Atmospheric Research 143 (2014) 142–175

Contents lists available at ScienceDirect

Atmospheric Research

j ourna l homepage: www.e lsev ie r .com/ locate /atmos

Marine fog: A review

Darko Koračin a,b,⁎, Clive E. Dorman c, John M. Lewis a,d, James G. Hudson a,Eric M. Wilcox a, Alicia Torregrosa e

a Desert Research Institute, Reno, NV, USAb University of Split, Croatiac Scripps Institution of Oceanography, San Diego, CA, USAd NOAA, National Severe Storms Laboratory, Norman, OK, USAe U.S. Geological Survey, Western Geographic Science Center, Menlo Park, CA, USA

a r t i c l e i n f o

⁎ Corresponding author at: Desert Research InstitutE-mail address: Darko.Koracin@DRI.edu (D. Korači

0169-8095/$ – see front matter © 2013 Published byhttp://dx.doi.org/10.1016/j.atmosres.2013.12.012

a b s t r a c t

Article history:Received 30 July 2013Received in revised form 18 December 2013Accepted 26 December 2013Available online 4 January 2014

The objective of this review is to discuss physical processes over a wide range of spatial scalesthat govern the formation, evolution, and dissipation of marine fog. We consider marine fog asthe collective combination of fog over the open sea along with coastal sea fog and coastal landfog. The review includes a history of sea fog research, field programs, forecasting methods, anddetection of sea fog via satellite observations where similarity in radiative properties of fog topand the underlying sea induce further complexity. The main thrust of the study is to provideinsight into causality of fog including its initiation, maintenance, and destruction. The interplaybetween the various physical processes behind the several stages of marine fog is among themost challenging aspects of the problem. An effort is made to identify this interplay betweenprocesses that include the microphysics of fog formation and maintenance, the influence oflarge-scale circulations and precipitation/clouds, radiation, turbulence (air–sea interaction),and advection. The environmental impact of marine fog is also addressed. The study concludeswith an assessment of our current knowledge of the phenomenon, our principal areas ofignorance, and future lines of research that hold promise for advances in our understanding.

© 2013 Published by Elsevier B.V.

Keywords:Marine fogReview

Contents

1. Prologue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1432. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1443. History of sea fog research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1454. Marine fog field programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

4.1. 1850s, Maury and ship observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1474.2. 1913 International ice patrol off Labrador . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

4.2.1. 1925–1927 Meteor Expedition in eastern Atlantic . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1494.3. 1910–1960 Oregon, California and Baja California . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1494.4. 1970s, California Coast, CEWCOM, US Navy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1494.5. 1980–2000s, ONR, other California coastal studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1504.6. Field programs after the 2000s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150

e, 2215 Raggio Parkway, Reno, NV 89512-1095, USA.n).

Elsevier B.V.

143D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

5. Causative factors for marine fog . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1505.1. Advection fog — warmer air flowing over a colder sea (cold sea fog or cold fog) . . . . . . . . . . . . . . . . . . . 150

5.1.1. Advection and air mass trajectory history . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1505.1.2. Advection over upwelled U.S. West Coast waters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1505.1.3. Fog during coastally-trapped disturbances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1525.1.4. Haar off British Isles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1525.1.5. Merging synoptic and local effects — Yellow Sea fog . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1525.1.6. Frequency and persistence of fog over the Yellow Sea . . . . . . . . . . . . . . . . . . . . . . . . . . . 1535.1.7. Importance of air–sea interaction for fog evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153

5.2. Advection fog — colder air flowing over a warmer sea (warm sea fog or warm fog) . . . . . . . . . . . . . . . . . 1535.2.1. Petterssen's and Pilié's viewpoints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1535.2.2. “Steam fog” — advected air much colder than the sea surface . . . . . . . . . . . . . . . . . . . . . . . 154

5.3. Air mass transformation leading to fog — processes at an elevated level . . . . . . . . . . . . . . . . . . . . . . . 1545.3.1. Inversion and cloud forcing leading to sea fog . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1545.3.2. Warm advection depressing the inversion — triggering mechanisms . . . . . . . . . . . . . . . . . . . . 1555.3.3. Synoptic forcing for fog formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1555.3.4. Local circulations and advection of aerosols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1575.3.5. Fog and precipitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157

