FLOODING 769

FLOODING
C A Doswell III, University of Oklahoma, Norman, OK, debris left behind when floodwaters recede can be
USA costly to clean up and also represent a health hazard,
Copyright 2003 Elsevier Science Ltd. All Rights Reserved. especially when there are decomposing bodies of
drowned wild and domestic animals in the debris. In
some situations, floods drive wild animals (including
Introduction invertebrates of all sorts) from their normal habitats
and into human habitations near and within
Flooding is arguably the weather-related hazard that is the flooded areas, which can create various pro-
most widespread around the globe. It can occur blems, especially when the animals are venomous or
virtually anywhere. A flood is defined as water aggressive.
overflowing onto land that usually is dry. Flooding is Although flooding has some large negative impacts
often thought of as a result of heavy rainfall, but floods on humans, it is also part of the natural processes
can arise in a number of ways that are not directly shaping the Earth. Floodplains along rivers and
related to ongoing weather events. Thus, a complete streams are among the most fertile regions known.
description of flooding must include processes that Most of the so-called ‘cradles of civilization’ are
may have little or nothing to do with meteorological within floodplains for this very reason (e.g., the Nile
events. Nevertheless, it is clear that in some ultimate River, the Tigris–Euphrates River, among others).
sense, the water that is involved in flooding has fallen Hence, humans have been affected by flooding both
as precipitation at some time, perhaps long ago. The positively and negatively since before historical times,
origins of flooding, therefore, ultimately lie in atmos- whenever they find themselves in the path of these
pheric processes creating precipitation, no matter natural events.
what specific event causes the flooding.
Floods produce damage through the immense
power of moving water and through the deposition
of dirt and debris when floodwaters finally recede.
People who have not experienced a flood may have
little or no appreciation for the dangers of moving
water. The energy of that moving water goes up as the
square of its speed; when the speed doubles, the energy
associated with it increases by a factor of four.
Flooding is typically coupled to water moving faster
than normal, in part because of the weight of an
increased amount of water upstream, leading to an
increase in the pressure gradient that drives the flow. In
most cases, the damage potential of the flood is
magnified by the debris that the waters carry: trees,
vehicles, boulders, buildings, etc. When the waters
move fast enough, they can sweep away all before
them, leaving behind scenes of terrible destruction
(Figure 1).
The effect of the water itself can be devastating on
structures and on the objects within them: books,
furniture, photographs, electronic equipment, and so
on can be damaged simply by being immersed in water,
even if they are not directly damaged by the water
movement. Moreover, floodwaters typically contain
suspended silt and potentially toxic microorganisms
and dissolved chemicals. This means that floods
usually compromise drinking water supplies, resulting
in short-term shortages of potable water, with the Figure 1 Damage resulting from the 1977 Johnstown, Pennsyl-
additional long-term costs in restoring drinking water vania, flash flood event. (r The Johnstown Tribune-Democrat,
service to the residents of a flooded area. The mud and used by permission.)
770 FLOODING
Floods as a Direct Result
of Precipitation
When the waters of a flood arise directly from
precipitation, atmospheric processes can be identified
as directly responsible for the event. That is, rainfalls
occur that are well beyond the average values for the
affected area. It is only when those rainfalls exceed the
average that land which is usually dry can be affected;
that is, a flood occurs. Thus, the rainfall amounts
needed for floods cannot be defined in absolute terms.
A precipitation event that causes a flood in one
location might be well within the bounds of what is
typical for another location. Generally speaking, the
threshold for flood-producing rainfalls increases as the
annual average rainfall for a region increases.
Flash Floods
Flash floods are defined as those flood events where the
rise in water is either during or within a few hours of
the rainfall that produces the rise. Therefore, flash
floods occur within small catchments, where the
response time of the drainage basin is short. Many
hydrological factors have relevance to the occurrence
of a flash flood: terrain gradients, soil type, vegetative
cover, human habitation, antecedent rainfall, and so
on. In steep, rocky terrain or within heavily urbanized
regions, even a relatively small amount of rainfall can
trigger flash flooding. These hydrological factors
determine the response of the catchment to the
precipitation event. Thus, a flash flood is clearly the
result of the concatenation of both meteorological and
hydrological circumstances.
Most flash floods associated with rainfall are
produced by thunderstorms; that is, deep, moist
convection. A single thunderstorm cell is unlikely to
produce enough rainfall to cause a flash flood, so the
typical flash flood is the result of several thunder-
storms moving successively over the same area, known
as ‘training’ thunderstorms (Figure 2), because it
resembles the passage of cars in a freight train. A
succession of thunderstorms results when new thun-
derstorms pass repeatedly over the same place while
the overall system of thunderstorms is very nearly
stationary. The infamous Johnstown, Pennsylvania
flash flood of 19–20 July 1977 was produced by such a
system. Thunderstorms forming in north-western
Pennsylvania moved south-eastward, only to be
replaced by newly formed thunderstorms, a process
that went on for several hours. The result was
torrential rainfall concentrated near Johnstown, with
amounts exceeding 400 mm. The ensuing flood was
responsible for 77 fatalities and $550 million (in 1999
dollars).
Occasionally, flash floods are created in conditions
that are not favorable for thunderstorms but which
still produce heavy rainfalls. This can occur when
moist air is forced upward over mountains by the wind
flow, called orographic precipitation. When the air
forced upward is very moist, the rainfall can be quite
heavy. The steep, rocky terrain also promotes rapid
runoff of the rainfall. Flooding along the West Coast of
the USA or in the European Alps is often of this type;
that is, not involving thunderstorms.
A characteristic of flash floods is the localized nature
of the heaviest rainfall. As shown in Figure 3, the most
intense rainfall is typically confined to a relatively
small area. When large amounts of this localized
precipitation fall within a small drainage basin, flash
floods can occur. Sometimes, the location where flash
flood damage occurs may actually receive little or no
rainfall. That is, the rainfall that causes the problem
can occur upstream of threatened areas. This separa-
tion between the rainfall and the flood can cause
confusion because it may not even be raining in an area
for which flash flood warnings are issued. Another
factor in the impact of flash floods is that the
precipitation causing the event often falls during the
night, when it can be difficult to get warnings to
sleeping residents. The central part of the USA is well
known for its heavy thunderstorm-produced rains
during nighttime hours. Worldwide, thunderstorms
are most common during the day, but on the central
plains of the USA (and in a few other places around the
world), the unique geography of the region favors
nocturnal thunderstorms. This setting promotes a
strong flow of moisture northward from the Gulf of
Mexico, called a low-level jet stream, during the warm
months of the year. Moisture carried by the low-level
jet stream helps to maintain thunderstorm systems
that often begin during daytime hours on the higher
terrain to the east of the Rocky Mountains. Because of
the low-level jet stream, such storms can persist well
into the nighttime hours, often forming clusters
of thunderstorms known as mesoscale convective
systems (Figure 4).
It is the rapidity of the event that makes flash floods
so damaging and dangerous. Flash floods involve
rapidly rising, fast-moving waters that can do im-
mense damage; the suddenness of the onset of the flood
can result in people being caught unawares. Most
fatalities result from drowning, with perhaps some
traumatic injuries from being carried along in the
debris-laden waters and being swept into standing
objects. The potential for loss of human life with flash
floods is high. Debris carried in flash floods can form
temporary ‘debris dams’ that typically fail as waters
back up behind them. Failure of these debris dams then
results in a ‘wall of water’ surging downstream. Debris
 FLOODING 771

Cell I

15

Cell II 10

10
Height (km)

Cell III 10
30
Cell IV 50
5
Cell motion

0
X
(A)

Cell II
Cell I
15

10
Cell III
30
10
Height (km)

Cell IV
30

Cell V
5
50

0
X
(B)

Cell II
Cell I

15
10
Cell III
Cell IV
10
10
Height (km)

Cell V
30
50
5

0
X
(C)

Figure 2 Schematic of the ‘training’ effect. (A) At this time, there are four numbered thunderstorm cells in various stages of development.
Cell I is mature, with both updrafts and downdrafts, and heavy rain is about to commence at point X. Cells II, III, and IV are still developing,
and have only updrafts. Cell II has precipitation forming aloft. The hatched contours are radar reflectivity, in standard units of dBZ, which is
related to the rainfall rate. (B) About 15 min later, Cell I’s updraft is dissipated, and it is now dominated by downdraft. Heavy rain continues
at X while Cell II is maturing and developing a downdraft. Cells III, IV, and now V are still immature. (C) About 15 more minutes have
elapsed. Cell I’s rainfall is continuing but it is now nearly dissipated, while Cell II is entering late maturity. It is still raining at X but now the
rainfall is from Cell II, and heavy rain from Cell II is descending from aloft. Now Cell III is developing its first precipitation aloft. Cell IV and V
are still immature. (Adapted from Figure 7 in Doswell CA III, Brooks HE and Maddox RA (1996) Flash flood forecasting: An ingredients-
based methodology. Weather Forecasting 11: 560–581.)
772 FLOODING

Figure 3 Observed total precipitation (mm) during the Johnstown, Pennsylvania (JST, located by an asterisk) flash flood event. For
reference, Pittsburgh, Pennsylvania (PIT, located by the plus sign) is also shown. (Adapted from Figure 14a in Hoxit LR, Maddox RA,
Chappell CF, Zuckerberg FL, Mogil HM, Jones I, Greene DR, Saffle RE and Scofield RA (1987) Meteorological Analysis of the Johnstown,
Pennsylvania, Flash Flood, 19–20 July 1977. NOAA Technical Report ERL 401-APCL 43, NTIS Accession No. PB297412.) NOAA,
National Oceanographic and Atmospheric Administration.

dam failure events can happen repeatedly during the
course of the flash flood. Not all flash floods are
characterized by a ‘wall of water’ but all of them (by
definition) involve rapidly rising floodwaters.
Because urbanized areas promote runoff of rainfall,
rather than permitting most of the rain to be absorbed
into the ground, flash flooding is more likely in cities
than in rural areas surrounding a city. It takes much
less rainfall in a city to create a flash flood situation
than in a rural area of comparable size.
Flash floods continue to be a major contributor to
loss of life, in spite of improved precipitation fore-
casting. Some noteworthy examples include events in
the Big Thompson Canyon in Colorado (1976 – 144
fatalities) and near the town of Biescas in the Spanish
Pyrenees (1996 – 86 fatalities).
Tropical cyclones often create devastating flash
floods as a result of torrential rainfalls. In late October
of 1998, Hurricane Mitch caused more than 9000
fatalities (the exact number is not known), mostly in
Nicaragua and Honduras, in Central America, from
flash floods and landslides associated with its rainfall.
It was the worst weather disaster in terms of casualties
in the Western Hemisphere during the twentieth
century.
River Floods
River floods, in contrast to flash floods, typically
unfold over days, or even months. This is because they
occur in large basins involving ‘main stem’ rivers like
the Missouri, or the Nile, and are usually the result of
many individual rainfall episodes spread out over
many days. In fact, within a river flood event, several
flash flood events can occur. Again, hydrological
factors often contribute to a river flood, but river
floods are not so sensitive to them as are flash floods.
Whereas individual thunderstorm systems can cause
flash floods, river floods are usually the result of a
stagnant synoptic-scale weather pattern. Localized
heavy rainfall events occur many times during a period
of days or even months, each contributing its share of

Figure 4 False-color enhanced infrared satellite image of a mesoscale convective system, with the light red colors indicating the coldest
(therefore the highest) clouds. Note that this image is from 17 August 2000 at 0345, local time, which corresponds to 0845 UTC.
rainfall to the tributaries, which then discharge into
the main stem of a river. The river rises gradually in
response to all the input rainfall. The river flood
potential of a situation can be increased by concurrent
snow melt and other factors besides rainfall.
The major flooding event during June and July of
1993 was the result of a weather pattern (Figure 5A)
that produced a storm track across the upper Mid-
western USA. Abnormally low heights of the pressure
surfaces (associated with cool temperatures) over the
northern Plains produced a pattern in which traveling
weather disturbances intensified in the Midwest after
crossing the Rocky Mountains. This pattern aloft also
produced an anomalously strong poleward flow of
low-level moisture from the Gulf of Mexico into the
Midwest. Mesoscale convective systems developed
almost every evening during the early summer, typi-
cally persisting through the night. These passed
repeatedly over nearly the same areas, resulting in
widespread significant rainfalls (Figure 5B) for the
FLOODING 773
~ 500 km
period over the lower Missouri and upper Mississippi
basins. In addition to these factors, considerable
rainfall over the region had fallen during the previous
several months, providing a hydrological setting that
favored runoff of the precipitation. This event pro-
duced disastrous flooding that persisted for many
weeks.
Owing to the long time scale of the rising waters,
river floods pose a lower risk of fatalities; people have
more time to take proper actions. Of course, some
casualties result from waiting until it has become too
late to respond to the threat. Levee and dam failures, as
well as intentional rapid release of impounded waters
to prevent the catastrophic failure of the flood control
structures, can produce rapidly rising water situations
embedded within a river flood, and these also can
contribute to loss of life.
Because of the large scale of river floods, the damage
figures may be enormous; easily into the billions of
dollars. Crop losses are a major factor in the costs of
774 FLOODING

54° N 303

51° N 306 306

306
48° N 309
309
45° N 312 309
312
315
42° N
315
39° N 318 312
318
36° N

33° N
30° N 315

27° N

160° W 150° W 140° W 130° W 120° W 110° W 100° W 90° W 80° W 70° W

−40 −30 −20 −10 10 20 30 40

(A) Height anomoly (m)

300
100

250

400

500
300 200
250 150
150
100 150
200 150
200
250

100 150 200 250 300 400 500

(B) mm

Figure 5 (A) Map of the 700 hPa heights (thin lines, in dam) and height anomalies (shading, in m) for June/July 1993, and (B) observed
precipitation for the same period (shading, in mm). Based on data supplied by J. Janowiak of the NOAA Climate Prediction Center.

river floods, whenever large tracts of prime agricul- damage and dislocations along the Upper Mississippi
tural land along floodplains are inundated. Levees are and Lower Missouri basins during the summer floods
often used to protect populated areas, so the failure of of 1993, during which several levees were breached,
those levees can generate major property losses. The illustrate the huge impact such events can have.

Floods Arising from Nonprecipitation
Events
Apart from floods resulting directly from rainfall,
there are many ways in which precipitation can cause
floods, perhaps long after it has fallen. When flowing
water is impounded by the construction of dams, there
is some risk that the dams will fail. Johnstown,
Pennsylvania, was inundated by a dam failure during
a rainfall event in 1889, for example. Such rapid
releases of stored water can be cataclysmic, manifest-
ing themselves as an enormous ‘wall of water’ choked
with debris.
Flood can also arise through the melting of snowfall.
In situations where the preceding winter’s snowpack is
deep, a sudden change to warm temperatures in the
spring can result in abnormally rapid melting and
runoff of the snow melt. The devastating flood created
in Grand Forks, North Dakota, in April of 1997 is an
example. Occasionally, warm rain falls directly onto
the melting snow, exacerbating such situations by
speeding the melting process and adding more liquid
water.
Deposits of snow and ice on volcanic peaks can melt
rapidly during eruptions. The resulting runoff, often
turned into a thick slurry by the inclusion of volcanic
ash, roars down the mountainside and is called a lahar.
A tragic example occurred with the Nevado del Ruiz
volcano in Colombia on 13 November 1985, which
killed more than 23 000 people, mostly in the town of
Armero. Another occurred in Iceland during 1996 on
the Vatnajökull glacier, with no fatalities owing to its
remote location. Lahars can continue occasionally for
years after an eruption, when heavy rains fall onto ash
deposited by the volcano.
During the winter and late spring, when ice can
build up on rivers in cold climates, the breakup of the
ice can create ice dams on the river. The ice dams cause
the waters to back up, sometimes flooding the land
upstream of the ice dam. Then, the breakup of the ice
dam can result in a flash flood wave that surges
downstream of the ice dam’s position.
Other flood situations can develop along the shores
of the world’s oceans and even with large freshwater
lakes. Tsunamis, typically caused by underwater
earthquakes and landslides, can flood the shorelines
with huge waves that break on the shallow waters near
the shore. Storms of all sorts, including tropical
cyclones, can drive the waters before the winds into
storm surges that inundate shore areas when the
storms are near the land. Large lakes can experience
flooding on their shores due to seiches, which are
surges of water (usually oscillatory) within enclosed
bodies of water. Seiches can be caused by earthquakes
or by atmospheric processes.
FLOODING 775
Societal Impacts and Their Mitigation
The results of floods on society worldwide are
substantial. Flooding is responsible for many drown-
ing fatalities in tropical cyclones, either from storm
surges or from freshwater rain-induced flash floods.
Flash floods and river floods typically produce more
fatalities every year than either tornadoes or hurri-
canes in the USA. In many parts of the world, flood
fatalities are associated with the most significant
weather-related disasters. Flood damage cost in the
USA is now on the order of several billion dollars
annually, and this figure continues to rise.
Many people now live and play in flood-prone
areas: for example, within floodplains of rivers and
their tributaries, as well as along coastlines that are
vulnerable to storm-caused flooding from tsunamis,
tropical cyclones, and nontropical storms. Develop-
ment of flood-prone areas for habitation and recrea-
tion has been increasing, with a corresponding
increase in the risks to life and property. The 1993
Upper Mississippi and Lower Missouri River floods
provided a grim reminder of the risks of building
permanent structures within floodplains, even when
flood-control measures have been taken.
In the case of flash floods, it is difficult to take
measures to protect property, owing to the rapidity
with which the event happens. However, prevention of
flash flood casualties is possible, provided warnings
can be issued and acted upon properly in a timely
fashion. Considerable attention has been paid to
increasing public awareness of the dangers of driving
into rapidly rising flood waters, for instance, as a result
of recent experiences with flash floods. Unfortunately,
situations can still arise where warnings are not issued
in time. People living and engaging in recreational
activities in places prone to flash floods need to be alert
during heavy rainfalls and be prepared to seek safety
even when they do not receive timely warnings.
For river floods and other relatively slow-developing
situations (such as rising snow melt or ice action events),
it may be possible to reduce the property damage as well
by removing the contents of structures. Obviously, any
structures (and their contents) built in flood-prone areas
are permanently at risk; the only way to guarantee their
protection from floods is to move them out of those
areas. Prevention of fatalities in river flood events is a
matter of heeding the warnings of danger and moving
residents out of the danger areas before the number of
options is reduced by the rising waters and by the failure
of levees or other flood-prevention structures.
Forecasting the details of flooding events is an
important part of mitigation. Knowing precisely when
and where a flood will occur would no doubt be
helpful, but it is also important to be able to anticipate
776 FLOODING
the magnitude of the flood. An example of this is the
tragedy of the 1997 Grand Forks, North Dakota case,
where the river level was only a few feet higher than
that forecast. Those few feet, however, had a large
impact, because the flood-control operations were
based on the lower forecast value. When the river rose
above that level, the flood-control measures failed
catastrophically. In reality, such a forecast can never be
a precise statement; uncertainty is implicitly a part of
every forecast, a point that perhaps needs greater
emphasis in the future.
Flooding, by its very nature, is usually a result of
both meteorological and hydrologic processes; the
character of a flood is determined both by the detailed
behavior of the precipitation and by the nature of
situation in which the event is likely to occur (soil
conditions, amount of antecedent rainfall, and so on).
It is not likely that precisely detailed forecasts of
flooding events will ever be possible, although it is
certainly well within our capability to anticipate the
possibility of most flood events. The challenge for
reducing the social impacts of floods is how best to
make use of the uncertain meteorological and hydro-
logical forecasts that are within practical means. The
challenge is to make effective use of whatever fore-
casting capability we have, even as we seek to improve
that capability.
Effects of Human Activities on
Flooding
In addition to the risks to lives and property that
people take by moving into flood-prone areas, devel-
opment for human use often involves clearing land of
its native vegetation and altering the characteristics of
the ground cover. Vegetation works together with the
soil to store rainfall, so when that vegetation is cleared,
rainfall runoff can increase substantially. Rather than
being absorbed by the soil and its natural vegetation,
in areas where that vegetation has been cleared (either
for construction or for agriculture), heavy rainfall is
more likely to run off and pour into streams and rivers,
increasing the potential threat from flash floods and
river floods. Construction of roads and buildings also
acts to increase runoff, and leads to an increasing
likelihood of localized urban flooding. Such construc-
tion dramatically increases the fraction of the rainfall
that runs off, regardless of antecedent rainfall. Hu-
man-caused fires can also produce at least temporary
increases in the runoff potential in the headwater
regions of streams and rivers. It is evident that human
activities are increasing the potential for floods around
the world.
Again recalling the Mississippi River floods of 1993
as an example, the issue of flood control through levees
and other structures was dramatically recalled to
public attention. The value of structural methods for
flood control (levees, flood control dams, breakwaters,
etc.) remains controversial, but the 1993 floods made
it apparent that structures such as levees can be
breached during major flooding episodes, even though
they may be able to contain lesser events. Structural
failures create rapidly rising waters (flash floods)
artificially within a river flood event, increasing the
hazards to human life as well as destroying property.
The decision about when and where to take structural
approaches will continue to be a challenge.
Finally, the use of flood-prone areas for human
activities puts lives and property at risk, although the
major flood events may be separated by many years.
The long time between events can lead to complacency
and subsequent disasters. The choices associated with
land use are a continuing challenge, now and in the
future. When humans live and play in ways that put
them in the path of potential floodwaters, major
societal impacts are inevitable.
See also
Air–Sea Interaction: Momentum, Heat and Vapor Flux-
es. Convective Storms: Convective Initiation; Overview.
Hurricanes. Hydrology: Modeling and Prediction; Over-
view. Mesoscale Meteorology: Cloud and Precipitation
Bands; Mesoscale Convective Systems. Palmer Drought
Index. Predictability and Chaos. Radar: Precipitation
Radar. Satellite Remote Sensing: Precipitation. Severe
Storms. Weather Prediction: Severe Weather Fore-
casting.
Further Reading
Agnone JC (ed.) (1995) Raging Forces: Earth in Upheaval.
National Geographic Society.
Barry JM (1997) Rising Tide: The Great Mississippi Flood of
1927 and How It Changed America. Simon and Schuster.
Cluckie ID and Collier CG (eds) (1991) Hydrological
Applications of Weather Radar. Ellis Horwood.
Dingman L (1994) Physical Hydrology. Upper Saddle River,
NJ: Prentice-Hall.
Hill CE (ed.) (1986) Nature on the Rampage: Our Violent
Earth. National Geographic Society.
Lorenz EN (1993) The Essence of Chaos. Seattle: University
of Washington Press.
Ludlam F (1980) Clouds and Storms. University Park, PA:
Pennsylvania State University Press.
Pruppacher HR and Klett JD (1997) Microphysics of Clouds
and Precipitation, 2nd edn. Dordrecht: Kluwer Academic
Press.
Ray P (ed.) (1986) Mesoscale Meteorology and Forecasting.
Boston, MA: American Meteorological Society.
Sarewitz D, Pielke RA Jr and Byerly R (eds) (2000)
Prediction: Decision-Making and the Future of Nature.
Washington, DC: Island Press.

