From: Report of the Proceedings (1994) of the U.S.-Spain Workshop on Natural Hazards (Barcelona, Spain, 8-11 June 1993), J. Corominas and K.P. Georgakakos, Eds., pp. 44-55. [Available from the Iowa Institute of Hydraulic Research, University of Iowa, Iowa City, Iowa 52242]

Extreme Convective Windstorms:

Current Understanding and Research



NOAA/ERL/ National Severe Storms Laboratory

Norman, Oklahoma U.S. A.


Convective windstorms are driven by downdrafts, the physics of which are relatively simple and correspondingly well understood. Damaging downdrafts are now often referred to as downbursts, and occur over a range of scales. Downburst also span a range of intensities, and even relatively mild convective windstorms can be a danger in some societal settings (e.g., recreational activities, aircraft operations, etc.). Moreover, convective windstorms can be isolated events or they can occur in large numbers within mesoscale convective systems. Those events which produced widespread damaging wind have recently come to be known as derechos.

Extreme convective wind events can be associated with either extreme intensity, or with large numbers of downbursts, or occasionally with both. Extremely intense events can attain windspeeds comparable to strong tornadoes: 75 m s-1 gusts are possible, with 50-60 m s-1 occurring virtually every convective season in the U.S. In derecho situations, damaging winds (exceeding 25-30 m s-1) can occur over areas of up to 2000 km2, and any one location might experience such winds for 30 min or more, with peak gusts approaching 60-70 m s-1. The societal threat from such storms is comparable in terms of devastation to tropical cyclone events.

It recently has been suggested that some extreme wind events may be the result of supercells embedded within the mesoscale convective system. Often such storms are accompanied by large hail (> 5 cm in diameter), which can increase the damaging effects of the strong winds. Recent research, including numerical cloud modeling and observations has suggested a consistent environmental structure associated with supercell-related windstorms. Thus, it appears that forecasting skill can be developed with such events.

1. Introduction

Convective wind events are a hazard to societies the world over, doing considerable damage and occasionally generating many casualties.[1] Most convection produces some straight-line wind as a result of outflow generated by the convective downdraft, and so anyone living in convection-prone areas of the world has experienced this phenomenon. On rare occasions, the intensity of the wind achieves the potential for doing damage. Whether or not damage actually occurs is the dependent on having structures in the path of the wind that can sustain damage. Although engineered structures typically are quite resistant to wind damage, many homes and outbuildings are quite vulnerable to damage from even relatively modest windstorms. In the United States, it is assumed that the potential for wind damage begins at around 25 m s-1 (50 kts). Of course, considerable damage occurs in situations where there was no anemometer, and so wind damage is graded according to its character: e.g., damage to tree limbs is considered non-severe, but uprooted trees is considered to represent a severe event. Figure 1 shows the recorded distribution of wind events considered to be severe in the United States.

Various human activities place people at risk from convective winds, notably aircraft operations and recreation. Most casualties from convective windstorms in the United States arise from such situations. Given the high vulnerability of aircraft operations during takeoff and landing procedures (the aircraft are operating on the margins of their flight "envelope" during such times), it does not take a particularly intense event from a meteorological standpoint to create many casualties. Commercial aircraft are less vulnerable than private aircraft, but their high occupancy means that rare events can have a large impact on casualty figures. Recreational boating also can account for many casualties in relatively modest windstorms, whereas most commercial craft are unlikely to be affected by marginal convective wind events.

Figure 1. Contours of average annual frequency of reported severe convective wind gust events, per 10,000 mi2 (25,900 km2) for the period 1953-1980. Maxima are denoted by "X" and minima by "N" (from Doswell 1985).

