101: Supercell Thunderstorms

Introduction to Supercells

Few meteorological phenomenon are as intense, dramatic, and dangerous as a supercell thunderstorm. What makes these storms "super" is their unique structure. While ordinary storm cells consist of an updraft and downdraft that essentially occupy the same air column, a supercell's updraft is entirely separate from a downdraft. When a downdraft occupies the same space as a storm's updraft, the downdraft will weaken the updraft until it collapses, resulting in the death of the storm cell. Without a downdraft to weaken it, a supercell's core updraft can maintain itself for a very long long time. Thus, while a typical thunderstorm might last 45 to 90 minutes, a supercell can exist for six to eight hours (or more).

Supercells occur across the world, wherever sufficient instability and wind shear are present. However, several hot spots include Bangladesh, Argentina, Japan, and most famously, North America east of the Rocky Mountains. All of the locations feature areas of intense daytime heating, unstable low-level air flow, relatively dry air flow aloft, and a significant amount of wind shear.

Because supercells are best studied using weather radar (Figure 1), this post will focus on storms in the US, since it is the only country that provides full radar data access to the public. In principle, most of this material can be applied worldwide (although some of the structure described may need to the mirrored in the Southern Hemisphere).

Figure 1. This 1953 radar image was the first case in which a supercell's structural feature was directly associated with an observed tornado.

Anatomy of a Supercell

To understand supercell thunderstorms, it helps to understand their key structural components. Below is an excellent diagram of an idealized classic Central Plains supercell as seen from a distance, looking towards the northwest (Figure 2), although real supercells (Figure 3) often deviate from this form to some degree. The order that those features are presented here vaguely follow the path that air flows through the storm.

Figure 2. This idealized diagram depicts what a supercell might look like when facing to the northwest.

Figure 3. A panoramic photograph displays most of the features depicted in Figure 2.

Inflow
The inflow supporting the main updraft of a supercell typically consists of warm, moist air moving northward (Figures 4 and 5). A key ingredient to this air is a decent amount of horizontal vorticity; which can be thought of as the inherent "spin" to the air with an axis parallel to the ground. In special circumstances, this can be seen on radar images as initially parallel lines that feed into the core of the storm.

Figure 4. On May 20, 2013, a line of supercells formed along a boundary in the southern Great Plains. While this storm was relatively benign, further north was the supercell that spawned the devestating EF-5 tornado in Moore, OK. The inflow of the storm (orange) displays the signature parallel lines of small cumulus clouds.

Figure 6. Striations along the underside of this supercell are due to inflow approaching the main updraft of the storm (in the distant background)

Updraft
The updraft of a supercell is the main engine that drives the storm (Figure 6). It is an area of intense upward motion so powerful, that it prevents all but the heaviest precipitation from falling through it. At upper levels, the distinctness of the updraft forces air from the surrounding environment to split. The result is a 'V' shape on radar and (sometimes) satellite imagery as precipitation and cloud material carried further downstream in the strong split flow, than in the weakened flow directly behind the updraft. The other function of the updraft is to tilt the horizontal vorticity into the vertical, thus providing the rotation that is at the heart of the supercell and one of its most defining characteristics.

Figure 6. The updraft region is highlighted by the small red area. On low elevation radar, the updraft of a mature supercell will often appear as a notch of low reflectivity. 

Overshooting Top
The updraft of a supercell is strong enough that it carries the condensed air (cloud) above the inversion that is capping the storm (Figures 7 and 8). The convection of a supercell is so deep and vigorous that it often extends through the entire depth of the troposphere. In those cases, the overshooting top actually punches into the typically cloud-free stratosphere.

Figure 7. The overshooting tops can often be seen on visible satellite imagery as a small rough area surrounded by comparatively smooth cloud tops. Note that this feature lies directly above the updraft indicated on radar.

Figure 8. The very top of the overshooting top can be seen in this photo of a distant supercell.

Anvil
While the overshooting top punches through the capping inversion, the rest of the storm's growth is impeded by it and spreads out horizontally instead. The anvil at the top of a supercell is generally comprised of thick cirrus that spreads out and downstream of the updraft (Figure 9). While anvils are common amongst thunderstorms, those produced by supercells are gigantic and can extend hundreds of miles downstream.

Figure 9. The anvil appears much smoother than the overshooting top and extends mainly in the direction of the upper-level flow.

Forward Flank Downdraft (FFD)
After ascending the updraft, precipitation laden air falls in two distinct downdrafts. The largest descends the downstream (or front) side of the supercell and is thus referred to as the forward flank downdraft (FFD) and typically extends roughly toward the northeast (Figure 10). The precipitation associated with the FFD is delineated by weight, with large hail falling close to the updraft and light rain falling furthest from the updraft. Some of the moist air produced by evaporated precipitation in the downdraft is pulled back into the updraft and may condense, creating what is often called a tail cloud. After the evaporatively cooled air reaches the surface, it spreads out in a gust front that can trigger new convection.

