0.PSL.0*****--0*****.nc is thousands of years of the CESM-LE.
TOGA
= tropical ocean global atmosphere

Agung: March 17 1963
El Chichon: April 4 1982
Pinatubo: June 15 1991

Analysis
NH DJF/JFM Circulation Response

Following the average of the three large volcanic eruptions during the 20th century, there is no significant shift in the probability distribution function (PDF) of the 1st year DJF or JFM Arctic Oscillation (AO, first EOF between 20N and 90N for all latitudes) in 41 ensembles of the CESM-LE when compared to the period of 1920-2005 (Fig. 1). To quantify the response in the North Atlantic sector, we compute the North Atlantic Oscillation (NAO, first EOF for 20N-90N, 90W-40E) and compare the PDFs of the NAO in the same way as was done in Fig. 1. The average response in the first DJF and JFM after these three eruptions is a more defined shift towards positive NAO-like conditions (Fig. 2). There is a 5% increase in the probability of +NAO conditions during DJF and a 2% increase during JFM.
Fig. 1: Probability distribution function (PDF) of Arctic Oscillation in 40 CESM-LE ensembles for the 1920-2005 period. AO defined as the first EOF of mean sea level pressure (20N-90N).


PDF of AO during DJF 1963-1964 (Agung)
PDF of AO during DJF 1982-1983 (El Chichon)
PDF of AO during DJF 1991-1992 (Pinatubo)


PDF of AO during JFM 1964 (Agung)
PDF of AO during JFM 1983 (El Chichon)
PDF of AO during JFM 1992 (Pinatubo)

Fig. 2: PDF of the North Atlantic Oscillation in 40 CESM-LE ensembles for the 1920-2005 period. NAO defined as first EOF of mean sea level pressure (20-90N, 90W-40E).


PDF of NAO during DJF 1963-1964 (Agung)
PDF of NAO during DJF 1982-1983 (El Chichon)
PDF of NAO during DJF 1991-1992 (Pinatubo)


PDF of NAO during JFM 1964 (Agung)
PDF of NAO during JFM 1983 (El Chichon)
PDF of NAO during JFM 1992 (Pinatubo)

In observations (one ensemble of reality), there is a mixed shift in the AO across the different eruptions. The average AO response across all eruptions is positive, but the average response in the NAO across all eruptions is neutral. The response to volcanic eruptions in reality is confounded by a lack of observable eruptions (n=3) compared to the number of eruptions in the CESM-LE (n=360).
1963-1964 DJF: AO: -0.48 NAO: -1.43
1982-1983 DJF: AO: 0.17 NAO: 0.95
1991-1992 DJF: AO: 1.095 NAO: 0.47
AVERAGE: AO: 0.262 NAO: 0.00

We can also visualize the circulation response in terms of the mean sea level pressure (SLP) anomaly in all of the ensembles of the CESM-LE (Fig. 3). From the ensemble average SLP anomaly in the first DJF after the Agung eruption, it is clear that the model is also simulating, on average -NAO/AO conditions, just like in observations. However, these signals are far weaker than in the observations. However, after El Chichon and Pinatubo the model is simulating lower pressure at the pole and somewhat higher pressure in the midlatitudes, which is consistent with a more +NAO response. Therefore, the ensemble average of the CESM-LE is in line with observations in terms of SLP, or NH DJF circulation. Interestingly, however, is that the circulation is different after Agung compared to both El Chichon and Pinatubo. We will seek to understand why this is the case, but speculate it will have to do with the pole-to-equator temperature gradient in the stratosphere. However, it is worth noting that AMIP (with tropical SSTs) replicates a winter warming temperature pattern similar to El Chichon and the Pinatubo eruptions, despite that pattern not appearing in observations after Agung (not shown).


Fig. 3: SLP anomaly in CESM-LE for the first DJF after the (a) Agung eruption, (b) El Chichon eruption, and (c) Pinatubo eruption.



