Imagine…if Hurricane Sandy had been forecast to go out to sea, as might have been the case as recently as a decade ago…or un-forecast entirely. Picture hundreds of thousands of people along an unprepared coastline waking up to the sound of surf and wind, the sight of flooding, and the realization that evacuation routes were closed off by the high waters.
__________________________________
To matter, a forecast should meet five criteria: (1) it predicts a departure from the expected. (2) Though unusual, it inspires confidence; people act on it. (3) When they do, history changes. (4) The forecast verifies; it proves accurate. (5) It is based on reality: on observation of present conditions, and on insights about how things work.
This five-fold yardstick applies to forecasts of every stripe: predictions of stock-market trends, sporting event outcomes, consumer product preferences…political elections, and the weather.
Presidential campaigns? The weather? Both are in play over the coming few days. At this writing they are intertwining in a way that could well prove historic.
Back to those forecasts. If any one of the five ingredients for greatness is missing, the forecast falls short. Take weather. It is a cliché of meteorology that “persistence is the best forecast.” Do you want a weather prediction that’s accurate two-thirds of the time? Just predict that tomorrow’s weather will be like today’s. Trouble is, that’s not particularly useful. You and I care primarily about the exceptions to this general rule.
What about the second test? What if people pay no attention to the forecast, or it isn’t timely? If a prediction doesn’t change what people decide and then do…then it might as well not been issued.
The third test asks: what was riding on those weather-related decisions? Were the stakes high? Or ordinary? A tornado bearing down on Joplin matters more than the same tornado moving across open prairie. A hailstorm threatening Dallas-Fort Worth airport creates both more risk and greater opportunity for reducing loss than that same hailstorm over a wheat field. With a good forecast, pilots, air-traffic controllers, and airlines can distance the aircraft from harm’s way. By contrast, that wheat is going to stay put.
Fourth, did the forecast prove accurate? Were the people who took action glad they did?
And finally, were the forecasters just lucky? Or did they know what they were doing? Would they likely be equally successful the next time out?
The D-Day weather forecast passed all those tests. In June of 1944 the Allies were poised to invade Nazi-held Europe and bring an end to World War II. They’d been planning and preparing for years. They’d massed hundreds of thousands of troops in England, and assembled a fleet of 7000 vessels to ferry those soldiers across the Channel to Normandy. More than 10,000 aircraft would provide support. They needed a full moon for visibility and for the highest-possible tide to get the troops past German riprap guarding the coast. The first window of opportunity would be June 4-6.
But they also needed good weather. Conditions were stormy on June 4th and 5th. But even as the storm raged on June 5th, Allied weather forecasters saw the prospect of a several-hour lull on June 6th (criterion 1). Eisenhower gave the orders to go (criterion 2). We know what happened. The Germans had forecasters too. But on their side all they anticipated was three days of unrelieved gale. In fact, the German leader, Field Marshall Erwin Rommel, concluded (wrongly) that weather conditions were so unfavorable he could afford a short break from the front.
The Allied’s Operation Overlord was successful; it marked the beginning of the end of Hitler’s regime. As it happened, had the Allied forces decided to wait two weeks for the next tidal window, they’d have been confronted with what Winston Churchill described as the worst Channel storm in decades (criterion 3). The historic outcome was achieved primarily by Allied might and resources, but materially aided by accuracy and specificity of the Allied weather forecast and its use (criterion 4). The Allied forces made use of some special observations. For example, a key data set for developing the forecasts came from the British frigate HMS Grindall, which had been transmitting weather reports from the mid-Atlantic every three hours since April. Allied weather support of aviation and naval operations was based heavily on theoretical ideas and concepts developed by the so-called Bergen school of meteorology. It continued to be strong and rapidly improve throughout the war (criterion 5).