FOG
P J Croft, University of Louisiana at Monroe, Monroe,
LA, USA
Copyright 2003 Elsevier Science Ltd. All Rights Reserved.
Introduction
Murky and perhaps even eerie, or pristine and serene,
are just a few of the descriptions of the often blinding
white–gray veil that comes to mind when people think
of fog. Whether over oceans or local waters or over
various landscapes, a certain uniformity and blank-
ness is associated with fog. Often hugging the terrain,
or simply masking the landscape, fog compromises the
integrity of our senses that are honed to a fog-less
environment. Sometimes it is like an unbending solid
wall, many times vanishing over short distances until
another patch is encountered, and even the beauty of a
fogbow belies the significance of fog.
Merely a collection of small-diameter suspended
water droplets in the air, fog may occur in a calm
atmosphere that is saturated, or nearly so, and may
also occur in cool and moist air moving quickly past
us. Although fog is commonly thought of as merely ‘a
cloud on the ground’, it is much more than this given
its very dynamics of formation, ‘intensity’ (or thick-
ness), nature of the droplets that it consists of, and
areal extent and duration. Fog occurs around the globe
for many different reasons and can be elusive when
predicting its exact occurrence.
Will fog form? Where will it form? When will it
form? How ‘thick’ will it be? How long will it persist?
These are only a few of the many questions forecasters
and people ask about the enveloping droplets of fog.
Yet to examine fog occurrence it is necessary to
consider first its impact on our activities and then learn
more about its characteristics and physical behavior.
Beyond fog impacts, the observation and study of fog
helps us to define its characteristics more completely
and thus aids in our understanding of the fog process.
With this information we are able to better predict its
occurrence, extent, intensity, and duration so as to
avoid or mitigate some of the hazards associated with
fog. In fact, such an examination provides us with an
opportunity to make use of fog in various agricultural,
military, and other applications. The significance of fog
and fog prediction includes impacts as well as benefits.
Impacts
Fog occurrence impacts a wide variety of human
activities worldwide. These impacts range from in-
FOG 777
convenience to annoyance and from high costs to
deadly consequences. Although mostly negative con-
sequences and perceptions are associated with fog,
there are positive benefits as well. These range from a
pleasing esthetic effect to fog harvesting for agricul-
tural and water supply applications. Thus repercus-
sions can lead to a variety of associated political,
social, and legal implications, depending upon the
precise impact and the person or peoples – and
economies – affected. The first and most obvious
impact of fog is related to the reduction of visibility.
This reduction hampers and restricts our navigational
abilities and thus increases our chances of judgment
errors in the operation of transportation vehicles.
Reduced visibility in fog quickly impacts our ability to
drive, move over water, fly, and transit land by train. In
each case the inability to see well, or to see an adequate
distance ahead, is compromised by both fog and the
speed of motion. It is further compromised by our own
ocular inability to distinguish objects given limited
brightness and contrast that occur with fog.
Land Transportation
Land transportation includes automotives, trucks,
and heavy machinery and is prone to disruption and
delay when fog is present. Near Windsor, ON (Can-
ada), a highway ‘pileup’ collision during morning fog
in September 1999 resulted in seven deaths as 62 cars
and tractor-trailers collided. In the United States in
Kingsburg, CA in November – and Waynesboro, VA in
April – heavy fog resulted in highway pileups that
killed 42 and injured 91 people as 40–65 vehicles
collided in mountain and valley regions in 1998. The
Virginia pileup is a ‘chain-reaction’ crash in a region
prone to ‘heavy’ fog, heavily traveled, and which
frequently experiences low visibilities. A tour bus and
truck collided in Asuncion (Paraguay) in March 2000
while traveling through early morning dense fog and
30 of 45 people on board were killed. A caravan of
buses transporting college students in Pennsylvania
(United States) traveling through dense fog overnight
collided with one another killing two and injuring 106.
In Bourg-Achard (France) in September 1997,
several chain-reaction crashes claimed eight lives and
injured 63 as over 100 vehicles were involved during a
mid-morning ‘heavy’ fog event. Witnesses and victims
reported that visibility was merely 45 yd (41.148 m)
when the crashes occurred. On the Ivory Coast in
Abidjan (Africa), ‘thick’ fog combined with dusty
winds from the Sahara Desert during the Harmattan
Season in December 1995 killed 14 and injured 86.
778 FOG
News reports indicated that a similar accident in
August killed 20 and injured 62, and that drivers in this
region are known for speeding and for their reluctance
to diminish their speeds even when weather conditions
are poor. In Lisbon (Portugal), four were killed and 70
hurt in a 100 car pileup in January 2000 that halted up
to 6000 cars in both directions of a 20 mile roadway
for 5 h.
In Mobile, AL in the United States, 193 vehicles
collided on the Mobile Bay ‘Bayway’ highway in the
country’s worst fog accident ever in March 1995,
sending 91 injured to hospitals and killing, miracu-
lously, only one person. Insurance losses were esti-
mated at over one million dollars at the time of the
accident, and witnesses and victims report sitting in
their cars and listening to the continuing crashes
behind them. Some report driving ‘into a wall’ of fog
with visibility immediately reduced from 0.5 mile
(0.8045 km) to near zero. The roadway was closed for
hours in both directions of travel. The event led to the
installation of fog sensors in the hopes of avoiding a
repeat of the accident. The same was done for a fog
warning system in Waynesboro at a cost of over five
million dollars.
Although many deaths are directly attributable to
collisions, a number are caused by fires ignited during
the collision process. Many factors lead to such serious
consequences during fog events. These include poor
visibility, vehicle speeds (posted as compared to
traveled), traffic volume, roadway design and surfac-
ing, driving habits (which include invincibility and
trust of braking systems, e.g., anti-lock systems),
roadway conditions (perhaps dry but sometimes wet
due to mist or drizzle, previous rains, or condensation)
that restrict braking ability, and windshield visibility
effects. Although fog has been cited as the primary
cause of an accident in generally less than 1% of all
accidents in a given region, it has been cited up to 10%
of the time in a fog-prone region, particularly in
multiple collisions. The average claim for one vehicle
is nearly $8000 (US) and over one million dollars for a
multiple-vehicle crash.
The most obvious threat is reduced visibility which
restricts a driver’s ability to navigate the roadway. This
is further diminished with increasing speeds and of
serious consequence resulting in many deaths and
injuries every year around the world. The visibility not
only restricts horizontal distance and depth percep-
tions but also reduces the ability of drivers to gauge
their own speed of travel. Through computer simula-
tions a psychologist was able to determine that
although drivers could learn to sustain their speeds
in simulation, the addition of fog distorted or de-
stroyed this ability. Ironically, many drivers are prone
not to check their speeds in a fog situation as they seek
desperately to maintain their scanning of the roadway
ahead to ensure their own safety.
In addition, the variation of fog intensity and
duration create a rapidly changing set of visibilities
during the course of travel and may be further
enhanced by hilly terrain and/or protected regions.
Although no criteria exist for safe driving in fog, it is
clear that visibilities under 1 mile (or 1.6 km) while
driving at speeds of near 60 miles per hour (i.e., one
mile per minute; or 96.54 kph) compromises seriously
a driver’s integrity and response time to hidden
hazards. This is often exacerbated by the distance
between vehicles and curved or inclined sections of
roadway. Thus the first lines of defense for navigating
fog is the reduction of speed, the use of headlights
and/or flashers, fog lights or fog-free lenses and
shields, and the ‘stop, rest, and wait’ approach. Other
alternatives include fog dispersion or mitigation tech-
niques discussed later.
Rail transportation may be impacted similarly by
fog conditions. In Badrshein (Egypt), on the Nile River
in December 1995 at about 8.00 a.m., one passenger
train plowed into an express coach which had slowed
due to ‘heavy’ fog. The wreck killed 75 and injured 76
as five train cars were destroyed and 40 damaged.
Reportedly the train’s driver could not see even a yard
(approximately 0.9144 m) to the front and had
apparently ‘stuck his head out of the window’ to try
to see better. Rail collisions have also involved motor
vehicles and marine vessels at various crossings.
Marine Transportation
The operations of ships and barges, pleasure craft, and
sailboats may all be hampered by fog. Fog events often
slow, and may even stop, marine operations with
significant economic costs. A well-known event is the
sinking of the Andrea Doria in July 1956 off the mid-
Atlantic coast of the United States after collision with
the Swedish liner Stockholm. One of the most serious
incidents occurred in the northern portion of Mobile
Bay, Alabama (United States) in September 1993.
During the early morning hours a barge collided with a
rail span, moving it out of alignment moments before
the arrival of an Amtrak passenger train. Rescue
operations were severely hampered due to inability to
reach easily the crash location and survivors as 47
people died. Three years later the country’s Coast
Guard implemented rules requiring towboats to be
equipped with radar, searchlights, radios, compasses,
and other navigational gear and that the crews knew
how to use them.
Shipping operations are stopped or slowed when
visibility is as low as 0.25 mile (0.4023 km). In a major
(or minor) port or cargo region, large financial losses
 FOG 779

can occur when operations are halted. In the case of for fog, a delayed or canceled flight may cost an airline
delays, reduced supplies and delivery of critical between $5000 and $25 000 per flight.
elements and products limit the productivity of
industry and commercial interests and their ability to Military, Rescue Operations, and Other Impacts
provide services to their clients. At the same time, ship
Although these are the most common types of impacts
operations require daily operating and maintenance
(Figure 1), many other impacts do occur for a variety
costs, salaries, and living expenses which decrease
of military and rescue operations and other activities.
profits. In the case of halting operations, these
For example, the D-Day invasion and other theaters of
considerations also include reduction or elimination
engagement have benefited or suffered from the effects
of the viability of product, particularly if perishables
of fog. The deployment of troops in Tuzla Bosnia-
or refrigerated goods are being transported. Losses
Herzegovina by the United Nations was put ‘on-hold’
may range from $10 000 to $25 000 per day per ship
for several days as persistent and continuous fog
and into millions for even a moderately active harbor
claimed the land in December 1995. The rough terrain
or port of call.
is known for its bad weather and thus allows only
small windows of opportunity for flight operations to
Air Transportation be made safely. In addition, rescue operations for a
cargo ship, which had collided with another off the
Aircraft impacted by fog include airplanes, blimps,
South Korean coast, were suspended in June 1996 as
balloons, and helicopters. Although fog-related crash-
persistent fog limited visibility to less than 10 yd
es have occurred for large aircraft in the past, this is
(9.144 m). Space shuttle operations, as well as delayed
much less frequent today given instrument flight
launches and landings, have been impacted by fog or
regulations and improved navigational beacons and
the prediction of its occurrence.
technology as part of air traffic control. Smaller planes
Although rare, fog has caused sporting event can-
however are more prone to difficulty as they lack such
celations (baseball and football) or suspensions and
systems. The more typical impact of fog today is the
class delays or cancelations at schools and colleges
delay and diversion of flights. In Hong Kong (China) in
(especially if a large number of commuter students).
March 1998, 10 000 passengers were stranded at the
Fog has also played a role in dangerous air pollution
airport as over 50 flights were canceled, delayed, or
episodes including Belgium in 1930, Pennsylvania in
diverted. Dense fog events in Minneapolis, MN
1948, and London in 1952. More recent events
(United States) in November 1997 and February
include fog combining with forest and oil fires,
2000, and at the Jackson Mississippi (United States)
chemical spills, and other emissions. These events are
International airport in June 1996 resulted in flight
sometimes aided by topographic variations (e.g., a
delays for thousands of travelers during the morning
valley location, river valley, or other water source),
hours. At LaGuardia Airport in New York
regional climate (e.g., coastal), and nature of human
(United States) in February 1996, delays claimed
activities (i.e., industrial or agricultural) in a stable
most travel plans for an entire day as visibility was
atmosphere (e.g., Mexico, Arctic).
reduced to 300 ft (91.44 m). Some passengers were
Air pollutants can also act as condensation nuclei
shuttled to nearby airports 10–30 miles (16–48 m)
and create lower visibilities through haze and fog. This
away to take alternate flights. Most flight operations,
was recognized both in early Roman history as well as
regardless of instrumentation, require at least
prior to that time. Some of the most intense fogs result
0.75 mile (1.207 km) visibility for takeoff and
when high concentrations of pollutants and other
landing.
aerosols are found in the air. London fog and its
Weather has been reported as responsible for one-
combination with pollutants, recognized as early as
fourth to one-half of all aviation accidents in the
the 1660s, was brought to the attention of King
United States, including fatal accidents, with an
Charles II and Parliament. In Donora, PA during
average of over 400 lives lost each year. The delays,
October 1948, nearly half the population of 14 000
diversions, and cancelations resulting from fog add to
became ill during a prolonged valley air pollution
the cost of major accidents. Delays and diversions may
episode that was accompanied at times by fog. The
result in greater costs due to fuel usage, passenger
enactment of air quality regulations has aided in the
discomfort and complaints, and the shuttling of
abatement, but not the elimination, of such impacts.
passengers to alternate flights or airlines. Cancelations
result in displaced passengers and flight crews and
Legal Implications and Mitigation
create additional costs of lodging, food, and alternate
transportation. Depending on plane size and passen- Although no direct fog insurance exists, the costs
ger loads, and regardless of an airport’s preponderance associated with fog events and disasters may be
780 FOG

The figure shows varying

terrain and patches of
fog (where written) with
various intensities in a
'big picture' perspective.
Airport/Rail Depot in
Valley Flat Land

Cold river
water

FOG
FOG
Airport
Industry FOG Rail
depot
FOG
FOG
FOG

FOG
FOG
Row Bridge
crops
FOG

FOG
FOG
Speed of
travel Brake lights
Roadway FOG
Driver visibility
FOG Mounted FOG
fog lights
Anti-lock FOG
Wet road braking FOG • Surface conditions
surface Headlights • Cloud microphysics

Figure 1 Fog transportation factors. The figure shows varying terrain and patches of fog.

covered according to damage, repair, and replacement
as well as personal suffering or loss. The liability may
necessarily carry high rates as a function of the
numbers of people involved and the nature of the
‘cargo’ being transported, particularly in shipping, or
the nature of the impact, such as in an air pollution
episode. Payment is made based on a judgment of
whether the situation was avoidable, whether actions
were prudent in the particular setting, and if the event
were predictable or an ‘act of God’ or a ‘once-in-a-
lifetime’ event.
The sociopolitical decisions and mandates that
come from these find their way into legislation or
standards of procedures as evidenced by aviation
instrument flight rules (IFRs), port regulations for
closure and/or delay/diversion, and the recommenda-
tion (or requirement) that drivers use low-beam
headlights when fog is present. Legal implications
may arise for a farmer who irrigates a crop (for water
or for frost protection) near a roadway and thus
enhances the fog and which results in an accident.
Clearly, liability is one of the reasons for more
conservative decisions on closings, delays, and can-
celations. It will also include those who clear fog at
airports and other locations if they do not meet their
obligations under contract and are taken to court for
damages.
Observations
The most obvious types of observations of fog are its
occurrence, reduction of visibility and/or fog’s ‘thick-
ness’ or ‘intensity’, color (of limited use in reporting),
duration, and extent. Each of these may be assessed
locally at an observation site and regionally using a
collection of sites. The traditional determination of
visibility in fog is based on the ability of a human
observer to see predetermined ground-based targets in
all azimuth directions about their site. Electronic
methods (e.g., transmissometer) have been applied to
roadways, harbors, and aerodrome runways to deter-
mine low visibility in the immediate location of the
sensor. Various methods and instrumentation used to
determine visibility are presented in the visibility
portion of this encyclopedia.