5.4. Environmental impact on terrestrial systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1575.4.1. Marine fog — hydrologic impacts on ecosystem dynamics . . . . . . . . . . . . . . . . . . . . . . . . . 1585.4.2. Marine fog — chemical and microbiological impacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1585.4.3. Marine fog — thermodynamic ecosystem impacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1595.4.4. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159

6. Microphysics of fog . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1597. Marine fog forecasting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163

7.1. Initial climatology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1647.2. Surge expansion with aviation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1647.3. After WWII . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1657.4. Growth of operational numerical models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1657.5. Satellite development and expansion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1667.6. The elusive sub-mesoscale and microscale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166

8. Remote sensing of marine fog . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1679. Epilog . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171

1 In the decades since Kenneth Ford made his estimates of this ratio, thereis convincing evidence that distances significantly smaller than the size of aproton exist, yet precise quantitative measurement of these distances isdifficult to confirm (Cary and Michael Hwang http://htwins.net).

1. Prologue

While studying atmospheric and oceanic processes, werealize that they are occurring on a wide spectrum of scalesintrinsically coupled through complex interactions. In a moregeneral sense, as we view the development of physics in thetwentieth century and the early part of the twenty-firstcentury, theorists have marveled at the interplay of the verylarge (the universe) and the very small (the elementaryparticles). To quote from the noted theorist John Wheeler:

Quantum mechanics is often described as the theory of thevery small. A true statement, as far as it goes. Quantummechanics is an absolute necessity, and an everyday tool, inexplaining how molecules, atoms, photons, electrons, andother particles behave. It is of no consequence in explainingmotion of spacecraft, planets, comets, and whole galaxies.So what does the quantum have to do with the universe?Perhaps everything, because in any fundamental theory ofexistence, the large and the small cannot be separated.

[(Wheeler and Ford, 1998, p. 329)]

A corollary to Wheeler's statement is the impossibility ofexplaining the phases of marine fog (existence, maintenance,and destruction) without accounting for the very small (themicrophysics of fog droplet formation) and the very large

(hemispheric circulations) as well as the intermediate scalesof motion. Since a typical dimension of fog condensationnuclei is 0.1 μm (10−5 cm) or less and the synoptic-scaleprocesses linked to marine fog are on a scale of 108 cm ormore, a conservative estimate of the ratio of length scales formarine fog is about 1013. For the universe as a whole, theratio of one of the smallest structures (the dimension of aproton: 10−13 cm) to the largest (the distance across theuniverse: 1029 cm) is 1042 (Ford, 1968).1

Fig. 1 schematically displays the collection of processesthat are central to the phases of marine fog — ranging fromthe large/synoptic scale (advection, subsidence, cloud) tosubsynoptic scale (land–sea breezes, sea surface temperaturestructure, and oceanic upwelling) to the mesoscale (radia-tion, surface fluxes) and to the smallest scale (droplets andaerosol within the foggy air mass). And precipitation, alwaysa challenging aspect of any geophysical problem, plays adouble role by moistening and cooling of the subcloud layerand generating cloud thickening and lowering of the cloudbase to eventual fog formation, but also inducing fog dissipationin some cases. The solution demands a coupled set of governing

Fig. 1. Schematic of the main processes governing the formation, evolution, and dissipation of marine fog.

Fig. 2. Sea fog approaching Scripps Pier.Photo by www.flickr.com.

144 D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

equations that describe the dynamics, moist thermodynamics,radiation, microphysics, and turbulence processes— a formida-ble mathematical/physical problem. It is no wonder that thisgeophysical problem attracted some of the most able scientificminds of the early twentieth century: Geoffrey Ingram (“G.I.”)Taylor (1915, 1917), Harold Jeffreys (1918), Anders Ångström(1920), and Ira Bowen (1926).

The difficulties and the challenges of marine fog investiga-tion, however, do not come without esthetic rewards. One onlyneeds to experience the splendor of a sea fog advance along theocean's shoreline to be captivated by the phenomenon. Fig. 2 is apicturesque example of the majesty of sea fog as it advancestoward the pier at the Scripps Institution of Oceanography in LaJolla, California. The turrets that form atop the fog bank giveevidence of the buoyant plume and the leading edge is dynamic.

