JOSH DARR – Senior Vice President and Lead Meteorologist, JLT Re, Chicago, IL
JEFF SCHMIDT – Catastrophe Exposure Consultant and Meteorologist, JLT Re, Philadelphia, PA
The 2017 and 2018 hurricane seasons have brought a significant level of insured and economic loss to the areas surrounding the Atlantic, from the U.S. to the Caribbean, and even in the U.K. as well as Europe. The last two years now rank the highest of all-time in terms of economic loss, even ahead of the infamous years of 2004-05 when accounting for inflation and exposure changes (data per The National Oceanic and Atmospheric Administration (NOAA)/ The National Hurricane Center (NHC) see Figure 1). The insurance protection gap for the flood peril is a main driver of 2017-18 overtaking 2004-05 from an economic loss perspective.
Figure 1: Economic losses from hurricanes associated with the most damaging back-to-back years on record. Latest statistics of 2017-18 accumulating over US$300 billion in economic loss exceeds that of 2004-05. (Source: The National Oceanic and Atmospheric Administration (NOAA) / The National Hurricane Center (NHC))
The three heaviest hurricane rainfall events in history have transpired in the last two seasons: Harvey impacting Houston in 2017, Lane’s impact on Hawaii in August 2018, and Florence in North Carolina just a month later. The meaningful gap between economic and insured losses highlights the opportunities and challenges posed by the flood peril (please see our specific analysis of the floods of Hurricane Florence in the following article). Yet with elevated losses, particularly after preseason conditions promoted expectations of lower activity from seasonal forecasts, the insurance industry is left with potentially more questions than answers after the extreme losses of the past two seasons.
Season Surprised Most Seasonal Forecasts To The Upside
What was generally expected to be an average to below average hurricane season surprised most forecasters to the upside; above average numbers of named storms, hurricanes, and Accumulated Cyclone Energy (ACE) compared to the 1981-2010 seasonal averages characterized the 2018 Atlantic hurricane season. Before hurricane season begins, numerous public and private entities release preseason hurricane outlooks, followed by incremental updates to their forecasts throughout the season as new information becomes available. What stands out, particularly in 2018, is that many of the initial preseason forecasts proved to be closer to reality than those that accounted for midseason adjustments during the summer. The general consensus for most forecasters was a decreasing expectation of activity, largely due to the slow start to the season – 2018 was the first year since 2013 without hurricane activity by August. However, a sharp uptick in activity was observed at the peak of hurricane season in September which persisted through the second half of the season.
Figure 2: Timeline of the 2018 Atlantic Hurricane season. (Source: JLT Re, NOAA, https://www.nesdis.noaa.gov/content/hurricane-florence-september-2018, https://www.nesdis.noaa.gov/content/hurricane-michael-makes-landfall)
So what made forecasters change their outlooks when the season turned out to be more active than anticipated?
- Tropical Atlantic sea surface temperatures (SSTs) cooled into the heart of summer, which is highly correlated with reduced activity.
- An increased probability of the formation of El Niño toward the beginning of the Northern Hemisphere summer is generally known to suppress Atlantic hurricane activity due to elevated wind shear.
Figure 3: Niño 3.4 index for the 2018 calendar year, indicating that a weak El Niño was probable leading into the peak of the Atlantic Hurricane Season. (Source: NOAA/CPC, https://www.cpc.ncep.noaa.gov/data/indices/wksst8110.for)
Together, these two factors would traditionally point towards lower than average activity, yet the activity of 2018 grew above average by less conventional means.
- Ocean heat content in the sub-tropical North Atlantic remained anomalously high compared to historical standards, which was displaced north of the tropics and in a distinctly different pattern than in 2017.
- This SST pattern led to a number of storms that formed in the subtropical North Atlantic. Storm genesis was largely outside of the Main Development Region (MDR, the area between Africa and the Lesser Antilles), whereas storm formation in the MDR was favored in 2017.
- 2018 set a new record of seven subtropical storms forming in the subtropical North Atlantic, a meaningfully different storm track than the year prior.
Michael and Florence: The Two Main Loss Drivers with Unique Attributes
Perhaps one of the most peculiar storms in recent years was Hurricane Florence. Florence was the first major hurricane of the 2018 season and formed within the MDR. Florence perplexed scientists with its persistent westward motion at a higher than average latitude, failing to recurve to the north and east as nearly all historical analogs had done. Strong blocking high pressure to the north resulted in a highly abnormal track westward to the Carolinas. After making landfall, Florence brought prolonged rainfall to the Carolinas and Virginia, causing extensive flooding and high economic rather than insured loss. Florence laid bare the broad protection gap for flood losses in the U.S., just as Harvey had a year prior.
