Paintings provide precise weather history
YOUR MISSION, should you accept it, is to visit 41 art museums in the United States and Europe, study 12,000 paintings for their meteorological revealings, and publish your results so that this fate of hitting museum marble in your Birkenstocks need not befall future generations. Is this a fleecing of America? Relax! This all happened in the 1960s, before our current zeal for budget balancing.
Our intrepid museum-goer was Hans Neuberger. He was a professor of meteorology at Penn State when I was a student there. I was a botanist at the time. Hans was a regular at our interdisciplinary gab sessions in the Botany Department.
We talked about deep things: Why does the amount of steam you could see from a compost pile depend on which side of the pile you are standing near? How do you get a dewbow (that is a terra-firma rainbow due to dewdrops on the ground)? Which way do fungal spores blow in a forest? Neuberger's results saw print in the British Royal Meteorological Society's journal Weather.
Hans was a devotee of paintings and a weather historian. His hypothesis was that the weather artists had "oiled" on their canvases over the last six centuries was a record of climate history. The task at hand was simple: Observe the weather shown on canvas skies in the great art of the world, record the date and location and then harvest data from weather station archives for these locations.
The 12,000 paintings he studied covered the period from 1400 to 1967, but the weather data was available only from the mid-19th century on. The test period he selected was 1850 to 1967. If the artists proved to be faithful recorders of the weather, then their works could tell us about the climate back to 1400. Here is how he did it.
Using three categories, pale, medium and dark blue, Neuberger estimated blueness of painted skies in each painting. This simple scheme minimized the confounding effect of the patina of the ages. From these data, he calculated the average blueness of the skies for each 50-year increment of the last six centuries.
With three visual distance categories (less than 2.5 miles, 2.5 to 11 miles, or more than 11 miles), Neuberger estimated the visibility or haziness of the air in each painting. Our European sojourner and weather and art geek visited the landscapes painted to see how far away the painted, visible things on the horizon were. This is the way the U.S. National Weather Service does it.
It was hard work but somebody had to do it. Cloudiness estimates were according to the British airways code categories: clear (less than 10 percent cloudiness), scattered (10 percent to 50 percent cloudiness), broken (50 percent to 90 percent) and overcast ( more than 90 percent).
Fifty-three percent of the 12,000 paintings contained weather information. Some of the outdoor paintings showed no sky and most of the indoor paintings had no sky visible. Neuberger enjoyed the other 47 percent of the paintings but he had to exclude them from the study. For the rest he found excellent correlation between observed weather and painted weather for the post-1850 period.
Observed visibility was 33 percent higher than painted visibility. Even with this bias, the correlation between observed and painted visibilities was 0.93. A perfect correlation gets a score of 1.0. The artist also put, on average, 26 percent too many clouds in the sky. However, given that painted cloudiness was highly correlated with observed cloudiness (with a score of 0.90), we may conclude that artists are biased but faithful observers of the skies.
Neuberger, well aware of the onset of the "Little Ice Age" (a.k.a. the Neoboreal climate period) which began around 1550 AD, divided his data into three periods: 1400 AD to 1550 AD, 1550 AD to 1850 AD, and 1850 AD to 1967. The Little Ice Age ended around 1850. He found that in the period before 1550 AD (the Pacific climate episode) that the skies were much bluer or more haze-free (65 percent compared to 50 percent), visibility was much greater and low clouds were much less common.
The brightest, bluest period was from 1400 to 1550. Even England had vineyards in those years. The cloudiest 50-year period was from 1600 to 1650. Cloud cover averaged 60 percent. By 1800-1850 the skies had cleared to 40 percent cloud cover. Thereafter, paintings show an increase in cloudiness to 49 percent. Airports in Europe and North America indicate a cloud cover increase since 1900 of about the same amount.
Artists, long-dead, are telling us that climate has been changing in a big way since 1400. Hooray for science in the art museums! Neuberger was a professor, so there is a take-home lesson. Interdisciplinary studies yield special rewards.
(Bruce Hayden is an environmental science professor.)