Was 2020 a Record-Breaking Hurricane Season? Yes, But. . .

Posted on Updated on

Chris Landsea and Eric Blake [1]

An Incredibly Busy Hurricane Season

The 2020 Atlantic hurricane season was extremely active and destructive with 30 named storms.  (The Hurricane Specialists here at the National Hurricane Center use the designation “named storms” to refer to tropical storms, subtropical storms, hurricanes, and major hurricanes.)  We even reached into the Greek alphabet for names for just the second time ever.  The United States was affected by a record 13 named storms (six of them directly impacted Louisiana), and a record yearly total of 7 billion-dollar tropical cyclone damage events was recorded by the National Centers for Environmental Information (https://www.ncdc.noaa.gov/billions/time-series/US).  Nearly every country surrounding the Gulf of Mexico, Caribbean Sea, and tropical/subtropical North Atlantic was threatened or struck in 2020.  Total damage in the United States was around $42 billion with over 240 lives lost in the United States and our neighboring countries in the Caribbean and Central America.

Track map of all 30 named storms during the 2020 Atlantic hurricane season.

The 30 named storms in 2020 sets a record going back to the 1870s when the U.S. Signal Service (a predecessor to the National Weather Service) began tracking tropical storms and hurricanes.  The only year that comes close is 2005 with 28 named storms.  It’s also apparent that a very large increase has occurred in the number of observed named storms from an average of 7 to 10 a year in the late 1800s to an average of 15 to 18 a year in the last decade or so – a doubling in the observed numbers over a century!  (The black curve in the figure below represents a smoothed representation of the data that filters out the year to year variability in order to focus on time scales of a decade or more).

Number of combined tropical storms, subtropical storms, and hurricanes each year from 1878 to 2020.

However, the number of named storms is only one measure of the overall measure of a season’s activity.  And indeed, for the 2020 season, other measures of Atlantic tropical storm and hurricane activity were not record breaking.  For example, the number of hurricanes (14) was well above average, but fell short of the previous record of 15 hurricanes that occurred in 2005. 

For overall monitoring of tropical storm and hurricane activity, tropical meteorologists prefer a metric that combines how strong the peak winds reached in a tropical cyclone, and how long they lasted – called Accumulated Cyclone Energy or ACE[2].   By this measure, 2020 was extremely busy, but not even close to record breaking.  In fact, with a total ACE of 180 units, 2020 was only the 13th busiest season on record since 1878 with seasons like 1893, 1933, 1950, and 2005 substantially more active than 2020.  One can also see that while there is a long-term increase in recorded ACE since the late 1800s, it’s quite a bit less dramatic than the increase seen with named storms.  There also is a pronounced busier/quieter multi-decadal (40- to 60-year) cycle with active conditions in the 1870s to 1890s, late 1920s to 1960s, and again from the mid-1990s onward.  Conversely, quiet conditions occurred in the 1900s to early 1920s and 1970s to early 1990s.

Accumulated Cyclone Energy (ACE), a measure which combines the number, intensity, and duration of tropical storms and hurricanes, each year from 1878 to 2020.

Technology Change and Named Storms

So why would the record for named storms be broken in 2020, while the overall activity as measured by ACE is not even be close to setting a record?                               

The answer is very likely technology change, rather than climate change.  Today we have many advanced tools to help monitor tropical and subtropical cyclones across the entire Atlantic basin such as geostationary and low-earth orbiting satellite imagery, the Hurricane Hunter aircraft of the U.S. Air Force Reserve and National Oceanic and Atmospheric Administration (NOAA), coastal weather radars, and scatterometers (radars in space that provide surface wind measurements).  In addition, the instrumentation and measuring techniques used by the satellites, aircraft and radars are continually improving.  These technological advances allow us at the National Hurricane Center to better identify, track, and forecast tropical and subtropical cyclones with an accuracy and precision never before available.  This is great news for coastal residents and mariners, since these tools help us provide the best possible forecasts and warnings to aid in the best preparedness for these life-threatening systems.

Such technology, though, was not available back at the advent of the U.S. Signal Service’s tropical monitoring in the 1870s.  Without these sophisticated tools, meteorologists in earlier times not only had difficulty in forecasting tropical cyclones, but they also struggled in even knowing if a system existed over the open ocean.  In the late 19th and early 20th Centuries, the only resource hurricane forecasters could use to monitor tropical cyclones were weather station observations provided via telegraph.  Such an approach is problematic for observing – much less forecasting – tropical cyclones that develop and spend most of their lifecycle over the open ocean.  Here’s a timeline of critical technologies that have dramatically improved tropical meteorologists’ ability to “see” and monitor tropical cyclones:

Technological improvements for monitoring tropical storm and hurricanes between 1878 and 2018.

The upshot of all of these advances in the last century is much better identification of the existence of tropical cyclones and their strongest winds (or what meteorologists call “Intensity”).  So, the further one goes back in time, the more tropical cyclones (and portions of their life cycle) were missed, even for systems that may have been a major hurricane. This holds for both counting named storms back in time as well as integrated measures like ACE.  Our database is incomplete and has – as statisticians would say – a severe undersampling bias that is much more prominent earlier in the record.  HURDAT2 – our Atlantic hurricane database – is an extremely helpful record which is a “by-product” of NHC’s forecasting operations, but it is very deficient for determining real long-term trends.  (It’s important to point out that many data entries in HURDAT2 for intensity and even the position of the named storms are educated guesses as opposed to being based on observations before the 1970s advent of regular satellite imagery). To be able to examine questions about any impact from man-made global warming (aka climate change) on long-term changes in the number of named storms, for example, one must first account for the massive technology change over the last century.

Fortunately, to help address this issue, researchers at NOAA’s Geophysical Fluid Dynamics Laboratory (GFDL) – (Gabe Vecchi and Tom Knutson in 2008’s Journal of Climate) have invented a way to estimate how many named storms were missed in the pre-geostationary satellite era (before the 1970s).  This was done by comparing the population of tracks and sizes of named storms that have occurred versus the density of observations from ships that were traversing the ocean.  If there were ships everywhere all of the time back to the 1870s (and these ships didn’t try to avoid running into tropical cyclones, which they certainly did), there would be very few named storms unaccounted for.  But the reality is that much of the Atlantic Ocean, Gulf of Mexico, and Caribbean Sea was sparsely traversed by ships from the late 19th Century until the middle of the 20th Century.  (The plots below indicate the amount of shipping traffic and weather observations from those ships – Orange/Red are numerous, Green/Yellow are moderate, Gray are few, and White are no measurements).

Plots showing the density of shipping traffic across the northern Atlantic Ocean between 1878-1914, 1915-1945, and 1946-1965. White and blue areas indicate little to no ship traffic, while oranges and red indicate a high level of ship traffic.

In addition to the issue of named storms that were previously missed, due to the lack of ability to observe them, technological improvements also have effectively allowed the standards for naming a storm to be refined resulting in better identification of weak (near the 39-mph/63-kph threshold) systems.   Tropical warnings for many of the weak, short-lived named storms in past eras were not issued, and thus these systems were not automatically included into the HURDAT2 database.  In the cases when forecasters in earlier years were either 1) not sure that the system possessed the required 39-mph/63-kph winds, 2) assumed that it would be too short in duration, or 3) thought that the system was non-tropical (i.e., with a warm to cold gradient of temperature across the system’s center), they usually did not issue named storm advisories, and therefore these systems did not get added into the historical database[3].

In research that the lead author had investigated (Chris Landsea and company in 2010’s Journal of Climate), we discovered that weak, short-lived (lasting less than or equal to two days) named storms – aka “Shorties” – had shown a dramatic increase in occurrence over time.  There were only about one a year in HURDAT2 up until the 1920s, about 3 per year from the 1930s to the 1990s, and jumping up to around 5 per year since 2000.

