Explosive volcanic eruptions that are significant in size reduce global temperature by creating a curtain of sulfur aerosols in the stratosphere that lasts for a few years. Cataclysmic eruptions, such as Mayon of the Philippines (1766), Tambora in Indonesia (1815) and Nicaragua’s Coseguina (1835), are known to have significantly lowered temperatures in the year of and the years that followed. They created years without a summer and cooled the Northern Hemisphere so that frosts occurred in all summer months in mid-latitudes. Histories tell of land surface temperature effects, but make no mention of the impacts on sea surface temperature (SST), which would be drastically altered as well.
I believe these resulting significant drops in SST are then replayed on a periodic basis (20 to 30+ years), in decreasing magnitude, by the phenomenon known as Thermohaline Circulation (THC) or Meridional Overturning Circulation (MOC). And since there are differing time scales (per ocean, per hemisphere, and per circulation area within each) for the circulations to run through their “cycles”, these replayed signals appear at different times in the SST records: sometimes offsetting and dampening one another, sometimes complimenting and amplifying one another, other times appearing simply as noise.
In the late 19th to early 20th centuries, there is a substantial dip in SST that appears in many of the SST anomaly signals. Refer to the sample in Figure 1, in which I’ve circled the drops in temperature, (as if you could miss them). To determine if those dips are a residual effect of volcanic eruptions earlier that century, I created a data set that combined the impacts of Total Solar Irradiance (TSI) and explosive volcanic eruptions as presented by the Lamb Dust Veil Index (DVI). Refer to Figure 2. This data set was discussed in the second part of this blog series.
Here are links to parts one and two of this series of posts:
While I haven’t proven that the late 19th to early 20th century drop in temperature results from those volcanic eruptions, I do present enough evidence in the following to suggest that it needs to be investigated further by those with better tools.
DECADAL SEA SURFACE TEMPERATURE VARIABILITY IN THE SUB-TROPICAL SOUTH PACIFIC FROM 1726 TO 1997
I found the data set titled Decadal Sea Surface Temperature Variability in the Sub-Tropical South Pacific from 1726 to 1997 A.D. at the NOAA World Data Center website. It more than covers the period in of the three large volcanic eruptions mentioned above.
ORIGINAL REFERENCE: Linsley, B.K., G.M. Wellington, and D.P. Schrag, 2000, Decadal Sea Surface Temperature Variability in the Sub-Tropical South Pacific from 1726 to 1997 A.D., Science v.290, pp1145-1148, 10 Nov 2000.
The first compilation within the data set I plotted was the curve of the Subtropical South Pacific that had been smoothed with an 8-year low-pass filter. Refer to Figure 3. It contains the requisite drop in temperature from around 1890 to 1920. The most prominent element of this graph is the peak temperature and the year it took place. The 1747 peak is more than 1.25 deg C higher than the 20th century peak in 1974. This is completely out of phase with the Little Ice Age, which is thought to have taken place from the 16th to the mid-19th centuries, though I have seen Little Ice Age reconstructions that do have a similar rise in temperature in the late 18th century.
Adding the Solar-Volcanic Aerosol Impact curve, it appears there could be a relationship. Refer to Figure 4. The three major volcanic eruptions agree with the timing of the downward trends of the SST oscillations. Two other points I found remarkable: The apparent agreement in timing of the two data sets, with the changes in SST occurring near to or at the same time as the major and minor eruptions, and the eruptions counteracting or amplifying cycles already in progress. Note: When considering the 1835 eruption, remember that Coseguina is in Nicaragua.
The next data set within the reconstruction that I plotted was SST, Figure 5. I found the monthly detail remarkable, but it was useless for the comparison. But, smoothed with an annual filter, Figure 6, it loses much of the noise.
In Figure 7, I added the Solar-Volcanic Aerosol Impact curve to compare it with the Subtropical South Pacific reconstruction data. The correlation between the two curves is extraordinary. What caught my attention: After the initial response to the 1766 eruption of Mayon, there are what appear to be two secondary dips in temperature, lagging the initial reaction by 26 and 32 years. These are well within the bounds of the time scales discussed in “Decadal Variability of Pycnocline Flows from the Subtropical to the Equatorial Pacific”, Wang and Huang (2005), Journal of Physical Oceanography, Volume 35.
There could be other repeated signals, but those two at 26 and 32 years appeared most prominent.
The repeated signal at 26 years may result from MOC in one of the major oceans, while the second at 32 years occurs in the other. Or they could reflect different hemispheres of the Pacific. I have no way to answer that part of the riddle. Further, it’s conjecture on my part that those two lagged drops in temperature are actually related to the volcano in 1766. Regardless, to determine if the combined initial and lagged effects of the volcanoes could create the oceanic responses in the smoothed reconstruction of Subtropical South Pacific SST, I attempted to duplicate the effect using EXCEL. Figure 8 illustrates a colorful spaghetti graph of raw and lagged Cumulative DVI, from 1727 to 2000. The initial lags are 26 and 32 years. The additional four lagged data sets are twice (56 and 64 years) and three times (78 and 96 years) the original lags. For the initial lags of 26 and 32 years, the multiplier used to simulate signal degradation is 0.5. For the second lagged signals of 56 and 64 years, the multiplier is 0.2. And the third lagged signals (78 and 96 years) are 0.1 times the raw DVI values.
In Figure 9, the raw and lagged Cumulative DVI values from Figure 8 were averaged annually. The resulting data was also smoothed with a 7-year running-average filter. The smoothed curve does not exactly match the filtered Subtropical South Pacific from Figure 3, but by adjusting multipliers, it’s possible I could get a better correlation. Due to limits of my time available for this blog, I’m not going to try.
(Realistically, I accept my role as climatology hack, not climatologist, and as such, if someone with credentials wants to take the idea farther, they would accept the preceding as a concept that needs additional investigation. Therefore, any effort on my part to better illustrate a correlation is actually a waste of time.)
However, this post does illustrate that the discussed effect is possible. To me, the most important segment of the smoothed curve in Figure 9 is the dip just after 1900. It increases gradually until its peak in 1942, mimicking the global temperature anomaly curve and many localized SST anomaly curves.
If someone were to mix in the effects of ENSO and solar and were to use the correct time lags and signal degradation multipliers for the repeated volcanic signals, who knows?