In general, then, convective wind events and wind-vulnerable human situations are both common, whereas it is only infrequently that the two components of a convective wind-induced disaster are concatenated. In most such situations, the damage is confined in time and space because convective windstorms operate on time and space scales associated with single convective storms. The effected areas are only a few km2 and the intense phase of the event is only a few minutes in duration. The opportunities for such space- and time-limited events to have a significant impact are few in number. On the other hand, occasionally, convective storms become prolific wind-producers, lasting for several hours and potentially affecting areas as large as 2000 km2. These extreme events clearly have enormous damage and casualty potential. Although extreme events are, virtually by definition, rare events, their impact can approach that of tropical cyclones and certainly exceeds that of all but extreme tornadoes. Therefore, it appears that, with the growth of populations at risk from convective wind events, an effort to mitigate their casualty production is worth pursuing. Perhaps relatively little can be done to deal with the damage potential, except perhaps where building practices can be modified to make structures less vulnerable than they now are to low-end convective windstorms; it may not be cost-effective to build ordinary structures to withstand the rare extreme events.


2. Basic Physical Processes

Convective wind is the result of convective downdrafts, so to understand convective windstorms, one must understand the nature of downdrafts. This can be done with the aid of the vertical momentum equation,

, (1)

where w is the vertical component of the wind (i.e., dz/dt, where z is the vertical coordinate), p is pressure, r is density, and g is the acceleration due to gravity. In this form, viscous forces have been ignored. Note that (d/dt) denotes the Lagrangian time derivative; i.e., following an air parcel. Suppose the variables are decomposed into a basic state, denoted by an overbar, and a perturbation, denoted by a prime, such that

, (2)

and assume that the basic state is defined by the following,

. (3)

This decomposition also can be thought of as representing a parcel with perturbed properties embedded within a hydrostatically-balanced environment having no vertical motion. It is useful to observe that when the perturbations are small compared to the basic state,

, (4)

so that when (2)-(4) are used in (1), it can be shown that the parcel's vertical momentum equation becomes

. (5)

If the Ideal Gas Law (p = rRTv) is used, where Tv is the virtual temperature (in deg K), then it is easy to show from (5) that

, (6)

where B is used to denote buoyancy.[2] Note that the virtual correction usually is rather small and to a good approximation, it can be ignored when computing buoyancy. Finally, the effects of precipitation loading on the vertical motion are parameterized[3] by including a term in (5) that decreases buoyancy as the liquid water mixing ratio (l ) increases, leading to the final form of the parcel's momentum equation:


This result contains the basic physical processes associated with the development of downdrafts (and updrafts, as well!). The first term in (7) is the effect of perturbation pressure gradients on vertical motion. In some storms, notably supercells, this term has a large effect on updrafts (Rotunno and Klemp 1982); however, there is not much reason to believe it has much of an impact on downdrafts, at least to a first approximation. Therefore, it will be ignored.

The second term is the effect of buoyancy on vertical motion. Clearly, in the case of downdrafts, we expect to find that B is negative: the parcel should be cooler than its environment. This cooling typically takes place as a result of phase changes (evaporation, melting, and sublimation). As shown by Kamburova and Ludlum (1966), precipitation particles that are small, but in large numbers, promote a maximum contribution to cooling and, hence, to creation of negative buoyancy. The major contribution to this process is from evaporation, so when I refer to "evaporation" in subsequent discussion, it should be understood that other phase changes can be involved, but that their contribution is usually minor.

The last term in (7) is that due to water loading. Whereas evaporation is promoted by large numbers of small droplets, it only takes a few large drops to contribute substantially to the downward acceleration of air parcels. This term is associated with storms having high precipitation rates. Comparing the effects of water loading to those associated with buoyancy, if a parcel has a liquid water mixing ratio of 1.0 g kg-1, this is roughly equivalent to about 0.3 deg K of negative buoyancy; alternatively, 1.0 deg K of negative buoyancy is about the same as 3.0 g kg-1; the latter is a large (but not extreme) value. Therefore, in general terms, negative buoyancy is typically the major contributor to downdrafts.

Finally, it should be observed that the contribution to negative vertical motion associated only with buoyancy (i.e., using pure "parcel theory") results in a prediction of the maximum downdraft of

where NAPE is the Negative Available Potential Energy,

, (8)

and where LFS denotes the Level of Free Sink for a descending parcel and SFC denotes the surface. This means that the maximum downward motion is associated with the integrated negative buoyancy. Even a relatively modest negative buoyancy can result in a substantial downdraft if it is maintained over a relatively large depth. A downward speed of 25 m s-1 results from the relatively modest NAPE value of 312.5 m2 s-2. To a first approximation, the maximum gust is roughly equal to the maximum downdraft speed.