The occurrence and reduction of visibility are easily
recorded based on human observation, but are often
electronically derived through the use of transmisso-
meters and other instruments (e.g., the ASOS used by
the National Oceanic and Atmospheric Administra-
tion’s National Weather Service as well as airports
around the world) and through remote sensing plat-
forms. The duration and extent are considered
through such platforms as well. The remote sensing
tools also include satellite, radar, and lidar. Satellite
observations (both infrared and visible) can indicate
fog location and extent and infer fog ‘thickness’ in a
manner far superior to a collection of surface obser-
vation sites across a region. The use of satellite imagery
is very helpful in completing the depiction of fog extent
and ‘intensity’ for a region. Visible imagery often
shows a sharp-edged boundary between a dull gray fog
region, whereas infrared imagery requires a tempera-
ture differencing technique to indicate fog areas. In
either case the ‘strength’ or signal ‘brightness’ shown
by the image is related to its continuous nature (e.g.,
patchy versus widespread) and thickness (strength of
signal) of the fog area.
However, satellite observations have limitations
including minimal temperature contrasts on infrared
imagery and intervening cloud layers on visible
imagery. In addition, satellite observations typically
are not current as processing and dissemination of
images may take an average of 10–45 min. In other
words, satellite imagery is of limited value in terms of
prediction of occurrence. Radar, although not a
detector of fog droplets, is useful in identifying
variations in the refractive index within a limited
radius of the radar site. This can be used to infer the
presence, or development, of an inversion and its
height which may indicate potential fog development.
The use of lidar is designed to detect even low
concentrations of small particles and can be used to
resolve fog formation and occurrence within a limited
distance of an observation site.
Definitions
Fog consists of suspended droplets, some of which
may be settling out and/or evaporating, which restrict
visibility and persist for some period of time. Fog may
vary in depth occurring within the lowest meter of the
atmosphere and extending up to 1000 m in height.
Occasionally the fog layer may be somewhat elevated
above the surface, particularly during the fog disper-
sion, breakup, or ‘burnoff’ process. For condensation,
the relative humidities of the air do not necessarily
need to be 100% and in fact may be as low as 80–90%
both during and after formation. These conditions are
mostly associated with a temperature-dew point
FOG 781
spread of up to 51F (approximately 31C) and a stable
layer of atmosphere. Although fog occurrence is
relatively rare near or below the melting point of
water, freezing or ice fog is possible. Other definitions
of fog relate the fog’s source and/or method of
formation to its name.
Fog may form in place, be transported from one
location to another, and may form in minutes or over
an hour, depending upon existing conditions. The two
basic processes responsible for formation, as well as
duration, are radiation and advection (which includes
vertical mixing of air). It is based upon these observa-
tions that fog is often referred to as ‘a cloud on the
ground’ and which consists of visible hydrometeors.
However, the fog formation process has important
differences from clouds (including, for example, total
moisture content, droplet size distributions, and
chemical contents). By international definition, fog
occurs (or is significant) when visibility is restricted to
less than 1 km (0.62 miles) and is distinguished from
the occurrence of mist (or drizzle).
Fog droplets range in size because of the various
condensation nuclei with which they form and ac-
cording to their resultant wettability and solubility
(and thus their physicochemical composition). The con-
densation process is best illustrated through Kohler
curves that show droplet radius versus saturation level
(i.e., relative humidity of below or near 100% through
supersaturation of up to 0.5%). As the droplet radius
grows, the saturation vapor pressure decreases and
the environment becomes supersaturated with respect
to the droplet. This allows the maintenance of
droplets in an equilibrium state. Because of these
relationships, maritime nuclei often have very small
droplet radii and thus little need for supersaturation
conditions.
Cited values for fog droplets range from 1 to 65 mm
in diameter with an average diameter of 10–20 mm
most often reported (although some sources state the
range as 20–40). Some observational studies differen-
tiate between small droplet fogs (often maritime in
origin) and larger droplet fogs (more typically conti-
nental). Maritime fogs are often observed to be more
continuous and ‘thick’ due to their smaller droplets
that allow a greater concentration within a parcel of
air. The occurrence of fog with light rain or light
drizzle is not uncommon and typically limits the fog’s
‘thickness’ and duration due to the fall of precipita-
tion, mixing, and wind flow that accompany them.
People tend to characterize fog as thick or shallow,
and some definition of this is found within the WMO
coded synoptic and METAR observations. Fog is
usually distinguished from haze according to visibility
considerations, relative humidity, and content. Haze
and fog may occur together and have nearly equivalent
782 FOG

impacts, but fog’s restriction of visibility typically
predominates. Somewhat by convention, fog ‘thick-
ness’ (or ‘intensity’) is defined according to the
restriction of visibility. Dense fog produces visibilities
of less than 1 km, moderate fog 1 to less than 5 km, and
light fog 5 to less than 11 km.These definitions rely on
sight distance as a surrogate for measurement of
droplet distributions and are exclusive of the occur-
rence of low stratus clouds.
Fog duration and coverage, although observed in
time and space, do not have a generally agreed upon
definition. Fog studies often differentiate between
long-lasting (i.e., several hours) and short-duration
fogs (less than 2 h), and in some cases examine
extended events (i.e., several days). Some studies
consider a minimum one-degree ‘square’ of fog to be
sufficient to depict it as a regional occurrence, with
smaller areas being defined as local coverage events.
Extended events lasting more than a couple of days
include late winter and springtime events in New
England and the mid-Atlantic states as well as winter-
time events over the western US valleys, the Gulf
Coast, central Europe, and the Po Valley in northern
Italy (Figure 2).
Climatologies
Global frequencies of fog are traditionally based on
surface observational data and consider the number of
days on which fog was reported. Few climatologies
focus on the time of day, extent, duration, or intensity
of fog which would make for a useful basis of
worldwide comparison. Fog frequencies tend to be
high in locations where moisture is plentiful (oceanic,
river/lake, and coastal regions as well as humid or
tropical locations) or cooling processes predominate
(mountain–valley locations and ocean currents). As
most frequencies are derived from surface-based
observations, they are necessarily skewed by popula-
tion centers and established habitable regions and thus
fog ‘hotspots’ are only approximations of the true
global occurrence and distribution. It must be noted
that even a region with a low fog frequency is not
immune to the devastating impacts of fog. Some

Satellite

Formation Other characteristics

• Condensation • RH 80−100% • Chemistry and pollutants
• T/Td spread 0 − 3°C (0 − 5°F) • Moisture content
• Droplet size distributions • Association with haze, drizzle, rain
• Radiative and Advective processes • Color
Visible
Fog intensity • Inversion level
Dense < 1 km Infrared
Moderate 1 to <5 km
Light 5 to <11 km
Airborne
radar

10 m 500 m Fog Dense fog

Moderate fog Visibility
Light fog
Visibility
Human sensor
Radome observer
Duration
Regional fog event ≤ 2 hours short-term
1° Lat./Long. area > 2 hours long-term
days extended event

Figure 2 Observations and definitions. The figure gives a perspective view, as in Figure 1, with focus on airport location and equipment
and varying fog intensities with spatial scale shown and varying depth of fog layer.

regions may experience an average of 100 days
per year with fog while others may only average less
than 10.
Hemisphere has the majority of land-based and
coastal zone fog occurrences given the spread of
continents around the Earth. Many regions in North
America, Europe, India, Africa, and Asia witness
debilitating fog events, some sporadic, others long-
lasting. A variety of climate zones are affected,
including polar, temperate, tropical and even dry to
moist. Locations with the highest frequencies are
found in the vicinity of cold ocean currents and/or
upwelling that stabilizes the atmosphere, provides
local moisture and aerosols, and can cool air quickly –
and do so for a prolonged period of time – to
condensation. For these locations fog may persist
for several days, cover an extensive area, and then
move en masse to another region. These fog regions
may occur any time of day and tend to be more
persistent during the daytime with little diurnal
fluctuations.
The Southern Hemisphere also has a variety of land-
based fog in regions such as South America, Australia,
South Africa, and the Antarctic. As strong winds and
mixing predominate over much of the Southern
Hemisphere oceans, fog regions tend to persist in the
coastal zones (e.g., Chile) and/or over the interior
where terrain varies considerably (e.g., Argentina). In
addition, the tropical rain forest regions provide ample
moisture and sufficient radiational cooling overnight
for the formation of large fog areas. In these cases, the
time of year and time of day vary from the Northern
Hemisphere. Further observations are much more
limited given smaller population bases and/or the lack
of observations or observational equipment in the
Southern Hemisphere. A wide variety of remote
sensing methods and tools have been used to refine
and localize these values but have not as yet provided a
complete view. For example, aerosol climatology
work and cloud climatologies based on satellite and
radar observations can more readily complete the
picture of fog occurrence, and thus frequencies and
climatology, for these parts of the world.
The importance of the foregoing discussion of fog
climatology is that it helps to identify fog-prone
regions, episodic regions, and assist in the forecasting
process. It also aids in the identification of common-
alities of fog formation and thus offers clues as to the
evolution and behavior of fog. Some fog regions tend
to be synoptically enriched or dominated and others
more by boundary layer processes. Each plays a role in
the extent, intensity, and duration of fog and helps
determine whether fog frequencies are bi- or tri-modal
in time and according to global circulation regimes
(e.g., central North America and the west coast of
FOG 783
South America as related to El Niño and La Niña
phenomena).
Location
A closer look at the common features of fog-prone
areas of the world provides a basic breakdown of
maritime polar climates, Mediterranean climates (e.g.,
France), and mountain/valley climates (e.g., Chile).
Although fog is also more likely to occur in the vicinity
of water sources such as ocean currents and river
valleys and warm water springs (e.g., in cold climates),
it may also be found in desert climates (e.g., Africa and
Australia). Radiational influences are maximized in
these desert regions as well as in valley or plateau
regions around the world (e.g., China and Mexico).
When advective influences predominate (such as
Newfoundland), various lifting and cooling processes
are more important. In addition, the intermediary
zone between ocean and land also plays a significant
role in the extent, intensity, and duration of fog.
Various combinations of these lead many to classify
fogs and develop a list of fog types that may occur. This
allows us to distinguish between the climatic predom-
inance of fog in some regions versus its origins, and
according to the prevailing synoptic flow in real time.
Clearly, this does not imply that other places do not
receive fog or that it is unimportant. It does provide a
context in which the scale of fog occurrence and extent
may be considered. These include synoptic-scale
features (e.g., high pressure), mesoscale variations
(e.g., in moisture distributions), and microscale (and
at times mesoscale) features that affect the occurrence,
extent, intensity, and duration of fog (Figure 3).
Scale
The role of synoptic versus mesoscale and microscale
variations is best understood through examination of
a site-specific climatology. However, this is less prac-
tical – and less informative – when the same principles
are applied to a collection of sites. Instead it is best to
examine the patterns and/or fluxes that are important
at each scale to explore the characteristics of fog.
These are also of use in the identification of fog
precursors which improve our understanding and
prediction of fog formation, extent, intensity, and
duration.
Synoptic-scale features include high- and low-pres-
sure areas (with/without rain or recent rainfall),
locations ahead of a warm front, the warm sector,
and behind a cold front. In the simplest case, high
pressure and sufficient near-surface moisture com-
bined with overnight cooling produce fog as radiative
processes dominate. In the case of low-pressure
systems and frontal regions, some advective and
 784 FOG
Alaska
Current North
FOG Atlantic
FOG Drift
Oyashio
FOG Labrador FOG
FOG N. Pacific Current California Current
am
Current Florida Stre
FOG
Current Gulf FOG
FOG
Kuroshio FOG Canaries
Current
FOG
N. Equatorial Current Guinea
Current

Equatorial Counter-Current
Brazil FOG
Current
S. Equatorial Current Benguela
FOG Current

Peru or FOG
Humbolt
Current
FOG
FOG Falklands
Current

Antarctic Circumpolar Current (West Wind Drift)

Figure 3 Fog climatology. Map of the world showing continents, cold ocean currents, topographic relief and some of the common locations favoring fog occurrence.

radiative processes will dominate. Fog occurrence is
possible, whether inclusive of precipitation-induced
and/or cooling effects. The surface character, over
which these features pass, then acts to promote or
moderate the cooling and/or lift. These include
upslope flow, cold ground (or frozen or snow or ice
cover), and vegetative contributions (in terms of
additional moisture in the local boundary layer).
Mesoscale variations modify the imposed synoptic
conditions and may grow with time to be more
significant if synoptic flow becomes stagnant (e.g.,
autumn) or blocked (e.g., spring) or simply ‘vanishes’
(e.g., low latitudes). Mesoscale variations in physiog-
raphy and weather conditions are obvious in areas
such as San Francisco, Death Valley, Salt Lake City,
and the Pacific Northwest (in the United States, West
Coast) and make a great deal of difference in
forecasting the location, duration, and intensity of
fog. These variations are further complicated by the
complexity of possibilities on the microscale where
boundary layer processes dominate. These tend to be
critical in identifying the precise location of formation,
timing, intensity, and duration. It is also critical with
regard to the microphysical aspects that involve
aerosols. This immediately distinguishes between
maritime and continental concentrations and types
of fog droplets as previously described.
Formation Mechanisms
Fog formation requires a variety of factors in different
combinations. Essential to fog formation are sufficient
moisture and the process of cooling and/or lifting
(inclusive of mixing). This gives us three basic ‘types’
of fog: radiational (cooling), advective (cooling and/or
lifting), and combinatorial (cooling and lifting, or
mixing). These may occur in several ways from the
synoptic to the local scale, as has been illustrated by
the different climates around the world which expe-
rience fog. The more critical factor in the atmosphere
is the presence of sufficient moisture, in terms of total
amount and depth, and its horizontal distribution.
Sufficient moisture may also be achieved by increasing
its ‘effectiveness’, in other words, making use of the
moisture present and realizing it through cooling
and/or lifting processes to cause condensation. In each
case, despite synoptic influences, moisture and its
realization is very much a function of mesoscale and
microscale conditions and variations.
In addition to moisture and cooling/lifting mecha-
nisms, fogs are observed to be associated with an
inherently stable atmosphere. This stability may
precede or occur after fog formation and often
increases with the advent of fog. Even fog that is
associated with strong winds, as is the case with some
FOG 785
advective fogs, occurs in relatively stable layers of the
boundary layer. Other relevant factors in fog forma-
tion which may be considered as secondary in nature
(yet significant in the prediction of fog) include cloud
microphysics, the vertical and horizontal distribution
of temperature and moisture, orographic effects,
sources and sinks of moisture and heat, and land use
and/or surface conditions (Figure 4).
Cooling
There are several means of cooling an air mass, or
parcel of air, that may lead to fog formation. The most
obvious and most prevalent (even in the presence of
cloud cover) is the diurnal loss of heat by the Earth’s
surface (i.e., radiational cooling). Other means include
the cooling of an air mass from below, adiabatic
cooling (or mixing), the cooling of an air mass itself
due to radiational release, and the evaporational
cooling of air due to precipitation through a dry air
layer which may induce cooling to saturation and thus
result in fog. Depending on the location, time of year,
and moisture availability, these cooling mechanisms
may lead to fog formation with varying persistence
and of varying extent and intensity.
Radiational cooling is primarily diurnal in nature
and is maximized overnight and during the early
morning hours, with minimum air and surface tem-
peratures often occurring at or near sunrise. Although
the diurnal cooling process occurs year-round, it is
favored during both dry and cold seasons of the year
when low-level moisture may be sufficient, relatively
undisturbed, and the cooling period lengthy. The
dryness of the atmosphere is most typically observed
above the boundary layer and allows great radiational
losses through an open atmospheric window, even in
the presence of middle- or high-level clouds. Radia-
tional fogs may be brief in duration (e.g., less than 1 h)
or may last several hours. The depth and intensity of
these fog events is a function of the cooling time,
extent, and amount of moisture available. It is not
unusual for such fogs to initiate dew deposition.
The other means of cooling are of varying impor-
tance to fog formation and duration. For example, the
cooling of an air mass from below is favored in
locations and seasons in which the active surface layer
is frozen and/or snow covered or when it experiences a
greater albedo (e.g., fallow versus the vegetative
growing season). Such fogs may form and persist for
hours or days at a time and cover a relatively large area
with significant intensity. Adiabatic cooling of air is
related to lifting mechanisms and therefore considered
in the next section. The cooling of an air mass itself due
to radiational release is typically a very slow process
and likely to be an important factor for persistent fogs
786 FOG
Cooling
Mixing
Cooling
Mixing
Figure 4 Fog formation mechanisms.
(e.g., sea fogs). The extent may be great but the
internal variations in coverage and intensity are large
given the interactions between the air mass and the
underlying surface features. Evaporational cooling
caused by showery precipitation falling through a
layer of dry air may be sufficient to lead to saturation
and fog, but is typically of short duration and of
limited intensity. In the case of a synoptic-scale warm
front, such fog may form and persist for several days
and become quite extensive and intense with minimal
local variations.
Lifting
The second basic means of cooling air to achieve or
sustain fog formation, or for realizing the effectiveness
of the moisture present in a parcel of air, is through
various lifting mechanisms. These include orographic
lift, frontal lift, adiabatic ascent, and mixing. In many
cases, these processes involve advection and thus give
rise to advective fog formation and transport. Al-
though this implies that there are many lifting situa-
tions in which fog may form, it is clear that most of
Other factors
• Wind
• Stability
• Horizontal/vertical temperature
and moisture
• Synoptic features
• Orographic features
• Source/sink moisture and heat
• Land use
• Surface conditions
• Cloud microphysics
Lifting
Cooling
these situations involve slow vertical lifting over large
horizontal distances or the relatively slow and shallow
vertical mixing of two distinct air masses in the
boundary layer. In the former case, long-lasting,
extensive, and intense fogs may be expected whereas
in the latter short term, shallow, and patchy fog of
varying intensity occur.
Slow vertical lift due to an upslope wind flow,
parallel with the elevation gradient, will result in
discrete levels of cooling and saturation with increas-
ing distance and transport. Although this process may
be slow in the initial formation of fog depending upon
the amount of moisture available in the air mass (e.g.,
several hours to nearly 24 h), it is a resilient process
that can produce extended events of widespread dense
fog (i.e., up to several days are possible). Similarly,
frontal lifting may produce similar conditions and may
persist for some time dependent upon the rapidity of
changes in synoptic features. Frontal lift is more
commonly warm in nature but may involve cold
frontal surfaces which are of lesser slope than a typical
cold front. In both orographic and frontal cases, the
formation, duration, extent, and intensity of fog

