Hurricane forecasters missed the bull's-eye with the 2006 season. Some would say that they didn't even hit the wall that the dartboard was on. Yet they'll toss their darts again next year, and people will anxiously await the results. Risk managers and other insurance professionals know better.
Last spring, forecasters called for a very active season, with upward of 17 named storms.
One forecaster, London-based Mark Saunders at Tropical Storm Risk, went so far as to predict two hurricane landfalls. William Gray's team at Colorado State University published a report calling for a "95 percent" chance of hurricane landfall on a U.S. coast.
In August--midseason--the hurricane prognosticators revised their forecasts down slightly, but still said the odds were in favor of an above-average season. Instead, the season played out more like the Spanish Armada. Most storms were wiped out at sea by unseen weather forces. None made landfall.
Forecasters got it wrong and overestimated the 2006 season, just as they had underestimated the number of storms the previous year.
The highest bidder in May 2005, the National Hurricane Center, called for 12 to 15 named storms and seven to nine hurricanes. Come August, the forecasters all upped their forecasts to catch up with what was an extraordinarily active early season. Even then, they still fell short. The highest August forecast came out of Tropical Storm Risk: 22.1 storms and 11.4 hurricanes. The actual totals when the season ended were 28 and 15, respectively.
The going explanation for why the forecasts overestimated this 2006 season is simple. Emily Schroeder, a meteorologist with Aon Impact Forecasting, says the season played out something like this:
Waves of thunderstorms did form early off Africa, but dry air over the Atlantic and perhaps dust from the Sahara Desert that dispersed across the Atlantic basin sapped those storms of their moisture and energy. In the second half of the season, El Niņo popped up in the Pacific, further emasculating the storms by whipping up winds in the Atlantic.
Any storm powerful enough to reach named status despite these factors then got caught up in a weak subtropical high, known as the Bermuda High, centered more toward Europe than it had been in the previous two years. The high directed storms northward while they were well east of the United States.
These hurricane-inhibiting factors weren't expected, or easily predictable, before the season started, Schroeder says.
But that's not news to risk managers and insurance professionals who follow storms every year. They've already learned to take hurricane season forecasting with a heaping of salt.
"I don't know how somebody could predict that," says Dennis Slabaugh, director of claims and senior risk management consultant at Gulfshore Insurance Agency Inc. in Naples, Fla. "I just don't think they have the tools and the resources--I just don't think they're available--to be able to predict accurately the number and the depth of the storms that we're going to have in any given year."
"Personally, to me, I find it somewhat humorous," says Wayne Salen, director of risk management at Labor Finders International Inc., headquartered in Palm Beach Gardens, Fla.
He's skeptical about how forecasters come up with a range for storms, and how the major forecasters get to "redoing" their forecasts in the middle of the season.
"From the standpoint that the forecasters are off, substantially for this year, it's not surprising to me because there are just too many variables for them to say with certainty that we're going to have so many storms, and so many named storms, and so many severe storms. And what are we finding out this year? It's almost nothing," Salen says.
The problem isn't entirely that hurricane forecasts are just "guesstimates," as Slabaugh calls them. The issue in the last few years--and of course more so since Hurricane Katrina--is the "overreliance upon and/or the hype, if you will, created by these forecasts," he says.
"Forecasters are all about statistics. They measure all sorts of stuff," says Bud Trice, vice president, catastrophe services, Crawford & Co. "But in the end, it's all about probability and relativity."
"When they state, '13 tropical systems, four hurricanes, two intense hurricanes,' they really don't believe they're going to achieve this level of exactitude. Rather, it's a sexier way to state the odds. And John Q. Public loves sexy information," Trice adds.
In other words, forecasters like the hype. But this hype, says Slabaugh, can sometimes cause a lot of people, business owners and consumers alike, to waste time and money in a fit to prepare for a bad season, or a bad storm--time and money that could be better spent on "planning rather than overreacting."
Experienced managers of risk and insurance claims, though, know better than to be affected by the false sense of urgency.
"I think everybody realizes that the predictions are just that-- they're a toss-up," Slabaugh says.
Salen in South Florida listens to the forecasts in so much only as a "measurement" or "kickoff" to review his company's plans for an emergency, which are in place no matter what a forecaster says. But Salen doesn't "shift gears" until actual storms organize over the ocean, when he has two to three people on staff to monitor them.
"They (the forecasts) do provide some reasonable value," he says, "but I discount them all to some degree and just simply try to stay atop of activity that's out there, because the reality is you've a good number of days, and even weeks, to prepare when there's some activity that starts to brew, especially off the coast of Africa."