Figure 10. The FFD is usually seen on radar as a vast area of precipitation with the strongest echoes forming a vague 'V' shape due to the split in winds caused the the updraft.

Rear Flank Downdraft (RFD)
The other downdraft of a supercell essentially wraps around the updraft on the upstream (of back) of the storm, and is thus called the rear flank downdraft (RFD). Compared to the FFD, the RFD is much smaller, but is much more intense and dynamic (Figure 11). There is often little precipitation associated with the RFD, since it tends to evaporate quickly in the rapidly descending air. When the RFD reaches the ground it creates a powerful gust front that can contain dangerous straight-line winds (Figure 12). Like the FFD, some moist RFD air may get pulled back into the updraft, creating a low cloud in the process.

Figure 11. The RFD, indicated by the blue outline, wraps around the updraft. In this case, its gust front has triggered new convection.

Figure 12. RFD winds are made visible here by the dust that is being lofted; some of which is even being pulled in the the updraft (to the right).

Flanking Line
In many cases, the RFD's gust front triggers the formation of convection immediately upstream of the supercell (Figure 13). These convective towers are known as the flanking line and are sometimes deep enough to produce their own precipitation (Figures 14 and 15).

Figure 13. The supercell featured in most of the previous diagrams is seen here near the bottom of the image, while the Moore tornado producing storm is the cell furthest north. What appears to be flanking line clouds extend immediately to the west of storm.

Figure 14. This supercell near Wichita, KS in 2012 occured at the same time as an inversion that allowed the radar to pick up low level features. The flanking line convection is evident in center of the reflectivity image (left) by the dark blue line. Striations caused by the inflow can be seen to the south of the storm, particularly in the radial velocity image (right), with  a few lines wrapping into the storm's core

Figure 15. An radar isosurface of the same storm as in the previous figure looking toward the northeast with the flanking line in the foreground, the RFD to the left (mostly invisible on radar), and the FFD in the background.

Mesocyclone
The mesocyclone is the rotating core of a supercell and their single most defining feature (Figure 16). The full structure and evolution of a mesocyclone is not completely understood and beyond the scope of this post. Essentially, though, it is co-located with the updraft to the extent that supercells are sometimes said to have a rotating updraft. The rotation is associated with the vorticity that the updraft ingested. As the air rapidly ascends the vorticity is said to be stretched, increasing the rotation in order to adhere to the conservation of angular momentum. In cases where there is relatively little cloud material associated with the supercell, corkscrew-like striations can be seen spiraling up the main core of the storm (Figure 17).

Figure 16. The yellow circle indicated the estimated location of the mesocyclone. Since the rradar is to the northwest of the storm cyclonic (counter-clockwise) rotation is evident by outgoing (red) velocities to the southwest of incoming (green) velocities.

Figure 17. This supercell near Northfield, TX on May 23, 2016 had little cloud material, allowing the striations caused by the rotating mesocyclone to be easily observed.

Wall Cloud
Sometimes a portion of the supercell's cloud base will descend below the mesocyclone (Figure 18). This rotating cloud feature is known as a wall cloud and is a feature storm chasers look for, as it often means the formation of a tornado is imminent.

Figure 18. While this wall cloud did not produce a tornado, a slight horizontal extension on its right indicates inflowing moisture from the FFD (distant far right). There might be hints of the RFD in the form of a sharp boundary seen near the surface on the left side of the image.  

Tornado
A tornado is the most violent manifestation of a supercell, as well as the most dangerous (Figure 19). Most supercell-spawned tornadoes descend from the wall cloud. As the winds inside the tornado increase, the pressure decreases, causing condensation and the formation of the iconic funnel cloud.

Figure 19. This high resolution image shows the 2013 Moore tornado (centered on the small red dot) just before it caused its first area of EF-5 damage. Much of the reflectivity (left) near the tornado is actually debris, while the sharp change in wind direction associated with the rotation is evident in the velocity image (right).

Lifecycle of a Supercell

Pre-Storm Environment
In order for supercells to form, several meteorological conditions must exist. First, the environment should be characterized as being conditionally unstable. This means that a low capping inversion prevents air parcels from ascending to any significant degree, but above that temperature decreases rapidly with height (ideally all the way up to the tropopause). Thus, if something were to erode the inversion, air parcels could rise to a considerable height.
Second, considerable wind shear must be present, with the most intense supercells forming when wind direction changes in both speed and direction with height (Figure 20). In the US, this condition is usually meet by steady low-level flow from the south to southeast and upper-level winds from the southwest to west. Additionally, the low-level flow should be considerably warm and moist.
Finally, the low-level flow should have considerable horizontal vorticity. Often this is provided by long parallel lines of small cumulus clouds that form early in the day called cloud streets (Figure 21). Vorticity is generated in these streets because the rising air in the cumulus updrafts is paired with the descending air between the streets to form. The tops of these cumulus correspond to the bottom of the inversion layer and that height is usually about a third the distance between the streets (Figure 22).

Figure 20. Earlier in the day of the Moore tornado, the wind profile showed considerable turning with height. Near the surface winds were from the south-southwest, while at upper levels they were from the west. 