events is also a function of the underlying surface and
its interaction with the lifted air. For example, the flow
of warm and moist air across frozen or snow-covered
ground – or simply upslope – increases the depth,
intensity, extent, and duration of fog.
This last process is an important aspect and illus-
trates how two diverse air masses, initially unsaturat-
ed, may mix to form a saturated air mass. The use of
saturation vapor pressure curves can be made to
compare air mass properties as a function of their
vapor pressures versus the absolute saturation vapor
pressure for various temperatures and pressures.
When a cool, and relatively lower vapor pressure, air
mass combines with a warm air mass with higher
vapor pressure, their mixing results in saturation. This
may be seen by plotting the original vapor pressures at
the individual air mass temperatures and connecting
the two points with a straight line. When the line
crosses the saturation vapor pressure curve, the two
mixed air masses will form a saturated air mass. The
manner in which these two air masses combine may be
through isobaric mixing or weak adiabatic mixing.
Lift that involves the adiabatic ascent and mixing of
air is greatly dependent upon the existing boundary
layer which evolves during the mixing process to
produce fog. Although of limited extent, turbulent
mixing through adiabatic ascent can result in fog
formation which is typically of very short duration
(i.e., less than a few hours), limited depth (e.g., ground
fog), and highly variable in coverage and intensity.
Such fogs may occur preceding and following the
passage of weak cold fronts with limited pressure and
air mass differences, and often following the passage
of scattered showers or light rain, and take place in a
conditionally stable boundary layer. These fogs tend to
be infrequent and of short duration as the dynamics
are more likely to lead to low cloud (and ceiling)
formation with drizzle. However, in some cases they
may persist and thicken over regions as the frontal
boundary decays and/or becomes stationary. The
processes of adiabatic ascent and mixing also play a
role in the formation of Arctic Sea Smoke and other
fogs in which the heat flux is rapid and results from a
temperature differential rather than a period of
radiational cooling.
Lifting and Cooling
Based on the preceding discussion, it is clear that there
are many possible combinations which may produce
fog. It is therefore understandable why so many ‘fog
types’ occur in the literature and are studied around
the world. For the same reason, it is clear that these
possibilities raise the question of whether fog is readily
predicted and whether one type is readily identified
FOG 787
over another or whether one type may evolve into
another. Since the orographic and frontal lifting
processes are typically a gradual cooling process over
long distances whereas the radiational cooling process
is gradual over time and specific to a location, it is
reasonable to consider various combination fogs in the
same manner. It is also reasonable to incorporate the
effects of cloud microphysics, the vertical and hori-
zontal distribution of temperature and moisture,
sources and sinks of moisture and heat, and land use
and/or surface conditions.
For example, given the features described above, the
longest-lasting, most intense, deepest, and potentially
most widespread fogs may occur near a coastal region
with a moist onshore flow in the vicinity of a warm
frontal (or topographic) or quasi-stationary boundary.
This would be further enhanced or favored if the flow
of moisture were sustained, the ground frozen or snow-
covered (and thus the source of cooling maintained),
and it was the cool season time of year. The formation
and advection of sea fog tends to meet these criteria to
varying degrees around the world and create some of
the foggiest regions known. Regardless of origin, that is
whether the sea fog formed first through radiational
cooling or other cooling and lifting processes, it is clear
that a variety of factors produce and sustain fog. This is
verified by observation of the movement of fog areas
and their passage from water to land.
Microphysical and Other Aspects
Although the identification of cooling and lifting
mechanisms is significant with regard to the conden-
sation process for fog formation, alone they are
insufficient if not considered with regard to the
nucleation process. A knowledge of drop size distri-
butions, condensation nuclei associated with fog
droplets, local nuclei populations, and the resulting
atmospheric chemistry are significant with regard to
the occurrence, extent, intensity, and duration of fog
events. Depending upon the population of condensa-
tion nuclei, the initiation of fog droplet formation, and
the actual drop size distribution, fog development may
occur within 5–15% of the saturation value of an air
mass (haze is typically within 35%). The role of nuclei
is determined by ‘how active’ they are in encouraging
or discouraging the process of condensation (i.e., how
hygroscopic or hydrophobic). The presence of con-
densation nuclei may be local in origin, advected, or
the result of both processes and is significant when fog
is considered in combination with smoke and pollut-
ants. The ‘proper’ combination of nuclei can lead to
long-lasting and devastating fog events.
For example, there are a number of well-known
cases in which fog combined with, and its formation
788 FOG
was aided by, industrial emissions. These created
unhealthy and dangerous air quality and low visibili-
ties over several hours (and even days), causing death,
illness, injury, and accidental losses. These can be
further modified according to the location in which
they occur. For example, marine environments typi-
cally produce a large number of small droplets
whereas continental locations are characterized by
large droplets. In marine environments then it is clear
that while haze is favored in the daytime (and salt
nuclei) and a high moisture source with mixing, fog
may still occur if these conditions are overcome (e.g.,
at night by cooling and with the introduction of
smaller droplets).
Regardless of the limitations to droplet size by the
concentrations of nuclei, it is the actual droplet
concentration that determines a fog’s opacity – and
that is often referred to as a fog’s intensity or thickness
(or severity). Because of this the lowest visibilities in
fog events are associated with high concentrations of
small droplets. Thus again a wide variety of fog
formations are possible, particularly when consider-
ing the cooling and/or lifting processes and when
considering the location, transport, and interactions of
various nuclei across a coastal zone.
The ‘Family of Fogs’
The foregoing discussion thus provides a ‘family of
fogs’ in terms of formation, extent, duration, and
intensity which may be enveloped in a conceptual
model of fog dynamics. Although similar in manner to
the conceptualizations of air mass thunderstorm to
supercell, or mesocyclone to wave cyclone, or various
levels of sea/land-breeze model conceptualizations,
the family of fogs is, at the moment, quite elusive.
Although generous amounts of research have been
accomplished and numerous modeling studies com-
pleted to reveal more explicitly the cooling and lifting
processes that may produce fog, they are as yet
incomplete. They do not incorporate adequately the
atmospheric chemistry and associated microphysics
that are significant to the fog process and are poorly
understood and not routinely observed. There is also a
need to consider the interactions and interface be-
tween the underlying surface over which fog forms and
these microphysical aspects of chemical and physical
behaviors.
Fog Dynamics
Given the basic knowledge of fog formation mecha-
nisms (or occurrence), and some knowledge of micro-
physical aspects, it is possible to explore the dynamic
processes involved and how they relate to fog intensity
(or thickness), extent (or coverage), and duration.
These are predominantly radiative processes as lifting
mechanisms are essentially cooling processes as well.
It is assumed that if the proper combination of factors
exists, and that moisture is available in sufficient
quantity and/or its effectiveness can be realized, fog
formation is possible. Adequate moisture is provided
through local evaporative fluxes, advective delivery, or
evaporation of falling precipitation. Ideally these are
all quantified operationally (whether modeled, ob-
served, or forecast) to make a precise determination of
fog occurrence, intensity, extent, and duration.
Radiative Considerations – Formation
and Growth
Rapid cooling, but with limited mixing, is best
accomplished through radiative heat loss by the
surface and by an air mass that is predominantly static
in nature and predisposed to stability. These condi-
tions are favorable to the potential for fog formation,
assuming sufficient cooling occurs and sufficient
moisture is available (or its effectiveness realized).
Commonly, such conditions are associated with clear
(and sometimes dry) air, light winds, and subsidence.
However, radiative cooling does occur in the presence
of cloud cover and may be enhanced or reduced by the
active surface over which air is present. Each of these is
maximized within high-pressure areas when winds are
near calm, the boundary layer is moist, and the mid-
and upper layers of the atmosphere are very dry with
strong subsidence. These conditions can produce
widespread, intense, and long-lasting fog. Less opti-
mal conditions (e.g., those present in the midst of a
weak low-pressure center with little pressure gradient)
may also produce fog that is of limited extent,
intensity, and duration.
Although radiative and advective process has been
initiated, the lowest portion of the boundary layer
nearest the ground becomes the coolest and thus
creates a microscale inversion. This inversion grows in
tandem with the rate of cooling and the net cooling
over time and, as the air reaches saturation, may lead
to fog formation. As the cooling process continues, the
inversion layer grows deeper and fog may grow or
develop upwards with time to several meters within a
few hours. At this point the presence of fog itself begins
to feed back into the radiative balance as warmth from
the surface may be absorbed by the droplets and the
rate of cooling slowed. In addition, the upper portion
of the forming fog and the fog layer itself continue to
cool, thus strengthening the inversion, while some
deposition and/or reevaporation of fog droplets occurs
nearer the surface. Often at this stage the radiative

processes immediately at the ground have slowed,
become less important, and reach a temporary equi-
librium in which temperature and moisture conditions
remain nearly constant.
Although these processes dominate in a general
sense, they are complicated by the nature of the surface
over which air lies (e.g., soil type), land use and cover,
heat and moisture sources, and vehicular and other
traffic that create local turbulence. It is the combina-
tion of these factors that dictates the areal extent and
initial intensity of fog and that accounts for variations
as the fog persists. Sand and clay soils radiate at
different intensities and thus can contribute to fog
formation at different rates. In the simplest case, the
ground surface may be conducting heat from subsur-
face layers and thus eroding fog formation from the
bottom after its initiation. The surface may also be
considered active if wet or dry, vegetated or barren,
paved or natural, and frozen or snow-covered. In the
case of a wet ground, more moisture and enhanced
radiative cooling are possible. For a frozen surface
there is a strong limitation on the radiative cooling of
the ground and a strong enhancement of the cooling of
the air itself. In the case of an asphalt or concrete
roadway, the radiative rates may favor more rapid fog
formation whereas vehicular flow (and turbulence)
may discourage or disperse it.
Duration and Extent
As the radiative cooling persists, the fog layer may
grow vertically and horizontally with time. In partic-
ular, it is not unusual for the fog on the surface to
diminish with time, through deposition (dew or frost)
and weak induced convective mixing, giving the
appearance of a lifting fog. In this process, the
moisture content and subsequently the dew point are
decreased within the lowest portion of the fog layer
and may allow for further radiative cooling and fog re-
formation. Thus some fogs appear to vary in their
occurrence, depth, and intensity with time. More
importantly, the middle and upper portions of the fog
layer are now the most ‘active’ in terms of their
behavior. In particular, these layers continue to cool,
resulting in an upward expanse of the fog, and thus the
inversion layer, and therefore become somewhat
independent of the surface over which they lie.
The horizontal formation and spread of the fog,
initially a function of the radiative properties of the
surface over which the air is found and which account
for the often patchy nature of fog formation, now
becomes a function of weak circulations or turbulence
above the ground. In the presence of high pressure and
light winds, this may create a fairly uniform fog in
terms of intensity and duration. For a weak low-
FOG 789
pressure system, or perhaps in advance of a warm
frontal boundary, both in the presence of light winds,
this may create large variations in fog extent, intensity,
and durations with frequent and rapid variations.
These conditions also imply slow transport of a fog
layer and modification as the fog travels across varying
terrain and surfaces and as it encounters variations in
condensation nuclei. Modifications include other
cooling processes, such as orographic lift, which may
reduce or enhance fog formation, maintenance, and
growth. For example, a fog that develops in the
vicinity of weak low pressure may move upslope in
response to a weak pressure gradient and either
precipitate out or intensify and deepen. The same fog
may encounter an industrial area with a variety of
nuclei, which may lead to changes in the drop size
distributions and either accelerate or defer fog forma-
tion.
Intensity
Once fog has formed, persisted, and grown, the
intensity or thickness of the fog is of greater practical
significance. Fog intensity is a function of the concen-
tration of small and large drops in terms of their ability
to reduce visibility to less than 1 km. When fog is in
place, there is a certain amount of droplet settling and
separation with time while cooling and condensation
occur at fog top and dissipation and deposition at fog
bottom (in many cases). While this process is effective
in maintaining fog, it is also effective in stratifying the
fog layer and generating variations in intensity. In
these situations the middle and upper fog layers
become the most opaque and thus the most difficult
to navigate in terms of transportation. These effects
may be negated to some extent given a greater degree
of turbulent mixing or the presence or introduction of
wind flow (and thus entrainment) near the top of the
fog layer.
In addition, the characteristics of condensation
nuclei, can lead to varying intensities. For example, a
variety of pollution-enhanced fogs owe their extreme-
ly low visibilities to specific chemical species. These
encouraged fog formation prior to saturation, reduced
the amount of deposition, and created a more uniform
drop size distribution often characterized by its color
or smell.
Dissipation
The dissipation of fog, is a function of the processes
that act against cooling and condensational effects
previously discussed. Therefore dissipation may be
considered in terms of the ‘prevention’ of the processes
that favor formation and growth, duration and extent,
and intensity. Many of these processes may reduce or
790 FOG
eliminate fog within less than 1 h without contraven-
tion, but more typically require several hours to
overcome the inertial presence of the fog layer as
well as any underlying or continuing fog formation
processes. Usually visibility improvements occur
within the first hour or two as dissipation processes
become dominant. Fog dissipation is typically longer
in valley and coastal regions and during the cool
season and over cold waters – and in some cases may
not occur.
The effects of cooling are mitigated or overcome
through direct solar heating of the ground surface (or
potentially the heating of fog droplets and the air layer
in which the fog is found, but this is of minimal
significance) and the destruction of a stratified or
inversion layer through turbulent mixing. Thus warm-
ing and mixing are vital to dissipate fog and may be
accomplished through a variety of frontal or convec-
tive processes. This dissipation will be most effective at
the bottom and top of the fog layer and around the
edges of the fog area where the air is not near
saturation. Thus fog is said to ‘burn off’ or ‘lift’ and
‘shrink’ with time. The pace of dissipation, which will
be greater under an imposed pressure gradient and
during the warm season, may last a few hours.
Prediction
The fog process is quite dynamic and requires a great
deal of physical knowledge and observation to be
understood completely and predicted successfully. In
practice however, this is difficult given the limited
nature of our present understanding of fog and the lack
of real-time observation, quantification, and modeling
of the chemical and microphysical behaviors that are
involved. This is made more difficult by the lack of
precise observation and modeling of the detailed
surface characteristics, the distribution of moisture,
temperature, and their interaction. However, if the
conceptual model of a family of fogs based on the
principles previously described is applied, the chances
for improved fog prediction may be increased. Ulti-
mately even a microscale observation network would
not be dense enough to provide essential details to
improve prediction. Instead the further application
and refinement of remote sensing tools and numerical
modeling will be necessary to better forecast fog
occurrence, extent, intensity, and duration in a wide
variety of circumstances.
Current forecast practices for fog include climato-
logical approaches, numerical guidance, observation-
al methods, modeling, and other methods (e.g.,
statistical or decision-tree methods and Artificial
Intelligence – or Expert Systems). Common to each
of these is the recognition of those processes and
factors important to fog formation including cooling,
lifting, and mixing (cooling and lifting) mechanisms;
surface and air mass moisture and characteristics; and
the chemical and physical behavior associated with
fog droplets. Some of these are summarized by many
authors and researchers according to a list of factors
such as the prior existence of fog, precipitation areas,
soil moisture and cover, temperature and stability,
boundary layer variations, orographic contributions,
the synoptic setting and flow regime, vertical and
horizontal wind flows, cloud cover, and advection.
Regardless of the list, it is essential to consider the
synoptic-dynamic regime as it determines the charac-
ter of the boundary layer – and how that changes with
time – to predict adequately fog occurrence, extent,
intensity, and duration.
Climatological
One traditional means of fog prediction is based on
climatological considerations. This approach, al-
though location specific, is a reasonable first approx-
imation in identifying and summarizing the factors
which produce fog in a given region. Typical climatol-
ogies assess fog frequencies according to the time of
day and year (or by season), intensity, duration, and
sometimes according to fog occurrence at multiple
sites. Unfortunately, the climatological approach
tends to oversimplify fog prediction by categorizing
events as fog ‘types’ with little regard to fog evolution.
In addition, the results are biased for the site at which
the climatology is based and can thus lead to predic-
tions of no fog for a region which may fail.
The use of conditional climatology adds some value
beyond simple climatology in that persistence and
contributing factors associated with a fog event can be
anticipated. Yet this approach is also limited in that it
does not address fog dynamics adequately and relies
heavily upon the data available. In total a forecaster
may be able to assimilate climatological information
and estimate local variations with some experience
for a region, but the process is pragmatic rather
than scientific and is only slightly more skillful.
Climatological predictions have only limited antici-
patory value: knowing the types of situations that
favor fog.
Numerical Guidance
Another approach to fog forecasting is through the use
of numerical guidance, both raw and processed. Raw
data from an operational model may be plotted or
examined across a region and yield specific informa-
tion with regard to moisture and cooling with regard
to fog formation. However, model output has a variety

