The dam that failed in central Michigan on Tuesday gave way for the same reason most do: It was overwhelmed by water. Almost 5 inches of rain fell in the area in the previous two days, after earlier storms had saturated the ground and swollen the Tittabawassee River, which the dam held back.

No one can say yet whether the intense rainfall that preceded this disaster was made worse by climate change. But global warming is already causing some regions to become wetter, and increasing the frequency of extreme storms, according to the latest National Climate Assessment. The trends are expected to continue as the world gets even warmer.

That puts more of the nation’s 91,500 dams at risk of failing, engineers and dam safety experts said.

“We should expect more of these down the road,” said Amir AghaKouchak, a professor of civil engineering at the University of California, Irvine. “It’s unfortunate but this is what the trend is going to be.”

Overall, he and others say, dams in the United States and elsewhere are unprepared for the changes coming in a warming world.

The dam that failed Tuesday, forcing the evacuation of about 40,000 people in and around Midland, Michigan, and threatening a chemical complex and toxic waste cleanup site, was designed a century ago, long before climate change was a concern.


The dam, at Edenville Township, about 30 miles upstream from Midland, had severe design problems: It had been cited for having spillways that were inadequate to handle a maximum flood, whether affected by climate change or not. (A second dam at Sanford, 10 miles downstream, was overrun by the arriving floodwaters but did not collapse.)

But the Edenville Dam was hardly alone in being outdated, with design or maintenance deficiencies or other problems that might make it unsafe. The American Society of Civil Engineers, in its latest report card on infrastructure in 2017, gave the nation’s dams a “D” grade.

The average age of dams in the United States is nearly 60. And nationwide, about 15,500 are classified as having a high hazard potential; in Michigan, more than 170 dams are in that category, as was the Edenville Dam. Repairing and upgrading high-hazard dams alone could cost tens of billions of dollars.

Since the mid-19th century there has been an average of about 10 dam failures a year in the United States, said Martin W. McCann Jr., a civil engineer who directs the National Performance of Dams Program at Stanford University. More than 90% of failed dams are less than about 50 feet high. (Edenville was 54 feet tall.)

Rivers and reservoirs swollen by rainfall are the cause of most of the failures. “It’s not a new thing per se,” McCann said.

But some recent dam episodes have been shown to have a climate change link. In February 2017, at Oroville Dam in California, the tallest in the nation, heavy mountain runoff into the reservoir led to the near-failure of an emergency spillway and severe damage to the main spillway. Nearly 200,000 people were evacuated as a precaution and repairs cost more than $1 billion.


A later study found that human-caused warming had increased early season runoff in the Sierra Nevada, contributing to the high water levels at the dam.

And there is little doubt that extreme rainfall events are getting more frequent. The fourth National Climate Assessment, issued in 2018, showed that the number of heavy precipitation two-day events have increased in all regions except the Southwest since the early 1900s. And since 1950, extreme events increased by more than 50% in the Midwest.

But Bill McCormick, who is in charge of dam safety for the state of Colorado and is the incoming president of the Association of State Dam Safety Officials, said short-duration extreme precipitation wasn’t the only problem.

Rainfall of longer duration but less intensity — an overall wetter climate, which climate models forecast for parts of the United States in coming decades — can contribute to the risk.

“They tend to saturate the ground,” McCormick said. “Then, if you do get these 4- or 5- inch rains, the ground can’t absorb it.”

That was evident in the Michigan disaster, he said, where even though the two-day deluge was not considered the most extreme possible rainfall event, it still overwhelmed the dam because the ground was already saturated from several days of rain.


Dam engineers have usually based their designs on past weather — what a decade or decades of data show about the maximum potential flood a dam would have to withstand. That would have been how the Edenville Dam was designed in the 1920s.

But there was no expectation that future weather patterns might be different.

This kind of thinking largely persists today, AghaKouchak said. “Still, our engineering design concept is based on the so-called stationary assumption — that things will stay the same,” he said.

“But as we get more and more evidence of changes in extremes, the question is if it’s reasonable to stay with this stationary assumption,” he said. “The answer is, probably not.”

Some designers are beginning to change their ways, said Robert Lempert, a principal researcher at the RAND Corp. who specializes in climate risk analysis. Legislation recently approved in California, for example, requires state engineers to take climate change into account when designing infrastructure projects.

“If you’re building a dam you want to pull in climate change from the very beginning,” he said. “How is climate change going to affect the design of the dam, or even whether I want a dam at all?”


For existing dams, operational changes might be called for, such as reducing the water levels behind the dam at certain times of year in anticipation of more extreme storms. “And you want to put climate change on the agenda for any maintenance and upgrades,” Lempert said.

Those upgrades might include changing spillway designs to incorporate the kind of rainfall pattern that occurred in Michigan, McCormick said. Rather than one designed to handle high peak inflow from a short, extreme storm, designers may opt for one that could cope with slightly lesser volumes over a longer time period.

“You need to look at how a given spillway is designed,” he said, “if the circumstances of the rainfall change.”