Fast forward to today. Hurricane Sandy meanders up the Atlantic coast of the U.S. Most hurricanes encounter westerly winds that steer the storm progressively further from shore and out into the open Atlantic. But computer models show Sandy doing something different, almost unique. As it moves north, it will encounter a blocking jet stream and a forming nor’easter. This is causing Hurricane Sandy to do a stutter step. It is scheduled to veer sharply toward the west, coming ashore somewhere on the south Jersey coast, between Monday and Tuesday morning, then slow, then turn north, and finally turn east only as it enters Canada (criterion 1). The outlook is for prodigious amounts of rain and high winds for perhaps two days, accompanied by extensive riverine flooding and coastal storm surge, at locations from Cape Hatteras to New England. Washington, Philadelphia, and New York City look to be hard hit.
Government forecasters and broadcast meteorologists are working in concert to get this message out. At the federal, state, and local level, emergency managers and public officials are building on the forecasts and working with private sector partners responsible for electrical power, communication, financial, and transportation infrastructure to discuss impacts and coordinate action. Millions of people across the Middle Atlantic States have stocked up on emergency supplies and hunkered down (criterion 2).
The timing could not be more portentous. The final days of a close, hotly-contested presidential election campaign, one held by both parties to be especially high-stakes. Coastal-Atlantic states are in play: Florida, Virginia, New Hampshire, and possibly others, depending upon your reading of the tea leaves. At a minimum, finely-tuned campaign schedules through these and other battleground states are being adjusted. The sitting president has cut back his campaign schedule in order to function as Emergency-Manager-in-Chief. His campaign will be carried on by surrogates. The president’s opponent finds himself needing to tread cautiously through this period. How both candidates handle these weather events will cement some voters’ views and change others’ minds. How many lives will be lost? What will be the extent of the property damage and business interruption? How extensive and how long-lasting will be any electrical power outages, and for whom? Could such disruption extend through Election Day itself and interfere with the actual voting? Will it look like the emergency response was adequate? Who will be praised, and who will be blamed? This is one of those campaigns where questions like these and every small success and every slight misstep will influence the final tallies and outcome…maybe change the course of American history for a decade or so. It’s this coincidence in timing that distinguishes Sandy from Snowmageddon or last summer’s derecho.
Consider a hypothetical; recall an historical analog. On 9/11/2001, then-President George W. Bush was caught without warning. When the news of the terrorist attack broke, he was reading My Pet Goat to a classroom of Florida school children. Over the next several hours, he and his advisors were flying over the country in Air Force One, in a state of agitation and confusion. Suppose he’d been a week away from re-election, versus at the start of his four-year term? In the same way, history is indeed being impacted by today’s events and their timing…and by the fact they were accurately forecast days in advance. (criterion 3).
That leaves us with criteria (4) and (5). Will the forecast verify? And has it been based on good observations and science?
Recall that this is a probabilistic forecast. It shows something like a 1-in-20 chance Sandy will blow out to sea. In this scenario success or failure might be difficult to judge. [But remember, as recently as ten years ago, that 5% probability might have been deemed the most likely outcome.] If this storm does proceed north for the next 24 hours and then veer dramatically onshore, anywhere between Long Island and Norfolk, the prediction does have to be seen as a scientific and technological triumph.
And much like that D-Day operation from 70 years ago, this success won’t have resulted just from the actions of the meteorological community over the span of a few short days. It’ll reflect a steady investment by the American people in Earth observations, science, and services over many years. Investments in satellite instruments and missions such as JPSS to monitor the weather from space and initialize the predictions. In networks of weather radars and surface sensors. In computers and the software needed to simulate the atmosphere. In research advancing our understanding of how that atmosphere works. Success will stem from years of partnership linking NOAA/National Weather Service leadership and staff with private-sector forecasters and broadcast meteorologists. And from partnerships extending far more broadly, joining such service providers with users: all those emergency managers and other state and local officials, and all those decision makers in private industry who need to work together if the wheels of the country are to continue turning. Partnerships that are making us a weather-ready nation.
Regrettably, that historically steady investment shows signs of fraying, under the pressures of the 2008 financial-sector meltdown, agency cost overruns, travel constraints, and the prospect of a sequester, as well as a possible end to all the tax cuts of recent years.