of errors, is often available in only limited time and
space resolution, and typically does not adequately
depict the boundary layer structure or behavior. When
the raw data are processed to produce graphical and
diagnostic analyses, surface and boundary layer plots,
or to generate statistical forecast guidance, it is of
much greater value. The graphical diagnosis of the
local environment according to model output is useful
in better identifying regions potentially favored for fog
development (e.g., moisture maxima, cool air pockets)
and thus can aid a forecaster in refining a prediction of
occurrence and extent. This provides a mesoscale
prediction which can be partially verified through
surface observations and satellite imagery.
When used to generate statistical guidance, model
output helps a forecaster assess the confidence level of
fog formation and extent and allows some speculation
as to its intensity and duration. This guidance makes
use of both current observations and model predic-
tions. For operational models with output statistics
from 12 to 60 h into the future, there is some predictive
value and skill improvement over climatology in the
anticipation of fog. However, most statistical guidance
used in this manner is derived from regression
relations that show current observations (or simply
persistence) to be the most important factor in fog
prediction. Thus numerical guidance is of best value in
anticipating fog during the first several hours, or
anticipating the synoptic setting conducive to fog
several days in advance.
Observational
The observational approach to fog prediction is
extremely limited in an anticipatory sense, but tends
to have the greatest skill in terms of fog extent,
intensity, duration, and ultimate dissipation. This
means that a forecaster has synthesized the prevailing
synoptic environment and its interactions with the
boundary layer to understand why fog is present (or
expected) and why it will continue to persist or
dissipate. Essentially a forecaster in this situation has
developed and applied a conceptual model frame-
work, based on theory, knowledge, and experience
which provides specification that no other current
method can match. The observational approach
involves a wide synthesis of sensible surface weather
elements, middle and upper atmospheric data and
analyses, and a review of satellite (and perhaps even
radar) images. Practical application of radiative laws
and parcel theory through the use of sounding
diagrams (e.g., a skew-T chart, and even isentropic
analysis) allow a forecaster to ‘run’ a conceptual
model of fog and to make a successful prediction.
FOG 791
Modeling
Many modeling efforts have been made to simulate as
well as to produce a fog environment and fog droplets.
Presently, no operational model is capable of com-
pletely generating these in real time for predictive use.
Modeling has more recently focused on drop size
distributions, deposition, and re-creation of the fog
environment as well as the physical representation of
fog and the fog process. Limited studies, including
statistical prediction, have focused on the prediction of
the visibility restrictions of fog. The First International
Conference on Fog and Fog Collection is indicative of
the new and renewed importance of fog study.
Much work is now focused on collection tech-
niques, particularly with regard to fog chemistry and
its impacts with regard to vegetation. In this regard,
fog modeling has served as an extension of agro-
meteorological and climatological study as well as
micrometeorology and cloud physics. Some profes-
sional consultants and companies now offer fog
forecast products, and even assessments, many of
which focus on predicting the occurrence and intensity
of fog for a specified location. A variety of other
methods, including statistical or decision-tree meth-
ods and Artificial Intelligence (or Expert Systems),
have found usefulness in the study of fog.
Benefits, Mitigation, and Special
Cases
Although a variety of severe impacts have occurred
with fog events around the world, efforts to mitigate
these impacts have been of limited success in part due
to the nature of human response (e.g., driving more
slowly through fog). However, fog has also yielded
some beneficial effects in terms of esthetics and
agricultural application. Fog has also been the subject
of literature and film and often accompanies the
depiction of paranormal and Halloween activities.
Fog systems are used routinely to create fog and other
special effects by the film industry. There has even been
a survey that found 28% of people engaged in sexual
relations more than once when it was foggy as
compared with only 11% who did when it was sunny.
Several facets of fog illustrate how its significance
extends beyond a simple reduction in visibility. For
example, fog water use for deposition collection is
critical to agricultural operations in Chile, Mexico,
and other regions of the world. Fog water may be of
use in providing a potable water supply for others.
Most often fog water is collected for agricultural
applications and has been studied with regard to its
effects on the growth of giant Redwood trees in
California (USA). Study also indicates that fog may
792 FOG
play a significant role in the physical interactions
found within plant canopies and their physiological
and growth conditions. These are also related to fog
water pH and the production of acid fogs. These and
other issues are being explored through various studies
around the world.
There also exist programs and methods for the
dispersion, or enhancement, of fog – particularly at
airport locations – and the reduction of acid fogs. The
basic methods tend to focus on heating of the fog layer
(to evaporate droplets), downwash mixing (to entrain
drier air), hygroscopic treatment (e.g., ice seeding) to
precipitate out, and the use of fog breaks (passive
control) to prevent formation or movement into an
area. The most effective methods tend to be those that
match the natural dissipative factors (i.e., mixing and
evaporation) and that promote improved visibility
within an hour. Unfortunately most techniques are
costly, labor intensive, esthetically intrusive, and must
be maintained until natural processes are capable of
continuing dissipation (often for at least several
hours). These measures are also impractical for
roadways and therefore drivers instead rely on vehicle
fog lights and fog-free lenses or shields. The enhance-
ment of fog, although seemingly undesirable, is an
industry with commercial application for party sup-
pliers and movie studios.
See also
Agricultural Meteorology and Climatology. Air–Sea
Interaction: Freshwater Flux; Gas Exchange; Momen-
tum, Heat and Vapor Fluxes; Sea Surface Temperature;
Storm Surges; Surface Waves. Anticyclones. Arctic
Haze. Aviation Weather Hazards. Boundary Layers:
Coherent Structures; Complex Terrain; Convective Bound-
ary Layer; Modeling and Parameterization; Neutrally Strat-
ified Boundary Layer; Observational Techniques In Situ;
Observational Techniques–remote; Ocean Mixed Layer;
Overview; Stably Stratified Boundary Layer; Surface Layer.
Chemistry of the Atmosphere: Chemical Kinetics; Gas
Phase Reactions; Ion Chemistry; Laboratory Studies;
Principles of Chemical Change. Cloud Chemistry. Cloud
Microphysics. Coastal Meteorology. Deposition. Di-
urnal Cycle. Humidity Variables. Hydrology: Ground
and Surface Water; Modeling and Prediction; Overview.
Isentropic Analysis. Land–Atmosphere Interactions:
Canopy Processes; Overview; Trace Gas Exchange.
Lidar: Atmospheric Sounding Introduction; Backscatter;
DIAL; Doppler; Raman; Resonance. Mesoscale Mete-
orology: Overview. Microclimate. Numerical Models:
Chemistry Models; Methods. Observation Platforms:
Balloons; Buoys; Kites; Rockets. Observations for
Chemistry (In Situ): Chemiluminescent Techniques;
Gas Chromatography; Ozone Sondes; Particles; Reso-
nance Fluorescence; Water Vapor Sondes. Observations
for Chemistry (Remote Sensing): IR/FIR; Lidar; Micro-
wave. Operational Meteorology. Orographic Effects:
Lee Cyclogenesis; Mountain Waves and Stratospheric
Chemistry. Parameterization of Physical Processes:
Clouds. Radar: Incoherent Scatter Radar; MST and ST
Radars and Wind Profilers; Precipitation Radar; Synthetic
Aperture Radar (Land Surface Applications). Radiative
Transfer: Absorption and Thermal Emission; Cloud-ra-
diative Processes; Non-local Thermodynamic Equilibrium;
Scattering. Satellite Remote Sensing: Aerosol Meas-
urements. Static Stability. Synoptic Meteorology:
Forecasting; Weather Maps. Thermodynamics: Moist
(Unsaturated) Air; Saturated Adiabatic Processes. Turbu-
lence and Mixing. Turbulent Diffusion. Weather
Modification: Inadvertant. Weather Prediction: Adap-
tive Observations; Data Assimilation; Ensemble Prediction;
Regional Prediction Models; Seasonal and Interannual
Weather Prediction; Severe Weather Forecasting. Wind
Chill.
Further Reading
Anderson JR (1985) Economic impacts. In: Houghton DD
(ed.) Handbook of Applied Meteorology. New York:
Wiley.
Air Weather Service (1979) General Aspects of Fog and
Stratus Forecasting. USAF AWS TR 239.
Croft PJ, Pfost R, Medlin J, and Johnson G (1997) Fog
forecasting for the Southern Region: a conceptual model
approach. Weather Forecasting 12: 545–556.
Eagleman JR (1991) Air Pollution Meteorology. Trimedia
Publishing Company.
George JJ (1960) Weather Forecasting for Aeronautics. New
York: Academic Press.
Houze RA Jr (1993) Cloud Dynamics. New York: Academic
Press.
Mason J (1982) Physics of radiation fog. Journal of the
Meteorological Society of Japan 60: 486–499.
Mayer WD and Rao GV (1999) Radiation fog prediction
using a simple numerical model. Pure and Applied
Geophysics, in press.
Schemenauer RS and Bridgman H (ed.) (1998) Proceedings
of the First International Conference on Fog and
Fog Collection, 19–24 July 1998, Vancouver, British
Columbia, Canada.
Online reference sources
http://meted.ucar.edu (COMET Module Radiation Fog)
http://www.cco.net/Btrufax/fluoride/fog.html

FORENSIC METEOROLOGY
L E Branscome, Climatological Consulting
Corporation, Palm Beach Gardens, Florida, USA
Copyright 2003 Elsevier Science Ltd. All Rights Reserved.
Introduction
Forensic meteorology is the study of past weather
events for the primary purpose of assisting a court of
law in adjudicating disputes in which weather may
have been a factor. The task of a forensic meteorologist
is to diligently investigate the weather conditions
pertinent to the litigation and objectively present to the
court the most accurate description possible of the
relevant meteorological events. Expert testimony by a
meteorologist often provides crucial information in
various types of civil and criminal proceedings.
Civil litigation is often initiated when a personal
injury or financial loss occurs, or when a contractual
dispute cannot be resolved. In many cases the weather
is a contributing factor, or even the primary cause of
the loss or dispute. For example, weather frequently
contributes to maritime cargo losses, aircraft and
highway accidents, property damage, construction
delays, air pollution emergencies, slip and fall injuries,
and structural failures. The party sustaining the loss or
injury may seek compensation through the legal
system from another party who is alleged to be
responsible for the incident. A judge, jury, or other
adjudicative body must determine the facts of the case
and make a decision in favor of one of the parties in the
litigation, either the plaintiff or the defendant. If the
weather was a factor in the incident or dispute on
which the lawsuit is based, then expert testimony
about the weather conditions relevant to the case must
be presented to the adjudicative body in a courtroom
or other judicial setting. Meteorological analysis and
testimony are sometimes necessary simply to rule out
weather as a contributing factor. Together with other
facts, testimony, opinions, exhibits, and evidence, the
adjudicative body considers the relevant weather
evidence and opinions in making its determination.
In addition to civil litigation, testimony about the
weather is occasionally required in criminal proceed-
ings. For example, the ability of an eyewitness to
identify an alleged assailant in low natural lighting is
occasionally an issue in criminal trials. In such cases a
meteorologist may be called to testify about the sky
conditions, phase of the moon, or sunrise and sunset
times. Other criminal proceedings have depended on
meteorological analysis and testimony as it relates to
the decomposition of the bodies of murder victims, the
FORENSIC METEOROLOGY 793
transport of odors from illegal drugs, the death of
children trapped inside motor vehicles exposed to
intense sunlight, and various types of criminal negli-
gence.
In a broader sense, the work of a forensic meteoro-
logist is not necessarily limited to civil litigation and
criminal prosecutions. For example, a windstorm that
causes property damage may result in a claims dispute
between an insured party and its insurer. A meteorol-
ogist may be hired by one of the parties in the dispute
to ascertain the peak wind speed during the storm. The
conclusions of the meteorologist may be sufficient to
allow the client to make a determination about the
claim and settle the dispute with the opposing party,
thus avoiding the effort and expense of a lawsuit.
Similarly, a forensic meteorologist may be hired to
provide information and expert opinion about past
weather events to parties involved in administrative or
regulatory disputes with government agencies.
The Practice of Forensic Meteorology
General Procedures and Considerations
Technical experts are often retained by attorneys
representing the parties involved in a lawsuit. The
immediate purpose of retaining a meteorologist is to
obtain information and expert opinion about the
weather conditions so that the attorney can evaluate,
with assistance from the expert, the importance of
weather to the case. The attorney will consider how
the weather facts and opinions may support or refute
theories for the causation of the loss that led to the
litigation. If the meteorological information and
opinions are relevant and helpful, the weather expert
may be asked by the attorney to prepare a formal
report, evaluate the reports and opinions of the
meteorologist (if any) working for the opposing legal
counsel, assist the attorney in evaluating the merits
and demerits of the case from a technical perspective,
develop exhibits for presentation at trial, and provide
expert testimony and opinions at deposition and trial.
The expert investigating and testifying about the
weather should be a meteorologist with sufficient
knowledge, skill, experience, and education to offer an
opinion about the particular weather conditions
related to the litigation. Although meteorologists are
not usually granted professional licenses like engineers
or architects, the American Meteorological Society,
for example, has a certification program for consult-
ants, identified as Certified Consulting Meteorologists,
794 FORENSIC METEOROLOGY
that involves extensive testing and board review.
Many attorneys find that the certification assists
them in identifying qualified experts. The forensic
meteorologist must also have a high degree of integrity
and composure under pressure, so that the expert testi-
mony that he or she gives in the courtroom is
trustworthy, unbiased and professional.
The admissibility of expert opinions in the federal
courts of the United States is governed by the ‘Daubert
test’ in which the judge assesses whether the reasoning
or methodology underlying the opinion is scientifically
valid and the offered testimony is relevant to the case.
Many non-federal courts still adhere to the earlier
‘Frye test’ in which expert opinions must be based on
principles and techniques that are ‘generally accepted’
as reliable within the relevant scientific community.
Before accepting work as an expert, the forensic
meteorologist should inquire about the nature of the
case and the parties and attorneys involved, in order to
avoid possible conflicts of interest. A conflict may arise
if the expert has had prior exposure to the facts and
legal arguments of the case or if the opposing attorney
is a regular client of the expert, particularly for cases of
a similar nature. In some instances the expert should
decline to accept the work if the case involves a special
area of meteorology in which the expert has limited
knowledge or training. For example, an expert in air
pollution meteorology may refer the client to an
aviation weather expert if the case involves an aircraft
accident in a severe thunderstorm.
After accepting the case, the weather expert should
obtain all the information that the attorney possesses
that might be relevant to the meteorological investi-
gation. This information will include, at a minimum,
the date, time, location, and nature of the event that
led to the litigation. It may also include reports and
statements describing the event that were prepared by
investigative agencies, law enforcement officers, or
eyewitnesses. For example, the National Transporta-
tion Safety Board of the United States produces
investigative reports that provide information about
the particulars and contributing factors in aircraft
accidents. The testimony of eyewitnesses regarding the
weather conditions can sometimes provide descriptive
details about a weather event that would not be
evident in official meteorological data. The expert may
also find it useful to visit the site of the incident to
assess the importance of exposure, surrounding ter-
rain, and site orientation to the analysis and interpre-
tation of the meteorological events.
The Value of Meteorological Data and Analysis
Although the attorney retaining the expert may have a
substantial amount of information about the event, a
forensic meteorologist cannot offer a reliable opinion
until weather data are obtained and reviewed. The
nature and quality of the data needed to formulate
expert opinions depends on the nature of the case and
the relative importance of weather to the arguments
and theories in the litigation. In the case of a slip and
fall injury, it may only be necessary to obtain hourly
rainfall data from a nearby weather station. In the case
of a fatal aircraft accident, a wide variety of data may
be needed to formulate a clear and complete descrip-
tion of the weather conditions at the time and location
of the accident. The data may include surface weather
observations, upper-air weather charts, satellite and
radar images, lightning strike data, aviation weather
forecasts, and pilot reports of weather conditions
aloft.
Forensic meteorological investigations often de-
pend on the analysis of severe weather phenomena
that have significant variations over small temporal
and spatial scales. For example, a severe thunderstorm
can create strong winds and large hail that cause
isolated property damage in a few seconds to minutes.
The ability to reconstruct past weather for forensic
purposes has been greatly enhanced by recent advanc-
es in observational techniques that target mesoscale
weather phenomena. For example, the National
Weather Service of the United States operates a
network of Doppler weather radar stations that
provide nearly complete and continuous coverage of
the country. Similar radar networks are found in other
countries. The radar monitoring of severe weather and
the archiving of the radar data have allowed forensic
meteorologists to develop more accurate and detailed
descriptions of severe weather events.
Similarly, the weather is monitored continually
from space by geostationary and polar-orbiting satel-
lites. Improvements in the resolution of the on-board
sensors and the addition of more observing channels at
infrared wavelengths allows forensic meteorologists
to investigate past weather events in greater detail. For
example, combinations of data from different infrared
channels permits the identification of potential avia-
tion hazards such as fog or airframe icing. Other
satellites are capable of measuring ocean wave heights
and estimating surface wind speeds at sea, which is
valuable information for the investigation of cargo
and ship losses during intense storms.
Other advances in meteorology are related to
improvements in computer models of the atmosphere.
The ability to computationally simulate or analyze
mesoscale weather phenomena has greatly improved.
The global re-analysis of historical weather data with
computer models that are also used in operational
weather forecasts has provided another new tool for
the forensic investigator.

Acquisition of Weather Data
The meteorological data used in forensic work are
commonly stored in government archives and can be
accessed by the general public for usually modest fees.
The primary sources of data in the United States are the
National Climatic Data Center, which is part of the
National Oceanic and Atmospheric Administration,
and Regional Climate Centers and State Climatolo-
gists. Similar government archives are operated by the
weather services of other countries. In addition to
national government sources, a number of local
government agencies, universities, and private com-
panies archive meteorological data and operate
weather observing stations in special monitoring
networks. While anyone involved in litigation can
obtain weather data, a forensic meteorologist is
usually more efficient in selecting and obtaining the
relevant data and organizing it into a coherent source
of information for the analysis of weather events.
The collection, archiving, and retrieval of meteor-
ological data had undergone considerable change in
recent years, affecting the practice of forensic meteor-
ology. In the past most meteorological data were
recorded and archived on paper or film. For example,
wind data were recorded on instrument recorder
charts, radar scope images on film, and surface
observations on handwritten forms. The data and
analysis techniques were often limited in quantity and
complexity. Forensic meteorologists usually ordered,
and in many cases continue to order, paper copies of
data and images that bear a fixed certification and seal
of authenticity from the government agency providing
the data. The physical seal and certification of the data
copy are still regularly used as a means of authenti-
cating the copy for admission into evidence at trial.
Production and delivery of the hard-copy data and
images by the data supplier often resulted in waiting
times of a few weeks to months. Analysis of the data by
a forensic meteorologist was labor-intensive, particu-
larly for cases in which years of data had to be
reviewed to determine normal climatological condi-
tions, since the data had to be extracted from the
certified paper forms.
The recording, storage, and delivery of data on
paper, microfiche, and film is rapidly diminishing.
Low-cost computer storage of data, data-intensive
observational systems such as Doppler weather radar
and multichannel satellite sensors, complex computer
models of the atmosphere and ocean, digital surface
weather sensors, computer networks, and the Internet
are radically changing the work of the forensic
meteorologist. Many government and private data
sets are now maintained on computer servers on the
Internet and are easily accessible at low or no cost to
FORENSIC METEOROLOGY 795
the end user. Even large data sets, such as satellite and
radar data, can be extracted from computer tapes at
government archives and delivered to the end-user by
file transfers over the Internet. Many data providers
are also providing large data sets on CD-ROM and
DVD. Large quantities of data can thus be retrieved
and analyzed on the computer workstations of foren-
sic meteorologists. Waiting periods for data delivery
have generally been reduced from weeks and months
to minutes and days. While the time required to
acquire and analyze the data on a per-unit basis has
been substantially reduced, the quantity and variety of
data available for forensic work has increased dra-
matically, so that the total amount of effort in a typical
investigation has not diminished. Even so, improved
access to data often allows experts to provide their
clients with preliminary analysis and opinions in a
time frame shorter than previously possible.
The physical certification of digitized data is often
impractical, if not impossible. The authenticity of
digitized meteorological data as admissible trial evi-
dence is usually not problematic, provided the testi-
fying expert can demonstrate that the data was
obtained from reliable sources and is customarily
used by meteorologists in their work.
Reports and Testimony
After the weather data has been obtained and ana-
lyzed, a report is made to the client either verbally or in
writing. A formal report that includes expert opinions is
sometimes required by the client or by the judicial rules
governing the litigation. The report becomes a basic
reference document for future testimony by the expert.
Before a case goes to trial, a discovery period occurs
when the attorneys in the litigation request copies of
the documents and information used or prepared by
the other side’s technical experts. The information
includes the data gathered and any written reports
prepared by the weather expert. Furthermore, the
expert is questioned in a deposition by the opposing
legal counsel to discover the expert’s opinions and the
foundational basis for those opinions. The discovery
and evaluation of the weather data and expert opinions
sometimes encourage a settlement between the oppos-
ing parties, particularly if the weather was the primary
cause of the loss or dispute that initiated the lawsuit.
As the trial approaches, the expert prepares exhibits
that display the data and information relevant to the
issues of the case. The exhibits are often directly
derived from diagrams, tables, and images found in the
expert’s report. The availability of digitized data has
led to improvements in the presentation of meteoro-
logical information in the courtroom. Radar and
satellite digital images and computer simulations of
796 FORENSIC METEOROLOGY
weather events can easily be annotated for trial
presentation and stored on a CD-ROM. Using a high-
resolution projector attached to a laptop computer, the
testifying expert can display the images in the court-
room using animation sequences, stop-action frames,
zoom views, and various enhancements. Presentations
of this kind allow the jury or judge to arrive at a better
understanding of the weather conditions associated
with the event that led to the litigation. In fact, many lay
people who serve on juries see animated satellite and
radar images on the weather segments of television
news shows and are somewhat familiar with basic
meteorological concepts related to these observations,
prior to seeing such data in the courtroom.
The expert should assist the attorney in preparing
for the expert’s trial testimony. The attorney needs to
have a thorough understanding of the expert’s opin-
ions, along with the basis for those opinions, so there
are no surprises in the courtroom. The testimony and
opinions of the expert, together with the trial exhibits,
are presented during the direct examination by
the attorney who retained the expert. The expert has
the responsibility of clearly and simply explaining the
weather elements of the case to the judge or jury.
Cross-examination of the expert by the opposing
attorney is standard practice. If some points of the
testimony need to be clarified following the cross-
examination, the attorney who retained the expert has
an opportunity to ask additional questions in redirect
examination. Once the expert’s courtroom testimony
is finished, the expert’s participation in the litigation
usually ends, unless appeals in the matter require a re-
hearing of the expert’s opinions.
Examples of Meteorological
Investigations
A tragic aviation accident, in which weather played a
critical role, was the crash of a commercial aircraft,
American Eagle flight 4184, near Roselawn, Indiana,
on 31 October 1994. The aircraft was in a holding
pattern at 3000 m altitude for about 30 minutes while
waiting for clearance to land at Chicago’s O’Hare
airport. Shortly after it was released by air traffic
controllers from its holding pattern, it descended
toward 2500 m, at which time the pilots could no
longer control the aircraft and it crashed into a field,
killing all 68 people on board.
Airframe icing can significantly degrade the per-
formance of an aircraft. Based on the findings of
detailed meteorological investigations of the Rose-
lawn accident, the cause for the loss of control is
strongly suspected to have been the accumulation of a
ridge of ice behind the leading edge of the wings.
Because the wings of the aircraft operating as Flight
4184 were located at the top of the fuselage, the pilots
were unable to see the formation of ice on top of the
wings. Relatives of the deceased passengers sued the
airline partly over the alleged inaction of the pilots with
respect to the hazardous icing conditions. Attorneys in
the litigation retained aviation weather experts and
much of the meteorological investigation focused on
the nature, timing, and severity of the icing conditions.
The microphysics of supercooled drizzle drops was an
important aspect of the investigation, since the atmos-
pheric conditions were indicative of the presence of
such drops in the area of the holding pattern. These
drops can be particularly hazardous to aircraft because
they can flow over and freeze behind the icing-
prevention devices on the leading edge of the wings
and create ice formations that seriously disrupt air flow
over the wings. Data and images from the National
Weather Service Doppler radar near Chicago (see
example in Figure 1) were carefully studied to deter-
mine when the aircraft most likely encountered super-
cooled drizzle drops during the holding pattern. The
meteorological analysis was partly intended to deter-
mine whether any early visible indications of icing were
present that might have given the crew an opportunity
to respond and take evasive action. A substantial
amount of time and effort was spent on the preparation
of trial exhibits related to the weather testimony, but
the litigation was settled shortly before trial.
Commercial and private aircraft are also subject to
hazardous weather associated with thunderstorms.
The occurrence of microbursts (intense small-scale
downbursts in the decaying stages of a thunderstorm)
are of particular concern on approach or departure
from an airport. Large changes in wind speed and
direction across a microburst can create sudden and
unexpected changes in lift. The ability of the pilots to
recognize and avoid the hazardous conditions associ-
ated with a microburst are usually at the center of the
litigation in such cases. Microbursts were the primary
cause of several major accidents, such as the Delta Air
Lines crash at Dallas, Texas, on 2 August 1985 and the
US Airways crash at Charlotte, North Carolina, on
2 July 1994. Litigation related to these crashes relied
not only on detailed analysis of the meteorological
data but on eyewitness statements from pilots of other
aircraft and air traffic controllers regarding the weather.
Severe turbulence can cause passenger injuries on
commercial flights and such incidents often lead to
lawsuits against the airlines operating the flights.
Typically the plaintiff will allege that the pilots should
have anticipated and avoided the turbulence. Aircraft
operation in and around thunderstorms increases the
likelihood of a severe turbulence encounter. Avoidance
of thunderstorms is made possible through direct
visual observation, on-board weather radar, and

