A Meteorologist’s Perspective of Hurricane Irma

Sep 18, 2017

After 12 years without a major hurricane (category 3 or higher) making landfall in the United States, we have now had two category 4 landfalls in the same year—a first in our recorded history for Atlantic hurricanes. Estimates of the economic impact of Hurricanes Harvey in Texas and Irma in Florida are in the hundreds of billions of dollars. The emotional impact on those people who lived through them is unmeasurable.

The GOES-16 Advanced Baseline Imager captures Hurricane Harvey’s outer rainbands reaching the Texas coast in this blended visible and infrared imagery. (Credit: NOAA/CIRA)

Hurricane Irma holds particular significance for Harris, which is headquartered in Florida and has more than 6,000 employees in the state who suffered through Irma’s battering winds, coastal storm surge, and flooding rains. But as bad as Irma was, we know it could have been much worse. Two key factors helped Florida avoid the total catastrophe experienced in portions of the Caribbean:

  1. Cuba.  As it approached the Caribbean islands, Irma reached category 5 status with peak winds of 185 mph, making it the strongest Atlantic basin hurricane on record outside the Gulf of Mexico and Caribbean Sea. Then came Cuba.

    Initially, the National Hurricane Center (NHC) official forecast track (black line in left graphic below) kept the eye of Irma over the open waters just north of Cuba, although the cone of uncertainty did include the possibility of Irma’s center tracking over northern Cuba. As it turned out, the eye of Irma spent about 24 hours bouncing right along the northern coast of Cuba, which weakened the storm to category 3. Irma barely regained category 4 strength once it emerged solidly back over warm waters on its way north toward Florida and never got back to the high category 4 or even category 5 level it might have been.

    Hurricane Irma forecast track and GOES-16 image
    Left: NHC forecast for Irma as of Thursday evening Sept. 7. (Credit: NOAA) 
    Right: Actual GOES-16 satellite imagery of Irma’s eye hugging right along the northern coast of Cuba (Credit: NOAA/RAMMB/CIRA)
  2. Southwest Florida landfall. After crossing over the Florida Keys, Irma’s ultimate landfall was the less-populated area of southwest Florida. The storm then made its way up Florida’s Gulf coast impacting Naples, Fort Meyers, and Tampa. While this caused great damage for those in that region, it was nothing on the scale of what we would have seen had Irma made landfall further east, closer to the densely populated and built-up Miami area.


Generally speaking, the forecast for Irma was quite good in that Florida was identified as the likely target days before landfall. By 5 a.m. Wednesday, September 6, most of Florida was squarely in the NHC cone of uncertainty, and both government officials and the public were already ramping up their preparations.

NHC forecast for Irma issued Wednesday, September 6 at 5 a.m. EDT.
NHC forecast for Irma issued Wednesday, September 6 at 5 a.m. EDT. (Credit: NOAA)

As Irma tracked across the Caribbean and approached Florida, forecasters were aided by NOAA’s GOES-16 satellite, which launched in November 2016 from Cape Canaveral, Florida. The Harris Advanced Baseline Imager on GOES-16 provided forecasters with updated imagery every minute or less, helping to locate the exact center of circulation and closely monitor intensification of the storm. This is the first hurricane season ever that forecasters have received satellite updates every minute and in some cases every 30 seconds.

Category 5 Hurricane Irma as seen by GOES-16 (left) and GOES-13 (right), on the morning of Sept. 5, 2017. The animation on the left shows the improved resolution and faster refresh rate of the GOES-16 Advanced Baseline Imager, with a new image every minute. (Credit: NOAA/CIMSS)

Despite the early warning of a Florida landfall, forecasts weren’t good enough to enable more precise planning and evacuation within the state. For example, one of the primary tools in the forecaster’s kit is the “model ensemble,” the result of running a particular model dozens of times, each with slightly different initial conditions as input, to account for the uncertainty and incompleteness in our observations of the atmosphere. Each model run is called a “member” of the ensemble. The more the different member tracks agree with each other, the higher the confidence we have in the forecast. The farther member tracks are spread apart, the lower the forecast confidence.

The large spread in the European model's ensemble member tracks shown below on the morning of Friday, September 8, indicates a fairly high uncertainty in the forecast track.

Forecast tracks, European model
Forecast tracks from the 50 ensemble members of the European model.
(Credit: StormVistaWxModels.com)

The uncertainty in the exact track of Irma led to a shifting forecast. In the graphic below, you can see how on the morning of Wednesday, September 6, the European ensemble model average (red), the U.S. ensemble model average (blue), and the official NHC forecast (orange) all favored a track along or near Florida’s east coast. In the end, however, Irma took a west coast track, very close to the NHC forecast track issued on the morning of Saturday, September 9 (right-most picture).

Hurricane Irma models Sept 6-9
European model ensemble average (red), U.S. ensemble model average (blue), and official NHC forecast issued each morning Sept. 6-9, 2017. (Credit: StormVistaWxModels.com)

Why did models struggle with Irma’s track forecast? One reason is the semi-constant sprawling area of high pressure about 18,000 feet up in the atmosphere known as the Bermuda High, the main steering influence of most tropical systems in the Atlantic Ocean, including Irma.

Tropical systems are typically steered from east to west by clockwise motion around the bottom side of the Bermuda High. When the Bermuda High is weaker and does not extend too far south and west, tropical systems can turn to the north around its west side—soon enough to recurve away from the United States. When the high is stronger and extends far enough south and west, as was the case with Irma, then tropical systems are forced to track further south and west, and are more likely to hit the country before recurving to the north and then to the east.

Bermuda High
Credit: University of Arizona

Although we have made great progress in observing and modeling the atmosphere, usually we still cannot predict the strength and expanse of the Bermuda High with the desired detail more than several days ahead of time. And even within that time frame, we can struggle, as we saw with Irma.


Harris continues to work with its government, commercial, and academic partners to advance the technologies that monitor and help predict hurricanes, tornadoes, winter storms, and everything in between.

In addition to designing and building the Advanced Baseline Imager on GOES-16 and processing all of the data from the satellite, Harris also designed and built the Cross-track Infrared Sounder (CrIS) that will fly on NOAA’s Joint Polar Satellite System-1 (JPSS-1) satellite scheduled to launch in November.  CrIS, already onboard NOAA’s Suomi National Polar-orbiting Partnership satellite, is one of the top contributors to weather forecast accuracy, providing three-dimensional profiles of temperature, pressure, and moisture that feed into weather models.

Data from GOES-16 and JPSS-1 is expected to be used in weather models starting in 2018. GOES-S, GOES-T and GOES-U, scheduled to launch from 2018 through 2025, will also carry Harris Advanced Baseline Imagers, with additional CrIS instruments to fly on future JPSS satellites.

So while we will always remain vulnerable to natural hazards like hurricanes, we are likely to see some improvement in being able to prepare for them—and thereby better protect lives and property—in the future. And that’s something positive to look forward to.


Dan Stillman is a meteorologist and senior marketing manager for Harris Corporation's Environmental Solutions business. He can be reached at dan.stillman@harris.com