Figure 21. Roughly three hours before the Moore tornado, cloud streets dominated the region. Note that the first bit of deep convection is just starting to form.

Figure 22. This diagram explains how vorticity is generated by the updrafts and downdrafts of adjacent cloud streets.

Storm Formation
Thunderstorms will begin to develop when something removes the capping inversion that is preventing air parcels from creating deep convection. Often, this mechanism is daytime heating. When the layer near the surface becomes very warm due to being in close proximity to the heated ground, near surface air parcels become extremely buoyant and begin pushing into the inversion and mixing in. If enough of this air mixes with the air in the inversion, the vertical temperature profile will shift until the inversion ceases to exist. It is because of this mechanism that most supercells form in the late afternoon or early evening.
Another common mechanism is forced lifting due to some boundary. This boundary might be outflow from a distant storm complex or a synoptic feature such as a front or dryline. When the warm low-level air meets the boundary, it is forced to rise (even through the inversion). Then, like the daytime heating mechanism, this forcibly lifted air mixes with the inversion layer until it dissipates.
Once the capping inversion is removed, convection often begins building extremely rapidly. Sometimes turning a nearly clear sky into one full of intense thunderstorms over the course of just an hour. It is because of this that an area is said to 'erupt' in activity (Figure 23).

Figure 23. This animation of the day of the Moore tornado shows how quickly deep convection forms once the capping inversion has been lifted.

Splitting Storms
Storm splitting is a key step in the formation of supercells, yet a full explanation is beyond the scope of this post, thus a simplified description will be given here.
As storms grow, their updrafts lift the inherent vorticity into the vertical (Figure 24). This can be envisioned as a rotating tube whose middle has been lifted up. Once lifted, the tube would have two vertical portions, spinning in opposite directions. In a thunderstorm, these vertical sections become the rotating cores of supercells. Through a series of complex dynamic processes, new updrafts form on either side of the storm (paired with the rotations). Meanwhile, precipitation associated with the original thunderstorm cell creates downward motion between the two new cores as well as a pool of cool air at the surface beneath it. The central downdraft works to sever the connection between the cores, while the expanding cold pool forces the new updrafts apart from one another. The end result is two storms with rotating updrafts that are separated from their respective downdrafts (portions of the original downdraft).
These two new supercells will move roughly 30 degrees to the left and right of the original storm's motion (Figure 25). However, when the vertical wind shear changes direction with height, dynamic interactions will cause one of the supercells to further develop and slow down while the other dissipates. In the US, the right moving cell nearly always dominates. For this reason, these supercells tend to have a greater eastward component to their motion compared to non-supercellular storms in the same environment (Figure 26). The remainder of this post will assume a right moving storm.

Figure 24. Vertical rotation is created when an updraft lifts horizontal rotation into the vertical (top). Opposing rotating cores and a central downdraft cause the storm to split in two (bottom).

Figure 25. Here, two supercells are formed by splitting. Notice on the right side of the image that the left mover has moved farther than the right cell and is somewhat smaller.

Figure 26. This animation shows the formation of a right moving supercell near Big Springs, TX on May 22, 2016. Notice how the storm makes a sharp turn to the right and slows down considerably (relative to the other storms to its north).

Weak Echo Region (WER)
As the storm intensifies, the angle at which the updraft ascends becomes increasingly more vertical. At the same time, the separation of the updraft from the downdraft (the FFD or its precursor) results in precipitation forming in the middle and upper levels of the updraft, but falling down-shear of it. Down-shear simply means downstream in the direction of the vertical wind shear that is driving the storm.
While the precipitation free portion of the updraft will appear visually as a rain-free portion of the storm cloud, on radar this region will have little or no reflectivity. A radar cross-section will depict the storm having an overhang, with high reflectivity located almost directly above a region of little to no reflectivity (Figure 27). Therefore this region is referred to as a weak echo region (WER). A storm with this signature feature should be closely monitored for severe weather and further development.

Figure 27. This cross-section through the supercell shown in Figure 17 displays a clear overhang with high reflectivity values (red and purple) located above much lower values (blue and green).

Hook Echo
Eventually, the RFD will form and begin to curve around the back side of the storm, carrying with it precipitation. When the precipitation associated with the RFD begins to make its way towards the south side of the storm, the supercell's radar signature will appear to form a pendant or hook. This is the infamous "hook echo" that indicates an intense supercell that likely has a robust mesocyclone core and has the potential for tornadic development (Figure 28).

Figure 28. The Moore supercell displayed a clear hook echo shortly before the tornado touched down. Note that this image is from the same time as the satellite image in Figure 13.

Bounded Weak Echo Region (BWER)
At its peak intensity, the supercell's updraft will become nearly vertical. This, combined with the precipitation wrapping around the updraft will result in cross-sections through the storm's core to appear to have a hole extending up into it (Figure 29). This is know as the bounded weak echo region (BWER) and can sort of be thought of as a portion of the WER that has been enclosed on all sides, versus just one (Figure 30).
The presence of a BWER is an indication that a large amount of mass (rain and hail) is being held aloft. Like the hook echo, a supercell with a BWER should be considered extremely intense and is likely to produce severe weather (if it hasn't begun to already). It is at this time that the largest hail reaches the ground.