Figure 1 Track of American Eagle Flight 4184 during part of its
holding pattern near Chicago, Illinois, on 31 October 1994, plotted
on a National Weather Service Weather Surveillance Radar
reflectivity image. Time associated with each aircraft position is in
minutes after 21:00 Universal Coordinated Time. The aircraft
occasionally encountered areas of light precipitation, most likely
supercooled drizzle drops, during its hold.
pre-flight and in-flight weather briefings. Other at-
mospheric phenomena, such as breaking mountain
waves and strong vertical or horizontal shear in jet
streams, can also cause severe turbulence, but such
conditions are usually more difficult to anticipate and
avoid than convective turbulence.
The examination of satellite, radar, and upper-air
data is usually a critical component in meteorological
investigations of turbulence incidents. Two passengers
who allegedly suffered turbulence-related injuries
during a flight across the south-western United States
in March 1996 sued the airline. The airline hired an
aviation meteorologist who concluded that there was
no meteorological evidence for the turbulence en-
counter. However, the meteorologist working for the
plaintiff obtained satellite photographs that showed
the existence of strong mountain waves at the time and
location of the incident, consistent with the plaintiffs’
allegation. The case was settled shortly after the report
of the plaintiff’s expert was submitted in the litigation.
FORENSIC METEOROLOGY 797
Sometimes unofficial weather measurements are at
least as valuable as official records. The construction
of a natural gas pipeline in South America was
interrupted by heavy rain and destructive flooding in
April 1998. The construction company was insured
against losses of this kind and filed a large claim for
extensive damage and delay associated with the event.
The insurer was uncertain about paying the claim since
rainfall totals from official weather stations in the
region were not sufficiently large to have caused the
flood damage claimed by the construction company.
The insured party produced an unofficial measure-
ment of 325 mm of rain in one day from a hotel located
near the project. The rainfall reading at the hotel was
many times larger than any of the official readings. In
order to resolve the validity of the claim, the insurer
hired a meteorologist to investigate the weather
conditions, as well as forensic civil engineers who
inspected the physical damage. An examination of
weather satellite data demonstrated that an isolated
severe thunderstorm with exceptionally high cloud
top heights (Figure 2) did occur along the segment of
the pipeline where the most severe damage was found.
A range of probable rainfall amounts was inferred
from satellite data and was generally consistent with
the high rainfall reading at the hotel. Although the
heavy rainfall did not cover as much of the pipeline
project as claimed by the construction company, the
insurer concluded that at least part of the claim was
valid.
In another example, the owner of a Florida resort
hotel claimed that extensive water damage to the
interior of the building was the result of roof leaks
during a very heavy rainfall event. The hotel owner
filed a lawsuit against a construction company that
had repaired the roof shortly before the alleged
damage. Attorneys for the construction company
retained a forensic meteorologist to determine wheth-
er a heavy rainfall event actually occurred at the
property during the period in which the water damage
occurred. Rain gauges in the area did not confirm the
occurrence of a rain event of the magnitude alleged by
representatives of the hotel. The amounts were far
below the quantity of water associated with the
interior damage. Attorneys for the hotel asserted that
since isolated heavy rainfall often occurs in Florida,
the rain gauges simply missed the ‘deluge’ at the hotel.
Examination of data from the National Weather
Service Doppler radar station that monitors the region
demonstrated conclusively that the rainfall amounts at
the hotel were similar to the amounts measured at the
rain gauges. The case reached trial and the meteorol-
ogist testified about the rainfall and radar analysis.
The jury found in favor of the defendant, i.e.,
the construction company, concluding that it was
798 FORENSIC METEOROLOGY
Figure 2 Color-enhanced infrared satellite image from Geostationary Operational Environmental Satellite 8 on 8 April 1998 at 11:45
Universal Coordinated Time near Puerto Suarez, Bolivia. Lower temperatures indicate higher cloud tops. Note the very cold temperatures
for the heavy thunderstorm near the center of the image. Heavy rain from this thunderstorm caused flooding and extensive damage to a
pipeline construction project.
impossible that rain could have caused the water
damage to the interior of the hotel.
The Outlook for Forensic Meteorology
The density and value of physical property is steadily
increasing in regions especially susceptible to weather
damage, such as coastal areas. While significant
improvements are being made in aviation safety
procedures, particularly related to the distribution
and analysis of weather information for pilots and air
traffic controllers, the frequency of air travel is also
expected to grow rapidly. Furthermore, judicial sys-
tems are generally increasing the monetary value
assigned to human life and health. As a result, the
amount of litigation related to weather is likely to
expand, increasing the need for forensic meteorolog-
ical services.
With the expectation of additional enhancements in
the quality and quantity of data from remote sensors
and computer models of the atmosphere and ocean,
the work of the forensic meteorologist will become
more complex and require continual upgrades in
professional skills and knowledge. On the other hand,
advances in observational and analytical techniques
will likely also reduce uncertainties regarding past
weather conditions and, therefore, enhance the value
of meteorological research and testimony with respect
to litigation.
See also
Air–Sea Interaction: Surface Waves. Aviation Weath-
er Hazards. Clear Air Turbulence. Convective
Storms: Convective Initiation. Flooding. Fog. Gust
Fronts. Hurricanes. Lee Waves and Mountain
Waves. Microbursts. Radar: MST and ST Radars and
Wind Profilers; Precipitation Radar. Satellite Remote
Sensing: Surface Wind. Synoptic Meteorology: Weath-
er Maps. Weather Prediction: Severe Weather Fore-
casting.
Further Reading
Bradley MD (1983) The Scientist and Engineer in Court,
Water Resources Monograph Series, 8. Washington DC:
American Geophysical Union.
Bronstein DA (1993) Law for the Expert Witness. Boca
Raton, FL: Lewis Publishers.
Falconer PD and Haggard WH (1990) Forensic meteorol-
ogy. In: Wecht CH (ed.) Forensic Sciences, Ch. 35B. New
York: Matthew Bender.
National Climatic Data Center (1999) Weather records in
litigation. Environmental Information Summaries, vol.
C-1. Asheville, NC: National Oceanic and Atmospheric
Administration, National Climatic Data Center.

FRONTOGENESIS
W Blumen, University of Colorado at Boulder,
Boulder, CO, USA
Copyright 2003 Elsevier Science Ltd. All Rights Reserved.
Introduction
Frontal formation is called frontogenesis and decay or
dissolution is referred to as frontolysis. The intensifi-
cation or weakening of the thermal gradient is usually
considered to be the measure of frontogenesis. The
instantaneous rate of change of the cross-front tem-
perature gradient may be expressed as a diagnostic
relationship that encompasses all processes that con-
tribute to frontogenesis. Contributions are associated
with particular configurations of both horizontal and
FRONTOGENESIS 799
air parcels and their surrounding environment. Close
to the Earth, where p  p0 and where vertical
displacements are relatively small, y may be replaced
by T. Frontogenesis depends on both the kinematic
properties of the wind field, and on the dynamic and
thermodynamic processes associated with the under-
lying physics of atmospheric flow. Consider, for
example, an increase in the thermal gradient in the
y-direction from the advection of temperature by the
confluent wind field in Figure 1A. Air parcels, confined
to a horizontal plane, are alternately compressed
along the y-axis and then stretched along the x-axis.
The rate at which the potential temperature gradient,
qy=qy, increases, following the horizontal motion of
an individual fluid element, DH =Dt, is related to this
confluent wind field by
  1  
vertical wind fields, cloud cover, which can allow a qy
  DH qy=Dt ¼ a
 
selective heating or cooling of the Earth’s surface,
either ahead or behind the front, evaporation and
condensation, among other non-adiabatic (i.e., diaba-
tic) processes and surface heterogeneity, particularly
coastal zones. Each contribution may be evaluated
independently by means of observations and data
from numerical model simulations. The relative
importance of each contribution to frontogenesis
depends on the type of front, surface-based or upper-
air, and its place in the time evolution from inception
to decay. Model simulations, which are based on a full
set of predictive equations, are required for a funda-
mental understanding of frontogenesis as an evolu-
tionary process that involves interactions between
different physical processes.
Frontogenesis
Frontogenesis is usually defined as an increase in the
three-dimensional spatial gradient of the potential
temperature y. The potential temperature y is defined
as
 R=cp
p0
y¼T ½1Š
p
where T is the absolute temperature, p is the pressure,
p0 ¼ 1000 hPa is a constant reference pressure, and
R=cp denotes the ratio of the gas constant to the
specific heat of air at constant pressure. The potential
temperature is a convenient thermodynamic variable
for the study of frontogenesis, because it is conserved
during adiabatic displacements of dry air parcels, and
changes in y may then be associated with diabatic
processes. The latter involve heat exchange between
qy qy ½2Š
where a denotes the rate of confluence. This expression
shows that, if a may be treated as a constant, the
potential temperature gradient increases exponential-
ly with time. It is not, however, sufficient to analyze
frontogenesis in this simple manner, because an
increase in the horizontal temperature gradient
requires an increase in the vertical shear of the
along-front horizontal wind speed. This relationship
between the horizontal thermal gradient and the
vertical wind shear is referred to as thermal wind
balance. The maintenance of thermal wind balance
requires the development of a thermally direct circu-
lation cell, referred to as a secondary transverse
circulation and displayed in Figure 2. The increase of
the thermal gradient from confluence is partially
reduced by adiabatic cooling of air which rises on
the warm side of the front, and adiabatic warming of
air which sinks on the cold side of the front. The
increase in the vertical wind shear, from the net
increase in the thermal gradient, is associated with the
horizontal branches of the circulation cell. The
Coriolis force deflects motion into the section aloft,
and out of the section below, to effect the necessary
increase in the vertical shear.
A shortcoming of the frontogenesis model, ex-
pressed by eqn [2], is the consideration of only a
confluent wind field to effect frontogenesis. Typical
values of a, observed near the Earth’s surface, reveal
that more than 2 days are required to increase the
temperature gradient by an order of magnitude. Yet,
comparable increases have been observed, in some
instances, during a period of 12 h or less. One way to
800 FRONTOGENESIS

− ∆


− ∆

y y


+ ∆

x
x

+ ∆

(A) (B)

+ ∆


− ∆

z z

+ ∆

y y

− ∆

(C) (D)

+ ∆


+ ∆



z
z
− ∆


− ∆

y y

(E) (F)

Figure 1 Frontogenesis mechanisms: (A) horizontal deformation by confluence and by convergence (thick arrows); (B) horizontal
deformation by horizontal shear and rotation; (C) vertical deformation; (D) vertical divergence; (E) tilting; (F) differential diabatic radiative
heating related to cloud cover.

resolve this discrepancy is to allow the convergent or frontogenesis function, defined as

wind field, also shown in Figure 1A, to collaborate
with confluence to enhance frontogenesis. A modified
F ¼ Djryj=Dt ½4Š
version of eqn [2] may be expressed as
  1  
qy
  DH qy=Dt ¼ a þ d
  where F is the rate of change of the three-dimensional
qy qy ½3Š
where d denotes horizontal convergence. The incor-
poration of d into this simple archetype provides a
feedback effect, because the convergent wind field
increases with the rate of frontogenesis. This mecha-
nism, among others, contributes to observed values of
rapid frontogenesis at lower levels of the atmosphere.
Kinematic contributions are not restricted to con-
fluence and convergence. Consider the frontogenetical
potential temperature gradient (r is the gradient
operator) following the motion of individual air
parcels, indicated by the operator D=Dt. Frontogen-
esis, as defined by F, is produced by physical processes
that do not depend on the translation of the front.
Therefore, the relationship between F and various
kinematic and thermodynamic frontogenetical proc-
esses is a convenient means to describe the relative
importance of contributions to frontogenesis, even if
evaluation is not made moving with the front.


+ 3∆


+ 2∆


+ ∆


ogeneity are all factors that contribute to diabatic (V)

