BAMS Paper

Intuitive Metric for Deadly Tropical Cyclone Rains

by

With hurricanes moving more slowly and climate models projecting increasing rain rates, scientists have been grappling with how to effectively convey the resulting danger of extreme rains from these more intense, slow-moving storms.

C_BosmaFlooding rainfall already is the most deadly hazard from tropical cyclones (TCs), which include hurricanes and tropical storms. Yet the widely recognized tool for conveying potential tropical cyclone destruction is the Saffir-Simpson Scale, which is based only on peak wind impacts. It categorizes hurricanes from 1, with winds causing minimal damage, to 5 and catastrophic wind damage. But it is unreliable for rain.

Recent research by Christopher Bosma, with the University of Wisconsin in Madison, and colleagues published in the Bulletin of the American Meteorological Society introduces a new tool that focuses exclusively on the deadly hazard of extreme rainfall in tropical cyclones. “Messaging the deadly water-related threat in hurricanes was a problem brought to light with Hurricanes Harvey and Florence,” says J. Marshall Shepherd (University of Georgia), one of the coauthors. “Our paper is offering a new approach to this critical topic using sound science methods.”

“One goal of this paper,” Bosma explains, “is to give various stakeholders—from meteorologists to emergency planners to the media—an easy-to-understand, but statistically meaningful way of talking about the frequency and magnitude of extreme rainfall events.”

That way is with their extreme rainfall multiplier (ERM), which frames the magnitude of rare extreme event rainfalls as multiples of baseline “heavy” rainstorms. Scientifically, ERM is the ratio of a specific location’s storm rainfall and the maximum amount of rain that has fallen most often at the location in two consecutive-year periods from 1981 through 2010—the baseline rain events that are relatively frequent at that location. A recurring baseline heavy rain amount is defined by the median (rather than the mean) annual maximum rainfall during the 30-year period and is used to weed out outlier events.

The authors are proposing the scale to

1. Accurately characterize the TC rainfall hazard;

2. Identify “locally extreme” events because local impacts increase with positive deviations from the local rainfall climatology;

3. Succinctly describe TC rainfall hazards at a range of time scales up to the lifetime of the storm system;

4. Be easy to understand and rooted in experiential processing to effectively communicate the hazard to the public.

Experiential processing is a term meaning rooted in experience, and ERM aims to relate its values for an extreme rainfall event to someone’s direct experience, or media reports and images, of heavy rainfall at their location. Doing this has the benefit of enabling them to connect, or “anchor” in cognitive psychology terms, the sheer magnitude of an extreme rain event to the area’s typical heavy rain events, highlighting how much worse it is.

Highest annual maximum ERMs (1948–2017) are indicated with colored markers and colored lines representing linear regression fit. A Mann–Kendall test for monotonic trends in annual maxima values did not reveal significant changes over time for either ERM or rainfall.
Highest annual maximum ERMs (1948–2017) are indicated with colored markers and colored lines representing linear regression fit. A Mann–Kendall test for monotonic trends in annual maxima values did not reveal significant changes over time for either ERM or rainfall.

 

The researchers analyzed 385 hurricanes and tropical storms that either struck or passed within 500 km of land from 1948 through 2012 and, through hindcasting, determined an average ERM of 2.0. Nineteen of the storms had ERMs greater than 4.0. And disastrous rain-making hurricanes in the record had ERMs directly calculated as benchmark storms. These include the most extreme event, Hurricane Harvey with an ERM of 6.4, Hurricane Florence as well as 1999’s Hurricane Floyd, which swamped the East Coast from North Carolina to New England, (ERMs: 5.7), and Hurricane Diane (ERM: 4.9), which destroyed large swaths of the Northeast United States with widespread flooding rains in 1955, ushering “in a building boom of flood control dams throughout New England,” says, coauthor Daniel Wright, Bosma’s advisor at UW-Madison.

Wright says that a major challenge in developing ERM was maintaining scientific accuracy while widening its use to non-meteorologists.

I’ve been reading and writing research papers for more than 10 years that were written for science and engineering audiences. This work was a little different because, while we wanted the science to be airtight, we needed to aim for a broader audience and needed to “keep it simple.”

In practice, these historical values of ERM would be used to convey the severity of the rainfall hazard from a landfalling storm. For example the authors successfully hindcast ERM values  in the Carolinas for Hurricane Florence, which inundated southeastern portions of North Carolina and northeastern South Carolina as it crawled ashore in 2018. With an active tropical storm or hurricane, the forecast value of ERM could be compared with historical hurricanes that have hit the expected landfall location.