Pete Fahrenthold, risk manager for Continental Airlines Inc., accounts for bad weather as well. The airline's heaviest traffic occurs during hurricane season, so Continental monitors storms and alters operations accordingly if a storm approaches. As for forecasts, Fahrenthold says he understands that they're "best guesses based on a lot of moving pieces."
"We do consider the hurricane predictions in our planning," he says, "but we don't usually make significant changes in our operations based on the predictions because we are so accustomed to dealing with adverse weather conditions."
Hart Hubbard, assistant vice president, catastrophe services, at risk management services provider GAB Robins North America, says essentially the same for his operations: "We approach each new storm season with the same intensity regardless of the forecast."
But "dud" seasons do have an effect, even for wary risk managers. Those who renew their coverages near forecasting season, for instance, could see an impact on their premium.
"The underwriters are taking a look at what the forecasts are, and they know they pretty much take them as they come out," Salen says. "It's unfortunate because those forecasts, as this year's really proven to be the case, can be way out of whack."
"There is enough subjective capability relative to underwriting," he says. "That's where it falls. Depending on their current status, from their prior five years, their reinsurance charges and indications from the reinsurers to what problems or difficulties they're going to have on laying off their risk--then you tell me for an underwriter or the carrier to say it doesn't have any impact, that's just naīve to even believe that that's the case," Salen says.
It's the timing of a risk manager's renewal, Salen says, that will "tell the tale as to whether or not you're seeing an immediate impact relative to what the forecasts are showing."
Labor Finders' renewal, Salen says, is at the year's end, well after the worst of the hurricane season, and well before prognostications come out for the next season.
THE FORECASTERS' END
At a recent symposium on hurricane forecasting put on by Risk Management Solutions Inc., the modeling firm's head of research, Robert Muir-Wood, stated in his opening remarks that the event--where academics and insurers mixed over soda, coffee, fruit and snacks--should give academics the chance to see how much business activity focuses on what they do.
He didn't have to make the point to insurance buyers in the audience, who know how an active season can often raises rates and a quiet season can keep them stable.
Nor did he have to remind the insurers and reinsurers, who nearly all raised premiums based on the two previous brutal seasons and now found themselves this year with billions of dollars coming in, and little going out in claims.
Even for all the capital swirling around storm forecasts Muir-Woods also said he hoped that insurance buyers and sellers would realize just how much argument goes on in academia about hurricanes.
For such disagreement, just take the explanation for what went wrong with 2006 forecasts.
There is the standard line, as explained by Schroeder. Then there's Saunders, who heads the forecasting team at Tropical Storm Risk at University College London. He's at a loss to explain why the 2006 season fizzled.
Saunders has called the 2006 season a "mystery." He has crunched the numbers for humidity, for water temperature, for wind sheer, for El Nino, and none seemed to correlate statistically with what happened.
"Maybe there is a factor here that's not understood," he said at the RMS symposium. Or perhaps it is in the nature of forecasting that an off-year will happen.
"2006 I feel is atypical of years since 1950 and should not reflect badly on the capabilities of seasonal outlooks," Saunders also said, and blaming forecasters for getting the 2006 season wrong is unfair.
Forecasters merely look at current climate conditions that are known to have an impact on hurricanes--water temperature or wind conditions, for instance--and then guess what these factors will be like during the upcoming season and search through data from past decades for seasons with similar conditions to see were those seasons above average, below average or, like this year, just average for storm activity.
They crunch the statistics, tweak to taste, and produce their storm ranges. "It is just based on statistics," Schroeder says. "There's no sure thing."
They've been doing it for decades, but it still isn't easy. "It's a real challenge, basically because there is a lot of inherent atmospheric/oceanic variability that we still don't understand," says forecaster Phil Klotzbach, referring to predicting El Nino.
Klotzbach recently took over the primary responsibility for probability forecasts from Gray, who's been in the hurricane forecasting business since 1984.
Over the long run, though, they bat better than Ted Williams. From 1984 to 2005, Schroeder points out, Gray's May/June forecast for named storms has been within two of the actual count 13 out of 22 years--a 0.591 average.
The prognosticators are even more skilled for August forecasts. "In early August, the skills are quite good," Saunders said.
When their skills aren't quite so good, they often do take the blame. Trice at Crawford recalls a season past when Gray, after a particularly off projection, produced and distributed a cartoon of him and his staff eating platefuls of crow.
And at least the scientists try.
"I'm not pointing fingers at them." Salen says. "Hurricanes are like people."
But it seems to him that the scientists are getting better every year at figuring them out. "Somewhere down the road, the variables become less variable."
is associate editor at Risk & InsuranceŽ.
December 1, 2006
Copyright 2006© LRP Publications