Figure 29. The Big Springs supercell exhibited a classic BWER radar signature.

Figure 30. This is the same radar isosurface as Figure 15, but looking toward the northwest. The little "cave" to the right of the center of the image shows where inflow is entering the BWER region.

BWER Collapse
As the supercell ages, and the RFD continues to further limit the inflow of low-level air into the updraft, the BWER often appears to begin to shrink. This BWER collapse phase is accompanied by an overall decrease in the storm's reflectivity and its apparent height. However, it is at this time that the supercell is at its most dangerous. During the BWER collapse is when tornadoes are most likely to form and when they will be their strongest.


Mesocyclone Occlusion
Eventually, the RFD wraps completely around the updraft, cutting the updraft off from the inflow in a process known as mesocyclone occlusion. On radar, it will appear that the mesocyclone is moving from the edge of the storm towards its center. Often tornadoes, which are fixed to the mesocyclone, will appear to track slightly to the left of their original heading. At this time a new updraft and mesocyclone is forming on the edge of the supercell and will replace the old one as the new core. Supercells will often undergo this cycle many times throughout their lives (Figure 31).

Figure 31. The Moore supercell's full lifetime is shown in this animation. The estimated location of the updraft is in the center of the white circle, which is 30 miles in diameter. The EF-5 tornado dissipates shortly before the storm passes very near the radar unit. Throughout the animation, several occlusion cycles occur, as indicated by the formation and disappearance of several hook echoes.

Supercell Varieties
Meteorologists often classify supercells into different modes, mostly based on the relative strength of the inflow into the updraft compared to the outflow from the downdrafts. The spectrum ranges from inflow dominant low precipitation supercells to outflow dominant high precipitation supercells. Often storms will switch between modes as they evolve and undergo internal changes such as mesocyclone occlusions.


Classic Supercells (CL)
Supercells whose inflow and outflow are roughly in balance are referred to as classic supercells (CL). These storms are the textbook type often seen across the great plains of the US. They often exhibit well defined hook echoes and tend to produce the strongest tornadoes (Figure 32).

Figure 32. The Big Springs supercell at a time it had a classic CL structure; complete with a hook echo and a 'V' shape of the highest reflectivities.

Low Precipitation Supercells (LP)
In drier air masses, such as is often observed in eastern New Mexico and western Texas, supercells tend to be dominated by their updraft and have little precipitation associated with them. The relative lack of precipitation and reduced amount of cloud material often allow many of the core structures of the storms to be visible. For the same reason, they have a very small radar signature and can be difficult to identify (Figure 33). The main hazard from these storms is very large hail, although tornadoes do occasionally occur. These storms have a tendency to transition to a more CL structure as they move eastward where there is often more moisture.

Figure 33. This Kansas LP supercell displays most of the primary features of supercells, although they are not well pronounced. The radial velocity image (right) does hint at the location of a mesocyclone.

High Precipitation Supercells (HP)
The most commonly observed supercell mode it the outflow dominant high precipitation supercell (HP). These are especially common in the southeast US during winter. HP supercells are particularly dangerous for several reasons. First, the intense amount of precipitation often obscures many of the key structures that help supercells be identified on radar (Figure 34). Second, tornadoes spawned from these storms are often hidden inside heavy precipitation. Finally, they are commonly active at night, further reducing the chance of noticing an approaching tornado. CL supercells will often take on HP characteristics temporarily after a mesocyclone occlusion.

Figure 34. A line of storms in central Texas contained a HP supercell. Many of the classic features of supercells are either absent or hard to detect in this storm. However, the radial velocity (right) clearly shows a rotating core. 

Miniature / Low-Topped Supercells
Under certain conditions, supercells can form far out of season or in places that otherwise do not see many supercells. These storms are typically much smaller than the other varieties and generally have much lower tops (Figure 35). Mini supercells are still considered dangerous as they can still produce hail and tornadoes (Figure 36). The fact that they are so small often makes them hard to identify on radar due to the radar's ability to discern their structure. Furthermore, since they form in regions unfamiliar to supercells, local meteorologists might not be able to readily identify them.

Figure 35. This mini supercell north of Seattle, WA on October 20, 2012 was covered in a previous post. A classic supercell shape is faintly recognizable, but the reflectivity values (left) are very low compared to typical supercells. Although there does appear to be rotation (right), it is not particularly robust.

Figure 36. Despite being small and weak, the supercell did produce a well defined wall cloud and at least one waterspout (possibly three).