Verification of the National Weather Service forecasts for the 3-day rainfall after landfall of Hurricane Florence (and ERM forecasts derived from these QPF estimates), issued at 1200 UTC 14 Sep 2018. Actual rainfall and 3-day ERM are based on poststorm CPC-Unified data.

Verification of NWS forecasts for the 3-day rainfall after landfall of Hurricane Florence (and ERM forecasts derived from these QPF estimates), issued at 1200 UTC 14 Sep 2018. Actual rainfall and 3-day ERM are based on poststorm CPC-Unified data.

 

In theory, the sound science is such that the ERM framework could be applied to other rain-producing storms.

“We think there is potential both for characterizing the spatial properties of all kinds of extreme rainstorms…and then also for examining how these properties are changing over time,” Wright says.

The researchers caution, however, that there are things that must be resolved before ERM can be used operationally as a communication tool. For example, ERM will need to be scaled to be compatible with NWS gridded rainfall products and generalized precipitation forecasts.  Forecast lead times and event durations also will need to be determined. And graphical displays and wording still need to be worked out to communicate ERM most effectively.

Nevertheless, the team argues:

…our Hurricane Florence ERM hindcast shows that the method can accurately characterize the rainfall hazard of a significant TC several days ahead in a way that can be readily communicated to, and interpreted by, the public.

D_Wright

Above, Daniel Wright, of the University of Wisconsin-Madison

When Hurricanes Become Machines…or Monsters

by

Officially, the Atlantic season is almost upon us. The season of tropical storms and hurricanes, yes, but more to the point, the season of heat-seeking machines and relentless monsters.

At least, that’s the metaphorical language of broadcast meteorologists when confronted with catastrophic threats like Hurricane Harvey in Houston in 2017. A new analysis in BAMS of the figures of speech used by KHOU-TV meteorologists to convey the dangers of this record storm shows how these risk communicators exercised great verbal skill to not only connect with viewers’ emotions, but also convey essential understanding in a time of urgent need.

For their recently released paper, Robert Prestley (Univ. of Kentucky) and co-authors selected from the CBS-affiliate’s live broadcasts during Harvey’s onslaught the more than six hours of on-air time for the station’s four meteorologists. The words the meteorologists used were coded and systematically analyzed and categorized in a partly automated, partly by-hand process. No mere “intermediaries” between weather service warnings and the public, the meteorologists—David Paul, Chita Craft, Brooks Garner, and Blake Matthews—relied on “figurative and intense language” on-air to “express their concern and disbelief” as well as explain risks.

As monster, the hurricane frequently displayed gargantuan appetite—for example, “just sitting and spinning and grabbing moisture from off the Gulf of Mexico and pulling it up,” in Paul’s words. The storm was reaching for its “food,” or moisture. The authors write, “The use of the term ‘feeder bands’…fed into this analogy.” Eventually Matthews straight out said, “We’re dealing with a monster” and Craft called the disaster a “beast.”

When the metaphor shifted to machines, Harvey was like a battery “recharging” with Gulf moisture and heat or a combustion engine tending to “blow” up or “explode.” Paul noted the lingering storm was “put in park with the engine revving.”

Other figurative language was prominent. Garner explained how atmospheric factors could “wring out that wet washcloth” and that the saturated ground was like “pudding putty, Jello.” The storm was often compared to a tall layered cake, which at one point Garner noted was tipped over like the Leaning Tower of Pisa.

In conveying impact risks, the KHOU team resorted frequently to words like “incredible” and “tremendous.” To create a frame of reference, they initially referred to local experience, like “Allison 2.0”—referring to the flood disaster caused by a “mere” tropical storm in 2001 that deluged the Houston area with three feet of rain—until Harvey was clearly beyond such a frame of reference. Then they clarified the unprecedented nature of threats, that it would be a storm “you can tell your kids about.”

The authors note, “By using figurative language to help viewers make sense of the storm, the meteorologists fulfilled the “storyteller” role that broadcast meteorologists often play during hurricanes. They were able to weave these explanations together with contextual information from their community in an unscripted, ‘off-the-cuff’ live broadcast environment.” They conclude that the KHOU team’s word choices could “be added to a lexicon of rhetorical language in broadcast meteorology” and serve as a “a toolkit of language strategies” for broadcast meteorologists to use in times of extreme weather.

Of course all of this colorful language was, perhaps, not just good science communication but also personal reality. Prestley et al. note: “The KHOU meteorologists also faced personal challenges, such as sleep deprivation, anxiety about the safety of their families, and the flooding of their studio. The flood eventually forced the meteorologists to broadcast out of a makeshift studio in a second-floor conference room before evacuating their building and going off air.”