2. Introduction

Marine fog is a worldwide phenomenon occurring at bothcoastal and open ocean areas, especially frequent over thenorthwestern Atlantic and Pacific Oceans (Fig. 3). Here thewarmwestern boundary currents in each of these oceans, theGulf Stream (Atlantic) and Kuroshio Current (Pacific), abutthe cold oceanic flows neighboring Labrador and Kamchatka,respectively. The processes that give rise to these foggy areasare discussed in Section 5.

Fig. 3 also indicates an area of significant marine fogfrequency off the U.S. West Coast. In this area, the coldupwelled ocean water along the coast in association with

strong warm-season subsidence and surface winds out of thenorth–northwest favor the formation of marine fog (Leipper,1994; Filonczuk et al., 1995). These high frequency areas ofmarine fog give evidence of the differing factors that lead tofog over the world sea.

Although we mention in the further text many of theinvestigations and associated areas of frequent fog occur-rence, here are some of the examples:

Fig. 3. Sea fog frequency [% of the observations indicating moderate to dense fog (not including mist)] in Jun, Jul, and Aug (1885–1933). Chart 53 from Atlas of theClimatic Charts of the Oceans (U.S. Department of Agriculture, 1938).

145D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

U.S. West Coast

Palmer (1917), Byers (1930), Anderson (1931),Leipper (1948), Oliver et al. (1978), Pilié et al.(1979), Hudson (1980), Koračin et al. (2001, 2005a,b), Lewis et al. (2003, 2004), Thompson et al. (2005),Johnstone and Dawson (2010) and O'Brien et al.(2013).

South American West Coast (Ecuador–Peru–Chile coast)

Schemenauer and Cereceda (1991) and Garreaud et al.(2008).

Atlantic Coast (U.S.–Canada)

Taylor (1917), Willett (1928), Gultepe et al. (2006c),Tardif and Rasmussen (2008), Toth et al. (2010) and Yanget al. (2010).

Yellow Sea

Gao et al. (2007), Zhang et al. (2009), Heo and Ha(2010), Kim and Yum (2010, 2012) and Zhou and Du(2010).

South China Sea

Huang et al. (2011).

Southern Africa

Taljaard and Schumann (1940), Cermak (2012) andvan Schalkwyk and Dyson (2013).

Arabian PeninsulaBartok et al. (2012).

From a pragmatic viewpoint, knowledge of marine fog cli-matology and the ability to forecast it is of utmost importance. Itis estimated that 32%of all accidents at seaworldwide and 40% inthe Atlantic occur in the presence of dense fog (Tremant, 1987).The disruption of trade and endangerment to life are always atthe forefront of its assessment (Johnson and Graschel, 1992;Croft et al., 1995; Garmon et al., 1996; Kim and Yum, 2010; Leemet al., 2005). And from an ecological viewpoint, changes in itsfrequency have devastating effects on ecology (Johnstone andDawson, 2010). Some have even suggested that the economicand human losses associated with fog and low visibilities arecomparable to the losses from other weather disasters such astornadoes and hurricanes (Gultepe et al., 2007b).

In this review,we pay particular attention to thewide rangeof small- and large-scale processes that are germane to thephenomenon of marine fog. A brief history of research ispresented in Section 3 and expanded upon at various juncturesin the paper. Section 4 reviews the worldwide field programsaimed at analyzing the fog and the centerpiece of the con-tribution is found in Section 5 — an explanation of the knowncausative factors associated with marine fog. The microphysicsof fog formation is discussed in Section 6, and an assessment offorecast capabilities and detection of fog via satellite observa-tions are found in Sections 7 and 8, respectively. The Epilogsummarizes our knowledge of this phenomenon and lists thoseareas where ignorance remains. Finally, we speculate on lines ofresearch that hold promise for advancement in our understand-ing of the phenomenon.