Number of tropical storms and subtropical storm “Shorties,” those which had a duration of 2 days or less, each year from 1878 to 2020.

Of the 30 named storms in 2020, seven were Shorties and a few more were just longer than two days in duration.  Of these seven Shorties, four are very unlikely to have been “named” before around 2000:  Dolly, Edouard, Omar, and Alpha.  (Of the remaining Shorties, Bertha and Kyle may have been named, while Fay likely would have been named).  These and other weak, short-lived systems since 2000 have been observed and recognized as tropical storms due to new tools available to forecasters including scatterometers, Advanced Microwave Sounding Units, the Advanced Dvorak Technique, and the Cyclone Phase Space diagrams.   The Hurricane Specialists here at the National Hurricane Center then are able to issue advisories on these named storms in real-time and then include them into the HURDAT2 database at the end of the season.

Examples of four “Shorties” in 2020 that were very unlikely to have been designated as named storms in the past.

From a warning perspective for mariners and coastal residents, it is very beneficial that the National Hurricane Center is now naming (and recording) these Shorties.  But without accounting for how technology affects our records, one can come to some unfounded conclusions about true long-term changes in named storm activity.  In addition, it is worth pointing out, but perhaps not too surprising, that it has been shown by the researchers at Princeton University and at GFDL (Villarini et al. 2011, Journal of Geophysical Research) that the observed increase in Shorties has no association with any environmental factor known to influence named storms including man-made global warming.  It is therefore reasonable to conclude that the dramatic increase in the number of these Shorties is simply due to better observational technology.

An “Apples-to-Apples” Comparison of the 2020 Long-Lived Named Storms with the Past

So how can we come up then with a more apples-to-apples comparison of how the number of named storms has actually changed over the last 100 years plus?  Here are the steps that were performed in the 2010 Journal of Climate paper, about Shorties, updated for data through the 2020 hurricane season:

(1) Start with the original HURDAT2 database of named storms from 1878 onward:

Number of combined tropical storms, subtropical storms, and hurricanes each year from 1878 to 2020.

(2) Remove all of the Shorties from the original database, leaving just the long-lived named storms:

Number of combined long-lived (more than 2 days) tropical storms, subtropical storms, and hurricanes each year from 1878 to 2020.

(3) Add in the best estimate of the number of missed long-lived named storms before geostationary satellite imagery and the Dvorak technique became available:

Number of combined long-lived (more than 2 days) tropical storms, subtropical storms, and hurricanes each year from 1878 to 2020, adjusted by adding “missed” systems.

The resulting final time series shows tremendous variability, with highest values of 23 in 2020 and 20 in 1887 and 2005, and lowest values of 2 in 1914, and 3 in 1925, 1982, and 1994.  Overall, there remains a modest upward trend in the database over the entire time series superimposed with quasi-cyclic variations seen in the ACE data as was discussed earlier: higher activity in the late 1800s, mid-1900s, and from the mid-1990s onward, but lower activity in the early 1900s, and in the 1970s to early 1990s. These cycles of higher and lower activity have been linked to a natural phenomenon called the Atlantic Multidecadal Oscillation (AMO) (see paper by Stan Goldenberg, Chris Landsea, and colleagues in 2001’s Science).  Recent controversial research, however, is calling into question whether the AMO actually exists (see paper by Michael Mann and company in 2021’s Science).  Regardless of the validity of the AMO, the bottom line is that the doubling in the number of named storms over a century is very likely due to technology change, not natural or man-made climate change.

(4) And finally, add in the uncertainty to these estimates with the reasonable largest number of missed long-lived named storms. This represents the 95% method uncertainty value, or in layman’s terms, the largest reasonable number of missed systems.

Highest reasonable number of combined long-lived (more than 2 days) tropical storms, subtropical storms, and hurricane each year from 1878 to 2020, adjusted by adding a high estimate of “missed” systems.

Note that after adding on the uncertainty to the missed number of long-lived named storms (blue coloring), we can conclude that 1887 and 2020 may be just as busy for the number of long-lived named storms. 

The New “Normal” for Named Storm Numbers

With the completion of the 2011 to 2020 decade, climatologists are updating records to provide a new “normal” (or average) to compare against new weather.  The previous 30-year based climate period to decide if a weather event or season was unusual or expected was 1981-2010.  For weather phenomena around the world, we’re now changing the years to compute normal conditions to 1991-2020.  (The 30-year normal concept is designed to provide a long enough time period to obtain relatively stable statistics, and to also have the time period reflect the most recent weather experienced over a human generation.  Thirty years is a good compromise between these two aspects.) It might seem odd to non-meteorologists to change the definition of “average” every ten years, but meteorologists/climatologists do so because climate is never stationary, i.e., the climate is always changing.  The climate has both natural variations (like El Niño/La Niña, effects from volcanic eruptions, and the Atlantic Multidecadal Oscillation) and man-made changes (like urban heat island, land use changes, and greenhouse gas emissions) that affect what’s been observed around the last three decades.  These revisions of new averages are done around the world in conjunction with the World Meteorological Organization.  Thus NOAA is updating the average of temperature, precipitation, and other meteorological parameters to reflect what has been observed.

This shift in the period used for the 30-year climate standard changes the definitions of average (or “normal”) levels of tropical cyclone activity to the following for the Atlantic Ocean, Caribbean Sea, and Gulf of Mexico (see this report by NOAA for more details):

System TypeOld 1981-2010 AverageNew 1991-2020 Average
Named Storms1214
Major Hurricanes33
Comparison of the number of named storms, hurricanes, and major hurricanes in the Atlantic basin using the old 30-year (1981-2020) averaging period with the new 30-year (1991-2020) averaging period.


These changes, therefore, reflect that most of the new 1991-2020 climatology period is within an active period that began in 1995 and includes the impact of the technology changes discussed above that have led to the National Hurricane Center more accurately diagnosing and naming more systems in the last couple of decades.    

Take Aways

The answers and conclusions to “Was 2020 a Record-Breaking Hurricane Season? Yes, but…”:

  • Doubling in the number of named storms over a century is very likely due to technology change, not natural or man-made climate change;
  • 2020 set a record for number of named storms, but given the limitations in our records it is possible that other years (such as 1887) were just as active for long-lived named storms; and
  • The boost in average or “normal” conditions from 12 to 14 named storms is due to a combination of a busy era that began in 1995 as well as the ability of the National Hurricane Center to observe and accurately diagnose more weak, short-lived named storms than had been done previously, mostly due to technology advancements.

A follow-on blog post, putting these observed changes of the number of named storms into context of what may be expected to occur in the future, is expected to be published in the near future.

[1] Christopher W. Landsea is the Chief of the Tropical Analysis and Forecast Branch at the National Weather Service’s National Hurricane Center in Miami, Florida. Eric Blake is a Senior Hurricane Specialist at the National Hurricane Center.  It should be noted that the following discussion is Chris’ and Eric’s opinions only and does not represent any official position of NHC, NWS or NOAA in general. Various scientists within NOAA have differing opinions about global warming’s impact on hurricanes and there is no official NOAA policy on the topic. Varying ideas on an issue often mean that it is a science in progress with no definitive answers. That is certainly the case with regards to global warming and hurricanes. Helpful comments on an earlier version of this writeup were provided by Neal Dorst, Stan Goldenberg, Robbie Berg, and Mike Brennan.

[2] Accumulated Cyclone Energy is calculated by squaring the named storm’s intensity – maximum sustained surface winds (expressed in knots)  – for every six hours that the system had at least a 39-mph (63-kph) intensity. 

