Volcanic Winters

Guest Post by Albert:  Weather is changeable but climate is forever. At least so it seemed, before the global warming began in earnest in the 1980s. The difference between weather and climate is often lost on us: we experience and remember the weather but changes over decades are hidden from us. Nature is different: flowers appear weeks earlier than they used to, in response to a warming we haven’t really noticed. But there are short term fluctuations which we do feel. These happen over six months to a few years, and change the weather notably, but are too brief to affect the climate. In the UK, the exceptionally mild winters of the last two years come to mind, or the three rather colder winters preceding them. In other places it can be years of drought. Often they are local events only: for instance, the last winter was very mild in Europe, very cold in parts of the US, balancing each other out. But sometimes they do trace a real, brief change in global temperatures. There are two main events which affect global temperatures on these time scales. One is the El Nino: it occurs irregularly and releases enough heat from the oceans to make a notable difference, lasting 6-12 months. The other is volcanic eruptions, with Pinatubo, Krakatoa and Tambora well-known examples. Pinatubo depressed global temperatures for up to two years. Tambora gave a larger change lasting a few years. But there are times when temperatures go up or down for a few years without a clear volcanic cause. And there is disagreement on how strong volcanoes need to be notable. I tried to find these effects in the global temperature records. The figure below shows the so-called Annual mean Land-Ocean Temperature Index which I consider the best one to use. (There are other temperature indices: some only measure temperatures on land which is closer to our experience but more prone to weather.) I have split it up over the northern hemisphere (above 24 degrees latitude), the southern hemisphere, and the tropics. Comparing temperature graphs immediately shows that the northern hemisphere, and in particular non-tropical latitudes, are most affected by variations. The southern hemisphere is much more stable. This shows the stabilizing effect of oceans: the southern hemisphere is mostly water, and water has a much higher heat capacity than either land or air. Land can respond quickly to changes, but oceans require a long time to heat up or cool down. The arrows at the bottom indicate larger El Nino’s. The very strong El Nino of 1998 is clearly visible. But how well are volcanic eruptions visible? The figures shows the main eruptions over this time. The largest ones are Krakatoa (1883), Novarupta (1912) (the name ‘Katmai’ is also used), El Chicon (1982), and Pinatubo (1991). I have also added the VE5 eruptions, as shorter lines to distinguish them. Furthermore, colour is used to show whether they erupted in the northern hemisphere (red), tropics (black), or southern hemisphere (blue). Pinatubo is clearly visible in the temperature record, in all three plots. El Chicon is not, but it coincided with a very strong El Nino which may have canceled its effects out. Novarupta had no effect. Krakatoa mainly affected the northern hemisphere. Some of the other volcanic eruptions coincided with a temperature excursion, but it is hard to be sure this was not a coincidence. The ones under suspicion of climate alteration are Tarawera (1886; southern hemisphere only), Bezymianni (1956) and Mount Agung (1963) (but the bad winter of 1962/3 came before this eruption and cannot be blamed on a volcano). The Cerro Azul (1932) eruption came before a cold year, but went off in the wrong hemisphere, so is above suspicion because of a good alibi. Hemisphere is important. It is very difficult for northern eruptions to affect the southern hemisphere and vice versa: the tropics act as a very effective barrier to the spread of stratospheric sulphur. Volcanoes in the tropics have an advantage: they can spread their pollution to both hemisphere. The rule of thumb is one of contrast: for a proper volcanic winter, you need a tropical volcano. Can we go back further in time? The temperature record becomes more difficult: there are too few direct measurements, and you can’t use say the long records from central England as they are far too localized: you would get weather rather than climate. However, there are other methods, such as tree rings, and Greenland ice core records, which can be used. They don’t quite measure temperature the same way and you wouldn’t expect a perfect agreement. But it has been shown to work well enough. The temperature record over the past 2000 years is shown below. The red line overplots the recent accurate measurements for the northern hemisphere. Note that all temperatures are shown against the 1951-1980 global average. The Little Ice Age, from about 1300 to 1800, is easily visible. There are also numerous brief excursions. Are any volcanic? To find out, one needs a list of major volcanic eruptions over the past 2000 years. Amazingly, this does not exist. Most major volcanoes erupted in places without a literate population, or, worse, wiped them out, and people just didn’t think of recording the ‘VE’ number for each eruption. Very inconsiderate. But ice cores come to the rescue. Volcanic ash and sulphur leave their traces in the ice, and can be identified both from Greenland and Antarctica. The years can be a little uncertain (it is not easy to accurately count that many layers), the eruption magnitude cannot be accurately inferred, and the location of the volcano is anyone’s guess. (Although it is possible to infer whether it was a tropical, northern, or southern volcano, depending whether it is seen in Greenland, Antarctica, or both.) A list of eruptions and associated volcanoes is in below. Before 1600 there is considerable uncertainty in date, identification, or both. The Kuwae eruption was possibly the largest of the millennium, exceeding Tambora. The 1808 eruption is unusual, in that a fairly accurate date is known (early Dec 1808) but there is no proposed location. Quite a few of these eruptions coincide with a temperature drop. Before 1452 the dates become more uncertain and you wouldn’t expect to see much of a correlation. The excellent fit in 535AD is a bit misleading, since the temperature decline was used to date the eruption! However, there are enough cases of real agreement to make the case for volcanic winters convincing. But not really winters. The sulphuric haze reduces the strength of the sun light, and this should have more effect on the summer temperatures (when the sun’s effect is strongest) than on the winter. A volcanic winter is typified by snow in summer. Extreme winters can rarely be traced to a volcano. The cause of the winter of 1740, when the Shannon froze over and a quarter of Ireland’s population died, lies out in the Atlantic, and not in sonme distant caldera. Not all weather excursions can be blamed on volcanoes.

Year	Volcano	Location	Indicative VE
1991	Pinatubo	Philippines	6
1956	Bezymianny	Kamchatka	5
1912	Novarupta	Alaska	6
1902	Santa Maria	Guatemala	6
1886	Tarawera	New Zealand	5?
1883	Krakatoa	Indonesia	6
1861	Dubbi	Eritrea	5?
1854	Sheveluch	Kamchatka	5
1831	Babuyan?	Philippines	4
1815	Tambora	Indonesia	7
1808	unknown	tropics	7?
1737	Tarumai	Japan	5
1667	Tarumai	Japan	5
1641	Awu	Indonesia	5
1600	Huaynaputina	Peru	5?
1570	Billy Mitchell	Papua New Guinea	6
1482	Mount St Helens	USA	5
1452	Kuwae	Vanuatu	7
1351	Tarawera	New Zealand
1257	Samalas	Indonesia	7
1026	Baitoushan	China	7
902	Ksudach?	Kamchatka
702	Bona-Churchill?	Alaska
535	Llopango	El Salvador
232	Taupo	New Zealand

• 1831: The identification with Babuyan seems doubtful. The dust/sulphur is seen only in Greenland, not Antarctica. Also the dry fog reported in Europe that year suggests a northern location, perhaps North America. There may have been two eruptions, one around 1830 and one around 1835.
• Long Island, Papua New Guinea, may be another candidate for the 1640 eruption. It had a large event sometime before 1700.
• The identification of the 1570 dust as from Billy Mitchell is tentative. The eruption itself is not accurately enough dated.
• The Baitoushan eruption has an uncertain date by +- 50 year.

Advertisement