With climate change causing a series of extreme weather events, scientists from the National Center for Atmospheric Research (NCAR) have found improved methods for predicting summertime heat waves up to 50 days in advance.

The idea is that patterns of contrasting warmer-than-average and cooler-than-average sea surface temperatures in the middle of the North Pacific Ocean will likely trigger a heat wave. In fact, researchers discovered that the odds that an extreme heat wave will strike during a particular week - or even on a particular day - can more than triple, depending on how consistent the pattern is.

Extreme heat waves in the U.S. have caused extensive damage and killed hundreds of people within the past decade. Since rising temperatures are only expected to get worse, there is a great need for earlier warning systems. Until now, however, most forecast models were not able to predict extreme weather more than 10 days out.

"Summertime heat waves are among the deadliest weather events, and they can have big impacts on farming, energy use and other critical aspects of society," explained Karen McKinnon, a postdoctoral researcher at NCAR and the lead author of the study. "If we can give city planners and farmers a heads up that extreme heat is on the way, we might be able to avoid some of the worst consequences."

For the study, researchers first divided the country into regions that tend to experience extreme heat at the same time. They then focused on the largest of the resulting blocks: a region that stretches across much of the Midwest and up the East Coast, encompassing both agricultural land and heavily populated cities. It was in this area that researchers found what they call the Pacific Extreme Pattern.

"Whatever mechanisms ultimately lead to the heat wave also leaves a fingerprint of sea surface temperature anomalies behind," McKinnon added.

In other words, the Pacific Extreme Pattern tends to form in advance of heat bursts. The next step was to test how well this fingerprint could predict future heat.

"We found that we could go back as far as seven weeks and still predict an increase in the odds of future heat waves," McKinnon said. "What's exciting about this is the potential for long-range predictions of individual heat waves that gives society far more notice than current forecasts."

In the summer of 2012, for example, seasonal forecasts predicted normal heat for the Northeast and Midwest. However, the summer ended up being especially hot, with three major heat waves occurring in late June, mid-July and late July. Using the Pacific Extreme Pattern to hindcast 2012, researchers were able to predict increased odds of extremely hot days for late June as early as mid-May.

"The results suggest that the state of the mid-latitude ocean may be a previously overlooked source of predictability for summer weather," McKinnon concluded.

Their study was recently published in the journal Nature Geoscience.