Why Matthew Matters

In 2012, just days before Hurricane Sandy made landfall on the US East Coast I wrote the following opinion piece:

Computer monitors.
That is how most atmospheric scientists see the subject of their careers. Across countless monitors flash high resolution satellite images of entire continents, data and maps just received from the latest computer model run, or maybe just some point observations. Your hometown, maybe even the state you live in might be represented by a single station model on a map or a handful of pixels on a satellite image. We atmospheric scientists sometimes have the rather unbecoming tendency to see weather at a large scale. Most of the time, weather appears as charts depicting the smoothed out distribution of parameters such as pressure or temperature. Even our computer models lead to disconnect; their resolution might be on the order of twenty or thirty miles, sometimes even more. All this can lead to the desensitization of what is actually happening on the ground. It can be all too easy to forget that below the anvils of supercells in an impressive squall line, there might be a tornado tearing through someone's house, or hail destroying a season's worth of crops. Tropical cyclones are even worse. Who could not marvel at a perfectly shaped spiral of brilliant white clouds spinning its way across the ocean? This is all perfectly fine when the cyclone is over open water, but what about when it makes landfall? Is the first thing that comes to mind that densely populated city on the coast in a country where people can't always afford the sturdiest of dwellings? A lot of people, including atmospheric scientists probably won't. This does not mean they're bad people; it’s just an unfortunate tendency of human behavior. On the other hand, those who study these various spectacles of weather should try to keep those affected firmly in mind. I myself am guilty of this sort of detached perspective; I might be mentally cheering on a storm, hoping to see it break some record, or become overly excited when I correctly forecast a storm. What it all boils down to is a reminder that when viewing a forecast made by a supercomputer or an image straight from the heavens, there are those beneath the clouds that might have a very different perspective on the weather.

Now, as Hurricane Matthew approaches Florida after already making landfalls in Haiti and Cuba, I believe this message still applies. However, this storm highlights not only the socio-meteorological issues, but geopolitical ones as well. It seems that no matter what the event, weather included, society tends to truly care primarily about what is happening in their "backyard". Sadly, the level of interest seems proportional to both racial and economic factors.

One telling factor is what level of damage a tropical cyclone needs to cause in order for its name to be retired by the World Meteorological Organization. It is not uncommon for Atlantic storm names to be retired despite only causing a few dozen fatalities. For example, Isidore was retired in 2002 after causing 19 fatalities and Igor was retired in 2012 with only 4. What these, and other relatively low fatality Atlantic storms typically have in common is their high price-tag. On the other side of the globe, typhoons often cause far greater loss of life, yet are not delisted. The 2010 typhoon season was a great example: neither Conson (111 fatalities) nor the incredibly intense Megi (69 fatalities) were retired. The fact is that neither of the storms caused more than 100 million US dollars in damage. In both cases The Philippines were the hardest hit, however the economic strength of that country is quite low, hence the low damage costs. Thus, it would seem that economic damage is at least as important for retiring a name as is the loss of life. By extension, it also appears that the more influential the affected country, the higher likelihood that the storm name will be removed.

As news agencies across the US provide near-continuous coverage of Matthew and its potential impact on the States, it seems much of the devastation the storm has already caused is swept under the rug. One has to do some research to find out that at least 142 people have already died from this storm; 136 in Haiti alone. The fact that relatively few people in the US die from tropical cyclones is forgotten. Really, the economic damage is what is devastating about US storms, where even a weak system can cause over a billion dollars in damage. At the end of the day, though, financial losses can be restored, human lives cannot.

Windstorms of the Pacific Northwest

Note: this post concerns regional windstorms, as opposed to localized windstorm that also occur in many Pacific Northwest locations.

Much of information and images in this post was gathered from a few key sources:

• An excellent book by University of Washington professor Cliff Mass, The Weather of the Pacific Northwest (2008)
• A related journal article: Mass and Dotson (2010): Major Extratropical Cyclones of the Northwest United States: Historical Review, Climatology, and Synoptic Environment. Mon. Wea. Rev., 138, 2499-2527, doi: 10.1175/2010MWR3213.1. 
• Visit Mass' weather blog here.

Figure 1: The University of Washington Atmospheric Sciences Department's logo: a depiction of the Thunderbird capturing an Orca.

The Thunderbird. According to the Quileute Tribe of the Washington State coast, the Thunderbird (figure 1) was a giant bird with wings twice as long as a war canoe and talons the size of oars who hunted Orca. With every flap of its wings, great winds would be unleashed. Clearly, The Quileute, along with other coastal tribes, knew the power of windstorms, something that is reflected in their practice of moving away from the coast during the winter months. As a hazard, Pacific Northwest windstorms pose a significant and common threat. The coastline from Oregon through British Columbia experiences hurricane-force winds as often (if not more often) than the Atlantic and Gulf coasts experience them from actual hurricanes. Based on NOAA’s Storm Data publication, it has been conservatively estimated that these storms have caused between 10 and 20 billion dollars in damage since 1950. Despite this, relatively little research or media attention has been paid to these storms relative to the East Coast's tropical cyclones. As a result, windstorms have a long history of surprising the region with little or no warning.

Aside from Native legends of great windstorms, written accounts from non-native settlers goes back to the Lewis and Clark expedition in 1805. Since, then many significant storms have been recorded (figure 2), with a few outlined below.