ENSO DJF/JFM Circulation Response
We examine the impact of these three volcanic eruptions on the El Niño-Southern Oscillation (ENSO) in the CESM-LE. We calculate the PDF of the Southern Oscillation Index (SOI, normalized difference in sea level pressure between Tahiti and Darwin) to quantify how the Pacific Walker Circulation responds to 20th century volcanic eruptions. Fig. 4 shows a moderate shift in the PDF of the SOI during both DJF and JFM towards El Niño-like conditions with a weaker than normal Walker Circulation. Finally, we examine the PDF of DJF and JFM Niño3.4 region sea surface temperature (SST) anomalies, shown in Fig. 5. These results are somewhat mixed. However, there is an increase in the probability of a 1-2.5 degrees Celsius positive SST anomaly in the Niño3.4 region for both DJF and JFM after a volcanic eruption. The probability of a weak La Niña episode decreases after these volcanic eruptions, but there is actually a slight increase in the probability of a large La Niña episode based on the SSTs. This may be the effect of aerosol-induced cooling, but then one would not expect the decrease in the probability of weak La Niña events. It has been found in a number of studies that there is a shift towards El Niño-like conditions after volcanic eruptions, so if one were to combine the signal of warming from an El Niño event and cooling from the aerosols, one would expect an increase in neutral conditions. Instead, there is an increase in the extremes after a volcanic eruption. This has yet to be explained.

Fig. 4: PDF of Southern Oscillation Index in 40 CESM-LE ensembles for the 1920-2005 period.


PDF of SOI during DJF 1963-1964 (Agung)
PDF of SOI during DJF 1982-1983 (El Chichon)
PDF of SOI during DJF 1991-1992 (Pinatubo)


PDF of SOI during JFM 1964 (Agung)
PDF of SOI during JFM 1983 (El Chichon)
PDF of SOI during JFM 1992 (Pinatubo)

Fig. 5: PDF of Niño3.4 temperatures in 40 CESM-LE ensembles for the 1920-2005 period.


PDF of N34 during DJF 1963-1964 (Agung)
PDF of N34 during DJF 1982-1983 (El Chichon)
PDF of N34 during DJF 1991-1992 (Pinatubo)


PDF of N34 during JFM 1964 (Agung)
PDF of N34 during JFM 1983 (El Chichon)
PDF of N34 during DJF 1992 (Pinatubo)


Conclusion: 20th century volcanic eruptions are more likely to shift the CESM-LE into an El Niño-like state than to shift the NAO/AO.


Eurasian Surface Temperature DJF /JFM Response
Consistent with a +AO/+NAO pattern, Eurasian DJF warming was observed after the El Chichon and Pinatubo eruptions (Fig. 6). We describe the observed temperature anomalies in the DJF following each eruptions:
Agung: Cold anomaly over the entire pole, warm anomaly over Alaska and northern Canada, Scandinavia, and eastern Russia (Siberia). Cold anomaly over most of the United States, Europe, the Middle East, and Central Asia. Moderate warming in the Niño 3.4 region.
El Chichon: Warm anomaly over most of North America, centered over Canada, cold anomaly over Greenland and far northern Canada. Warming over Europe and Eurasia. Cooling over northern Africa and the middle East as well as southern China and parts of southeast Asia. Strong warming in the Niño 3.4 region and Niño4 region, south Africa, the Indian Ocean, and Australia.
Pinatubo: Warm anomaly over most of North America, centered over Canada, cold anomaly over Greenland and far northern Canada. Warming over Europe and Eurasia. Cooling over northern Africa and the Middle East. Strong warming in the Niño3.4 region and moderate warming in the Niño4 region. Warming over South Africa and parts of Antarctica.
El Chichon and Pinatubo have very similar anomalies compared to Agung. Agung is the only eruption which features a cold anomaly over the entire polar region and then weaker warming over the high latitude continents which is consistent with a +AO/+NAO. Agung is also the only eruption with very weak El Niño-like conditions, which actually makes sense. Given the propensity for El Niño-like conditions to affect the North Atlantic region by inducing a dipole-like anomaly very similar to the -NAO, it is possible that the weaker El Niño conditions allow for a more positive +NAO than otherwise possible compared to the other cases. It becomes clear that we need to look at the pressure patterns to understand how El Niño and the AO/NAO may be influencing circulation.
Looking at the CESM-LE it becomes clear that something is amiss between the model and the observations, as continental warming is not simulated over North America and most warming is actually simulated over the pole as opposed to the high latitude continents.