3. History of sea fog research

As in all modern science, collection and analysis of obser-vations is the starting point for research. Further, in the presenceof the great challenge to predict sea fog — its origin, main-tenance, and decay — the climatology of sea fog assumesparamount importance. The first known “archive” of sea fog

146 D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

comes from the two Icelandic Sagas that include records of seavoyages from Iceland to northeastern North America in the AD1050s through the early AD 1100s (over nine hundred yearsago) — a new world called the “Wineland” by Leif Eriksson andthe other Viking explorers (Bergthórsson, 2000). Bergthórsson, arenowned Icelandic meteorologist and protégé of Carl-GustafRossby, pays particular attention to the meteorological andoceanographic events encountered during these voyages. Fur-ther, he reconstructs weather charts from information in theSagas that give evidence of the sea fog that the explorers en-countered (Bergthórsson, 2000, Ch. 2). And as shown on theclimatic charts of fog over the world sea that came from theextensive logs at Seewarte and the Naval Observatory in late-19th century, the preponderance of sea fog along the track of

Fig. 4. Relative monthly frequencies (in percentages) of sea fog hoursThese data were extracted from Filonczuk et al. (1995). The plot is repr

these voyages from Iceland toNewfoundland exhibits the largestfrequency in theworld—≈30% frequency over the Grand Banksof Newfoundland during the warm season (Fig. 3). Augmenta-tion of these data resources continues with collection at theHadley Climate Centre in England in coordination with COADS(Coupled Ocean and Atmospheric Data Set). Excellent local seafog climatology based on COADS has appeared in publicationssuch as those from the Scripps Institution of Oceanography in LaJolla, California [see, for example, Filonczuk et al. (1995)]. Fig. 4presents information on themonthly frequency of sea fog off theCalifornia coast that has been extracted from data in Filonczuk etal. (1995).

Taylor made the most celebrated study of sea fog duringthe summer of 1913 over the Grand Banks of Newfoundland.

along the California coast by regions for the period 1949–1991.oduced from Lewis et al. (2004). For more details, see Koračin et al. (2005b).

147D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

The study was made in response to the RMS Titanic disasterwhen the British Government and some shipping companiesequipped an expedition on an old 230-ton wooden whalingship, the S.S. [steam ship] Scotia (see Lewis et al., 2004 fordetails). Although the primary purpose of the expedition wasto track icebergs from the glaciers of Greenland, the analysis ofmeteorological conditions that led to sea fog was also a centraltheme. One of Taylor's recollections of the expedition follows:

“… of the 806 occasions on which meteorological observa-tions were taken during the voyage of the whaling shipScotia over the Banks, there was fog 141 times…It wouldhave been impossible to examine every case in detail to seehow the fog arose, but in certain cases I succeeded in raisingthe kite to explore the upper air, and in most of thoseobservations I traced the cause of the fogwhich prevailed atthe time of the ascent. In every one of these cases, it turnedout that the fogwas due to air blowing offwarmwater on tothe cold water of the Banks.”

[(Taylor, 1917, p. 250)]

Fig. 5 shows the track of air for a particular case in July 1913and the associated thermodynamic structure in the presence ofa foggy layer over the Banks. In this case, Taylorwas able to trackthe surface air for a period of nine days through the judicioususe of observations from merchant vessels. The sea surfacetemperatures were obtained from weekly charts published bytheBritishMeteorological Office. Analysis of the kite sounding atthe terminal point of the trajectory alongwith knowledge of thesea surface temperature along the trajectory led Taylor to reasonthat the air column, characterized by a well-mixed adiabaticstate immediately above the sea surface on 18 July, appearsabove 700 m on 25 July. Extension of this profile to the surfacegives a temperature of 26 °C (79 °F) that is close to the seasurface temperature measured by a merchant ship at the originof the trajectory on 18 July. He concluded that after 18 July, theair traveled over progressively colder water and cooled as theresult of turbulence (shear turbulence in thepresence of a stablestratification). From themeasurements of relative humidity andthe saturated state in the lower atmosphere, Taylor concludedthat the fog extended to 700 ft (210 mASL). The high frequencyof sea fog in the northwestern Pacific Ocean displayed in Fig. 3stems from the same action — air flowing over the Kuroshiocurrent to a location above the cold waters of the OyashioCurrent that emanates from the Bering Sea. In Taylor's study, thecold water stemmed from the Labrador Current that flows overthe Grand Banks from its Arctic Ocean point of origin.