Not once, but twice, Hurricane Florence underwent rapid intensification from a category 1 to 4 hurricane. According to scientific studies, over 70% of major hurricanes undergo a period of rapid intensification, but rarely twice in the storm lifetime. Rapid intensification events have been prevalent in recent years: Harvey, Irma, and Maria of 2017, and this year Florence, Michael, and Yutu in the Western Pacific, among others. Beyond that, recent storms seem to be rapidly intensifying at an even higher acceleration than the threshold (30 knots in 24 hours). Hurricane Michael’s maximum wind speed increased from 74 miles per hour to 155 miles per hour, just shy of a category 5, in one 24-hour period. Michael was one of the strongest intensification events in Atlantic history due to exceptionally warm SSTs and lower wind shear. Recent academic research on the changes in hurricane activity reasons that the increase in such rapid intensification events is likely due to shifts in the climate. In the case of Michael intensifying to the coastline, the outcome for the Florida panhandle and southwest Georgia was catastrophic. As the strongest hurricane to make landfall in this region on record, Michael’s extreme storm surge caused massive inundation. The Florida panhandle, most notably the towns of Mexico Beach and Port St. Joe, also endured high end category 4 winds resulting in severe wind damage.
Figure 4: Depiction of the difference in SST anomalies between the start of the 2017 and 2018 hurricane seasons. The 2018 season commenced with cooler than average tropical Atlantic SSTs from Africa west to the Caribbean, as well as being markedly cooler than the start of the 2017 season. Meanwhile, sub-tropical SSTs were at near record warm levels and warmer than observed in 2017. (Source: Riskpulse)
Figure 5: The non-traditional track of Hurricane Florence, which reached category 4 status in the Western Atlantic before weakening to category 1 prior to landfall. The track of Florence relative to all category 3 and 4 hurricanes since 1851 within 250 miles of the storm shows that most historical analogs tracked perpendicular, rather than parallel, to Florence, particularly close to landfall. (Source: NOAA/NHC, JLT Re, https://catalog.data.gov/dataset?tags=hurricane+tracks)
Topics to Address Post 2017 & 2018 to Advance Hurricane Catastrophe Risk Assessment
With the 2017 and 2018 seasons in the rearview mirror, many researchers are contemplating what the implications of an ever-changing climate might have on tropical systems. While one must take caution when making conclusions based on two recent data points, a general trend can be observed with the recent data, highlighting the occurrence of storms that are both stronger in pressure and wind speed at the time of landfall. Hurricanes Irma, Maria, and Michael visibly stand alone in the graphic below, where stronger landfalls are in the bottom right of the chart. Typhoon Yutu in the Western Pacific, which made landfall in Guam, tied Camille as the second strongest U.S. tropical cyclone to make landfall, based on central pressure at the time of landfall.
Figure 6: A data visualization of the lights in Florida on October 6 before Hurricane Michael struck (top), and after on October 12 (bottom). Scientists filtered out light from anything with a natural glow – like from moonlight, or fires. And they paired it with neighborhood info from OpenStreetMap. (Source: Joshua Stevens/NASA Earth Observatory and Image: NASA Earth Observatory; Data: Miguel Román and Ranjay Shrestha, NASA/GSFC)
Figure 7: 2005, 2017, and 2018 hurricane landfall pressures and wind speeds compared to all U.S. landfalling hurricanes since 1851. (Source: AOML/NHC/NOAA, http://www.aoml.noaa.gov/hrd/tcfaq/E23.html)
As already seen on the back of 2017, a wealth of claims data is helping to inform improved risk assessment at the local level for all lines of business. Clearly emerging is the importance of data characteristics that will inform the next generation of damage assessment well beyond the occupancy, construction, square footage, and the year a building was constructed or modified. For the insurance community, understanding the strengths and shortcomings of catastrophe models as well as the historical record being leveraged for future views of risk in a changing climate are at the forefront of broadening our collective understanding. Several of these emerging topics are actively being investigated by the academic and insurance sectors, highlighted in Figure 8.
Figure 8: Impacts of the observations in the 2017-18 hurricane seasons that may alter future views of risk in hurricane catastrophe models. (Source: JLT Re)