[3] There is on-going research into updating and revising the HURDAT2 database for the seasons of 1851 to 1999 in order to improve and make more complete the records that currently exist.  This is done by obtaining the original named storm observations from ships, weather stations, Hurricane Hunter aircraft, radars, and satellites and using today’s best meteorological analyses to revise the positions, intensities, and statuses in the database.  This work also adds in newly discovered named storms that were not identified as such at the time. Currently, the reanalysis project has added 35 years (1851 to 1885) to our official records and has revised the 1886 through 1965 hurricane seasons. 


Lifesaving NOAA Support Following the Sinking of the Bourbon Rhode

Posted on Updated on

Visible satellite imagery of Hurricane Lorenzo over the eastern North Atlantic Ocean on September 26, 2019, from the MODIS instrument aboard NASA’s Terra satellite.  Credit: NASA Worldview, Earth Observing System Data and Information System.

The National Oceanic and Atmospheric Administration (NOAA) provided critical assistance during the international search and rescue (SAR) and recovery efforts that followed the sinking of the M/V Bourbon Rhode in Hurricane Lorenzo last fall.  This intra-agency NOAA effort (see figure below) included Hurricane Hunters from the Aircraft Operations Center (AOC), scientists from the National Environmental Satellite, Data, and Information Service (NESDIS) and the Hurricane Research Division (HRD), and marine forecasters from the Tropical Analysis and Forecast Branch (TAFB) of the National Hurricane Center (NHC)

An organizational chart of NOAA offices that provided assistance during SAR and recovery efforts following the sinking of the M/V Bourbon Rhode in Hurricane Lorenzo.


On the morning of Thursday, September 26, 2019, French authorities received a distress signal from the M/V Bourbon Rhode, an offshore tugboat that was en route from Las Palmas, Canary Islands, to Georgetown, Guyana, with 14 crew members on board.  The Bourbon Rhode had made a dangerously close approach to the eye of rapidly intensifying Hurricane Lorenzo in the central Atlantic Ocean, and water was entering through the rear of the vessel.  At 0600 UTC (2 AM Atlantic Standard Time [AST]) September 26, Lorenzo was a Category 2 hurricane with 95-kt (110-mph) winds and seas 12 feet or greater that extended 240 to 330 nautical miles (275 to 380 statute miles) outward from its center. A 1200 UTC (8 AM AST) TAFB sea state analysis, issued around the same time as the last automatic identification system (AIS) signal from the Bourbon Rhode, showed peak significant wave heights in Lorenzo up to 41 feet.  By 1800 UTC (2 PM AST), Lorenzo had strengthened to a Category 4 hurricane with maximum sustained winds of 115 kt (130 mph).  The Bourbon Rhode ultimately sank on September 26 in the central Atlantic Ocean.  

File photo of M/V Bourbon Rhode.  Credit: Bourbon Offshore


Since Hurricane Lorenzo was a major hurricane that posed no imminent threat to land, both of NOAA’s P-3 aircraft were preparing to fly dedicated research missions into the storm. As NOAA43 (nicknamed Miss Piggy) transited from Lakeland, Florida, to Barbados on September 26, the French Government and the United States Coast Guard (USCG) reached out and requested SAR assistance.  Meanwhile, the nearest marine vessel to the incident site — a bulk carrier named SSI EXCELLENT – was redirected toward the last-known position of the Bourbon Rhode.  Later on September 26, NHC/TAFB was contacted by the USCG Rescue Coordination Center (RCC) Miami to begin providing spot forecasts for surface wind and wave conditions that would impact vessels aiding in the SAR efforts.  The first TAFB point forecast for the rescue detailed the dangerous marine conditions that were still ongoing in the wake of Lorenzo, with gusty tropical-storm-force winds and combined seas of 20 feet near the incident site.

NOAA aircraft fleet in Barbados for Hurricane Lorenzo research missions. The NOAA P-3 aircraft provided critical SAR support for the Bourbon Rhode incident.  Credit: LCDR Sam Urato, NOAA Corps


Flight track from NOAA43 during the Bourbon Rhode SAR mission over the central Atlantic Ocean on Sept. 27. Credit: Jon Zawislak, NOAA/AOML/HRD

NOAA43 departed Barbados on September 27 with a crew (see list at the end of the post) of AOC personnel as well as HRD and NESDIS researchers.  As requested, they planned to fly over the locations of the last Bourbon Rhode distress signals and report any findings.  If nothing was sighted, the crew would continue on with the planned research mission into Lorenzo.  NOAA43 was the first SAR-capable asset to reach the incident site, but the crew did not find anything upon arrival.  With growing concern about the fate of Bourbon Rhode crew members, the NOAA43 crew quickly decided to abandon the Lorenzo research mission and continue SAR support.  With little information besides the last-known location of the Bourbon Rhode, they quickly adapted to the situation and developed a SAR flight plan.  Crew members rearranged themselves by any available window and called out locations of suspected targets or debris while surveying in the vicinity of the last known Bourbon Rhode position.  Poor visibility, extremely large waves, and turbulence from strong rainbands posed difficult challenges as NOAA43 received sporadic emergency beacon signals.  With only minutes left before the plane needed to head back to Barbados due to fuel limitations, crew members spotted debris and what appeared to be a life raft.  This information was relayed to SSI EXCELLENT, which was en route to the SAR area.

Convective cell in an outer rainband of Hurricane Lorenzo, taken during a NOAA43 SAR mission on September 27, 2019.  The bulk carrier SSI EXCELLENT is also pictured. Credit: Kelly Ryan, NOAA/AOML/HRD


Flight track from NOAA42 during the Bourbon Rhode SAR mission over the central Atlantic Ocean on Sept. 28. Credit: Jon Zawislak, NOAA/AOML/HRD

On September 28, NOAA42 (nicknamed Kermit) flew a SAR mission in coordination with SSI EXCELLENT and other supporting marine vessels across the search area.  As Hurricane Lorenzo moved farther away, improving weather and marine conditions allowed the plane to fly as low as 200 feet above the ocean surface.  The NOAA42 crew (see list at the end of the post) conducted visual searches while listening for emergency beacon signals, guided by previous reports from NOAA43 as well as new information from supporting ships.  Crew members located a large debris field and the remains of several sailors, and they directed ships to these locations so the victims could be recovered.  The dedicated efforts of NOAA personnel significantly narrowed the search region and guided ships toward the area where a life raft was discovered later that day.  Three Bourbon Rhode survivors were rescued from that life raft in the Atlantic Ocean.


Aerial photo of the life raft carrying three surviving Bourbon Rhode crew members on September 28, 2019.  Credit: Marine Nationale (French Navy) via Facebook


NOAA assets played a pivotal role in early SAR efforts, which were led by the Maritime RCC Fort-de-France on the island of Martinique.  As the international search efforts continued, TAFB provided six-hourly forecast updates on wind, wave, and weather conditions.  From September 26 to October 5, 2019, TAFB produced 35 spot forecasts (see example below) that were shared with RCC Miami and MRCC Fort-de-France in support of this unprecedented SAR operation.  Over two weeks, 21 ships and four aircraft searched over 110,000 km2 (about 42,500 mi2) of the central Atlantic Ocean for survivors.  Four bodies were recovered, and seven others were declared lost at sea after SAR efforts were officially called off on October 5, 2019.

The first of 35 spot forecasts that NHC/TAFB marine forecasters sent to USCG and international partners in support of Bourbon Rhode SAR efforts.


