Hurricanes Katia, Irma and Jose on Sept. 8, 2017. (NASA)

Last year’s hurricane season was more than twice as active as a typical season, and an onslaught of storms left deep wounds across the Caribbean and southern United States that have not yet healed. But, ready or not, the 2018 Atlantic hurricane season begins on June 1.

As it has done every year since 1984, Colorado State University has released its initial predictions for the upcoming season. Its forecast is for a total of 14 named storms, seven hurricanes and three major hurricanes, which is slightly above the long-term average of 12 named storms, six hurricanes and two major hurricanes. Last year, we saw 17 named storms, 10 hurricanes and six major hurricanes.

These 2018 predictions will be updated on May 31, July 2 and Aug. 2.

Two primary factors critical for determining how active the upcoming Atlantic hurricane season will be whether El Niño develops and the configuration of North Atlantic sea surface temperatures.

El Niño

When El Niño conditions are present and ocean temperatures in the tropical Pacific are warm, the Atlantic hurricane season tends to be less active. The reverse is true when La Niña, the opposite phase, prevails and the tropical Pacific waters are cool. Right now, the verdict is still out about which conditions, if any, will take control by the time hurricane season is underway.

Similar to what occurred last winter, the tropical eastern and central Pacific cooled to weak La Niña conditions this past winter. These weak La Niña conditions have continued to persist, although there has been rising of sea surface temperatures in the eastern and central tropical Pacific in recent weeks.


Current SST anomalies across the Pacific basin. Generally SST anomalies are negative (cool) across most of the eastern and central tropical Pacific, indicative of weak La Niña conditions. (NOAA)

During the Northern Hemisphere spring, the future state of the El Niño and La Niña is notoriously difficult to predict — a phenomenon known as the springtime predictability barrier.

The low-level winds that force ocean circulation changes that drive El Niño and La Niña events are usually at their weakest during this time, and consequently, small changes in the atmosphere can have significant effects on whether El Niño develops. As a result, computer model forecasts are typically least reliable during the Northern Hemisphere spring. This lack of skill was clearly demonstrated last year, as most of the forecast models called for El Niño to develop by the peak of the Atlantic hurricane season. Instead, we ended up with neutral to weak La Niña conditions, which supported the very active season.


Predictions from a suite of ENSO models from March 2017 — the information that CSU had available at the time of its early April forecast last year. The black stars are the observed SST values.

The latest suite of computer model guidance indicates a chance for El Niño development by the peak of the Atlantic hurricane season, but the majority of the model guidance calls for neutral conditions, that is, neither El Niño or La Niña and sea surface temperatures near their long-term average.

At this point, CSU is anticipating that neutral conditions in the tropical Pacific are the most likely scenario for the summer/fall of 2018.

In summary, the Pacific Ocean may not give us many predictive clues about the Atlantic hurricane season.


Latest suite of ENSO forecast models from March 2018. The black oval highlights the peak of the Atlantic hurricane season (August-September-October). (International Research Institute for Climate and Society)

North Atlantic sea surface temperatures

Should the forecast for neither El Niño or La Niña hold true in the Pacific, the configuration of Atlantic sea surface temperatures becomes very important for predicting how active the hurricane season will be. Generally, the presence of warm water in the North Atlantic would tend to portend a busy season.

The western North Atlantic Ocean is currently somewhat warmer than normal, while the far North Atlantic and eastern tropical Atlantic are slightly cooler than normal.

Overall, sea surface temperatures dropped much faster than normal during this past winter, in large part because of a strong subtropical high pressure that dominated the North Atlantic during January and February, driving stronger winds that caused more mixing and churning up of the sea surface and, thereby, promoting cooling. Consequently, high sea surface temperatures, compared to normal, that dominated the North Atlantic last fall were replaced by cooler waters by late winter/early spring.


Current North Atlantic sea surface temperature anomalies. The far North Atlantic and portions of the eastern tropical Atlantic are colder than normal, while most of the remainder of the North Atlantic is warmer than normal. (NOAA)

The current sea surface temperature pattern is fairly similar to the pattern that prevailed in early April of last season. However, during last spring/summer, the waters warmed much faster than normal across most of the Atlantic, and by the peak of the Atlantic hurricane season, the sea surface temperatures were much warmer than normal across most of the tropical Atlantic, providing increased fuel for the hurricanes that formed last year (most notably, Irma, Jose, and Maria).

Several climate models including recent predictions from the National Weather Service Climate Forecast System model indicate that the tropical Atlantic will become warmer than normal by August-October. If this holds true, at least a somewhat active hurricane season would be likely.


Recent prediction of August-October global SSTs from NOAA’s Climate Forecast System model. The Climate Forecast System model is predicting a warmer-than-normal tropical Atlantic in August-October, with neutral ENSO conditions in the tropical Pacific. (NOAA)

Past preseason hurricane outlook accuracy

The skill of Colorado State’s seasonal forecasts improves as the season approaches. Looking at past predictions, the early April forecast has modest skill, while the skill increases considerably by June and especially by August.

The modest skill in early April is to be expected, given that much can change in the atmosphere/ocean system between early April and the peak of the Atlantic hurricane season. This was demonstrated clearly last season, where early seasonal forecasts issued by CSU and other agencies predicted a near- or even slightly below-normal season, given the anticipated El Niño conditions. However, by early August, most groups correctly predicted a well above-average Atlantic hurricane season, as El Niño chances diminished and the tropical Atlantic anomalously warmed.


Hindcasts from the early April statistical forecast model from 1982-2010, with real-time forecasts from 2011-2017. Accumulated Cyclone Energy is an integrated measure that takes into account tropical cyclone frequency, intensity and duration.

What could go wrong?

As is typically the case with early seasonal outlooks, the primary area of uncertainty lies in whether El Niño develops. Should a significant El Niño develop, Atlantic seasonal hurricane forecasts would likely be reduced. Alternatively, should the tropical Pacific have neutral conditions and the Atlantic were to markedly warm, seasonal forecasts would likely increase.

Many groups will issue seasonal hurricane forecasts over the next few weeks, and they will be displayed on a seasonal hurricane forecast compilation website jointly developed by the Barcelona Supercomputing Center, XL Catlin and CSU. A more detailed discussion of the 2018 outlook from CSU as well as archives of past outlooks are available here.

The 2018 list of storm names will be the same as 2012’s, with the exception of Sandy, which was retired and replaced with Sara.