As water entered the building, Matthews told viewers, “There are certain things in life you think you’ll never see. And then here it is. It’s happening right now.”

The new BAMS article is open access, now in early online release.

 

Active Hurricane Seasons: Maybe For 2020, But Not Necessarily in a Warmer Future

by

For a fifth consecutive year, NOAA is forecasting an above-average number of tropical cyclones (TCs) in the Atlantic, with 13-19 named storms expected in 2020. The number of TCs includes both tropical storms and hurricanes. This is in line with recent hurricane season forecasts by The Weather Channel, Penn State, Tropical Storm Risk, and others.

NOAA-2020-outlook

The recent spate of highly-active TC seasons, however, contrasts sharply with future trends in a majority of climate models, which simulate decreasing annual numbers of TCs as Earth’s climate continues to warm. That’s one of a number of findings in a recent paper by Tom Knutson (NOAA) and colleagues in the Bulletin of the American Meteorological Society.

In the paper, a team of tropical meteorology and hurricane experts led by Knutson assessed model projections of TCs in a world 2°C warmer than pre-industrial levels. The authors indicated mixed confidence in a downward TC frequency trend, even though 22 of 27 climate models the authors reviewed indicating the decrease. Some reputable models, though a minority, showed the frequency in named storms will instead increase in a warmer world, which lowered confidence in this particular finding.

As noted in Knutson et al. (2019, Part I of their two-part study: “Tropical Cyclones and Climate Change Assessment”), there is no clear observational evidence for a detectable human influence on historical global TC frequency. Therefore, there is no clear observational evidence to either support or refute the notion of decreased global TC frequency with climate warming. This apparent discrepancy between model projections and historical observations could be due to a number of factors. Among these are the relatively short available global TC records, the relatively modest expected sensitivity of global TC frequency to global warming since the 1970s, errors arising from limitations of model projections, differences between historical climate forcings and those used for twenty-first-century projections, or even observational limitations. However, the growing TC observational databases may soon provide a means of distinguishing between some highly divergent modeled scenarios of global TC frequency.

An average hurricane season in the Atlantic, which includes storms forming in the Caribbean Sea and Gulf of Mexico, sees 12 named storms with 6 becoming hurricanes. Of those hurricanes, typically three strengthen their sustained winds above 110 mph, becoming major hurricanes.

NOAA’s forecast cited warmer-than-usual sea surface temperatures, light winds aloft, and the lack of an El Niño, which tends to shear apart hurricanes, as factors for this year’s potentially active season. “Similar conditions have been producing more active seasons since the current high-activity era began in 1995,” NOAA stated in a release Thursday.

Knutson and his colleagues explain that the reason or reasons for a future decrease in TC frequency is uncertain, even as a warmer world would mean a continuation of warming seas. One possibility, the team entertains, is a decrease in large-scale rising air, termed “upward mass flux,” in the future. Its mechanism, however, is unclear, they find. Another is a reduction in saturation of the middle atmosphere in the models. Both are unfavorable for TC genesis.

The authors state that projections of TC frequency in different TC basins are “less robust” than the global signal. Comparing basins, they did find that the southwest Pacific and southern Indian oceans had greater TC decreases than the Atlantic and the Eastern and Western Pacific oceans.

They conclude this portion of the study stating that “reconciling projection results with theories or mechanistic understanding of TC genesis may eventually lead to improved confidence in projections of TC frequency.”

Knutson’s team found greater certainty in other facets of future TCs in the same study. For example, they expressed medium-to-high confidence that hurricanes will become stronger and wetter by the end of the twenty-first century.

New Assessment Is Confident Global Warming Brings Stronger, Wetter Tropical Cyclones

by

Even with a modest amount of global warming, future hurricanes will become nastier. They’ll push ashore higher storm surges, grow into superstorms like Hurricanes Dorian and Irma more often, and unleash inundating rains similar to Hurricanes Harvey and Florence more frequently.

That’s the assessment of published, peer-reviewed research in the past decade, according to an assessment by Thomas Knutson (NOAA) and colleagues, recently published in the Bulletin of the American Meteorological Society. It’s the second in a two part study conducted by the author team, 11 experts in climate and tropical cyclones (TCs). Part 1 found there are indeed already detectable changes in tropical cyclone activity attributable to human-caused climate change. Part 2, in the March 2020 BAMS online, project changes in the climatology of these storms worldwide due to human-induced global warming of just 2°C.