Figure 2: Some tracks of historical windstorms, from Mass and Dotson (2010).

The Windstorm of 1880
The first well documented major windstorm occurred on 9 January, 1880. The storm, which was described by newspapers as being “the most violent storm…since its occupation by white men”, killed one person and destroyed or severely damaged hundreds of buildings, uprooted trees, and damaged telegraph lines throughout Oregon’s Willamette Valley. Near the coastal town of Newport, every barn was destroyed, and in Coos Bay, a schooner dragged its anchor, was blown onto the beach, and broke in two.

The Great Olympic Blowdown
A few decades later, another major windstorm swept through the region, although this one’s damage was focused in Washington instead of Oregon. The Great Olympic Blowdown of 29 January, 1921 was notable for bringing hurricane force winds to a long stretch of Washington and Oregon coasts. The North Head Weather Bureau station, located on the northern tip of the mouth of the Columbia River measured a peak gust of 130 kt before the anemometer was destroyed. The exact wind speed will never be known since at the time, wind was measured using four cup anemometers, instead of the now standard three cup versions. Along the entire coast of the Olympic Peninsula at least 20% of all trees were blown down, with some areas seeing as much as a 40% loss (figure 3). Due to the devastation, roads into the backcountry became impassible, resulting in billions of board feet of timber being unable to be salvaged by the logging industry. By comparison, the 1980 eruption of Mount Saint Helens leveled only an eighth as much forest.

Figure 3: Olympic Peninsula timber damage from the 1921 windstorm from Mass and Dotson (2010).

The Windstorm of 1934

Over the course of the next few decades, two more historical windstorms impacted not just the coast, but also the interior regions of the Pacific Northwest; the Puget Sound region in particular. The 21 October, 1934 windstorm brought gusts of 50-60 kt to Puget Sound where it kicked up waves 6 m high and brought down roofs across the Seattle area. At Boeing Field, four aircraft were destroyed when winds lifted a hanger which then landed upon them. Fires occurred throughout the interior cities and spread rapidly by the wind, resulting in the busiest day in the Seattle Fire Department’s history up until then. A total of twenty-two fatalities were attributed to the storm.


The Windstorm of 1958
On 3-4 November 1958, a storm tracked through the region near Olympia, WA, and Mt. Rainier. This unusual westerly track resulted in one of the few windstorms to produce both northerly and southerly high winds. Northerly winds of 40 to 55 kt occurred over the central and northern Puget Sound region, while southwesterly winds of 60 to 75 kt were observed in areas south of Olympia. Along Oregon’s coast, nearly every major roadway was blocked by downed trees.

The Columbus Day Storm
In early October, 1962, Typhoon Freda (figure 4) formed east of the Philippines and recurved to the northeast while undergoing extratropical transition. After heading east across the Pacific during the following days, it made a sharp turn northward while roughly 1900 km northwest of Los Angeles and began to rapidly intensify. On 12 October, the storm reached its lowest pressure of 956 hPa (1 hPa equals 1 mb) while 480 km southwest of Brookings, OR. Over the next 18 h, the storm tracked nearly parallel to the coast before making landfall (figure 5) on the northwest tip of the Olympic Peninsula without any significant change in intensity. Heavy winds were experienced from northern CA, to southern BC, with wind gusts reaching 179 kt at Cape Blanco's Loran Station in southern OR, and 139 kt at the Naselle radar site in southwest WA. Meanwhile, the interior regions of the Willamette Valley and Puget Sound experienced gusts between 80 and 110 kt.

The Columbus Day Storm was likely the most damaging windstorm since the arrival of European settlers and quite possibly the strongest non-tropical storm to strike the contiguous US in the past century. In total, 46 people died and another 317 required hospitalization, while one million homes lost power. Along with the direct effects on humans, 15 billion board feet of timber was downed, 53,000 homes were damaged, and thousands of utility poles were toppled. The total damage was about a quarter of a billion 1962 dollars; if the storm were to strike today’s Pacific Northwest, the damage would likely be in the billions or tens of billions. The Columbus Day Storm also highlighted another significant difference between Pacific Northwest windstorms and tropical cyclones: inland wind speed. While even the strongest tropical cyclones typically do not produce heavy winds very far inland, the Columbus Day Storm (along with some other windstorms) produced winds greater than 85 kt at sites at least 80 km from the coast.

Figure 4: Path of the Typhoon Freda/the Columbus Day Storm. The triangle points are when the storm was extratropical. Insert is a satellite image of Typhoon Freda.

Figure 5: Surface analysis of the Columbus Day Storm, from Mass and Dotson (2010).

The Hood Canal Storm of 1979
On occasion, the terrain can even induce mesoscale-like lows in their lee. This was the case on 13 February, 1979 when a rather typical large scale winter windstorm caused a localized low center to form near Port Townsend. The low resulted in a small area of very high southerly winds over the Hood Canal that reached over 85 kt. A 975 m section of the Hood Canal floating bridge (figure 6) was lost (at the cost of 140 million 1979 dollars) and up to 80% of timber in the area was blown down.