Just like a hurricane, that fiscal cliff also making landfall. Although it’s still a month or so out, its effects on unemployment are already being felt, much as surf and swell precede a hurricane’s arrival.
May we summon the grace to address these larger challenges with the same discipline, vigor, and common purpose with which we face a weather threat. May we remember the debt we owe to all those who labored in the past and who are working today to ensure that Hurricane Sandy is an October event…not an October surprise. May we thank all those emergency responders, utility linemen, and myriad others from both government and private sector who are working in the face of danger to keep the rest of us safe.
And may God have mercy on all in harm’s way.
Pingback: The most important weather forecast since D-Day? | Finger on the Weather | Wichita Eagle Blogs
The description of the forecasts for Overlord has passed into mythology, but it is incorrect. Schwerdtfeger’s memoirs of the event (published in Weather in 1986) go into detail. The German forecast was more accurate than the Allied forecast. Harry Butcher, a naval aide to Eisenhower, wrote about the US Navy’s experience on D-Day in his memoirs “they didn’t like the bloody wind, which didn’t go down as the weathermen had predicted.” WGF Jackson also discussed the problems with the weather in his book on the invasion. Losses before getting onto the landing ground were higher than anticipated.
As proof of the quality of the German forecasts, none of the forecasters involved in the forecasts were removed from their positions and, in fact, received medals and/or promotions in the two months that followed.
It could be argued that the D-Day forecasts are the best example of the quality-value paradox. Many other factors obviously went into the decision to invade, but, in effect, both sides acted as if their forecasts were correct. The Germans assumed no invasion would occur given the conditions. Rommel left Normandy for his wife’s birthday and the German response to the landings was slow. The Allies invaded, losing more troops to non-combat causes (waves, drownings) than they expected, but they got enough troops on shore to make the landings successful. If the German forecast had been less accurate, they might have been better prepared for the landings. If the Allied forecast had been more accurate, they might not have invaded.
Thanks, Harold. As you point out, there have been many takes on the history and “success” of this forecast over the years since. I despaired of doing thorough scholarship and so your points are especially welcome. Some readers might be interested to know that the European Center for Medium-range Weather Forecasting (ECMWF) did a reanalysis (and forecast) of the weather for D-Day; here’s the link:http://www.ecmwf.int/research/era/dday/
A quibble about “A tornado bearing down on Joplin matters more than the same tornado moving across open prairie.”
Forecasting can only tell us that conditions are favorable for tornado formation. Their creation and travel tends to be so fast that infrastructure informed by those forecasts, as well as the general weather sense we learn growing up in the area, is what warns people. Case in point, the 2011 Joplin tornado formed in a scant 30 seconds. But we had a lot of warning as these things go, although plenty of people didn’t pay (sufficient) attention to that. The hospital that got hit did pay attention and lost only one visitor and 4-5 of the people who were on respirators.
Moderate that with the observation on gaining weather sense by growing up in the area; most of us stepped outside after the warning, and many of us including me said, yeah, looks bad, head for shelter now. Which saved me from being pelted by rocks hurtled into my apartment from the northern edge of EF-5 core that passed around 500 feet to the south.
Oh, I’d add that there’s precious little “open prairie” around here. Farmers need warning as well, although I’ll bet there are holes in Doppler RADAR coverage for them.
Thanks, Harold…insightful comment, especially about that “open prairie.” Confess to having had a flicker of doubt about that when I wrote it, and contented myself with the idea that such prairie, though diminishing, still exists.
And thanks as well for the reality check about the time scales involved in tornado formation.
Pingback: “Superstorm” Sandy Devastates New York, New Jersey, Connecticut
Pingback: “Superstorm” Sandy Devastates New York, New Jersey, Connecticut - Conservative Read
Pingback: Hurricane Sandy: Landmark forecast. Good emergency response. Now for the hard part. | Living on the Real World