Highest confidence among the experts was in storm surge flooding. Rising sea levels due to warming and expanding oceans, responding to atmospheric warming and glacial ice melt, are already making it easier for hurricanes and even tropical storms to drive greater amounts of seawater ashore at landfall. And this will only worsen.

With CO2 levels climbing to about 414 ppm in March, as measured atop Mauna Loa in Hawaii, Earth is on track to reach a 2°C average global temperature increase by mid century. Already global average surface temperature has risen 1.2°C since the Industrial Revolution began.

In the assessment the authors have medium-to-high-confidence that rainfall rates in tropical cyclones will increase globally by 14% due to the increasing amount of water vapor available in a warmer atmosphere. They project a 5% global increase in tropical cyclone intensity along with an increase in the number of Category 4 and 5s ̶ although the range of opinions among the experts involved is 1-10%. In the Atlantic Basin, which includes the Caribbean Sea and Gulf of Mexico, the number of storms is projected to decrease while intensity as well as the number of intense hurricanes increases.

Other studies found that hurricanes will slow down, making them even more prolific rainmakers, among other changes. Authors of the new assessment discussed these additional changes, but cited less confidence in general and that different tropical basins around the world had different projections:

Author opinion was more mixed and confidence levels generally lower for some other TC projections, including a further poleward expansion of the latitude of maximum intensity of TCs in the western North Pacific basin, a decrease of global TC frequency, and an increase in the global frequency (as opposed to proportion) of very intense (category 4–5) TCs. The vast majority of modeling studies project decreasing global TC frequency (median of about −13% for 2°C of global warming), while a few studies project an increase. It is difficult to identify/quantify a robust consensus in projected changes in TC tracks across studies, although several project either poleward or eastward expansion of TC occurrence over the North Pacific. Projected TC size metric changes are on the order of 10% or less, and highly variable between basins and studies. Confidence in projections of TC translation speed is low due to the potential for data artifacts in the observed slowdown and a lack of model consensus. Confidence in various TC projections in general was lower at the individual basin scale than for the global average.

Summary of TC projections for a 2°C global anthropogenic warming. Shown for each basin and the globe are median and percentile ranges for projected percentage changes in TC frequency, category 4–5 TC frequency, TC intensity, and TC near-storm rain rate. For TC frequency, the 5th–95th-percentile range across published estimates is shown. For category 4–5, TC frequency, TC intensity, and TC near-storm rain rates the 10th–90th-percentile range is shown. Note the different vertical-axis scales for the combined TC frequency and category 4–5 frequency plot vs the combined TC intensity and TC rain rate plot. See the supplemental material for further details on underlying studies used.
Summary of TC projections for a 2°C global anthropogenic warming. Shown for each basin and the globe are median and percentile ranges for projected percentage changes in TC frequency, category 4–5 TC frequency, TC intensity, and TC near-storm rain rate. For TC frequency, the 5th–95th-percentile range across published estimates is shown. For category 4–5, TC frequency, TC intensity, and TC near-storm rain rates the 10th–90th-percentile range is shown.

Website Tracks Public Understanding of Tornadoes

by

Imagine you live in a part of the country where few people have experienced tornadoes. It would make sense that your neighbors wouldn’t know the difference between a tornado watch or warning, or know how to seek safety.

A new, openly available online tool shows exactly that, by combining societal databases with survey results about people’s understanding of weather information. But there are some surprising wrinkles in the data. For example, the database drills down to county-level information and finds “noteworthy differences” within regions of similar tornado climatology.

How is it that Norman, Oklahoma, residents score higher in what people think they know of severe weather information than those in Fort Worth, Texas? And why is there a similar gap between what people actually do know, as tested in Peachtree City, Georgia, versus Birmingham, Alabama?

“Differences like this create important opportunities for research and learning within the weather enterprise,” say Joseph T. Ripberger and colleagues, who describe the weather demographics tool in a recently published Bulletin of the American Meteorological Society article. “The online tool—the Severe Weather and Society Dashboard (WxDash)—is meant to provide this opportunity.”

For example, in one key set of metrics, the WxDash website looks at survey data on how well people receive and pay attention to tornado warnings (reception), how well they understand that information (both “subjective” comprehension—what people think they know—and “objective” comprehension—what they actually know), and response to tornado warnings.