The Bourbon Rhode incident is just one example of how TAFB has evolved to provide impact-based decision support services (IDSS) to the USCG, its primary core governmental partner.  Last year, TAFB forecasters produced 56 spot forecasts for 13 marine incidents including SAR missions, distressed vessels, and even a medical rescue. In July 2019, the USCG and U.S. Air Force coordinated a rescue operation of two critically injured people off a disabled fishing vessel in the eastern North Pacific Ocean.  TAFB provided spot forecast support for the rescue operation and subsequent transport of the injured people to a Mexican naval medical clinic on Socorro Island.  “This information is truly impacting operations,” said Douglas Samp, Search Mission Coordinator for RCC Alameda (USCG District 11).

Additionally, TAFB forecasters prepare and deliver live briefings to USCG District leadership when tropical cyclones threaten USCG SAR regions and U.S. ports.  In 2019, TAFB delivered 42 tropical briefings combined to USCG District 7 and District 8, including 25 briefings for Hurricane Dorian.  “I cannot overstate how much your [NHC/TAFB] insight into the storm’s effects is vital to our planning and response efforts,” commented Captain Eric Smith, Chief of the Incident Management Branch for USCG District 7.

Tragedies like the Bourbon Rhode highlight the importance of TAFB standing ready to provide year-round IDSS support to core partners.  In this case, the dedicated IDSS provided by TAFB forecasters, combined with the valiant efforts of NOAA AOC crew members and HRD and NESDIS researchers, played a critical role in the international rescue efforts that ultimately saved three lives.

— Brad Reinhart


Crew of NOAA43 September 27th Flight

All female science crew aboard NOAA43, pictured before the research flight was diverted to provide Bourbon Rhode SAR support. L-R: Kelly Ryan, Jezabel Viraldell, Ashley Lundry, Zorana Jelenak, and Heather Holbach. Credit: Paul Flaherty, NOAA

Cmdr. Pat Didier – Aircraft Commander
Lt. Cmdr John Rossi – Co-pilot
Lt. Cmdr Dean Legidakes – Co-pilot
Lt. Cmdr Peter Freeman – Navigator
Mr. Joshua Sanchez – Flight Engineer
Mr. Chris Lalonde – Flight Engineer
Mr. Paul Flaherty – Flight Director
Mrs. Ashley Lundry – Flight Director
Mr. Dana Naeher – Data Technician
Mr. Joe Greene – AVAPS Technician
Mr. Todd Richards – System Engineer
Mr. Damon San Souci – Avionics Technician
Dr. Zorana Jelenak – Principle Investigator (Scientist)
Jezabel Viraldell Sanchez – NESDIS Scientist
Heather Holback – Lead Project Scientist / Radar Scientist
Kelly Ryan – Dropsonde / Radar Scientist

Crew of NOAA42 September 28th Flight

Cmdr. Nathan Kahn – Aircraft Commander
Lt. Cmdr Adam Abitbol – Co-pilot
Lt. Cmdr Robert Mitchell – Co-pilot
Lt. Cmdr Brian Richards – Navigator
Mr. Paul Darby – Flight Engineer
Mr. Ken Heystek- Flight Engineer
Mr. Mike Holmes – Flight Director
Mr. Mike Mascaro – Data Technician
Mr. Joe Greene – System Engineer
Mr. Nick Underwood – AVAPS Technician
Dr. Jon Zawislak – Lead Project Scientist
Trey Alvey – Radar Scientist
Kathryn Sellwood – Dropsonde Scientist
Joe Sapp – NESDIS scientist

Acknowledgments: Special thanks to Zorana Jelenak, Kelly Ryan, and Joe Sapp for sharing their personal accounts of this experience with the author. Additional thanks to Jonathan Shannon, Shirley Murillo, Jon Zawislak, Nathan Kahn, Patrick Didier, and Erica Rule for their helpful input and feedback.

Additional References:

Storm Surge: Planning for the Risk

Posted on Updated on

Mobile home park on the north side of U.S. Highway 98 in Mexico Beach, Florida, washed away by the storm surge and wave impacts of Hurricane Michael. Nov. 2, 2018 (NOAA)

If you were asked to define the word “risk”, what would your response be? What are the first things that come to mind? What questions are you asking? What are some things you would consider to be risky?  Risk can take many forms, be it financial, personal injury, or even simple decisions that we all have to make on a daily basis. Consider two simple questions: 

  • Should I change the tires on my car?
  • Should I change the tires on my car today

From a risk perspective, how has the question changed? The answer to the first question is a simple one because it’s general in nature. Yes, the tires should be changed somewhat routinely over the lifetime that you own or lease the car. Question two requires more thought, however, because it narrows the task down to a specific time frame. If your tires were changed recently and are in good shape, then it’s most likely business as usual. If you’re overdue, what are the potential consequences of not having your tires changed? Are you risking personal injury to yourself, your family, or others by not changing them? The risk profile changes based on myriad variables that exist. As this example conveys, the idea of risk is relatively simple in itself but can become complex quickly. Let’s define risk for now as the potential of gaining or losing something of value. That is to say:

Risk = Probability x Consequence 

Yes, we threw some math at you, but this concept is relatively easy to grasp. Using the previous example, the overall risk profile is determined by multiplying the probability of a given event by the resulting consequence if that event occurs. So if there is a high chance of an event occurring, or the consequence is severe, then the risk to you would be high.  One way to look at risk is by using a risk matrix, as shown below in Figure 1.  Your risk increases if the probability of the event goes up or if the consequence of the event goes up.

Figure 1.  Risk matrix showing different levels of risk based on the probability of an event and the consequence if that event occurs.

The event in our car example is blowing a tire on the interstate, and a potential consequence would be having a fatal accident due to the blown tire. That consequence is so severe that your risk is quite high.  But let’s take the example a little further.  Risk is further compounded by vulnerability. Let’s consider a new equation:

Risk = Probability x Consequence x Vulnerability

Using the same example, what are variables that could increase the vulnerability, and thus the risk of a fatal accident, in this scenario? Are there kids in the car? What speed is the car travelling? If the tire pops while backing out of the driveway, isn’t that much different than the tire popping while travelling 70 mph down a busy interstate? This is just one of many examples that we all encounter on a daily basis. If there is a consequence to an action you might take, then you are making a risk-based decision.


Risk Perception and Risk Tolerance

There are two other topics related to risk that we should touch on:  risk perception and risk tolerance.  First, risk perception is the subjective judgement people make about the severity and probability of a risk. Why is this important? Well, there are two main reasons:

  1. Actual risk usually doesn’t equal perceived risk
  2. Perceived risk varies from person to person

Why does this complicate matters? When people have to make decisions, it’s important that how they perceive the risk be equal to the actual risk of the event, or at least as close to equal as possible. This is especially true when there is a desired response or action that needs to be taken to protect life and property, as is often the case with weather. We can use a simple example to further understand this. If a tornado is on the ground moving towards a community, the desired action for people living within that community is to seek shelter. In this scenario, the cost of persons within that community not seeking shelter is very high given that their lives are potentially in peril. If people dismiss what that tornado could do to their community (e.g., “tornadoes always pass our city to the south”), then that can be a recipe for disaster, especially when the cost is human lives.

This leads us nicely into risk tolerance. How an individual responds to risk is governed by their risk tolerance, which is unique to both the situation and the person. Risk tolerance is the amount of risk that an individual is willing to accept with respect to a given event occurring. Turning back to the tornado example, let’s consider two hypothetical people that live in the community threatened by the approaching tornado; we’ll name them Sara and Monika. For the sake of example, let’s assume that Sara has a family with two young children and lives in a mobile home. Monika, on the other hand, lives by herself in a well-built house. If Sara and Monika have the same perception of the imposing risk, do these life factors change their respective risk tolerances? In the real world it’s difficult to say, but in this idealized example, let’s assume that it does. Monika does not have anyone in her care and also lives in a home that could withstand stronger winds than Sara’s mobile home. Is Monika’s tolerance for risk higher than Sara’s? It certainly could be, couldn’t it? Again, we understand that assumptions are being made here, but this is simply a hypothetical scenario to demonstrate how risk tolerances may change across individuals and circumstances unique to those individuals. All of this is to say that humans are complicated and risk perceptions and tolerances vary across all of us. 

