| Above: Volunteers help rescue residents and their pet from their flooded home during Hurricane Florence September 14, 2018 in New Bern, North Carolina. Image credit: Chip Somodevilla/Getty Images. |
Climate change made Florence’s most intense rains over North Carolina more than 50% greater in magnitude than they would have been otherwise, according to a first-of-its-kind experimental “pre-attribution” study done last week as the storm was approaching landfall. Florence was also likely 8 – 9% greater in areal size due to climate change, and the storm was more likely to stay at a high Saffir-Simpson category as it approached landfall, the researchers found, using a climate model that generated forecasts with and without climate change factored in.
The research, led by Dr. Kevin Reed of Stony Brook University’s Climate Extremes Modeling Group, used techniques that researchers have been performing with increasing confidence and sophistication since the first “attribution studies” linking specific weather events to climate change began appearing about five years ago. The Hurricane Florence research was unique because it used a forecast of an event that had not yet occurred, instead of an after-the-fact look at a past weather event.
 |
| Figure 1. Predicted precipitation for Hurricane Florence made by the Community Atmosphere Model (CAM) version 5, initialized at 0Z September 11, 2018. Left: simulation done in the current climate. Right: simulation done in an atmosphere where human-caused climate change had not occurred. The predicted rainfall in the bullseye of accumulation on the southeast coast of North Carolina was more than 50% higher in the current climate simulation. The actual rainfall in Florence was even heavier than the "With Climate Change" simulation found. This is to be expected in part because it is the ensemble average rainfall from a set of ten different runs of the model that is plotted in the figure. Given that there is some difference in track location throughout the ensemble members, you can expect that the average rainfall will "smooth out" the extremes in a spatial distribution plot. Image credit: Reed et al., 2018, The Human Influence on Hurricane Florence. |
Increased rainfall from hurricanes in a warmer climate is to be expected
As Hurricane Florence approached landfall in North Carolina, the storm was invigorated by sea surface temperatures (SSTs) that were about 2 – 4°F above average. Climate change was responsible for a good deal of this warming. As I discussed in great detail in a post earlier this year, Extreme Hurricane Rainfall Expected to Increase in a Warmer World, one of the more confident predictions hurricane scientists can make on the future of hurricanes in a warmer climate is that they will dump heavier rains, due to increased moisture in the atmosphere from warmer oceans. There is a growing body of literature showing that heavy precipitation events of all kinds—including those from tropical cyclones (which include all hurricanes, tropical storms, and tropical depressions)—have already grown more common. Four papers published in 2017 and 2018 have found that human-caused global warming significantly increased the odds of heavy rains such as those Hurricane Harvey brought to Texas.
One of the co-authors of last week’s Hurricane Florence study, Dr. Michael Wehner of Lawrence Berkeley National Laboratory, was also a co-author of one of these Hurricane Harvey studies: Attributable Human‐Induced Changes in the Likelihood and Magnitude of the Observed Extreme Precipitation during Hurricane Harvey. Using only observational data, the study found that human-induced climate change increased the chances of the observed 7-day precipitation accumulations during Hurricane Harvey in the most affected areas of Houston by a factor of at least 3.5, and that precipitation accumulations in these areas were increased by around 38% (lower bound, 18%). Their analysis showed that there had been a clear increase in the probability of extreme 7-day precipitation events along the Texas coast since 1950, with a 1-in-100-year event now being more like a 1-in-25-year event.
The 38% increase in Hurricane Harvey’s heaviest rains is lower than the 50+% increase found for Florence’s rains, so this result bears additional scrutiny. However, in an email to me, the lead author (Kevin Reed) pointed me to research he had co-authored showing that extreme precipitation events can increase by over 60% for a one degree Celsius increase in sea surface temperatures. Thus, a greater than 50% increase in Florence’s heaviest rains due to climate change is possible. Further research is needed to investigate this finding, though.
How much did long-term #GlobalWarming contribute to the hot sea surface temperatures that boost hurricanes? Here an analysis of the Atlantic temperature trends, from Caesar et al., Nature 2018. https://t.co/YEgxI3FiGR #Florence pic.twitter.com/LdWv3NKOq9
— Stefan Rahmstorf (@rahmstorf) September 9, 2018
Other modeling studies on storm size and climate change: mixed results
Less convincing is the finding that Hurricane Florence’s size may have increased by about 8 – 9% due to global warming. That number is small enough that it could be a peculiarity of the particular model used. As I document in great detail in an April 2018 post, Will Global Warming Create Larger Hurricanes?, there are at least four modeling studies in recent years that have shown mixed results on whether or not global warming will create larger hurricanes. To some extent, the result appears to depend upon the model chosen. For example, Schenkel (2018) ran three separate models for a moderate global warming scenario (RCP4.5). Two of their models showed an 8 – 9% increase in size of Atlantic tropical cyclones by 2100, as measured by the radius of 8 m/s (18 mph) or greater winds. Their other model showed no significant change in tropical cyclone size.
Should scientists do more pre-attribution studies?
In the wake of this first-ever pre-attribution study, it’s worth asking if more of these studies should be done. On the one hand, it is relevant to gain a more informed answer to what degree an extreme event, as it is happening, is being influenced by climate change; on the other hand, doing so using a forecast gives higher uncertainties than if one were to study the event after-the-fact, since a post-storm analysis will give more reliable results. Post-storm attribution studies also have that advantage of being subjected to peer-review—the process by where other experts review the research, suggest improvements, and help make the study more scientifically rigorous.
But post-storm attribution studies occur weeks or months after the event, long after public attention has moved on, meaning people never become aware of the science linking a specific extreme event to climate change. Because of this, I think more pre-attribution studies should be done, with the emphasis that the approach is experimental research potentially subject to considerable error—similar to the large errors we see in April pre-season forecasts of the coming season hurricane activity. Future pre-attribution studies should be done using more than one model, so that the individual idiosyncrasies of a particular model are less likely to create unrealistic results. If two or more independent groups of researchers using different models were involved, that would be even better.
Dangerous climate change is here now
In an interview with climateprogress.org, Dr. Michael Wehner, one of the co-authors of the Hurricane Florence study, said:
"The most important message from this (and previous) analyses is that “Dangerous climate change is here now!” It is not a distant threat in the future but today’s reality. Event attribution has shown this for heat waves, floods, certain kind of droughts and tropical cyclones."