3. Convective Storm-Scale Events

On the scale of a convective storm, downdrafts typically develop as storm-relative winds move precipitation out of the upper portions of the updraft, where it develops, and that precipitation then falls into relatively dry air. This process produces negative buoyancy through evaporative chilling. Also, the water loading effect contributes to downdrafts, as noted above. In the presence of storm-relative wind, which is present in most cases involving convection, the precipitation cascade region is somewhat removed from the updraft. This precipitation cascade is roughly coincident with the downdraft (see Fig. 2).

Figure 2. Schematic depiction of the mature stage of a convective cell, showing the precipitation (drop hatching) and the outflow (stippling).

Downdrafts lag the updrafts that produce them by about half a convective time scale (a useful value for which is the time needed for a parcel to rise from the condensation level to the equilibrium level, or about 20 min). The duration of the maximum gusts with any individual downdraft is at most a few minutes, typically occurring shortly after the downdraft first reaches the surface. Since the spatial scales of most convective cells are on the order of 10 km, the size of the outflow during the time of maximum horizontal wind gusts is only a few km. Such small-scale events have come to be called microbursts (Fujita 1978). It has become clear recently that microbursts, even when their intensity is below the arbitrary "severe" limit of 25 m s-1, are still a considerable hazard to aviation (see Fujita and Byers 1977; Fujita and Caracena 1977, Caracena et al. 1989).

The so-called "dry" microburst event is one created by evaporation, but it is ultimately driven by the initial negative buoyancy. That is, evaporation produces the negative buoyancy, but the environments in which such events occur, namely with deep layers of nearly dry adiabatic lapse rates and low relative humidities (Fig. 3) permit unimpeded descent of the chilled parcels. The circumstances leading to dry microbursts are relatively well-understood and they are fairly easy to forecast in the sense that the conditions in which they arise are fairly restricted (see, e.g., Wakimoto 1985; Caracena and Flueck 1988). On the other hand, the "wet" microburst appears to develop with some contribution from evaporation, and the negative buoyancy is maintained by continuing evaporation. Wet microbursts occur in situations with nearly moist adiabatic environments, having deep surface-based layers of high relative humidity. The key to developing strong downdrafts is the continuing evaporation of condensed water during descent to maintain saturation. Without such evaporation, the descending parcel would warm at the dry adiabatic rate and quickly lose its negative buoyancy. Water loading also may play a larger role than in dry microbursts. Much less is known about how wet microbursts arise and they are correspondingly more difficult to forecast than dry microbursts.

Figure 3. A composite of five 0000 UTC soundings at Denver, Colorado on days that produced dry microbursts on the Front Range area of Colorado (from Brown et al. 1982).

It is important to recognize that the conditions under which downdrafts are unstable do not necessarily coincide with those producing strong updrafts (Johns and Doswell 1993). This means that a convective event with comparatively weak updrafts still can produce strong downdrafts. Some situations are capable of producing both strong updrafts and strong downdrafts, which often occurs with supercells, but not all convective wind events are caused by supercells. This means that severe and hazardous weather predictors keyed to locating conditions for strong updrafts will not detect all cases involving downbursts.

Owing to the small space and time scales of isolated convective storms, it is unlikely that they can produce extreme wind events. Occasionally, a moderately intense wind event is associated with an isolated "pulse-like" convective storm, but meteorologically speaking, such events do not pose much of a threat of a disaster, except when they occur in combination with a particularly vulnerable human activity. Hence, we turn now to events associated with systems of convection.


4. Events Associated with Convective Systems

Conditions in which convection becomes organized into convective systems of various sizes and durations increase the chances for a convective wind-produced disaster, simply because of the increasing time and space scale. The events producing the wind may not be substantially more intense than their more isolated counterparts, but the sheer number of them makes the chances for interacting with humans during their intense phases greater than with isolated storms.