Circling back to the initial equation that we used to define risk, let’s establish a baseline for what the potential consequences are with respect to storm surge by looking at history. Storm surge is the abnormal rise of the ocean produced by a hurricane or tropical storm, and normally dry land near the coast can be flooded by the surge.  Historically, about 50% of lives lost in landfalling tropical cyclones in the United States have been due to storm surge (Figure 2):

Figure 2.  Causes of death in the United States directly attributable to Atlantic tropical cyclones, 1963-2012 (Rappaport 2014).

The mission statement of the National Weather Service charges us to “provide weather, water, and climate data, forecasts and warnings for the protection of life and property and enhancement of the national economy.” There is no clearer way to illustrate what the consequences are in this equation: the loss of human lives. Because the cost here is so high – arguably the highest – our risk tolerance for your safety is extremely low. We have absolutely no appetite for someone losing their life from a weather event. This idea directly informs many of the products that we use to communicate risk prior to and during landfalling storms, and we therefore use near-worst-case scenarios to encapsulate the full envelope of storm surge risk to communities. One life lost during a storm is one life too many. The remainder of this blog post will discuss two such products used by the National Hurricane Center and emergency managers to understand storm surge risk.

MOMs and MEOWs

Can we all agree that “MOMs” are extremely important? Well, yes, those moms are important in our lives, but that holds true for storm surge MOMs as well. Have you ever wondered how officials decide what areas should evacuate ahead of a hurricane? Look no further. MOMs (Maximum of the Maximums) are the rock from which the nation’s storm surge evacuation zones are built upon.  MOMs are generated ahead of time. That is to say that these are precomputed maps meant for planning and mitigation purposes well ahead of a landfalling hurricane. In fact, one can view these any time as they are hosted on the National Hurricane Center’s website at https://www.nhc.noaa.gov/nationalsurge/. MOMs are generated by hurricane category (think 1-5) and depict the maximum storm surge height possible across all storm surge attributes. Attributes include things such as forward speed, storm trajectory, and landfall location, just to name a few. Because this product is designed for planning, you can think of the MOM as a worst-case scenario for a given category of a storm. MOMs do have limitations, however. Remember at the beginning of this post we asked the question “should I change the tires on my car?”  MOMs are similar to that question because they are general in nature in that they lump all types of hurricanes into a single category.  They can tell you what type of storm surge risk you would have from a category 3 hurricane, for example, but they’re not quite as helpful if you know that the category 3 hurricane will be moving toward the west (and not north or northeast for instance).  

For this reason, the MOMs have a slightly more refined counterpart – MEOWs (Maximum Envelope of Water). MEOWs are like the second question we asked:  “should I change the tires on my car today.”  Since we said “today,” we know a little bit more about the actual situation we’re dealing with to make a better-informed decision.  Similarly, once a storm or hurricane forms and is within 3–4 days of impacting the coast, we have at least some idea of how strong it could get, how fast it’ll be moving, and in what general direction it’s headed. We are able as forecasters to whittle down the worst-case MOM such that we only consider storms moving toward a particular direction at a particular forward speed — not all directions and forward speeds. Similar to the Maximum of Maximums, a MEOW is a worst-case for storms of a certain strength (for example, Category 3 hurricanes), but it’s more representative of what the storm surge at individual locations could be based on the attributes and forecasts of the active tropical cyclone. At 3 days out, there is still considerable forecast uncertainty, so the MEOW is meant to supplement the MOM, not replace it. As some might say, you can’t go wrong if you always trust your mom. The same adage goes for a hurricane storm surge MOM.

Hurricane Florence

To help us better understand these products, let’s look at how they could have been used in practice during a past landfalling hurricane. Hurricane Florence made landfall along the North Carolina coast on Friday, September 14th of 2018 and presented numerous forecast challenges, as many landfalling tropical cyclones typically do. One benefit of using MOMs and MEOWs to plan, especially at longer lead-times, is that they provide stability in a situation where the forecast of the storm itself can often change quickly from advisory to advisory. Let’s take a look at what Florence’s forecast looked like about 5 days out from an expected landfall. Figure 3 is taken from the official forecast from the National Hurricane Center on September 8th at 11 pm AST.

Figure 3.  NHC five-day forecast track and cone of uncertainty issued for Tropical Storm Florence at 11 PM AST September 8, 2018 (Advisory 39).

At this point, the information we know is that a potential major hurricane is roughly 5 days away from impacting some portion of the Mid-Atlantic or southeast coast. This is a good point to begin looking at the MOMs. Florence at this point is forecast to be a category 4 hurricane at landfall, so a good rule-of-thumb to follow is to look at a MOM one category higher than the forecast intensity. Let’s take a look at the Category 5 MOM to get an idea of a worst-case storm surge scenario for a portion of the North Carolina coast. You can find that image below in Figure 4.

Figure 4.  Category 5 storm surge Maximum of Maximums (MOM) for portions of eastern North Carolina.

Given how strong Florence could be, it’s no surprise to see potential inundation that would be catastrophic. Remember what we are looking at here and also that this is still a planning tool. This graphic is showing you the worst-case scenario from a Category 5 hurricane. That is to say that these are the highest possible inundations at each individual location for any given storm attribute. We actually shouldn’t expect to see these types of inundation values across the entire area, but given the uncertainties in the storm, all locations in this area should be prepared for these types of inundation values. It would also be prudent to consider looking at other MOMs as well, for context. For example, viewing the Category 3 and 4 MOMs gives context if Florence was to reach the coast at a lower intensity.

Good. Now let’s fast forward by 2 days. We are now 3 days out from a potential landfall. Forecast confidence has increased, but the fine-scale details are still quite blurry regarding the exact location of landfall and how strong Florence will be. But this is when you can begin to turn to the MEOWs. From this point, we can begin to whittle down the MOMs to generate a more realistic potential scenario based on the information currently available. Below in Figure 5 is the forecast from September 10th at 11 pm AST.

Figure 5.  NHC five-day forecast track and cone of uncertainty issued for Hurricane Florence at 11 PM AST September 10, 2018 (Advisory 47).

As you can see, there are some updates to the forecast track. The official forecast now calls for Florence to slow down significantly as it approaches the North Carolina coast. Let’s now talk about which MEOWs we should be looking at and explore how we select them. This is an important junction in the forecast because right now we need to evaluate what we do know, what we don’t know, and what we can and cannot rule out. Remember that MEOWs are generated individually for a particular storm category, forward speed, trajectory, and initial tide level. At this point, is there anything that we can rule out in terms of unrealistic directions that Florence could potentially make landfall? It’s ok to acknowledge that there remains some subjectivity here, but it needs to be an informed decision with an understanding that our risk tolerance is low. That being said, let’s go ahead and rule out some storm directions.  Since the forecast track in advisory 47 reflects a northwestward trajectory at landfall, we’ll select that direction, as well as the two surrounding it (west-northwest and north-northwest) to account for uncertainty.  How about the intensity?  The latest forecast still shows Florence reaching the coast as a category 4 hurricane, so we still need to account for the possibility that it makes landfall one category stronger (category 5).  Lastly, let’s consider the speed at which Florence is moving and will be moving near its landfall. The tropical cyclone forecast discussion from advisory 47 explicitly mentions that Florence is expected to decrease in forward speed as it approaches the coast:

After that time [48 hours], a marked decrease in forward speed is likely as another ridge builds over the Great Lakes to the north of Florence.”

