We can no longer rely on historical data to predict extreme weather conditions

A satellite image of Hurricane Harvey on August 24

A satellite image of Hurricane Harvey on August 24 (NASA / NOAA /)

Floods and other dangerous weather extremes are only intensifying and becoming more frequent as our climate warms. Historically, we have always been able to predict these extremes by examining the frequency with which they have occurred in the past. But a new study released Wednesday Scientists progress reveals how many of these forecasts are actually insufficient. In just a decade, the results suggest that the climate has changed so dramatically that the frequency of past extreme events is no longer a reliable predictor.

These predictions help us to map flood plains and design infrastructure so that it can withstand even intense events. But if our forecasts are wrong, it means that we can no longer plan for new homes, roads and bridges based on past storms. Rising extremes – such as tropical cyclones, heat waves and heavy storms – will force us to modify our plans and design structures capable of withstanding these changes.

It is difficult to understand the influence of human-caused warming on extreme events. The atmosphere is chaotic by nature, and extreme records are by definition rare, giving scientists few data points to understand them. Noah Diffenbaugh, Earth System Scientist at Stanford University, and a team of climatologists have incorporated a recording of extremely hot, humid and dry weather events from 1961 to 2005 in a climate prediction framework. This framework integrates both historical forecasts based on events and climate models, which integrate the expected future warming into their estimates.

Over the next decade, however, humans continued to burn fossil fuels and record weather events hammered regions around the world. Seven of the hottest 10 years on record hit between 2006 and 2017, and huge storms like Hurricane Harvey in 2017 caused more destruction than ever.

Given the additional warming, Diffenbaugh wanted to test the extent to which historical data could predict recent extreme events. He used data between 1961 and 2005 to develop probabilities of hot, humid, and dry extremes in the northern hemisphere between 2006 and 2017. Separately, Diffenbaugh also used climate models to compare heat waves. , real-world record storms and droughts with their in the past, as well as projections based on climate models.

Diffenbaugh's forecasts based on historical data have turned out to be poor for this decade. The results underestimated extreme events, especially hot and humid events. Compared to the projection based on historical data, extreme hot days have increased by at least 50% in Europe and East Asia. And the wet extremes observed were also 50% more frequent in the United States and Europe compared to historical prediction. "I was very surprised," says Diffenbaugh. "I had the impression that the framework that my group had developed over the past few years had some flaws."

But that was not necessarily the problem. The setting is perfect for predicting extreme events that occurred in the latter part of the 20th century. But over the past decade, the additional warming we have generated is so significant that extreme weather conditions diverge from those of the past. At the same time, the climate models tested by Diffenbaugh were able to accurately predict the frequency of record events between 2006 and 2017. "Short-term climate models for the future include what actually happened" says Diffenbaugh. "Even if they were future predictions at the time."

"Noah Diffenbaugh's paper is innovative and combines both models and observations to demonstrate that extreme probabilities are changing rapidly due to global warming," said Erich Fischer, climatologist at ETH Zurich, who doesn’t was not involved in the study. "The document has implications for risk management."

Diffenbaugh's findings have huge implications for designing new climate change infrastructure and updating existing structures. It seems that we cannot now estimate the probability of a 500-year flood simply on the basis of past floods, this is how we have tended to do our risk planning in the past. These results show that we need to use a combination of historical information and climate modeling to design the future under the effect of climate change.

Indeed, although climate models are effective at predicting changes in large areas, they will not tell an urban planner how often a particular river crossing their city can be flooded. These local events are just too difficult to predict at the moment. Some states are already trying to cope. In California, the 2016 Climate Security Bill has been enacted to develop a planning process for new roads, bridges and other structures so that they can withstand climate change. Diffenbaugh is part of the task force established by the bill. He says we need to use the results of climate models and local historical data to better prepare for future extremes. Fischer adds: "The climate and therefore the extreme probabilities that occur are different today than they were 10 or 20 years ago. To prepare, we need to use both past observations and future forecasts to literally keep our cities out of the water.