Figure 6: The Hood Canal Bridge after its collapse in 1979.

The Twin Storms of 1981
On 13-15 November, 1981, a series of two windstorms followed near identical paths, making landfall on central Vancouver Island. The first storm (figure 7), which was by far the stronger one, brought high winds to locations along the Oregon coast, including an unofficial gust of 100 kt. Offshore, two ships reported gusts of 85 kt and another ship reported 10 m seas. In the Seattle area, the Evergreen floating bridge (SR 520) recorded gusts up to 65 kt and received thousands of dollars worth of damage that required an eleven hour closure. The second, much weaker storm (figure 8) brought lower wind speeds overall, but, strangely, caused far more damage to trees. In total, thirteen fatalities were attributed to the storms, while four thousand residents lost power.

What these two storms highlighted was the incredible difficulty in forecasting massive windstorms. Even just 24 hours before the first storm, NWS forecasts had little indication that an incredibly rapid intensification would occur. The only warning came eight hours before the storm when a local television meteorologist rushed a videotape of an animation of satellite images showing the intensifying storm over to the Seattle NWS office. Thankfully, weather forecast offices have since been equipped to make their own animation loops.

Figure 7: A visible image from about 14 November, 0000 UTC of the first of the two 1981 storms. This was captured soon after the storm's explosive intensification.

Figure 8: A visible image from about 15 November, 2100 UTC of the second of the two 1981 storms.

The Inauguration Day Storm
On 20 January, 1993, the day of President Bill Clinton’s inauguration, a storm blasted through Oregon and Washington that has been considered to be the third most damaging in the past 50 years. At 0000 UTC on the 20th, the low was located roughly 1000 km west of northern California. By 1500 UTC (figures 9 and 10), while the low was offshore the Columbia River outlet, the pressure had dropped 14 hPa to its lowest value of 976 hPa. The storm brought hurricane force winds to the Puget Sound area and the atmospheric sciences building at the University of Washington in Seattle measured a record wind speed of 76 kt. In Washington, six people died from the storm, while about 870,000 people lost power. The insured losses from the storm were estimated at 159 million 1993 dollars, with 79 homes and 4 apartment buildings being destroyed, and another 581 dwellings were severely damaged.

One of the aspects that made the Inauguration Day Storm significant was that it was the first major windstorm to be skillfully forecasted. As a result, the National Weather Service was able to issue a high-wind watch at 2130 UTC on 19 January and a high-wind warning 0600 UTC on the 20th. As is typical of most Pacific Northwest windstorms, however, the forecast was widely ignored by the media at large.

Figure 9: IR image of the Inauguration Day Storm at 1500 UTC on 20 January as it was making landfall on the Washington coast.

Figure 10: This the surface analysis of the storm, also at 1500 UTC, from Mass and Dotson (2010).

The Windstorm of 1995
Possibly the most skillfully forecasted windstorm occurred on 12 December, 1995 (figure 11). The storm brought hurricane-force gusts along a stretch of the west coast extending from San Francisco all the way to southern British Columbia. Five fatalities and over 200 hundred million dollars in damages were attributed to the storm, along with the loss of power to a total of over 1.3 million people.
What was fortunate about the storm was that forecast models predicted it up to four days in advance. Furthermore, it occurred during a major research program, COAST (Coastal Observation And Simulation with Topography experiment), which included surveillance of the storm's structure by one of the planes (a P-3 Orion) typically used for hurricane reconnaissance.

Figure 11: A visible image of the 1995 storm at 1930 UTC on 12 December.

The South Valley Surprise
Unfortunately, forecasts are still far from perfect, as was made evident on 7 February 2002, when a rapidly developed storm (figure 12) brought strong winds to Southern Oregon with practically no warning. The system resulted in the toppling of thousands of trees and power lines. The winds experienced in the southern Willamette Valley were likely the strongest since the Columbus Day storm.

Figure 12: A visible image of the South Valley Surprise at 2300 UTC on 7 February.

The Hanukkah Eve Storm
The strongest windstorm to strike the region since the Columbus Day Storm occurred on 14-15 December, 2006. This storm took a much more westerly track than most windstorms, making landfall (figure 13) along the central coast of Vancouver Island. Most of the damage from the Hanukkah Eve Storm occurred in western Washington, where 1.3 million people lost power, some for well over a week. Across Washington and Oregon, at least 13 fatalities were recorded and the total damage was estimated to be between 500 million and one billion 2006 dollars. In magnitude, this storm was comparable to the Inauguration Day Storm, yet the damage in western Washington was significantly greater. Much of this increase in damage was due to the record breaking amount of precipitation during the preceding November that left soils highly saturated, allowing trees to be uprooted far easier. Another factor influencing the damage total was the incredible amount of population growth that had occurred in the Seattle area since 1993, leaving far more people vulnerable.