− ∆


− 2∆

z
y
Figure 2 Thermally direct circulation cell. The dashed lines are
isotherms of potential temperature, and the horizontal jet flow is into
the section at low levels and out of the section at upper levels.
A mathematical representation of the atmospheric
processes that produce an instantaneous value of F
provides a diagnostic expression, called the fronto-
genesis equation. The essence of this equation may be
represented as
F ¼ horizontal deformation ðIÞ
þ vertical deformation ðIIÞ
þ vertical divergence ðIIIÞ
þ tilting ðIVÞ þ diabatic ðVÞ
The first four contributions relate to kinematic
configurations of the wind field that interact with the
temperature field, and the fifth encompasses all
diabatic processes that contribute to frontogenesis.
The horizontal deformation term (I) is composed of
four contributions, which are illustrated in Figure 1.
The confluent flow in Figure 1A, restricted to the
horizontal plane, concentrates the thermal gradient so
as to increase F; difluence acts in the opposite sense.
The convergent flow in Figure 1A also increases the
thermal gradient in the horizontal plane as does a
confluent flow, but horizontal convergence and diver-
gence also produce motion normal to the plane.
Horizontal shear, as depicted in Figure 1B, both
concentrates the isotherm pattern and rotates the
alignment to increase frontogenesis in the y-direction.
Vertical deformation (II), associated with the vertical
shear of the horizontal wind, is shown in Figure 1C.
The contribution displayed is frontogenetical, increas-
ing both the vertical and horizontal temperature
gradients. Vertical divergence (III) only acts on vertical
gradients of temperature. This term is illustrated by
the convergent wind field, in Figure 1D, which
produces frontogenesis along the z-axis. The tilting
term (IV) is a kinematic effect produced by differential
vertical motion. The motion displayed in Figure 1E
FRONTOGENESIS 801
rotates isotherms in the vertical cross-section, to
increase the temperature gradient along the y-axis.
Finally, cloud cover, precipitation, and surface heter-
frontogenesis or frontolysis. The principal physical
processes included in the diabatic contribution are
associated with differential radiative heating, as illus-
trated by the extensive cloud shield in Figure 1F, and by
latent heating and evaporative cooling. A secondary
factor is differential heating across a frontal zone
produced by frictional dissipation of kinetic energy
close to the Earth’s surface. The effect of this process
on frontogenesis is unclear since measurements are
lacking. Turbulent transfer of heat acts to reduce
temperature gradients, but molecular heat diffusion is
not considered to make a significant contribution to
frontogenesis.
The relative importance of these contributing fac-
tors will be examined for three types of frontogenesis.
They are surface frontogenesis, coastal frontogenesis,
and upper-level frontogenesis. Each may be delineated
with synoptic-scale observations that may be supple-
mented with special observational platforms, such as
instrumented aircraft, to explore relatively small-scale
details that cannot be measured by conventional
means.
Surface Frontogenesis
Surface frontogenesis refers to frontogenesis that
occurs in the horizontal cross-front direction over
relatively level terrain, away from coastal zones. Close
to the ground, where the vertical motion is relatively
small, temperature changes in the cross-front direc-
tion, jrH yj, are primarily associated with horizontal
deformation (I) and with horizontal gradients of
diabatic heating or cooling (V). The observed field of
horizontal deformation overlaying surface potential
temperature isotherms, shown in Figure 3, produced
dry nighttime frontogenesis in the Great Plains over a
5 h period. The idealized patterns of Figures 1A and B
all appear in observed low-level deformation flows
that produce surface frontogenesis, but are not easy to
separate into component parts in real flows.
A few examples will illustrate situations in which
differential diabatic heating or cooling contribute to F.
Figure 1F depicts a daytime situation which enhances
frontogenesis. Relatively little temperature change
will occur in cloud-capped cold air behind the front;
low-level heating is significant on the warm side of
the front, jrH yj increases. Alternatively, frontolysis
would be evident at night under the same conditions
shown in Figure 1F, since radiational cooling of the
low-level air on the warm side would take place, with
802 FRONTOGENESIS
277
279
281
283
285
Figure 3 Streamlines of horizontal flow, with isotherms of
potential temperature (K) denoted by dashed lines, during a period
of frontogenesis observed in the Great Plains. (Adapted with
permission from Ostdiek V and Blumen W (1997). Journal
of the Atmospheric Sciences 54: 1490–1502. Boston: American
Meteorological Society.)
little temperature change in the cold air expected to
occur. The formation of clouds also contributes to
frontogenesis by means of differential heating associ-
ated with the release of latent heat. Cumulus convec-
tion, for example, in advance of a cold front, increases
the cross-front temperature gradient by condensatio-
nal heating in the warm air. Evaporative cooling will
also produce frontogenesis when precipitation falls
into an unsaturated layer of cold air behind the front.
Consider the vertical deformation (II) term, as
displayed in Figure 1C, to be associated with positive
wind shear as a consequence of boundary layer friction
or drag. This effect on frontogenesis would be difficult
to measure, since it acts at low levels in consort with
horizontal deformation (I). Close to the ground,
however, contributions to frontogenesis by vertical
divergence (III) and tilting (IV) of the front in a
vertical cross-section, which arise from gradients of
vertical velocity, are expected to be relatively small.
Yet, these terms become more significant aloft, where
they contribute to both the frontal slope and to the
vertical component of the temperature gradient. It is
not only kinematics of the vertical motion field,
however, that needs to be addressed. Convergence at
low levels, for example, contributes to frontogenesis
by horizontal temperature advection. At the same
time, rising motion on the warm side of the front
would cool by dry adiabatic ascent, and dry adiabatic
descent on the cold side would produce warming. The
temperature change, brought about by this thermo-
dynamic process, is frontolytical. This example is one
of many which illustrates that the purely kinematic
approach to frontogenesis is limited. It is necessary to
encompass both thermodynamic and dynamic pro-
cesses, described by the complete set of predictive
equations, to understand completely frontogenesis
and frontolysis.
Coastal Frontogenesis
Coastal frontogenesis is surface frontogenesis that
owes its existence to the particular conditions associ-
ated with a coastal zone. Coastal frontogenesis can be
particularly prominent, for example, along the east
coast of the United States in wintertime. Favorable
conditions are set up by the low-level thermal gradient
between the relatively warm air over the ocean and the
colder continental air inland. The shallow mass of cold
continental air is blocked to the west by the Appala-
chian Mountains, which are oriented approximately
parallel to the coastline. This topographic constraint is
called cold air damming.
Differential diabatic heating (V) establishes a pre-
ferred location for the onset of coastal frontogenesis.
The conditions are set up for frontogenesis to proceed
when air flows southward from a cold anticyclone,
which is situated north of the front in the vicinity of the
south-eastern coast of Canada. This flow is modified
by convective heat flux from the warm ocean and
crosses the United States coastline as an easterly flow.
Convergence occurs along the coastal zone as the low-
level airstream encounters rougher inland terrain, and
experiences an increased surface drag. Cold air also
streams southward, parallel to topographic contours,
in the corridor between the coast and the Appalachian
chain. As a consequence, the low-level flow turns from
northerly to easterly across the frontal zone, providing
a horizontal deformation (I) flow field that reinforces
the preexisting thermal gradient. Air convergence in
the frontal zone also provides a vertical uplifting of
warm, moist air that produces clouds and precipita-
tion on both sides of the front. The sign and magnitude
of this secondary diabatic (V) contribution to fronto-
genesis may vary from case to case. Further, frontolysis
may occur in less than 24 h when the horizontal
deformation field, which reinforces the thermal gra-
dient, is altered by an evolving large-scale synoptic
flow pattern.
Upper-Level Frontogenesis
Upper-level frontogenesis refers to frontogenesis that
is initiated aloft in the vicinity of midlatitude jet
streams, and may not appear to be directly coupled to
frontogenesis accompanying surface-based activity.
Figure 4 displays a synoptic analysis that illustrates a
 FRONTOGENESIS 803

100 km

390
km hPa
40 0 390
40 20
380
370
11.8 200
360
350
340
330 0
9.2 320300
J 310
80 300

60

5.6 500
290

20
3.0 700

280
1.0 900
270
(A) 270

100 km

km 60 360 370 hPa

40 350 60 40 20
370
11.8 360 200
340 350
340
330
320
9.2 300
J 310
100
330
20 300
5.6 80
0
500
320
290

3.0 700
310
300 280
1.0 900
(B) 290 280 0

Figure 4 Cross-section of an upper-level frontal system during a period of frontogenesis: (A) early stage; (B) mature stage reached 24 h
later. The tropopause is denoted by thick solid lines. The thin dashed lines are isotherms of potential temperature (K), and the thin solid
lines are isotachs (m s 1). J denotes the axis of the jet stream directed out of the section. Pressure levels and standard heights are shown
on the abscissa. (Adapted with permission from Reed RJ (1955) Journal of Meteorology 12: 226–237. Boston: American Meteorological
Society.)

case of upper-level frontogenesis, which evolved into a
mature front exhibiting a significant thermal gradient
that extended down to the 800 mbar level, or about
2 km above the Earth’s surface. As in Figure 2, a
thermally direct circulation cell is required in order to
maintain thermal wind balance at upper levels. The
schematic representation, presented in Figure 5, dis-
plays a prominent clockwise circulation cell, which is
confined to the warm side of the frontal zone. The
circulation on the cold side may not be well defined
and, accordingly, is omitted from the figure.
This cross-front circulation represents a response to
horizontal deformation (I), which provides an en-
hancement of the thermal gradient upstream, and a
diminution of this gradient downstream, along the
axis of the jet stream. The relative position of the
circulation cell and its intensity are also determined by
cold air advection along the jet axis. The secondary
circulation that develops in response to changes in the
thermal gradient serves a dual purpose. First, thermal
wind balance is maintained in the presence of fronto-
genesis. Second, this secondary circulation converts
potential energy into kinetic energy in order to
maintain the upper-level jet stream frontal system.
The tropopause, which is low over the cold air and
high over the warm air, represents the relatively sharp
transition to the lower stratosphere, where large static
stability resists vertical displacements. Accordingly,
the secondary circulation tends to steepen the tropo-
pause and draw a relatively thin slice of stratospheric
804 FRONTOGENESIS
z for this process to occur are above and below the jet
J
Direct
cell
Figure 5 Schematic illustration of the transverse secondary
circulation associated with an upper-level front, as shown in Figure
4. (Adapted with permission from Danielson ER (1968) Journal
of the Atmospheric Sciences 25: 502–518. Boston: American
Meteorological Society.)
air downward to lower levels and create a tropopause
fold. The upper-level frontal zone is delineated by the
concentrated horizontal and vertical potential tem-
perature gradients appearing below the 300 mbar level
in Figure 4. The development and alignment of this
characteristic thermal field is intimately associated
with the development and evolving structure of the
circulation cell. In particular, the subsiding branch of
the circulation provides adiabatic compressional heat-
ing below the jet axis, which is responsible for the
cross-front thermal gradient, and its descent to low
levels of the troposphere. The elapsed period between
the relatively early stage of frontogenesis in Figure 4A
to the mature stage in Figure 4B is 24 h.
The vertical deformation field (II) provided by the
circulation cell is frontogenetical at upper levels, but
becomes frontolytical below where the front weakens
and terminates. The contribution to vertical diver-
gence (III) cannot, however, be easily assessed. Rela-
tively large gradients of vertical motion could be
expected to exist below the tropopause, where vertical
motion is suppressed. The sign and magnitude of these
gradients is only available from model simulations and
from measurements obtained by instrumented air-
craft. These sources do not provide sufficient infor-
mation for a definitive evaluation of the overall effect
of vertical divergence on frontogenesis. The large
magnitude of subsidence on the warm air side of the
front is, however, consistent with the frontal slope
produced by tilting (IV) of horizontally aligned
isotherms at upper levels into the more vertical
orientation that characterizes the mature front in
Figure 4B.
The diabatic (V) contribution to frontogenesis has
generally been associated with turbulent dissipation at
upper levels. This process is related to the breakdown
or instability of relatively laminar horizontal flow into
small-scale turbulent eddies that ultimately dissipate
into heat at molecular scale. Flights carried out by
instrumented aircraft reveal that the favored locations
stream and in the frontal zone, where vertical shears of
the horizontal wind are very large. These regions are
more susceptible to large vertical displacements and
overturning eddies, even when the buoyancy force
tends to oppose this type of motion. Small-scale
mixing is usually viewed as a frontolytical process,
acting to smooth out sharp gradients. Yet observations
do reveal the formation of relatively sharp tempera-
ture gradients in the atmosphere that appear to
develop from mixing in thin layers separated by
more laminar flow. However, the relative importance
of this type of differential mixing, and the contribution
made to frontogenesis by dissipation of these turbulent
eddies into heat, has not been adequately determined.
Frontogenesis in the Wind Field
Surface heterogeneity can fix the position of a front,
and differential diabatic heating can provide a contri-
bution to frontogenesis, but ultimately frontogenesis is
a process driven by a favorable wind distribution. The
kinematic approach to frontogenesis does not, how-
ever, reveal the complex nature of nonlinear interac-
tions that sharpen gradients. It is necessary to solve the
basic equations of dynamic meteorology to develop an
understanding of how these interactions operate.
Frontogenesis in the wind field may be simply
illustrated by a prototype, which exhibits a physical
process that has been identified in more complex
atmospheric models. The evolution of the velocity
normal to the front, u, can be described by
qu qu
þu ¼0 ½5Š
qt qx
where t is time and x is the direction normal to the
front. The initial distribution of u, shown in Figure 6,
displays a convergent wind field. According to eqn [5],
each value of u travels at velocity u. At later times, a
given value of u is displaced a distance ut from its
previous position. This nonlinear self-advection will
continue until qu=qx ¼ 1. A dissipative process in
nature will, however, prevent the occurrence of a
discontinuity in the velocity field. A finite frontal zone
could be expected to occur, exhibiting a balance
between advection and dissipation.
The nonlinear process described by eqn [5] is only
one of many that contribute to frontogenesis in
atmospheric flows. The process of nonlinear

ut
u
ut
x Hoskins BJ (1982) The mathematical theory of frontogen-
Figure 6 Horizontal velocity u as a function of the cross-front
coordinate x . The distribution of u is shown at the initial time (solid)
and at a later time (dashed).
self-advection is a particularly important feature of
small-scale frontogenesis, when the Earth’s rotation
may be neglected, along-front variability is not signif-
icant, and vertical velocity is not a factor close to the
Earth’s surface.
FRONTS
W Blumen, University of Colorado at Boulder, Boulder,
CO, USA
Copyright 2003 Elsevier Science Ltd. All Rights Reserved.
Introduction
Atmospheric fronts are three-dimensional zones that
represent a rapid transition of the thermal field,
accompanied by a marked shift in the wind direction.
In some cases, a significant transition in the moisture
field may also occur across the frontal zone, and an
increase in the magnitude of the wind may occur in
collaboration with the wind shift. The thermal gradi-
ent is usually largest at the Earth’s surface, and
weakens as the zone slopes upward with relatively
warm, less-dense air overlying relatively colder air of
higher density. The transition zone may extend over a
distance of 1000 km or more along the Earth’s surface,
and a typical cross-front to along-front ratio is 1:10.
Fronts are observed at all longitudes in the extratrop-
ical latitude belt and are associated with most of the
significant weather events that occur during the late
autumn through the early spring period. Although
many fronts are surface-based, thermal gradients may
be enhanced at levels up to the tropopause producing
FRONTS 805
See also
Boundary Layers: Neutrally Stratified Boundary Layer.
Cyclogenesis. Cyclones, Extra Tropical. Fronts.
Synoptic Meteorology: Weather Maps. Turbulence
and Mixing.
Further Reading
Bluestein HB (1993) Synoptic-Dynamic Meteorology
in Midlatitudes, Vol. 2, Observations and Theory of
Weather Systems. New York: Oxford University Press.
Carlson TN (1991) Mid-Latitude Weather Systems.
London: Harper Collins Academic.
esis. Annual Reviews of Fluid Mechanics 14: 131–151.
Keyser D (1986) Atmospheric fronts: an observational
perspective. In: Ray P (ed.) Mesoscale Meteorology and
Forecasting, pp. 216–258. Boston: American Meteoro-
logical Society.
Keyser D and Shapiro MA (1986) A review of the structure
and dynamics of upper-level frontal zones. Monthly
Weather Review 114: 452–499.
Orlanski I, Ross B, Polinsky L, and Shaginaw R (1985)
Advances in the theory of atmospheric fronts. Advances
in Geophysics 223–251.
fronts in the troposphere, which appear to be uncon-
nected to surface-based frontal features.
The existence of a front or, equivalently, a frontal
zone, is an intrinsic property of the physical laws that
govern atmospheric motions. Fronts are produced in
numerical model simulations of the atmosphere in
which radiation, cloud cover, and surface heterogene-
ity are not a consideration. These latter processes
provide, however, some of the interesting variants that
characterize observed atmospheric fronts.
Surface Fronts
An atmospheric surface front, in the present context, is
a well-defined phenomenon that may be distinguished
by the network of global meteorological observations.
The frontal intensity, measured by changes in the
temperature, humidity, and wind fields that occur
across the frontal transition zone, is most pronounced
at or near the Earth’s surface, and diminishes with
altitude. Fronts that fall into this category are synop-
tic-scale events, which may retain their individual
identities for up to a week or more. The physical
processes that give rise to identifiable frontal features
are controlled by four principal forces: the buoyancy
806 FRONTS

force, the pressure gradient force, the Coriolis force,
and the frictional force, or drag, whose influence is
primarily restricted to the lowest 1–1.5 km of the
atmosphere, the planetary boundary layer. Gravity
acts downward, toward the Earth’s surface, and is
opposed by an upward directed or vertical pressure
gradient force. Hydrostatic balance occurs when the
two forces are equal in magnitude. In most cases there
is a small imbalance that provides a buoyancy force.
The direction of the buoyancy force controls the
direction of vertical motions associated with fronts.
The Coriolis force is directly proportional to the
Earth’s rotation rate and to the magnitude of the
horizontal wind vector. The Coriolis force, directed to
the right of the horizontal wind in the Northern
Hemisphere, is opposed by the horizontal component
of the pressure gradient force, which is directed
toward low pressure. These forces tend to be in
approximate balance above the planetary boundary
layer (the so-called geostrophic balance), providing a
counterclockwise circulation around a low-pressure
center. The drag force reduces the wind speed within
the planetary boundary layer, and lessens the magni-
tude of the Coriolis force, which leads to cross-isobaric
flow toward low pressure.
A midlatitude wave cyclone, a low-pressure system,
is depicted in Figure 1. A cold front, a warm front and
an occluded front emanate from the center of the low
at various stages of the 36 h period depicted. The
designations, cold front, warm front, and occluded
front, emerged from the Norwegian cyclone model,
introduced after World War I, and have their basis in
military terminology. In this analogy polar air attacks
southward, the cold front; warmer air counterattacks
northward, the warm front. The occluded front
develops when the cold front makes a sharp turn and
attacks the warm air in its flank and, ultimately, the
two fronts merge with warm air rising above the
colder, surface-based air. Frontal modifications that
arise from passage over and around relatively large
topographic features, or from the passage across a
large body of water, modify this simple picture.
Cold Front
A cold front represents the advance of relatively cold,
dense air into a region previously occupied by relati-
vely warm less dense air. The cold front in Figure 1
is delineated by the triangular teeth that point in
the direction of the advancing cold air. The finite

Map view 992

1000
99
0 99

6
100 6
Cool air
99 L
6 2
99 B B
99 Cold air
4
Occluded front

L L
Cool air
A Cool air
Cold air
Cold air
A
A
B
Warm air Warm air

Warm air
Cold front Warm front

Cross-sectional view

Warm air
Warm air
Warm air
Cold air Cold air Cold air Cool air
Cool air Cool air
A B A B A B
Cold front Warm front Occluded front Occluded front

(A) (B) (C)

Figure 1 Stages in the development of an occluded front, and its relationship to a wave cyclone in middle latitudes. (A) Mature wave
cyclone. (B) Partially occluded wave cyclone. (C) Occluded wave cyclone. The thin solid lines in the upper panels are isobars (hPa), and L
designates the low-pressure center. The cold, warm, and occluded fronts are described in the text. (Reproduced with permission from
Lutgens FK and Tarbuck EJ (1992) The Atmosphere, 5th edn. Englewood Cliffs, NJ: Prentice-Hall.)
 FRONTS 807

transitional zone, generally 100 km or less, is reduced
to a line on synoptic weather maps for two reasons: the
surface observational network is too widely
spaced to delineate the transition zone between
disparate air masses, and the map scale essentially
reduces the transition zone to the thickness of a broad
line.
Some cold fronts may be relatively shallow, as is
indicated in Figure 1, where the extent of the cold air is
limited to 1–2 km above the ground. In other cases, as
in Figure 2, the cold front retains its identity up
through most of the troposphere. The frontal slope is
designated as dz=dy, where z denotes height above the
ground and y is normal to the front, increasing toward
the cold air. Typical slopes of cold fronts are usually in
the range dz=dy  1 : 50–1 : 150, and the frontal
slope in Figure 2 falls within this range. Close to the
ground, however, frontal translation is restrained by
the drag force, which has the effect of steepening the
frontal slope and deepening the wedge of cold air
behind the front. The cold air aloft may, in some cases,
overrun the warm air ahead of the front, but the nearly
hydrostatic balance that usually prevails for midlati-
tude frontal situations requires high-density cold air to
reside under the sloping transition zone, as displayed
in both Figures 1 and 2.
The increase in wind through and above the frontal
zone may be represented by the thermal wind relation,
which incorporates both hydrostatic and geostrophic
balance. The appropriate expression is provided by
qu 1 qT
¼ ½1Š
qz  qy
fT
where u is the along-front geostrophic wind, normal
to the direction of the temperature gradient, T the
absolute temperature (T  is an average value, inde-
pendent of y) and f ¼ 2O sin f, where O is the angular
velocity of the Earth’s rotation and f denotes latitude.
The marked vertical change, or shear, of the geo-
strophic velocity is a direct consequence, according to
eqn [1], of the relatively large horizontal temperature
gradient through the frontal transition zone.
Typical features associated with the passage of
a cold front past an observer at a fixed location may
be described by reference to the first two panels of
Figure 1. The temperature decreases relatively abrupt-
ly as the frontal zones passes, and the wind veers in the
clockwise direction. In addition, the surface wind
speed may increase in response to an increase in the
pressure gradient force, and it may also become gusty.
The temperature contrast across the frontal zone is
maintained by the horizontal advection of tempera-
ture toward the front, which is consistent with the
observed wind shift. Figure 1 shows that the front lies
in relative minimum or trough in the surface pressure
field, which produces a pressure drop as the front
approaches. Then the pressure increases as cold, dense

−64°C
−56°C 40
100
30
20
10

−64°C
−56°C
−48°C
−48°C 250
10
−40°C
Pressure (hPa)

J −32°C
80 70
Height (km)