In the United States, widespread convective wind events are sometimes referred to as derechos (Johns and Hirt 1987). Derechos invariably arise in association with mesoscale convective systems (MCSs). An MCS is composed of a number of individual convective cells, often arranged as a squall line of intense convection and a trailing "stratiform" precipitation area (Houze et al. 1989; Loehrer and Johnson 1993). As the convection evolves from the initial cells, new convective cells are initiated along the leading edge of the outflow from preceding cells and the system as a whole can maintain itself as long as sufficient moisture and lapse rates are present in the inflow. Note that convective systems of this sort can have a substantial inflow at low levels by virtue of their motion.

In these situations, it appears that a significant role is played by a system-relative, rear-to--front flow that begins in the lower mid-troposphere and descends to near the surface at the leading edge of the system, where the deep convective cells are. Although the origins of the rear inflow are not entirely clear, such a flow maintains a supply of subsaturated air to drive downdrafts by the evaporation of precipitation brought out of the convective region by a front-to-rear flow that begins ahead of the convective line and extends into the system's anvil region. These features are summarized in Fig. 4. Within such a system, the

potential exists for numerous intense downdrafts, as the rear inflow interacts with the strong convection on the leading edge. As shown in Fig. 5, the damaging winds can be distributed over a large area.

Figure 4. Conceptual model of a Mesoscale Convective System (MCS) with both a leading squall line and an area of trailing "stratiform" precipitation, shown in cross section parallel to the motion of the system (from Houze et al. 1989).

Many important windstorms are associated with bow-shaped radar echoes (Fujita 1978). These structures can arise on scales from 15 to over 150 km (Johns and Doswell 1993) and appear to arise from the development of strong rear-to-front wind flow. Weisman (1993) has suggested that the so-called "bookend" vortices (Fig. 6 ) often observed at opposite ends of a bow-shaped echo arise from tilting of ambient horizontal vorticity; his interpretation of their role is that they can enhance the rear-to-front flow between the vortices by as much as 30% through the development of favorable pressure gradients. This interpretation remains to be validated by observations, although the basic morphology of observed bow echo systems does, in fact, agree with the numerical simulations (Smull and Weisman 1993).

Figure 5. Damage swath produced by the derecho of 5 July 1980; three-hourly squall line positions are indicated by the dash-double dot lines, with UTC hours indicated. Officially measured wind gusts are indicated by the wind barbs, with a full barb indicating 5 m s-1 and a flag indicating 25 m s-1. Personal injuries are indicated by dots and deaths with an "x."

Figure 6. Schematic evolution of low-level radar reflectivity structure in a bow-shaped echo convective system (from Fujita 1978).


5. Supercell Events

It already has been noted that supercells sometimes arise in environments that exhibit the potential for both strong updrafts and downdrafts. This is a consequence of having high Convective Available Potential Energy (CAPE), where

, (8)

with LFC denoting the Level of Free Convection and EL denoting the Equilibrium Level for an ascending parcel. This is the updraft equivalent of the NAPE. High values of CAPE are invariably associated with (i) large values of low-level absolute humidity and (ii) low relative humidity in the lower mid-troposphere. The latter condition allows the development of step lapse rates in the lower mid-troposphere, which are necessary to develop high CAPE (see Doswell et al. 1985). Clearly, a layer with low relative humidity and steep lapse rates favors the development of downdrafts by the processes already described.

On occasions when supercells develop, they contain (by definition!) a mesocyclone, at least in middle levels, if not near the surface. This paper is not the forum for a discussion of the origins of the supercell mesocyclone; interested readers should consult Davies-Jones and Brooks (1993) for a review. Recently, Brooks and Doswell (1993) have suggested that some supercells, which have considerable precipitation within their mesocyclones, thus becoming what is known as High-Precipitation (or HP) supercells (see Moller et al. 1990), can develop extremely intense and persistent surface winds. The existence of precipitation in the mesocyclone may arise as a result of having a strong mesocyclone at mid-levels in the presence of weak storm-relative flow at those levels (Brooks and Doswell 1993). In such a situation, the precipitation would not be carried far from the updraft by storm-relative flow and the mesocyclonic circulation would entrain the nearby precipitation, thus wrapping large amounts of liquid water around to its rear, where it would encounter the dry, unstable mid-tropospheric air. This evolution would favor the continuous production of evaporatively-chilled air on the rear side of the mesocyclone and, as long as the mid-level mesocyclone could survive being undercut by the low-level outflow, a quasi-continuous strong outflow would develop. Given the downdraft instability and the persistence of such events, a large swath of wind damage could result; these are potentially extreme convective wind events, especially when the wind is accompanied by large hail, as has been observed on occasion. Damage swaths of up to 2000 km2 (or more!) are not inconceivable.