This is reflected in the official forecast which slows Florence down to less than 10 mph near the coast. While this certainly complicates the forecast, the beauty of using MEOWs is that it allows you to compensate for this forecast uncertainty. In this case, it’s fair that we could eliminate the MEOW forward speeds of 15, 25 and 35 mph, given forecaster confidence in Florence’s slow down. This leaves us with a forward speed of 5 mph (only a certain set of speeds is actually available to select). 

Let’s quickly recap the parameters that we’ve settled on to generate our MEOW:

Intensity: Category 5
Direction/trajectory: Storms that are moving West-Northwest, Northwest, or North-Northwest
Forward Speed: 5 mph
Tide-level: High (this will always be the assumption)

Using those parameters, Figure 6 shows the potential storm surge inundation that could occur across eastern North Carolina:

Figure 6.  Composite storm surge Maximum Envelope of Water (MEOW) over portions of eastern North Carolina for a category 5 hurricane moving west-northwest, northwest, or north-northwest at 5 mph at high tide.

To take this one step further, let’s zoom in around the New Bern, North Carolina, area and do a quick comparison of the category 5 MOM that we initially used 5 days out and compare it to the composite of MEOWs (Figure 7). 







Figure 7.  Comparison of MOM (left) and composite MEOW (right) from Figures 4 and 6 above, zoomed in on the New Bern, North Carolina, area.

Remember that at this point in the forecast process, we are looking at synthetic or simulated storms to get an idea of what the near-worst case storm surge inundation could be within an environment characterized by forecast uncertainty that’s very high.  What differences do you notice when you compare the two pictures above?  Don’t worry–you’re eyes aren’t deceiving you.  You probably don’t notice much difference at all.  That’s because, unfortunately, slow-moving storms moving in a generally northwestward direction are likely some of the worst types of storms for the New Bern area.  Essentially, they’re the storms that are most likely to be causing the storm surge heights you see in the MOM.  Our confidence in the hurricane’s forecast has increased since we’re 2 days closer to landfall, but the storm surge risk really hasn’t gone down at all.  While that might be a sobering thought, this process allows emergency managers to be as efficient as possible, appropriately assess their risk, and focus on the most at-risk areas.  This is a powerful and informative process when used properly! 

In the end, while all of eastern North Carolina did not experience the type of storm surge flooding shown in Figure 6 above (which we didn’t expect anyway), some areas did.  Areas around New Bern, for instance, had as much as 9 feet of storm surge inundation above ground level (red areas in Figure 8 below).  Even though Florence’s peak winds decreased while the storm moved closer to the coast, the MOM and MEOW risk maps accounted for Florence’s increasing size and slow movement (which both contribute to more storm surge) and appropriately prepared emergency managers in the area for a severe storm surge event days before Florence even reached the coast.

Figure 8.  Post-storm model simulation of storm surge inundation caused by Hurricane Florence around the New Bern/Neuse River area of North Carolina.

It’s important to note at this point that MOMs and MEOWs are predominantly used during the period before storm surge or wind-related watches and warnings are in effect for the coast (more than 48 hours before wind or surge is expected to begin).  Once we get to within 48 hours when watches or warnings go into effect, another suite of storm surge products–specifically the Potential Storm Surge Flooding Map and the Storm Surge Watch/Warning graphics–become available.  These products refine the storm surge risk profile even further because they are based on the characteristics of the actual storm, not on the simulated storms used in MOMs and MEOWs.  We plan to create another blog post addressing these products in the near future.

To really bring this home, let’s circle all the way back around to the initial discussion of risk. How does risk tolerance and risk perception affect how these products are used? We know that these products are used by a wide range of people and organizations, all of which have varying tolerances of risk. It is unrealistic to assume that we at the National Hurricane Center could know how these tolerances change across our entire user base. That being said, it is our job to gently guide the decision-making in accordance with our own risk tolerance. Said another way, we work with emergency managers and the Hurricane Liaison Team (HLT) to hopefully bring those risk perceptions more in line with the ACTUAL risk for a given storm. Emergency managers have the resources at their disposal to view MOMs and MEOWs to build out their assessment of risk tailored to their local areas. They possess the intricate knowledge specific to their area which makes them invaluable partners to us at the NHC. During a storm, we sometimes provide advice on types of MOMs and MEOWs to consult to ensure that our partners fully capture a reasonable envelope of risk. These decisions can be stressful, especially when they have to be made in line with a risk tolerance that needs to be low by necessity. Remember what the cost is again here: human lives. It’s imperative that we capture the full breadth of the risk during every storm because the cost of not doing so is immense. We are comfortable accepting that our low risk tolerance can result in some areas not experiencing the potential storm surge that was conveyed prior to a hurricane making landfall. That is, by definition, what having a low tolerance for risk means, but it’s also by design.  To us, one life lost is one life too many. 

— Taylor Trogdon and  Robbie Berg



Rappaport, E.N., 2014: Fatalities in the United States from Atlantic Tropical Cyclones: New Data and Interpretation. Bull. Amer. Meteor. Soc., 95, 341–346, https://doi.org/10.1175/BAMS-D-12-00074.1

Skill or Luck?: How NHC’s Hurricane Track Forecasts Beat the Models

Posted on Updated on

Skill or Luck?

There’s one thing that many of us are missing right now while we’re occupying ourselves at home:  sports.  We should have been all set for the playoffs in major league hockey and basketball, and we would be excited about the beginning of the major league baseball and soccer seasons.  We also would have been eagerly anticipating some of this spring and summer’s major sporting events, including the Olympics.  So let’s dream a little…

When we set out to write this blog post for Inside the Eye, we wanted to show how National Hurricane Center (NHC) forecasters use their skill and expertise to predict the future track of a hurricane.  And then it got us thinking, how does luck factor into the equation?  In other words, when meteorologists get a weather forecast right, how much of it is luck, and how much of it is forecasters’ skill in correctly interpreting, or even beating, the weather models available to them?

Investment strategist Michael Mauboussin created a “Skill-Luck Continuum” where individual sports, among other activities in life, are placed on a spectrum somewhere between pure skill and pure luck (Figure 1).   Based on factors such as the number of games in a season, number of players in action, and number of scoring opportunities in a game or match, athletes and their teams in some sports might have to rely on a little more luck than other sports to be successful.  On this spectrum, a sport like basketball would be closest to the skill side (there are a lot of scoring opportunities in a basketball game) whereas a sport like hockey would require a little more luck (there are fewer scoring opportunities in a hockey match, and sometimes you just need the puck to bounce your way).  Fortunately for hockey fans, there are enough games in a season for their favorite team’s “unlucky” games to not matter so much.

The Skill-Luck Continuum for Sports

Figure 1.  The Skill-Luck Continuum in Sports, developed by investment strategist Michael Mauboussin.


Where would hurricane forecasting lie on such a continuum?  There’s no doubt that luck plays at least some part in weather forecasting too, particularly in individual forecasts when random or unforeseen circumstances could either play in your favor (and make you look like the best forecaster around) or turn against you (and make you look like you don’t know what you’re doing!).  But luck is much less of a factor when you consider a lot of forecasts over longer periods of time, where the good and bad circumstances should cancel each other out and true skill shines through (just as in sports).  At NHC, we routinely compare our forecasts with weather models over these long periods of time to assess our skill at predicting, for example, the future tracks of hurricanes.

An International Friendly?