0 60 −24°C
50 −16°C

−40°C 40 −8°C
500
5 30
−32°C
0°C
−24°C
20
−16°C 8°C
10
−8°C 20 16°C

0
OM CB NA AT CH
1
Figure 2 Distribution of isotherms (1C), denoted by thin solid lines, and isotachs (m s ), denoted by thin dashed lines, in a vertical
cross-section through a cold front. The jet stream axis, denoted by J, is directed into the section. Heavy solid lines indicate the slope of the
surface-based cold front, and the upper-level tropopause. The distance from Omaha (OM) to Charleston (CH) is approximately 1700 km.
(Reproduced with permission from Wallace JM and Hobbs PV (1977) Atmospheric Science, An Introductory Survey. New York: Academic
Press.)
808 FRONTS
air moves through the observation site. There is often a
decrease in the moisture content of the air, measured
by the dew point temperature, through the frontal
zone. This feature is particularly noticeable if the
warm air sector, ahead of the front, is relatively close to
a large body of water, such as the Gulf of Mexico. The
extent of cloudiness and precipitation is determined by
the extent of moisture in the air, and by the extent of
vertical uplift at and behind the front. Yet the front is
not a material surface, and some of the air that
converges at the front in low levels may pass into the
frontal zone and recirculate without significant uplift.
Some may also rise, reverse direction, and then move
away from the front, producing little or no cloudiness.
A characteristic feature, indicated in Figure 1, is for the
air to rise, cool, and then condense, forming cumulus-
type cloudiness. The characteristic precipitation
patterns are varied, but intermittent shower
activity originating from large cumulus cells and
banded structures aligned parallel to the front are
not uncommon. Severe activity associated with cold
frontal passages includes hail, lightning, and tornado
inceptions.
A backdoor cold front designates a cold front that
moves opposite to the pattern depicted in Figure 1: its
movement is westward in the Northern Hemisphere.
In the USA, for example, a New England backdoor
cold front is associated with the westward expansion
of a cold surface high-pressure system situated near the
North Atlantic coast during winter. The thermal
gradient is reversed in this case, and the frontal zone
is characterized by fog and low stratiform clouds
fueled by moisture from the ocean. A similar event
may also occur in spring and summer, when relatively
cool maritime air from the Atlantic Ocean moves over
the eastern seaboard. This cold air incursion, also
called a back-door cold front, moves southward along
the eastern slope of the Appalachian Mountains,
which serve as a barrier to inland penetration.
The southerly buster or southerly burster, defined as
a squally wind change, is an intense summertime cold
front that arrives at the south-eastern tip of Australia
from the Southern Ocean. Arrival of this front in the
afternoon can be accompanied by temperature chang-
es of 10–151C over a period of a few minutes, but
precipitation is not usually associated with a southerly
burster. The front travels equator-ward, acquiring a
characteristic S-shape as its movement is inhibited by
the east coast mountain chain, but movement inland
and along coastal waters is less restrained.
Warm Front
A warm front represents the advance of relatively
warm, low-density air into a region previously occu-
pied by relatively cold, high-density air. The leading
edge of the warm front in Figure 1 is delineated by the
semicircles that point in the direction of movement of
the warm air. The receding cold air is not, however,
pushed northward by the advancing warm air. The
north-easterly retreat of the cold air is associated with
the dynamics that governs the coupled system of the
surface-based wave cyclone and westerly flow aloft.
The transition zones of warm fronts are not nearly as
sharp and distinct as cold frontal zones, and may
extend over a few hundred kilometers. There are at
least two reasons for this lack of sharpness. First, the
receding cold air has progressed much further from its
source region than the relatively fresh surge of cold air
behind the cold front. As a consequence, exchange of
heat and moisture between the surface layer and the air
above has produced a more significant modification of
the receding cold air, and thereby reduced the contrast
between the adjacent air masses. Second, other air near
the surface, embedded in the planetary boundary
layer, is subject to a drag force. Air motion above the
boundary layer is essentially frictionless flow. The
result is to reduce the frontal slope, typically
dz=dy  1 : 300, and the depth of the receding cold
air. As a consequence, the relatively shallow depth of
the receding air mass is also more prone to modifica-
tion by vertical exchange with the underlying surface.
Figure 1A and B also indicate a sequence of change
as a warm front passes an observer at a fixed location.
The temperature increases through the transition
zone, but the magnitude of change is less than the
magnitude generally accompanying cold frontal pas-
sages. The wind direction also veers in the clockwise
direction, but the wind speeds are relatively light and
the wind shift may be obscured by local topographical
influences on the wind. The circulation must, however,
maintain a net horizontal advection of temperature
toward the front in order to maintain frontal con-
trasts; otherwise the front would begin to lose its
identity. The warm front lies in a trough of low
pressure, but the sign and magnitude of the pressure
change associated with the frontal passage is depend-
ent, to a large extent, on the direction of frontal
progression. A change in the dew point temperature
may also be difficult to observe, particularly if an
extensive band of precipitation falls into the cold air
mass. In the absence of precipitation, the dew point
temperature would normally increase after frontal
passage. The warm, less-dense air, which is not
entrained into a frontal zone circulation, tends to
rise above the surface-based cold air and move with
the upper-level flow more rapidly than the translation
speed of the front. The type of cloudiness that may be
observed in advance of the front tends to be of the
stratiform variety. An extensive overcast may precede
 FRONTS 809

the warm front, with low stratus and fog character- Occluded Front
izing conditions within and near the frontal zone. The
The development of an occlusion or occluded front is a
type and extent of precipitation that may fall from
process that increasingly cuts off the low-level supply
stratiform clouds into the receding cold air depends on
of relatively warm air by vertical uplift, as depicted in
the moisture content, the vertical profile of air
Figure 1. The occluded front is denoted on the surface
temperature, and the northward extent of the cloud
map by alternating triangles and semicircles that point
cover. A particularly treacherous situation occurs
in the direction of movement. According to the
when rain falling into cold air freezes and produces
Norwegian model, depicted in Figure 1, a cold
ice forms that coat structures, such as power lines and
occlusion occurs when the advancing cold air mass,
motorways.
behind the cold front, overtakes the less-cold receding
Northward advection of warm air, as depicted in
air mass ahead of the warm front. The receding cool
Figure 1, is the most common circulation pattern
air is less dense than the advancing cold air and is
associated with warm fronts, but not the only one.
forced to rise, as shown in Figure 1, to maintain a
Westward advection of relatively warm air from the
hydrostatically stable environment. At low levels, cold
Atlantic Ocean, for example, followed by a recircula-
air continues to advance as a cold front and replaces
tion can produce a warm front that progresses in a
cool receding air. As a consequence, the wind temper-
southerly direction, which is referred to as a backdoor
ature and pressure changes across the frontal zone are
warm front. Another type of warm front can be
as with a cold frontal passage, although less pro-
produced by adiabatic warming of air that descends
nounced. The warm front lifted aloft, designated as the
along the eastward-facing lee slopes of the Rocky
occluded front in Figure 1, advances with the advance
Mountains. The leading edge of this descending air,
of the surface-based cold air mass. The precipitation
referred to as a Chinook front, will either remain along
pattern is related to stratiform clouds, as in the case of
the Front Range of the Rocky Mountains or progress
a warm frontal passage, but tends to be displaced over
into the Great Plains as a warm front, if the cold air
the cold surface air. This type of synoptic evolution
ahead of it is displaced eastward.
would tend to be most evident when cold continental
air behind the cold front overtakes receding cold air
Stationary Front that either has been warmed by a longer residence time
over the continent or had originally acquired charac-
A stationary front represents a transition zone be-
teristics of a relatively warmer maritime environment.
tween cold and warm air masses that does not exhibit
The counterpart of the cold occlusion in the
significant motion toward either air mass. It may also
Norwegian model is the warm occlusion, depicted in
be referred to as a quasi-stationary front. This type of
Figure 3. The thermal characteristics are now re-
front may simply reflect a change in the synoptic scale
versed: the advancing cold air is warmer and less dense
circulation pattern that halts the translation of either a
than the receding cold air and, upon merger, the cold
cold or a warm front, or the heterogeneity of the
front is lifted aloft. This process constitutes a warm
Earth’s surface may provide conditions that fix a
front occlusion, and the cold front aloft is designated
frontal transition zone to a preferred location. Coastal
as the occluded front. The surface characteristics tend
zones, zones separating relatively warm oceanic air
to reflect a warm frontal passage, but the cold front
from colder continental air in winter, provide an
environment that promotes the formation of a quasi-
stationary front, referred to as a coastal front. Favored
locations are along the east coast of the United States,
from Maine to the Carolinas, and the Texas coast.
Although these fronts exhibit transition zones that are
comparable to those of a cold front, they are of limited
extent, ranging from 200–600 km, and of limited
duration, lasting up to a day or less. The frontal zone is
maintained by convergence of relatively warm moist
air from the ocean toward the relatively dry continen-
tal air, with a clockwise shift in the wind across the Cool air
Cold air
front. This situation often results in a band of
precipitation, parallel to the front, with the maximum
Occluded front
precipitation occurring on the cold side, and possibly a
transition from frozen precipitation to rain on the
warm side. Figure 3 Warm front type of occlusion.
810 FRONTS
aloft may, however, produce more upper-level con-
vective activity than is usually associated with a typical
warm front. This type of occlusion could be expected
when cold air of maritime origin enters northerly
portions of western coastlines as a cold front, and then
overtakes colder continental air further inland. The
Pacific Northwest, and the west-coastal regions of
northern Europe, represent favored locations for this
to occur.
It has, however, become apparent in recent years
that the Norwegian model of an occluded front does
not represent the only three-dimensional atmospheric
structure when cold fronts and warm fronts appear to
merge. Other synoptic processes can provide surface
signatures that resemble the occlusion depicted in
Figure 1. A cold front in a marine cyclone, for
example, may separate or fracture from its union
with a warm front in the low center, and then travel
along the surface position of the warm front. This
process duplicates the increasing separation of the low
center from the triple point, where the cold, warm, and
occluded fronts meet. Another process that may
appear to be an occlusion occurs when an upper-level
front, not originally part of the surface wave cyclone
system, descends into the region where the cold and
warm fronts are starting to merge. Passage of the
upper-level front above the receding cold air mass
provides a synoptic structure similar to a warm front
occlusion. The upper-air observing network is not
sufficiently dense to provide an unambiguous distinc-
tion between the various scenarios presented. Refine-
ment of the Norwegian model of an occlusion by the
documentation of other physically consistent physical
processes that have the appearance of an occlusion,
their preferred locations, and their frequency of
occurrence are topics of current research interest.
Upper-Level Front
An upper-level front, or an upper tropospheric front, is
a transition zone exhibiting a sharp thermal contrast
that may extend from the tropopause down to 2 or
3 km above the ground. Upper-level fronts are not
usually associated with the type of weather that
characterizes surface fronts. Clear-air turbulence and
tropospheric-stratospheric exchange are often defin-
ing characteristics that distinguish upper-level fronts.
Temperature changes within an upper-level frontal
system are associated primarily with dry adiabatic
compression; condensation and radiative heating are
considered to be secondary factors in the maintenance
of the mature front displayed in Figure 4. The potential
temperature y is introduced to characterize the dry
adiabatic processes associated with this frontal sys-
tem. It is defined as
 R=cp
p0
y¼T ½2Š
p
where T is the absolute temperature, p the pressure, p0
a constant reference pressure, and R=cp denotes the
ratio of the gas constant to the specific heat of air at
constant pressure. Potential temperature is conserved
under dry adiabatic displacements of air parcels:
temperature increases with descent to higher pressure
and decreases during ascent to lower pressure. Poten-
tial temperature increases with decreasing pressure
(increase in altitude) to maintain a hydrostatically
stable atmospheric environment. As a consequence,
stratospheric air can be identified with relatively large
values of y.
The mature upper-level frontal structure displayed
in Figure 4, shows a frontal zone depicted by the
concentration of potential temperature isotherms. It is
approximately 100–200 km wide, and exhibits a slope
dz=dy  1 : 100. This structure is not described as
either a cold or a warm front, but exhibits qualities of a
quasi-stationary front that moves parallel to itself.
Superposed on this front and jetstream system is a
transverse or cross-front circulation, which is dis-
played in Figure 5. This circulation cannot be ob-
served, because the magnitude of the velocities is
smaller than can be resolved by current measurement
techniques. Its existence is based on theoretical work,
and the position of the circulation cell and its
properties are extracted from numerical model simu-
lations. The circulation on the cold side of the front is
not well defined by these simulations and, accordingly,
is omitted from the figure. The circulation displayed in
Figure 5 does provide a consistent explanation of how
the upper-level front is maintained, and why strato-
spheric constituents, such as ozone and high-level
radioactivity from nuclear explosions, can be ob-
served in the lower troposphere.
The subsiding branch of the circulation cell main-
tains the most prominent characteristics of the upper-
level front that are revealed in Figure 4. In particular,
the descent of stratospheric air within a narrow
pocket, called a tropopause fold, provides observa-
tional evidence of strong subsidence that extends
downward to low tropospheric levels. This downward
motion is also responsible for the thermal contrast
across the front and for the slope that characterizes the
fold and the potential temperature isotherms that are
contained within it. The thermal contrast across the
front is maintained by adiabatic compressional heat-
ing below the jet axis in the subsiding branch of the
cell, with relatively less vertical motion and corre-
sponding temperature change on the cold side of the
 FRONTS 811

100 km

km 60 360 370 hPa

60 40
40 350 20 370
11.8 200
360
340 350
340
330
320
9.2 300
J
310
100

330
20 300
5.6 80
0 500
320
290

3.0 700
310
300 280
1.0 900

290 280 0

Figure 4 Cross-section of an upper-level front. The tropopause is denoted by thick solid lines. The thin dashed lines are isotherms of y
(K), and the thin solid lines are isotachs (m s 1). J denotes the axis of the jetstream, directed out of the section. Pressure levels (hPa) and
standard heights (km) are shown on the abscissa. (Adapted with permission from Reed RJ (1955) A study of a characteristic type of upper-
level frontogenesis. Journal of Meteorology 12: 226–237.)

front. This feature of the circulation has the effect of
concentrating the thermal gradient within the fold.
The sloping alignment of the isotherms arises because
the differential vertical motion field tilts the potential
temperature isotherms from a horizontal alignment, at
upper levels, into the one that displayed in Figure 4.
Thermal wind balance, expressed by eqn [1],
provides the explanation for the high-speed jet flow
above the frontal zone. In this example, the sharp
thermal gradient across the front produces a relatively
strong jet that flows southward. Two characteristic
features of an upper-level front exhibited in Figure 4
are cyclonic shear, a wind shift that promotes a
counterclockwise turning of the wind across the front,
and the intersection of y and p-surfaces, which charac-
terizes a baroclinic atmosphere. These two features are
always associated with the development and enhance-
ment of a prominent upper-level jet and frontal system,
and with surface-based fronts that extend to the upper
troposphere (see Figure 2).
significant factor, but the nonhydrostatic acceleration
may be important in the dynamics of these fronts.
Sea Breeze Fronts
See breeze fronts are most prominent in the warm part
of the year, when daytime heating creates higher
temperatures over the land than over the water. A
pressure gradient, which develops in response to this
differential heating, drives an onshore flow at low
levels, with a return flow at about 1–2 km above the
surface. The sea breeze front is characterized by both a
sharp temperature drop of a few degrees centigrade or
more, and a marked increase in the humidity, that can
occur over a horizontal distance of a kilometer or less.
Inland penetration of the sea breeze front, 10 km or
more, may be opposed by an offshore wind ahead of
the front, and by turbulent convective mixing over
land, which tends to smooth the temperature and
humidity differences across the frontal zone.

Small-Scale Fronts
Small-scale fronts occupy a relatively limited horizon-
tal domain, have relatively short lifetimes and are
surface-based phenomena. Because they exist only for
a few minutes to a few hours, the Coriolis force is not a
Density Fronts
Density fronts represent the leading edge of density or
gravity currents, driven by the upstream release of
relatively cold high-density air. A pressure gradient
that is directed away from the source drives the
motion. The situation is similar to the instantaneous
812 FRONTS

Oz
on Ra
e dio
ac
tiv
ity
Activ
e m
ixing
and
exch
a nge
layer

Figure 5 Schematic illustration of the transverse secondary circulation associated with an upper-level front, as shown in Figure 4.
(Adapted with permission from Danielson EF (1968) Stratospheric–tropospheric exchange based on radioactivity, ozone and potential
vorticity. Journal of the Atmospheric Sciences 25: 502–518.)

release of water from a dam. Two prominent density
fronts are gust fronts and drainage fronts.
Gust front The gust front develops from evaporative
cooling associated with precipitation below large
convective clouds. Its vertical extent is limited by the
height of the cloud base, usually below 2 km. Temper-
ature changes as high as 101C over a few tens of meters
can occur during severe thunderstorm activity, and the
wind gusts may represent a danger to aircraft that
attempt to land on runways where gust fronts are
evident.
A gust front will last only minutes or tens of minutes
if the cold air moves away from its source. It may,
however, persist for a few hours or more if the cold air
below the cloud base moves with the convective
system. Relatively warm, moist air ahead of the front
moves up and along the frontal surface to the cloud
base, to provide the necessary fuel to continue the
convective activity and sub-cloud precipitation that
ultimately drives the gust front.
Drainage front A drainage front is the leading edge
of downslope drainage of cold air from high eleva-
tions. It may then be referred to as a density front as it
continues to progress along relatively level terrain.
Drainage currents develop at night, and the most
favorable conditions for occurrence are usually met
during the fall when clear skies and calm or relatively
light ambient wind conditions prevail. A pressure
differential develops between the air over the slope and
air at the same level over the level terrain. This pressure
gradient force, together with gravity, drives a down-
slope cold current of air with a temperature differen-
tial at its leading edge. The distinguishing charac-
teristics of the topography and the depth of radiatively
cooled air along the slope will determine the frontal
characteristics. Compressional heating during descent
will also modify the cold temperatures behind the
front, but temperature drops of 51C are not uncom-
mon after frontal passage. Humidity changes are
usually not considered a significant factor in this type
of front, but relatively light and gusty winds of about
5 m s 1 are often encountered.

See also
Boundary Layers: Neutrally Stratified Boundary Layer.
Clouds: Cumuliform. Cyclogenesis. Cyclones, Extra
Tropical. Density Currents. Frontogenesis. Gust
Fronts. Jet Streaks. Stratosphere–Troposphere
Exchange: Local Processes. Synoptic Meteorology:
Weather Maps. Turbulence and Mixing.
Further Reading
Bluestein HB (1993) Synoptic-Dynamic Meteorology in
Midlatitudes, vol. 2. Observations and Theory of Weath-
er Systems. New York: Oxford University Press.
FRONTS 813
Carlson TN (1991) Mid-Latitude Weather Systems. Lon-
don: HarperCollins Academic.
Keyser D (1986) Atmospheric fronts: an observational
perspective. In: Ray P (ed.) Mesoscale Meteorology and
Forecasting, pp. 216–258. Boston: American Meteoro-
logical Society.
Keyser D and Shapiro MA (1986) A review of the structure
and dynamics of upper-level frontal zones. Monthly
Weather Review 114: 452–499.
Orlanski I, Ross B, Polinsky L, and Shaginaw R (1985)
Advances in the Theory of Atmospheric Fronts. Advances
in Geophysics, pp. 223–251. New York: Academic
Press.