From the BAMS article, a figure showing knowledge and response to average person percentile (APP) estimates of tornado warning reception, subjective comprehension, objective comprehension, and response by county warning area (CWA). The inset plots indicate the frequency distribution of APP estimates across CWAs. These estimates compare the average percentile of all adults who live in a CWA to the distribution of all adults across the country. For example, an APP estimate of 62 indicates that, on average, adults in that CWA score higher than 62% of adults nationally. The range of APP scores is wide. CWAs range from 38 to 61 on the reception scale, 32 to 69 on the subjective comprehension scale, and 37 to 60 on the objective comprehension scale. Response scores vary less. Not surprisingly, all categories broadly reflect the higher frequency of tornadoes in middle and southeastern CWAs.
From the BAMS article, a figure showing knowledge and response to average person percentile (APP) estimates of tornado warning reception, subjective comprehension, objective comprehension, and response by county warning area (CWA). The inset plots indicate the frequency distribution of APP estimates across CWAs. These estimates compare the average percentile of all adults who live in a CWA to the distribution of all adults across the country. For example, an APP estimate of 62 indicates that, on average, adults in that CWA score higher than 62% of adults nationally. The range of APP scores is wide. CWAs range from 38 to 61 on the reception scale, 32 to 69 on the subjective comprehension scale, and 37 to 60 on the objective comprehension scale. Response scores vary less. Not surprisingly, all categories broadly reflect the higher frequency of tornadoes in middle and southeastern CWAs.

 

WxDash combines U.S. Census data with an annual Severe Weather and Society Survey (Wx Survey) by the University of Oklahoma Center for Risk and Crisis Management. The database then “downscales” the broader scale information to the local level, in a demographic equivalent to the way large scale climate models downscale to useful information on regional scales.

The site also provides information on public trust in weather information sources, perceptions about the efficacy of protective action, vulnerability to beliefs about a variety of tornado myths, and other weather-related factors that can then be studied in light of regional and demographic factors.

Some of the key findings seen in the database:

  • Men and women demonstrate roughly comparable levels of reception, objective comprehension, and response, but men have more confidence in subjective warning comprehension than women.
  • Tornado climatology has a relatively strong effect on tornado warning reception and comprehension, but little effect on warning response.
  • The findings suggest that geography, and the community differences that overlap with geographic boundaries, likely exert more direct influence on warning reception and comprehension than on response.

Even the relatively expected relation of severe weather climatology to severe weather understanding is problematic, Ripberger and colleagues write.

Tornadoes are possible almost everywhere in the US and people who live on the coasts can move—both temporarily and permanently— throughout the country. These factors prompt some concern about the low levels of reception and comprehension in some communities, especially those in the west.

In addition to interacting with these data, you can download one of the calculated databases for community-scale information, the raw survey data, and the code necessary to reproduce the calculations.

The idea is social scientists can dig in and figure out why what we know about weather isn’t nearly as closely correlated with what we experience as we might think. The hope is an improvement in public education and risk communication strategies related to severe weather.

Japan’s “Gosetsu Chitai” (Heavy Snow Area) Illuminates Sea- and Lake-effect Precip Processes

by

Snow WallNorth American meteorologists, welcome to the snow climate of western Japan. Every year in winter lake effect-like snow events bury coastal cities in northern and central Japan under 20-30 feet of snow. Above is the “snow corridor” experienced each spring when the Tateyama Kurobe Alpine Route through the Hida Mountains reopens, revealing the season’s snows in its towering walls. The Hida Mountains, where upwards of 512 inches of snow on average accumulates each winter, are known as the northern Japanese Alps.

The tremendous snow accumulations largely occur from December to February during the East Asian winter monsoon when sea-effect snowbands form behind frequent cold outbreaks. But their snowfall isn’t just pretty to look at and play in — extreme snowfalls combined with dense populations in cities adjacent to the Sea of Japan such as Sapporo (pop. 1.95 million) are public safety hazards, turning exceptionally deadly every year. On average 100 people die and four times that number are injured from snow and ice in Japan, not only from snow removal but also from “roofalanches” — masses of snow sliding off roofs onto people.

Similar to their counterparts downwind of North America’s Great Lakes, the Sea of Japan snowbands invite research from Japanese scientists and those in many other locales where bodies of water enhance snowfall over populated lands. A new paper in BAMS by Jim Steenburgh (University of Utah) et al. not only highlights what’s known about the Japanese snow events but also is designed to “stimulate increased collaborations between sea- and lake-effect researchers and forecasters in North America, Japan, East Asia, and other regions of the world” who can collectively realize the “significant potential to advance our understanding and prediction of sea- and lake-effect precipitation.”