From our experience of talking to people about hurricanes and weather models, it seems to be almost common “knowledge” that only two models exist – the U.S. Global Forecast System (GFS) and the European Centre for Medium Range Weather Forecasts (ECMWF) model.  It’s true that those two models are used heavily at NHC and the National Weather Service in general, but there are many more weather models that can simulate a hurricane’s track and general weather across the globe.  (Here’s a comprehensive list showing all of the available weather models that are used at NHC today, if you’re interested:  https://www.nhc.noaa.gov/modelsummary.shtml.)  We’ve also heard and seen people compare the GFS and ECMWF models and talk about which model scenario might be more correct for a given storm.  This blog entry summarizes the performances of those models and discusses how, on the whole, NHC systematically outperforms them on predicting the track of a storm.

Below are the most recent three years of data (2017, 2018, and 2019) of Atlantic basin track forecast skill from NHC and the three best individual track models:  the GFS, ECMWF, and the United Kingdom Meteorological Office model (UKMET) (Figure 2).  Track forecast skill is assessed by comparing NHC’s and each model’s performance to that of a baseline, which in this case is a climatology and persistence model.  This model makes forecasts based on a combination of what past storms with similar characteristics–like location, intensity, forward speed, and the time of year–have done (the climatology part) and a continuation of what the current storm has been doing (the persistence part).  This model contains no information about the current state of the atmosphere and represents a “no-skill” level of accuracy.

Figure 2.  NHC and selected model track forecast skill for the Atlantic basin in 2017, 2018, and 2019.


On the skill diagrams above, lines for models or forecasts that are above other lines are considered to be the most skillful.  It can be seen that in each year shown, NHC (black line) outperforms the models and has the greatest skill at most, if not all, forecast times (the black line is above the other colored lines most of the time).  Among the models, the ECMWF (red line) has been the best performer, with the GFS (blue line) and UKMET (green line) trading spots for second place.

Yet another metric to estimate how often NHC outperforms the models is called “frequency of superior performance.”  Based on this metric, over the last 3 years (2017-19), NHC outperformed the GFS 65% of the time, the UKMET 59% of the time, and the ECMWF 56% of the time.   This means that more often than not, NHC is beating these individual models.  So the question is, how do the NHC forecasters beat the models?

Keep Your Eyes on the Ball

Forecasters at NHC are quite skilled at assessing weather models and their associated strengthens and weaknesses.  It is that experience and a methodology of using averages of model solutions (consensus) that typically help NHC perform best.  If you ever read a NHC forecast discussion and see statements like “the track forecast is near the consensus aids,” or “the track forecast is near the middle of the guidance envelope,” the forecaster believed that the best solution was to be near the average of the models.   Although this strategy often works, NHC occasionally abandons this method when something does not seem right in the model solutions.  One recent example of this was Tropical Storm Isaac in 2018.  The figure below (Figure 3) shows the available model guidance, denoted by different colors, at 2 PM EDT (1800 UTC) on September 9 for Isaac, with the red-brown line representing the model consensus (TVCA).

Figure 3. NHC forecast (dashed black line) and selected model tracks at 2 PM EDT (1800 UTC) September 9, 2018 for then-Tropical Storm Isaac.  The solid black line represents the actual track of Isaac and the red-brown line represents the model consensus.


Although the models were in fair agreement that the storm would head westward for some time, a few models diverged by the time Isaac was expected to be near the eastern Caribbean Islands, mostly because they disagreed on how fast Isaac would be moving at that time.  Instead of being near the middle of the guidance envelope, NHC placed the forecast on the southern side of the model suite (dashed black line) at the latter forecast times since the forecaster believed that the steering flow would continue to force Isaac westward into the central Caribbean.  Indeed, NHC was correct in this case, and in fact, for the entire storm, NHC had very low track errors.

In some cases all of the models turn out to be wrong, which usually causes the official forecast to suffer as well.  That was the case for a period during Dorian in 2019.  Figure 4 shows many of the available operational models at 8 PM EDT on August 26 (0000 UTC August 27) for then-Tropical Storm Dorian.  As you can see by noting the deviation of the colored lines from the solid black line (Dorian’s actual track), none of the models or the official forecast (colored lines) anticipated that Dorian would turn as sharply as it did over the northeastern Caribbean Sea, and no model showed a direct impact to the Virgin Islands, where Dorian made landfall as a hurricane.


Figure 4. NHC forecast (dashed black line) and selected model tracks at 8 PM EDT on August 26 (0000 UTC 27 August), 2019 for then-Tropical Storm Dorian.  The solid black line represents the actual track of Dorian.


Figure 5 shows many of the operational models at 2 AM EDT (0600 UTC) on August 30 when Dorian, a major hurricane at the time, was approaching the Bahamas.  You can see that all of the models showed Dorian making landfall in south or central Florida in about four days from the time of the model runs, and none of them captured the catastrophic two-day stall that occurred over Great Abaco and Grand Bahama Islands.  NHC’s forecast followed the consensus of the models in this case and thus did not initially anticipate Dorian’s long, drawn-out battering of the northwestern Bahamas.

Figure 5.  NHC forecast (dashed black line) and selected model tracks at 2 AM EDT (0600 UTC) on August 30, 2019 for Hurricane Dorian.  The sold black line represents the actual track of Dorian.


The Undervalued Player?  A Consistently Good Field-Goal Kicker

In American football, probably one of the most undervalued players on the field is the kicker.  They don’t see much action during the majority of the game.  But at the end of close games, who has the best chance to win the game for a team?  A dependably accurate field goal kicker.  In that vein, it’s not just accuracy that can make NHC’s forecasts “better” than the individual models.  Another important factor is how consistent NHC’s predictions are from forecast to forecast compared to those from the models.  We looked at consistency by comparing the average difference in the forecast storm locations between predictions that were made 12 hours apart. For example, by how much did the 96-hour storm position in the current forecast change from the 108-hour position in the forecast that was made 12 hours ago (which was interpolated between the 96- and 120-hour forecast positions)?  Figure 6 shows this 4-day “consistency,” as well as the 4-day error, plotted together for the GFS, ECMWF, UKMET, and NHC forecasts for the Atlantic basin from 2017-19.  It can be seen that NHC is not only more accurate than these models (it’s farthest down on the y-axis), but it is also more consistent (it’s farthest to the left on the x-axis), meaning the official forecast holds steady more than the models do from cycle to cycle.  We like to say that we’re avoiding the model run-to-run “windshield wiper” effect (large shifts in forecast track to the left or right) or “trombone” effect (tracks that speed up or slow down) that are often displayed by even the most accurate models.

Figure 6.  96-hour NHC and model forecast error and consistency for 2017-2019 in the Atlantic basin (change from cycle to cycle).


NHC’s emphasis on consistency is so great that there are times when we knowingly accept that we might be sacrificing a little track accuracy to achieve consistency and a better public response to the threat.  An example would be for a hurricane that is forecast to move westward and pose a serious threat to the U.S. southeastern states.  Sometimes, such storms “recurve” to the north and then the northeast and move back out to sea before reaching the coast.  When the models trend toward such a recurvature, the NHC’s forecast will sometimes lag the models’ forecast of a lower threat to land.  In these cases, NHC does not want to prematurely take the southeastern states “off the hook”, sending a potentially erroneous signal that the risk of impacts on land has diminished, only to have later forecasts ratchet the threat back up after the public has turned their attention and energies elsewhere if the models, well, “change their mind”.  That would be the kind of windshield wiper effect NHC wants to prevent in its own forecasts.  Now, there are times where the recurvature does indeed occur.  Then, NHC’s track forecasts, which have hung back a little from the models, could end up having larger errors than the models.  But, NHC can accept having somewhat larger track forecast errors than the models in such circumstances at longer lead times if in doing so it can provide those at risk with a more effective message–achieved in part through consistency.