Fig. 6. Surface temperature anomaly using GISSTEMP observations and using CESM-LE (42 ensembles).




Model Representation of Teleconnections

I calculate the following correlation coefficients:

Fig. 7: (a) AO and surface temperature, (b) NAO and surface temperature, (c) AO and SLP, (d) NAO and SLP for DJF.

Fig. 8: (a) AO and surface temperature, (b) NAO and surface temperature, (c) AO and SLP, (d) NAO and SLP for JFM.

Fig. 9: (a) SOI and surface temperature, (b) Niño3.4 and surface temperature, (c) SOI and SLP (d) Niño3.4 and SLP for DJF.

Fig. 10: (a) SOI and surface temperature, (b) Niño3.4 and surface temperature, (c) SOI and SLP (d) Niño3.4 and SLP for JFM.

Based on this analysis, it is clear that Eurasian warming in DJF/JFM occurs as a result of a strongly positive AO/NAO pattern. However, the warming over North America is less clear. Is the model relationship between the Northern Hemisphere SLP during JFM and Niño3.4 SSTs / the SOI consistent with reality? Based on the NOAA-20CR, it appears to bear a resemblance to the expected spatial pattern in reality. However, the ‘center of action’ in reanalysis is closer towards the pole. CESM-LE is biased towards a weaker Icelandic low than a weaker tropospheric polar vortex. Additionally, in reanalysis the connection between El Niño-like conditions and lower pressure over the North Atlantic extends closer to Europe compared to the CESM-LE. This implies a more robust connection between ENSO and European surface weather in reanalysis compared to the model. This does not explain the weaker Eurasian warming in the CESM-LE, as one would expect the opposite if there was a weaker ENSO-NAO connection in the model. This is because the ENSO-NAO connection is that El Niño-like conditions results in a -NAO like pattern. This leaves a huge question – what is causing the winter warming to become stronger in years with much stronger El Niño events?

One potential idea is that while stronger volcanic forcing causes both stronger +AO forcing and stronger +ENSO forcing, the +AO forcing is more able to deflect an increase in vertically and poleward propagating planetary waves as a result of the +ENSO conditions, allowing for the +AO tropospheric conditions to be maintained. Thus, we would expect for the effect of the +AO to overcome any effect of ENSO on the AO/NAO system. This is a hypothesis that I must figure out a way to test, but any analysis of vertically and poleward propagating waves is confounded by the effect of a strong polar vortex’s ability to deflect waves away. We should expect that the differential temperature gradient in the stratosphere after Agung will be significantly less than after El Chichon and Pinatubo. This would explain the more robust +AO/NAO patterns in El Chichon and Pinatubo despite the El Niño-like conditions. I will also investigate the potential increase in vertically and poleward propagating waves as a result of a volcanic El Niño compared to a regular El Niño. We will see if this metric is a function simply of Niño3.4 SSTs or other factors which could be influenced by volcanic aerosols (Such as increased stability in the upper troposphere from heating of aerosols). Stenchikov et al. (2002) may be instructive and should be reviewed, as it conducts some sensitivity tests on this very issue and also finds that ozone destruction from volcanic aerosols may play an important factor in the dynamics of the wintertime NH circulation.

In conclusion, my hypothesis is that if you are forcing a +AO due to an increase in stratospheric heating in the mid-latitudes, this will be enough to deflect the vertically and poleward propagating waves that increase in frequency as a result of the volcanically-induced El Niño. Now, I will seek to test this through a series of analyses. First, I must show that there is an increase in stratospheric heating that is greater in El Chichon/Pinatubo compared to Agung.


Fig. 11: Temperature anomaly profile and zonal wind anomalies in CESM-LE after (a) Agung, (b) El Chichon, (c) Pinatubo.


(a) JFM 1964

(b) JFM 1983

(c) JFM 1992


Fig. 12: Temperature anomaly profile and zonal wind anomaly regression in CESM-LE as a function of (a) AO, (b), NAO, (c) SOI, (d) Niño3.4 SSTs during JFM.

(a) Regression of AO-T and AO-U
(b) Regression of NAO-T and NAO-U
(c) Regression of SOI-T and SOI-U
(d) Regression of N34-T and N34-U