Figure 13: An IR image of the Hanukkah Eve Storm at 0300 UTC on 15 December. Note the scale of the storm compared to most of the others

The Summer Windstorm of 2015
While windstorms of various intensities typically occur at least once a year, the 23 August, 2015 storm (figure 14) stands out by being the only recorded windstorm to occur during the summer. Most Pacific Northwest storms occur between November and March, when large extratropical cyclones dominate the region's weather. During the summer, high pressure usually sets up, bringing very dry conditions and clear skies. While the storm was far weaker than the other historic storms covered here, gusts over 35 kt were recorded at the Evergreen floating bridge before its instruments went offline. Like many early season storms, this oddity likely began its life as a tropical system that then became extratropical.

Figure 14: A visible image of the 2015 summer storm at 1600 UTC on 29 August.

Meteorologically speaking, there are several factors contribute to intensity of Pacific Northwest windstorms. Perhaps the most important is the effect of local topography. The many mountain ranges of the region often channel or enhance wind speed. It is for this reason that many of the record wind measurements, such as those at North Head (figure 15), were made at sites atop mountains or along the coast, where coastal mountains enhance winds along their windward slopes. This is especially true of the southwest flank of the Olympic Mountains, which receives the full force of the southwesterly winds. 

Figure 15: The North Head Lighthouse, where many of the record winds have been recorded. Notice how it sticks out from the coast and at the top of a pronounced slope.

A second major factor in Pacific Northwest windstorms is the location of highest winds relative to the low center. Unlike tropical cyclones whose maximum winds are found adjacent to the eye, or inland mid-latitude cyclones whose wind maxima are often found associated with a cold front, studies have found that these lows have maximum winds near a trough, well behind the cold/occluded front. This structure, sometimes called the poisonous tail of the bent-back occlusion (figures 16 and 17) typically rotates cyclonically around the south side of the low in the unstable air of the cold sector.

Figure 16: A surface model of the Hanukkah Eve Storm from the same time as figure 13 showing the bent-back trough extending south from the low’s center with very closely spaced isobars. From Mass and Dotson (2010).

Figure 17: A model of  near surface wind speed from the same time as figure 13 and 16. The black, red, and white areas show the strong winds associated with the bent-back trough. From Mass and Dotson (2010).

During a Pacific Northwest windstorm, the greatest danger to both lives and property are falling trees and branches. This type of damage is so prevalent, that tree damage, as inferred via tree ring width data, has been used to estimate the timing of major windstorms as far back 1701. A study in 2009 found that between 1995 and 2007, non-convective winds were associated with 143 fatalities, or 35% of all deaths due to wind related tree failures in the United States (figure 18). Of these, 28 were in Washington alone, making the state the second deadliest (after New York) in terms of all fatalities associated with tree failures due to high wind.

Figure 18: Fatalities due to non-convective wind tree failures between 1995 and 2007.

Trees typically fail in a windstorm either by snapping off the upper portions and branches or by being uprooted altogether (figures 19 and 20). In the Pacific Northwest, trees regularly reach 50 m or higher and are fast growing. These factors make them highly susceptible to windsnap due to their large exposure to the wind. The wet autumns and winters of the region often cause the soil to become saturated, making the ground less able to hold the trees’ roots. In more recent decades, tree failures have become more common as developers clear land for buildings. In doing so, they expose parts of the forest that had previously been protected by surrounding trees; the long-time structure of the forest having grown shorter, hardier trees along the edges and weaker, taller trees in the interior. Perhaps an even greater issue is that a few trees are often left standing as scenic backdrop. These isolated groups of trees are highly susceptible to wind damage. Since this pattern is often found in neighborhoods, it is not uncommon for roofs to be heavily damaged by large pieces of failing trees. In cases where trees become completely uprooted (figure 21), the heavy trunks can literally slice a house in two.

Figure 19: Extensive tree damage from a smaller scale windstorm in 2007. From Mass and Dotson (2010).

Figure 20: This picture (and the remaining images) is of damage caused by the 2006 Hanukkah Eve Storm.

Figure 21: A house has been sliced in half by an uprooted tree.

Finally, there is the issue with electrical power. In the Pacific Northwest, most power lines are above ground and thus easily broken by falling trees (figure 22). This is such an issue that most power lines are equipped with explosive charges that cut-off power when large current is created due to the grounded line. The flashes and booms accompanying these explosions can travel for miles and thus are often heard and/or seen repeatedly throughout a windstorm. Since above ground power lines pose such an issue, why are they not placed underground? The answer is: nobody knows. After every windstorm, the suggestion is commonly made by the public, but after a few months, it seems to be forgotten...until the next windstorm.

Figure 22: Trees brought down countless power lines during the 2006 storm.

Windstorms in the Pacific Northwest are a fact of life. Most residents keep multiple battery-powered lamps available; some even invest in personal electrical generators. As regular as these events may be, it is important to remember that they are both dangerous and expensive, with an estimated average cost of 112 million dollars per event, based on data from 1952-2006. Despite this, they are too often underestimated and under-researched; only since the 1990s has there been any real skill in forecasting these storms. Clearly, much still has to be done in understanding, and mitigating the effects of the windstorms of the Pacific Northwest. Until then, the Thunderbird will continue to dominate its domain.