Stanford researchers are using AI to make sense of extreme precipitation days in the Midwest.

Extreme precipitation days in the Midwest account for over half of all major US flood disasters. Researchers at Stanford University have developed a machine learning tool that helps them analyze the causes of long-term changes in weather and the mystery of why extreme precipitation days in the Midwest are becoming more frequent. (Stanford News)

Weather extremes in a changing climate

Extreme weather events are occurring across the world at a faster rate amidst global warming.

A landmark climate report released Monday by the U.N.’s Intergovernmental Panel on Climate Change (IPCC) warns that human influence has warmed the atmosphere, ocean and land, and that the changes inflicted on the planet, especially oceans, would be “irreversible for centuries to millennia.”

It also said continued warming will lead to an acceleration of “extreme events unprecedented in the observational record.”

U.N. Secretary General Antonio Guterres said there was irrefutable evidence that greenhouse gas emissions from fossil fuel burning and deforestation were choking the planet.

He called the report a ‘code red’ for humanity.

Though their cause can be attributed to global warming, extreme events are triggered by a chain of complex factors and it is difficult to predict or prepare for them.

Though it is known that global warming will cause much heavier rain and snowfall (precipitation), it is difficult to model these events on a regional basis.

Worse flooding

“We know that flooding has been getting worse,” said study lead author Frances Davenport, a PhD student in Earth system science in Stanford’s School of Earth, Energy & Environmental Sciences (Stanford Earth). “Our goal was to understand why extreme precipitation is increasing, which in turn could lead to better predictions about future flooding.”

Using AI

The researchers used publicly available climate data to calculate the number of extreme precipitation days in the upper Mississippi watershed and the eastern part of the Missouri watershed from 1981 to 2019. This is a highly flood-prone region of the US.

They then trained a machine learning algorithm designed for analyzing grid data, such as images, to identify large-scale atmospheric circulation patterns associated with extreme precipitation (above the 95th percentile).

“The algorithm we use correctly identifies over 90 percent of the extreme precipitation days, which is higher than the performance of traditional statistical methods that we tested,” Davenport said.

Davenport added that the Stanford approach could be expanded to more broadly understand extreme weather events, as well it could be applied to regions other than the US Midwest.

Related Story:  AI Can Now Forecast Ozone Levels Upto Two Weeks Ahead