The superior accuracy and higher levels of consistency of the NHC forecasts are both important characteristics since emergency managers and other decision makers have to make challenging decisions, such as evacuation orders, based on that information.  It is not surprising to us that NHC’s forecasts are more consistent than the global models, since forecasters here take a conservative approach and usually make gradual changes from the forecast they inherited from the previous forecaster.  Conversely, the models often bounce around more and are not constrained by their previous prediction.  And, unlike human forecasters, the models also bear no responsibility or feel remorse when they are wrong!

Filling Out Your Bracket

Accuracy, consistency, and luck are important factors in one particularly favorite sport:  college basketball.  We just passed the time of year when we should have been crowning champions in the men’s and women’s college basketball tournaments.  But before those tournaments would have kicked off, “bracketologists” (no known relation to meteorologists!) would have made predictions on which teams would make it into the tournaments and which teams would have been likely to win.

Think of it this way:  a team can be accurate in that they have a spectacular winning record during the regular season, but does that mean they are guaranteed to win the tournament, or even advance far?  Nope.  As is often said, that’s why they play the game.  An inconsistent team—one whose performance varies wildly from game to game—has a higher risk of having a bad game and losing to an underdog in the first few rounds, even if their regular season record by itself suggests they should have no problem winning.  The problem is, they could have been very lucky in the regular season, winning a lot of close games that could have easily swung the other way.  If that luck runs out, the inconsistent team could have an early exit from the tournament.  With a consistent team, on the other hand, you pretty much know what kind of performance you’re going to get—good or bad—and that increases confidence in knowing how far in the tournament the team would advance.  You’d want to hitch your wagon to a good team that is consistent and hasn’t had to rely on too much luck to get where they are.

The same can be said for hurricane forecasts from NHC and the models.  NHC’s track forecasts are more accurate and more consistent than the individual models in the long run, and that fact should increase overall user confidence in the forecasts put out by NHC.  Even still, there is always room to improve, and it is hoped that forecasts will continue to become more accurate and consistent in the future.  It is always a good idea to read the NHC Forecast Discussion to understand the reasons behind the forecast and to gauge the forecaster’s confidence in the prediction.  For more information on NHC forecast and model verification, click the following link:  https://www.nhc.noaa.gov/verification/

— John Cangialosi, Robbie Berg, and Andrew Penny

National Hurricane Center Decision Support Services for the United States Coast Guard

Posted on Updated on

United States Coast Guard Cutter (image courtesy of uscg.mil)

Semper Paratus (Always Ready): A Shared Mission of Watching Over a Vast Blue Ocean

The National Hurricane Center (NHC) has the responsibility for issuing weather forecasts and warnings for a wide expanse of the Atlantic and eastern North Pacific Oceans.  Within NHC, the Hurricane Specialist Unit (HSU) issues forecasts for tropical storms and hurricanes in these regions, issues associated U. S. watches and warnings, and provides guidance for the issuance of watches and warnings for international land areas.  NHC’s Tropical Analysis and Forecast Branch (TAFB) makes forecasts of wind speeds and wave heights and issues wind warnings year-round for the eastern North Pacific Ocean north of the equator to 30°N, and for the Atlantic Ocean north of the equator to 31°N and west of 35°W (including the Gulf of Mexico and Caribbean Sea).  These wind warnings include tropical storms and hurricanes as well as winter storms, tradewind gales, and severe gap-wind events (for example, the “Tehuantepecers” south of Mexico).

The United States Coast Guard (USCG) has areas of responsibility (AORs) that extend well beyond those of NHC, with potential weather hazards affecting the fleet and their missions over the ocean, inland U.S. waterways, and flood-prone U.S. land areas. Although the USCG is responsible for search and rescue missions that may occur due to weather hazards, they are also vulnerable to severe weather and must also protect their own fleet and crews from these hazards.

USCG Search and Rescue Regions (SRRs) cover vast ocean areas affected by tropical cyclones. Superimposed on the Pacific SRRs is the NHC tropical cyclone area of responsibility, which overlaps with two eastern Pacific USCG SRRs as well as all Atlantic SRRs. The number of briefings provided by NHC to each USCG district in 2018 are shown. (Map images courtesy of uscg.mil)

One of the USCG’s oldest missions and highest priorities is to render aid to save lives and property in the maritime environment.  To meet these goals, the United States’ area of search-and-rescue responsibility is divided into internationally recognized inland and maritime regions.  There are five Atlantic USCG Search and Rescue Regions (SRRs) (Boston, Norfolk, Miami, New Orleans, and San Juan) and two Pacific USCG SRRs (Alameda and Honolulu) that overlap with NHC’s hurricane and marine areas of responsibility. The other eastern Pacific regions north of the Alameda SRR do not typically, if ever, experience hurricane activity. The multi-million square mile area of the agencies’ overlap allows NHC to provide weather hazard Decision Support Services (DSS) for the USCG.

Building Partnerships with the Districts

The National Weather Service (NWS) signed a Memorandum of Agreement (MOA) with the USCG to provide them with weather support. Over the past couple of years, staff at NHC have had numerous discussions with several of the USCG districts in order to build stronger partnerships. These discussions, primarily involving how NHC can better serve the USCG, established criteria for requiring TAFB to provide weather briefings to key decision makers within the USCG. When criteria are met, TAFB provides the relevant USCG District with once- or twice-a-day briefing packages detailing the weather impacts on their area of responsibility. This information provides the USCG districts with the details necessary to make efficient and effective decisions about potential mobilization of their fleet.

Example of a briefing slide of NHC’s earliest reasonable arrival time of tropical-storm-force winds graphic, which is one of the USCG’s most-desired decision support tools provided by NHC.  This example, from Hurricane Michael, illustrates the timing of the earliest reasonable onset of tropical-storm-force winds at a given location. This information is critical for fleet mobilization, as once these winds arrive preparations become difficult, if not impossible, to complete.

2018 Hurricane Season Briefing Support

During the 2018 hurricane season, TAFB provided 30 briefings to USCG Districts 5 (Norfolk), 7 (Miami), 8 (New Orleans), and 11 (Alameda) for the several tropical storms and hurricanes that affected them. These interactions helped to build the relationships between NHC and the USCG districts and aided the districts in making decisions regarding fleet mobilization, conducting search and rescue missions, and preparation for USCG’s land-based assets and personnel. Some of these briefings occurred during rapidly evolving high impact scenarios, including Hurricane Michael. Michael was forecast to become a hurricane within 72 hours of developing into a tropical depression and was forecast to make landfall within 96 hours of its formation. Ultimately, Michael rapidly intensified into a category 5 hurricane only 3½ days after formation, before making landfall on the Florida Panhandle. Hurricane Michael’s track across the east-central Gulf of Mexico straddled the border of USCG Districts 7 (Miami) and 8 (New Orleans), leading to both Districts taking action in advance of the hurricane.

Support for District 5 (Norfolk)

The NWS’s Ocean Prediction Center, the NHC (through TAFB), and the NWS National Operations Center have worked together to provide weekly high-level coordination briefings to USCG District 5 on upcoming hazards focused on the Atlantic Ocean north of 31°N over the following seven days.  Each Monday (except Tuesday if Monday is a holiday) by noon Eastern Time, the NWS provides a briefing that covers the mid-Atlantic region from New Jersey through North Carolina.  Typically, the briefing covers the area to roughly 65°W, though the exact area covered can vary based on the week’s expected weather hazards.  The USCG, in turn, has been sharing the information with mariners, port partners, and industry groups for situational awareness and critical decision-making.

Future Support

NHC’s TAFB is ready to provide decision support services to the USCG Districts for the 2019 hurricane season. Plans are being developed to continue this type of support for many years to come.

— Andy Latto