A new study has now shown that colours in sunset paintings by famous artists can provide valuable information about atmospheric dust following volcanic eruptions.
Famous artists such as Turner and Gainsborough were well known for faithfully reproducing the colours they saw in the landscapes they painted.
The high-quality digitised images of the work of these artists that were available from the Tate Gallery in London, and other major galleries, allowed researchers to make careful analysis of the pictures, finding more than artistic merit – these paintings have a lot to say about our atmospheric history.
Sky Colors Over the Years, Recorded in Paintings
Over the centuries the colours changed markedly, with differing amounts of reds and greens, particularly in the sunset paintings. Now a team of scientists led by Prof. Zerefos of the University of Athens has examined the works of the old masters to see whether they might obtain actual quantitative information.
This follows on from an earlier study (Zerefos et al., 2007) which suggested the feasibility of the procedure. This is not the first time that art has been a source of information to atmospheric scientists – they also learned about a period known as the little ice age. During this time, the winters in Europe were exceptionally cold, and rivers such as the Thames in London froze over.
By comparing the number of paintings in which the Thames was pictured frozen compared with its current thawed state, they could deduce an indication of the winter temperature. However, the current study is the first time that researchers have obtained quantitative data about volcanic aerosols.
Why do Sunset Colours Change?
One of the major impacts on sunset colour is the amount of particles or aerosols in the atmosphere. At sunrise and sunset, the long sloping path of the sun through the atmosphere samples a large volume of aerosols which scatter the light more widely as the aerosols increase. This shifts the visible colours away from blue during the peak of the day to the more usual red sunset.
When volcanic aerosols are present, the shift in wavelength of light is even further to the red. The effect is quite subtle but by measuring the wavelengths in the atmosphere itself, we can get information about the aerosol optical depth – the amount the light is decreased by the aerosols at different wavelengths or colours.
Sunset Analysis Method
In the new study, to be published on 25 March 2014, in the Journal of Atmospheric chemistry and Physics, the scientists looked in painstaking detail at over five hundred digitised paintings, including over 50 large volcanic eruptions, to determine the ratio of reds to greens in the setting sun between the years 1500 and 2000.
Many of the sunsets could be correlated against known atmospheric conditions at the time. Professor Zerefos pointed out that the red to green ratios measured at sunset in the paintings correlated well with the volcanic aerosols in the atmosphere independent of the artist or the style of painting.
For example, after the Tambora eruption in Indonesia in 1815, the colour of the sky changed as far away as Europe. Red and orange sunsets persisted for almost three years afterwards and the British artist J.M.W. Turner recorded them at the time. I like Turner, and anyone who knows Turner’s work knows that a lot of his material is pretty vague (e.g. ‘Burial at sea’), but his colours were accurate.
The results also correlated well with other atmospheric proxies such as ice core and volcanic explosivity data and support the previous work in this area published by the team in 2007. The researchers obtained further support for the data analysis method by getting an artist to paint the sunset affected by a Saharan dust storm in 2010, which also produced a slightly more orange sky than observed in the Island of Hydra in Greece, as shown in the picture. Measurements of the aerosol optical depth agreed with the method of analysis applied to the painted sunsets.
Climate Proxy Data
These new data form part of a valuable group of data known as proxy data.
Ideally, we would like past measurements of temperature, volcanic aerosols etc. in the same way as we currently measure them, but this is an unrealistic aim if we are to investigate the atmosphere prior even to the 1970s when satellite data first became available.
The idea of proxy data is that a measurement is made of something, such as the width of a tree ring, and this is related to the measurement that we want, say temperature.
The relationship is of course not usually direct, but by having a period of overlap between the ‘proxy data’ and the ‘real data’ we can set up correlations between them – although the correlations can never be exact.
This was one of the issues in the so-called Climate Gate affair in which tree ring data for recent years needed careful interpretation.
Once a correlation is set up between the proxy data and the real data for an overlap period for which both sets of data exist, the correlation is extended into the period when only the proxy data exist. For example, we have tree ring data providing temperatures back millennia. We also have ice core data, another type of proxy going back even longer.
We now have a valuable proxy for aerosol amounts as revealed by the recent study by Prof. Zerefos’ team. “We wanted to provide alternative ways of exploiting the environmental information in the past atmosphere in places where, and in centuries when, instrumental measurements were not available,” Zerefos tells us.
Volcanic Aerosol Impacts
Volcanic aerosols affect the climate by reflecting some of the radiation incident from the sun. For a period of a year or two following major volcanic eruptions, the global surface temperature could drop by 0.4 oC as it did following the eruption of Mt. Pinatubo in 1991. Such cooling would temporarily halt the increase in global temperature due to human-produced greenhouse gases.
Simulations of observed weather conditions following volcanic eruptions provide important tests of our climate models. The new data on major volcanic aerosols now provides valuable data on historical eruptions for more detailed simulations of the past. With these more careful comparisons with the past, climate models will need to be more accurate, and this will mean that simulations of the future atmosphere will be more trustworthy.