"Decision-making under meteorological uncertainty" for D-Day's Famous Forecast

by

The success of the D-Day Invasion of Normandy was due in part to one of history’s most famous weather forecasts, but new research shows this scientific success resulted more from luck than skill. Oft-neglected historical documentation, including audio files of top-secret phone calls, shows the forecasters were experiencing a situation still researched and practiced today: “decision-making under meteorological uncertainty.”
New research recently published in BAMS into that weather forecast for June 6, 1944, which enabled the Allies in World War II to gain a foothold in Europe, answers questions about three popular perceptions: were the forecasts, which predicted a break in the weather, that good? were the German meteorologists so ill-informed, missing that weather-break? and was the American analog system for prediction so great and better than what the Germans had?
The “alleged” weather break
An expected ridge and fair weather between two areas of low pressure, one departing and one arriving over the area, didn’t materialize. The departing low instead lingered and created a lull in visibility and lifted the cloud ceiling height, but it didn’t slow winds much. They blew at Force 4-5 (~13-24 mph), creating very choppy seas that sickened many troops prior to the invasion.

Synoptic analyses at 00 UTC from 5 to 8 June 1944. The low that was supposed to move northeast to southern Norway remained over the North Sea for some days. On 6 and 8 June the observed winds in the Channel were force 4 and occasionally force 5.
Synoptic analyses at 00 UTC from June 5-8, 1944. The low that was supposed to move northeast to southern Norway remained over the North Sea for some days.

 
A blown German Forecast?
Because the invasion came as a complete surprise to the Germans it has been surmised their weather forecast for June 6 had to be bad. German forecasters prior to the war were the best at “extended” forecasts, and their synoptic maps and forecast for that day were more realistic than the Allies, with a less optimistic speculation of any break in the weather.
The German's European-Atlantic map at 00 UTC June 6, 1944, where the analysis over the North Atlantic appears not to be based on observations but intercepted American coded analyses.
The German’s European-Atlantic map at 00 UTC June 6, 1944, where the analysis over the North Atlantic appears not to be based on observations but intercepted American coded analyses.

 
A historically debated forecast
The analog weather prediction system employed by the Allies for the invasion was claimed by its creators to have correctly identified the weather break. But historical analysis and review doesn’t bear this out. What it does find, though, is that the system correctly identified a transition from zonal to meridional flow, which delivered the break the Allies needed for success. History’s finding: The forecast was “Overoptimistic.”
The 1984 Fort Ord meeting about the D-Day forecast got coverage in the local Monterey newspapers. The invasion was said to have occurred in a "break" or a period of a "brief lull" in the weather.
The 1984 Fort Ord, California, AMS meeting about the D-Day forecast got coverage in the local Monterey newspapers. The American forecasting group was led by Lt. Col. (Dr.) Irving Krick of Caltech. The president of the Naval Post Graduate School, Robert Allen, Jr., at the time an Air Force officer conducting high-level weather briefings at the Pentagon, also spoke at the meeting.

 
As a lesson learned from this most famous of weather forecasts, the paper’s author, Anders Persson of Swedin’s Uppsala University, concludes:

It was 75[+] years ago and the observational coverage has improved tremendously since then, both qualitatively and quantitatively. Our understanding of the atmosphere is much better,and the forecast methods have reached a standard that could hardly have been dreamt of in 1944. However, there’s one element that has a familiar ring to it and is of great interest today. That is when Air Marshall Tedder [Deputy Supreme Commander of the Invasion under General Eisenhower] asks about an assessment of the confidence in the forecast he has just heard … This illustrates that the D-day forecast is a significant early example of decision-making under meteorological uncertainty.

“Decision-making under meteorological uncertainty” for D-Day’s Famous Forecast

by

The success of the D-Day Invasion of Normandy was due in part to one of history’s most famous weather forecasts, but new research shows this scientific success resulted more from luck than skill. Oft-neglected historical documentation, including audio files of top-secret phone calls, shows the forecasters were experiencing a situation still researched and practiced today: “decision-making under meteorological uncertainty.”

New research recently published in BAMS into that weather forecast for June 6, 1944, which enabled the Allies in World War II to gain a foothold in Europe, answers questions about three popular perceptions: were the forecasts, which predicted a break in the weather, that good? were the German meteorologists so ill-informed, missing that weather-break? and was the American analog system for prediction so great and better than what the Germans had?

The “alleged” weather break

An expected ridge and fair weather between two areas of low pressure, one departing and one arriving over the area, didn’t materialize. The departing low instead lingered and created a lull in visibility and lifted the cloud ceiling height, but it didn’t slow winds much. They blew at Force 4-5 (~13-24 mph), creating very choppy seas that sickened many troops prior to the invasion.

Synoptic analyses at 00 UTC from 5 to 8 June 1944. The low that was supposed to move northeast to southern Norway remained over the North Sea for some days. On 6 and 8 June the observed winds in the Channel were force 4 and occasionally force 5.
Synoptic analyses at 00 UTC from June 5-8, 1944. The low that was supposed to move northeast to southern Norway remained over the North Sea for some days.

