A brown cloud of pollution over Phoenix, Arizona. Brown clouds of aerosol pollutant particles could be overwhelming the expected changes in rainfall arising from increasing greenhouse gas levels in the air. Credit: Flick/Flickr
Contrary to previous predictions and measurements, rain patterns have got more uniform as the world has warmed over the past 70 years. So say Michael Roderick and his teammates from Australian National University, Canberra, who’ve developed an ‘accounting system’ that looks closely at where and when rain fell. And the reason could be aerosols – clouds of pollutant particles – produced by humans. “The existing dogma is that increasing greenhouse gas concentrations in the atmosphere have raised rainfall variability,” Michael told me. “In that context, our results emphasise the importance of taking a whole system approach in trying to understand how something complex, like rainfall, is changing in different places.”
When scientists want to understand how climate has been changing over large areas, they usually look at maps of long-term average data that ignore patterns of change in time, Michael explained. When they want to look at how it’s changed over time, they usually either look at a single place or a worldwide average, which ignores patterns in where the changes are. But Michael, along with fellow scientists Fubao Sun and Graham Farquhar, wanted to find a way to link place and time.
To do this Fubao started from a common statistical test called Analysis of Variance or ANOVA. Normally it’s used to compare the effect of different “treatments” – such as a variety of temperatures – on the yield of a crop, for example. In such cases each treatment must be repeated more than once, giving different “replicates”, for the test to be valid. ANOVA can be used to give a value for variance – a measure that shows how spread out an experiment’s measurements are.
A grand plan
Australian National University’s Michael Roderick and his teammates have developed an “accounting method” for variability in climate records. Credit: Australian National University
In 2010, along with four other scientists, Fubao, Graham, and Michael used a similarity between this design and how climate records are organised. For example, were the experiments put through the ANOVA test done in dishes, a scientist could organise those dishes in a grid in their lab. To know which is which, they could sort them in treatment order along one side and replicate order along another. Climate measurements can be organised in a similar way, replacing the treatment with place information, and replicate information with dates. So the team replaced replicates and treatments in the ANOVA test with place and date, in the process changing the nature of the analysis.
“It is not a test,” Michael said. “It is better described as an accounting procedure where variations, either through space, or through time, both contribute to the overall variance. By formulating it as an accounting procedure one can disentangle the sources of variation and develop a new way of summarising the overall variability of the precipitation. The basic method would work with any two – or more – dimensional data set, like those with variation in space and time, including temperature or radiation. It can also work with socio-economic data, such as GDP by country.”
This overall, or “grand” variance is especially interesting for rain as, when averaged across the entire planet, rainfall hardly changes from one month to the next. That link means that when some people experience unusually wet summers, others can expect unusually dry ones, Michael explained. “One consequence is that if a place receives more rain than usual in a given month, then another place must receive less,” he said.
Pollution possibility
The US National Oceanic and Atmospheric Administration explains how climate models suggest rainfall will get more variable, with wet areas becoming wetter and dry areas drier. Credit: NOAA
Michael, Fubao, and Graham subjected detailed data on rainfall in 1,967 grids covering over two thirds of the Earth’s land area to their procedure. They compared data from 1940-2009 from all seven publicly available rainfall databases and cross-checked the results between each of them, finding the results almost identical. “We found a reduction in global land precipitation variability because wet places/times of the year, tended, on average to get a little drier whilst dry places/times of the year, tended, on average, to get a little wetter,” Michael said. “That change has not previously been documented.”
As rain is a critical ingredient in growing food, among many other things, understanding how it has changed and will change with humans adding chemicals to the air is important. But this finding, published in the scientific journal Geophysical Research Letters last month, contradicts previous climate models predicting that rain should get more variable with higher CO2 levels in the air. It also disagrees with measurements made at sea saying that wet places have got wetter and dry areas drier with climate change. However Fubao and Michael point out that these previous measurements look at the difference between evaporation and rainfall, while their new study only looks at rainfall.
The Australian scientists also note the largest changes in rainfall variability seem to be where clouds of pollutant particles are being produced. Climate models have also shown that these aerosols can lower rainfall variability – and Michael speculated that they will play a big part in explaining what they’ve seen. “Model results have suggested that aerosol emissions are important in changing the dynamics of rainfall,” Michael said. “Our results suggest that aerosol emissions have likely dominated the changes in rainfall dynamics over the past 70 years.”
Journal Reference:
Fubao Sun, Michael L. Roderick, Graham D. Farquhar (2012). Changes in the variability of global land precipitation Geophysical Research Letters DOI: 10.1029/2012GL053369
