CSU meteorology team underestimated activity levels of 2017 hurricane season, report shows

Samantha Ye

Flood waters rose into the Lakes on Eldridge North neighborhood near the Addicks Reservoir West Houston, Texas on Wednesday, Aug. 30, 2017. Hurricane Harvey, one of two Category Four hurricanes to hit the continental U.S. this year, inundated the Houston area with several feet of rain. (Tom Fox/Dallas Morning News/TNS)

The activity levels of the 2017 Atlantic hurricane season were considerably higher than what was predicted by Colorado State University’s Tropical Meteorology Project forecast team.

In their 2017 seasonal hurricane forecasts summary and verification report, authors Philip Klotzbach, research scientist, and Michael Bell, associate professor from the department of atmospheric science, attributed above-average sea surface temperatures in the tropical Atlantic and reduced levels of vertical wind shear, or the change of wind direction and speed with height, in creating a season defined by devastating hurricanes such as Harvey and Irma.


“The 2017 hurricane season was extremely active,” said Klotzbach in an article by SOURCE. “Overall, our predicted numbers from our early July and August issue dates for named storm and hurricane formations were relatively close to what was observed, but our early season predictions and our predictions for integrated metrics such as Accumulated Cyclone Energy were far too low.”

The forecast team’s first extended range seasonal forecast, issued in April, projected below-average hurricane activity in the Atlantic basin, based on predictions that the El Niño-Southern Oscillation would transition to either weak or moderate El Niño conditions by the peak of the Atlantic hurricane season.

According to SOURCE, El Niño conditions include increased levels of vertical wind shear, which “tear apart” developing hurricanes.

But, the dynamic and statistical models for El Niño did not materialize. Instead, La Niña conditions developed, reducing vertical wind shear.

As a result, the April report underestimated hurricane activity by a significant amount, projecting only 11 named storms, four hurricanes and two major hurricanes. According to the final report, the observed season had actually 17 named storms, 10 hurricanes and six major hurricanes.

Of the six major hurricanes, both Harvey and Irma made landfall in the continental United States, the first time since 1851 that two category-four hurricanes hit the nation in one year.  

The second extended range seasonal forecast, issued in June, came closer with 14 named storms, six hurricanes, and two major hurricanes, after factoring in diminishing prospects of an El Niño and a warming Atlantic.

The forecast team continued to refine their forecasts with July and August updates, increasing activity predictions with each. According to the final report, the six consecutive two-week forecasts issued August through October — the peak months of the Atlantic hurricane season — generally verified well.

Combinations of precursor signals such as ENSO, Atlantic sea surface temperatures and sea level pressures among others, can create forecast potentials of around 50-60 percent. The forecast team has 70 years’ worth of such data to go off of, which they consistently adapt to new forecast schemes.

“Keeping up with the changing global climate system, using new data signals, and exploring new physical relationships is a full-time job,” the authors wrote in the final report. “Success can never be measured by the success of a few real-time forecasts but only by long-period hindcast relationships and sustained demonstration of real-time forecast skill over a decade or more.”


The Tropical Meteorology Project will release their first outlook for the 2018 hurricane season Dec. 13. 

Collegian reporter Samantha Ye can be reached at news@collegian.com or on Twitter @samxye4.