 

A blown German Forecast?

Because the invasion came as a complete surprise to the Germans it has been surmised their weather forecast for June 6 had to be bad. German forecasters prior to the war were the best at “extended” forecasts, and their synoptic maps and forecast for that day were more realistic than the Allies, with a less optimistic speculation of any break in the weather.

The German's European-Atlantic map at 00 UTC June 6, 1944, where the analysis over the North Atlantic appears not to be based on observations but intercepted American coded analyses.
The German’s European-Atlantic map at 00 UTC June 6, 1944, where the analysis over the North Atlantic appears not to be based on observations but intercepted American coded analyses.

 

A historically debated forecast

The analog weather prediction system employed by the Allies for the invasion was claimed by its creators to have correctly identified the weather break. But historical analysis and review doesn’t bear this out. What it does find, though, is that the system correctly identified a transition from zonal to meridional flow, which delivered the break the Allies needed for success. History’s finding: The forecast was “Overoptimistic.”

The 1984 Fort Ord meeting about the D-Day forecast got coverage in the local Monterey newspapers. The invasion was said to have occurred in a "break" or a period of a "brief lull" in the weather.
The 1984 Fort Ord, California, AMS meeting about the D-Day forecast got coverage in the local Monterey newspapers. The American forecasting group was led by Lt. Col. (Dr.) Irving Krick of Caltech. The president of the Naval Post Graduate School, Robert Allen, Jr., at the time an Air Force officer conducting high-level weather briefings at the Pentagon, also spoke at the meeting.

 

As a lesson learned from this most famous of weather forecasts, the paper’s author, Anders Persson of Swedin’s Uppsala University, concludes:

It was 75[+] years ago and the observational coverage has improved tremendously since then, both qualitatively and quantitatively. Our understanding of the atmosphere is much better,and the forecast methods have reached a standard that could hardly have been dreamt of in 1944. However, there’s one element that has a familiar ring to it and is of great interest today. That is when Air Marshall Tedder [Deputy Supreme Commander of the Invasion under General Eisenhower] asks about an assessment of the confidence in the forecast he has just heard … This illustrates that the D-day forecast is a significant early example of decision-making under meteorological uncertainty.

Snowflake Selfies as Meteo Teaching Tools

by

Undergrads at Penn State recently took to their cellphones to mingle with and snap pics of tiny snowflakes to reinforce meteorological concepts. The class, called “Snowflake Selfies” and described in a new paper in BAMS, was designed to use low-cost, low-tech methods that can be widely adapted at other institutions to engage students in hands-on field research.

In addition to photographing snow crystals, students measured snowfall amounts and snow-to-liquid ratios, and then gained meteorological insight into the observations using radar data and thermodynamic soundings. The goal of the course was to reinforce concepts from their other undergraduate meteorology courses, such as atmospheric thermodynamics, cloud physics, and radar and mesoscale meteorology.

As a writing intensive course at Penn State that meets the communication skills requirement of the AMS guidance for a Bachelor’s Degree in Atmospheric Science, “Snowflake Selfies” also was designed to help students communicate meteorological science. Students shared their observations with the local National Weather Service office in State College and also wrote up their work in term papers and presented their pics and findings to the class.

Snow crystal photographs taken by students in the "Snowflake Selfies" class.
Snow crystal photos taken by students in the “Snowflake Selfies” class.

 

Of course to have such a class, you need snow, and “the relative lack of snowfall events during the observational period” in winter 2018 was definitively a challenge for students, the BAMS paper states. Pennsylvania’s long winters often see many opportunities to photograph snow, but the course creators caution that perhaps a longer observational period is needed in case nature doesn’t cooperate. It also would allow students enough time to closely observe snowflakes while juggling their other classes and activities.

A survey conducted at the end of the class found that “Snowflake Selfies” was well received by students, engaging them and encouraging their introduction to field science. And they “strongly agreed [it] helped reinforce their understanding of cloud physics and physical meteorology compared to” a previous such course where students designed, built, and deployed their own 3-D printed rain gauges to measure precipitation.

Actually, that previous course sounds like a lot of fun, too!

Observations without Fear: NOAA's Drones for Hurricane Hunting

by

Nowhere is it more dangerous to fly in a hurricane than right near the roiling surface of the ocean. These days, hurricane hunting aircraft wisely steer clear of this boundary layer, but as a result observations at the bottom of the atmosphere where we experience storms are scarce. Enter the one kind of plane that’s fearless about filling this observation gap: the drone.
NOAA’s hurricane hunter aircraft in recent storms has been experimenting with launching small unmanned aircraft systems (sUAS) into raging storms–and these devices show promise for informing advisories as well as improving numerical modeling.