There are two lines of reasoning that support this hypothesis for how such storms arise. First of all, such storms have been observed to arise (Smith 1993; Cummine et al 1992, Brooks and Doswell 1993) and the conditions in which they have been observed seem to fit this model (Fig. 7). In the limited sample of events consider to this date, the CAPE values associated with such storms are quite comparable to those with classic tornadic supercells, but the storm-relative wind speed profiles are quite distinct from tornadic storms and similar from case to case. The other form of supporting evidence is that a numerical simulation using one of the environmental soundings from the cases produces the sort of "wall of wind" that is observed in these extreme events.

Given that the early results of this work continue to hold up to further scrutiny, the fact that such storms arise in a reasonably limited form of environment, it should be possible to anticipate such storms. The extreme events do not happen often, but their large impact means it would be valuable to be able to forecast the threat of such a potentially devastating convective windstorm.


6. Summary and Discussion

Convective winds are a relatively common phenomenon, although most such events are not very intense. On the scale of the convective storm itself, peak outflow winds have short time and small space scales, reducing the threat to society, even when the peak winds become relatively strong. However, convection often is organized into mesoscale systems in which numerous convective cells develop, mature, and decay. In some situations, convective systems can produce a large number of convective wind events, leading to a much larger effected area than that associated with any individual cell. When storms influence a large area, the chances for significant hazards increase. The majority of windstorms in a convective system are of marginal severity, with only isolated events reaching high intensity. Nevertheless, the large area covered by such storms can result in major property losses.

Figure 7. Storm-relative environmental wind speed profiles from an intense wind event (as described in the text) and a tornadic supercell, showing the weak storm-relative flow associated with the intense wind event.

The most threatening situation would be for a very intense convective wind event that also affected a large area. It appears that a few times each year in North America, extreme convective wind events of this sort do occur. To date, no such storm has struck a major city during a vulnerable time (e.g., the morning or evening rush hours). However, it is only a matter of time until this sort of unfortunate concatenation actually occurs. Given that the area affected can approach that of a tropical cyclone's damage swath, and certainly far exceeds that affected during a tornado outbreak (while not being as intense, of course), it is uncomfortable to imagine the potential devastation. When such storms are accompanied by large hail (e.g., > 5 cm in diameter), the damage potential soars to even greater heights than when the wind occurs alone. The occurrence of hail has resulted in some of the costliest storms in United States history; coupling a fall of large hail with winds approaching 50 m s-1 could produce incredible damage in a populated area. Of course, economic losses to agriculture from such storms are already high, but do not attract much public attention, and such losses would be very difficult to mitigate with a 20-30 min warning.

A timely forecast may not be able to do much to mitigate the property loss, but could reduce the casualties. It appears possible to forecast these extreme events with some skill, but further research needs to be done to test the existing hypothesis about the interaction between the convective storm and its environment that produces the extensive swath of high winds.

Acknowledgments. I would like to thank my colleagues, Dr. Harold Brooks and Mr. Ken Howard, for their help in developing the figures.


Brooks, H.E., and C.A. Doswell III, 1993: Extreme winds in high-precipitation supercells. Preprints, 17th Conf. Severe Local Storms (St. Louis, MO), Amer. Meteor. Soc., 173-177.

Brown, J.M., K.R. Knupp, and F. Caracena, 1982: Destructive winds from shallow high-based cumulonimbi. Preprints, 12 th Conf. Severe Local Storms (San Antonio, TX), Amer. Meteor. Soc., 272-275.

