Man-made climate change has been influencing extreme rain and precipitation events for decades, according to new research using machine learning.
Climate models have long predicted that extreme precipitation (rain, snow or hail that dramatically exceeds normal levels) will happen more often as anthropogenic climate change advances. But this new study, published today in Nature Communications, analysed changes in precipitation from global observational datasets spanning 1982 to 2015 and found evidence that anthropogenic climate change had already intensified extreme precipitation over that time.
Extreme precipitation becomes more common in a warming world because warmer air can hold more water vapour, resulting in more intense rain, hail or snowfall. According to Steven Sherwood, an expert in meteorology and climate dynamics at UNSW who was not involved in the study, the amount of water vapour in the atmosphere increases by about 7% per degree Celsius of warming.
“In warmer climates, the most extreme events become better at squeezing the available water vapour out of the air, at least in model simulations,” he says.
The data already suggests that extreme precipitation has been on the rise since at least the 1980s, with the US experiencing nine of its top-10 years for extreme single-day precipitation since 1996.
Identifying the smoking gun for causes of extreme weather has historically proved challenging because weather has such natural variability, and extreme precipitation events lack large datasets since, by definition, they aren’t a common occurrence.
To approach this problem, the team of researchers, led by Gavin Madakumbura from the University of California, Los Angeles, trained an artificial neural network (ANN) to recognise an ‘anthropogenic signal’ – evidence of a human-origin influence on precipitation – from climate and geographic datasets that chart annual maximum daily precipitation around the globe.
The authors identified these ‘anthropogenic signals’ across various global datasets, and suggest that observed long-term changes in extreme precipitation can be consistently attributed to humans.
Extreme precipitation can have devastating impacts on communities and industry, causing flooding, soil erosion, destroying agriculture and leading to water-borne or sanitation-based diseases. In March 2021, for example, New South Wales suffered devastating flooding that temporarily displaced 24,000 people and killed two. While this event was attributed partly to existing, cyclical weather patterns, experts warned that similar catastrophes would become increasingly common as the climate warmed.
Sherwood says that as these events occur more regularly, society will have to learn to adapt.
“It is really important for any new infrastructure to plan for these predicted increases, to avoid flooding and landslide issues down the track,” he says.
Sherwood also says that more detailed research is necessary to tease out the specifics of what we can expect from future extreme precipitation events.
“It’s important to know how strong the increases will be for different event return periods,” he says. “We still don’t know, and this new study doesn’t really help to put a number on it, only to show that some change is detectable in the observations.”