Lead author Joe Cione of NOAA's hurricane research division holds a Coyote sUAS. The drones are being launched into hurricanes from the P-3 hurricane hunter aircraft in the background.
Lead author of a new paper in BAMS, Joe Cione of NOAA’s Hurricane Research Division, holds a Coyote sUAS. The drones are being launched into hurricanes from the WP-3D Orion hurricane hunter aircraft in the background.

 
The observations were made by a new type of sUAS, described in a recently published paper in BAMS, called the Coyote that flew below 1 km in hurricanes. Sampling winds, temperature, and humidity in this so-called planetary boundary layer (PBL), the expendable Coyotes flew as low as 136 m in wind speeds as high as 87 m s-1 (196 mph) and for as long as 40 minutes before crashing (as intended) into the ocean.
In the BAMS article, Joe Cione at al. describe the value of and uses for the low-level hurricane observations:

Such high-resolution measurements of winds and thermodynamic properties in strong hurricanes are rare below 2-km altitude and can provide insight into processes that influence hurricane intensity and intensity change. For example, these observations—collected in real time—can be used to quantify air-sea fluxes of latent and sensible heat, and momentum, which have uncertain values but are a key to hurricane maximum intensity and intensification rate.

Highs-lows
Coyote was first deployed successfully in Hurricane Edouard (2014) from NOAA’s WP-3 Orion hurricane hunter aircraft. Recent Coyote sUAS deployments in Hurricanes Maria (2017) and Michael (2018) include the first direct measurements of turbulence properties at low levels (below 150 m) in a hurricane eyewall. In some instances the data, relayed in near real-time, were noted in National Hurricane Center advisories.
Turbulence processes in the PBL are also important for hurricane structure and intensification. Data collected by the Coyote can be used to evaluate hurricane forecasting tools, such as NOAA’s Hurricane Weather Research and Forecasting (HWRF) system. sUAS platforms offer a unique opportunity to collect additional measurements within hurricanes that are needed to improve physical PBL parameterization.

Coyote launch sequence: (a) Release in a sonobuoy canister from a NOAA P-3. (b) A parachute slows descent. (c) The canister falls away and the Coyote wings and stabilizers deploy. The main wings and vertical stabilizers have no control surfaces; rather, elevons (i.e., combined elevator and aileron) are on the rear wings, controlled by the GPS-guided Piccolo autopilot system with internal accelerometers and gyros. (d) After the Coyote is in an operational configuration, the parachute releases. (e) The Coyote levels out after starting the electric pusher motor, which leaves minimally disturbed air for sampling at the nose. The cruising airspeed is 28 m s-1. (f) The Coyote attains level flight and begins operations. When deployed, the Coyote’s wingspan is 1.5 m and its length is 0.9 m. The 6-kg sUAS can carry up to 1.8 kg. Images were captured from a video courtesy of Raytheon Corporation.
Coyote launch sequence: (a) Release in a sonobuoy canister from a NOAA P-3. (b) A parachute slows descent. (c) The canister falls away and the Coyote wings and stabilizers deploy. The main wings and vertical stabilizers have no control surfaces; rather, elevons (i.e., combined elevator and aileron) are on the rear wings, controlled by the GPS-guided Piccolo autopilot system with internal accelerometers and gyros. (d) After the Coyote is in an operational configuration, the parachute releases. (e) The Coyote levels out after starting the electric pusher motor, which leaves minimally disturbed air for sampling at the nose. The cruising airspeed is 28 m s-1. (f) The Coyote attains level flight and begins operations. When deployed, the Coyote’s wingspan is 1.5 m and its length is 0.9 m. The 6-kg sUAS can carry up to 1.8 kg.
Images were captured from a video courtesy of Raytheon Corporation.

 
The authors write that during some flights instrument challenges occurred. For example:

thermodynamic data were unusable for roughly half of the missions. Because the aircraft are not recovered following each flight, the causes of these issues are unknown. New, improved instrument packages will include a multi-hole turbulence probe, improved thermodynamic and infrared sensors, and a laser or radar altimeter system to provide information on ocean waves and to more accurately measure the aircraft altitude.

Future uses of the sUAS could include targeting hurricane regions for observations where direct measurements are rare and models produce large uncertainty. Meanwhile, the article concludes, efforts are underway to increase sUAS payload capacity, battery life, and transmission range so that the NOAA P-3 need not loiter nearby.