Caracena, F., and J.A. Flueck, 1988: Classifying and forecasting microburst activity in the Denver area. J. Aircraft, 25, 525-530.

______, R.L. Holle, and C.A. Doswell III, 1989: Microbursts: A Handbook for Visual Identification. Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402, 35 pp.

Cummine, J., P. McCarthy, and M. Leduc, 1992: Blowdown over northwestern Ontario. Preprints, 4 th Workshop on Operational Meteorology (Whistler, British Columbia, Canada), Atmos. Environ. Service/Canadian Meteor. and Oceanogr. Soc., 311-317.

Davies-Jones, R., and H.E. Brooks, 1993: Mesocyclogenesis from a theoretical perspective. The Tornado: Its Structure, Dynamics, Prediction, and Hazards (Geophys. Monogr. 79), Amer. Geophys. Union, 105-114.

Doswell, C.A. III, 1985: The Operational Meteorology of Convective Weather. Vol. II: Storm Scale Analysis. NOAA Tech. Memo. ERL ESG-15, Available from the author at National Severe Storms Laboratory, 1313 Halley Circle, Norman, OK 73069, 240 pp.

______, F. Caracena, and M. Magnano, 1985: Temporal evolution of 700-500 mb lapse rate as a forecasting tool -- A case study. Preprints, 14 th Conf. Severe Local Storms (Indianapolis, IN), Amer. Meteor. Soc., 398-401.

Fujita, T.T., 1978: Manual of downburst identification for project NIMROD. Satellite and Mesometeorology Res. Paper No. 156, Dept. of Geophys. Sci., Univ. of Chicago, 104 pp.

______, and H.R. Byers, 1977: Spearhead echo and downburst in the crash of an airliner. Mon. Wea. Rev., 105, 129-146.

______, and F. Caracena, 1977: An analysis of three weather-related aircraft accidents. Bull. Amer. Meteor. Soc., 58, 1164-1181.

Houze, R.A., Jr., S.A. Rutledge, M.I. Biggerstaff, and B.F. Smull, 1989: Interpretation of Doppler weather radar displays of midlatitude mesoscale convective systems. Bull. Amer. Meteor. Soc., 70, 608-619.

Johns, R.H., and W.D. Hirt, 1987: Derechos: Widespread convectively induced windstorms. Wea. Forecasting, 2, 32-49.

______, and C.A. Doswell III, 1993: Severe local storms forecasting. Wea. Forecasting, 7, 588-612.

Kamburova, P.L., and F.H. Ludlam, 1966: Rainfall evaporation in thunderstorm downdraughts. Quart. J. Roy. Meteor. Soc., 92, 510-518

Loehrer, S.M., and R.H. Johnson, 1993: The surface pressure features and precipitation structure of PRE-STORM mesoscale convective systems. Preprints, 17 th Conf. Severe Local Storms (St. Louis, MO), Amer. Meteor. Soc., 481-485.

Moller, A.R., C.A. Doswell III, and R. Przybylinski, 1990: High-precipitation supercells: A conceptual model and documentation. Preprints, 16 th Conf. Severe Local Storms (Kananaskis Park, Alberta, Canada), Amer. Meteor. Soc., 52-57.

Rotunno, R., and J. B. Klemp, 1982: The influence of the shear-induced pressure gradient on thunderstorm motion. Mon. Wea. Rev., 110, 136-151.

Smith, B.E., 1993: The Concordia, Kansas downburst of 8 July 1992: A case study of an unusually long-lived windstorm. Preprints, 17 th Conf. Severe Local Storms (St. Louis, MO), Amer. Meteor. Soc., 588-592.

Smull, B.F., and M.L. Weisman, 1993: Comparison of the observed and simulated structure of a bow-shaped mesoscale convective system. Preprints, 17 th Conf. Severe Local Storms (St. Louis, MO), Amer. Meteor. Soc., 557-561.

Wakimoto, R.M., 1985: Forecasting dry microburst activity over the high plains. Mon. Wea. Rev., 113, 1131-1143.

Weisman, M.L., 1993: The genesis of severe, long-lived bow echoes. J. Atmos. Sci., 50, 645-670.