Taylor was an expert in the theory of turbulence (Taylor,1915) and he clearly understood the action of turbulence atthe air–sea interface. It was Petterssen (1936, 1938, 1939) whoexpanded the viewof turbulence as a factor in sea fog generation.He studied sea fog generation off the southern California coast-line and realized that it was buoyancy of low-level air thatstemmed from contact with a warmer sea surface (air temper-ature initially colder than sea surface). If the lifting condensationlevel (level where buoyant air reaches condensation) is belowthe level of the inversion, then stratus forms. Andwith longwavecooling at the top of the stratus that mitigates solar warming,negative buoyancy can be created and lead to cooling of thesubcloud layer and subsequent sea fog formation (the stratusthickening/lowering phenomenon for fog creation). It was

Douglas (1930) that first postulated and measured thisnegative buoyancy as it led to the haar, sea fog over theNorth Sea. Others who have made early contributions tounderstanding the influence of radiation as a contributor tosea fog are Emmons and Montgomery (1947) and Leipper(1948, 1994). Again, expanded discussion of contributionsby these investigators is found in Lewis et al. (2004).

As a complement to this brief history that has emphasizedcontributions from the United Kingdom and the United States,we refer the reader to the book by Wang (1985). AlthoughWang does not view the development of sea fog research fromahistorical perspective, he includes a broad-based bibliographyand presents a host of interesting analyses of sea fog in the EastAsian area. Further valuable introduction to sea fog processesfrom a pedagogical viewpoint is found in Roach (1995).

4. Marine fog field programs

Field programs designed to establish a climatic base for seafog effectively began with the international agreement oncollection of ship observations in the 1850s. The first surfacemeasurements and kite soundings in fog-prone areas thatsupported this climatological effort took place in 1913 as partof the International Ice Patrol over the Grand Banks. It was thesame program discussed in the previous section in which G.I.Taylor collected meteorological measurements associated withfog. In addition to the measurements discussed earlier, Taylordeveloped fog diagrams that related surface variables to fogoccurrence (discussed further in Section 5.1.1). In the followingdecade, the German survey vessel Meteor took surface observa-tions and made kite soundings over the Atlantic in the latitudeband 20 °N to 60 °S. This monumental field exercise becameknown as the Meteor Expedition (German: Deutsche AtlantikExpedition) of 1925–1927. As part of this field program, ob-servations of inversion-base height were made in the easternsubtropical zone of the south Atlantic and in the easternboundary upwelling zones where fog is extensive. In the period1910–1960, data were collected from lighthouses along the U.S.West Coast in conjunction with aircraft soundings and finallyballoon soundings. These data were used to identify mesoscalecharacteristics important to fog formation. The 1970s saw single-purpose scientific ships along theU.S.West Coast that focused onidentifying small-scale atmospheric and oceanic structures in theupwelling zones that occurred in the presence of coastal fog. Andagain along the U.S. West Coast during the 1980–2000 period, aseries of instrumented aircraft flights complemented observa-tions from automated coast and buoy stations to study coastalfog. After 2000, satellite microwave measurements of foggyareas were used to improve microphysical parameterizationfor numerical forecast models. Highlights from the variousfield programs follow.

4.1. 1850s, Maury and ship observations

Themost basic climatology, including fog for 71% of the Earth,was directed by Matthew Fontaine Maury when he served assuperintendent of the United States Naval Observatory between1842 and 1861. The systematic collection and analysis of surfacemeteorological and oceanographic observations from ships' dailyweather logs were the basis for this climatology (discussed inMaury's well-known book, The Physical Geography of the Sea,

Fig. 5. Figures from Taylor (1917): (a) path history of air, and (b) temperature and relative humidity profiles obtained from an instrumented kite where the seasurface temperature is indicated by the arrow along the abscissa.From Lewis et al. (2004).

148 D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

1861; reviewed in Lewis, 1996). Maury organized the first trueinternational meteorological conference in Brussels in 1853 forthe world-wide collection of marine data. As part of thecollection strategy, the world oceans were divided into

1° × 1° latitude–longitude squares that were used togenerate monthly analyses — a practice that continues tothis day. Results are found in atlases such as The Atlas of theClimatic Charts of the Oceans (U.S. Dept. of Agriculture,

149D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

1938), later the U.S. Navy Marine Climatic Atlas of theWorld(U.S. Navy, 1955), then a CD-ROM version (http://gcmd.nasa.gov/index.html) that is available on the Internet as TheInternational Comprehensive Ocean–atmosphere Data Set(ICOADS). The ship weather observation was formatted intothe “ship synoptic code” to be complete and systematic fortransmission (U.S. Weather Bureau, 1930) which includes var-iables directly related to fog occurrence and its nature withfactors such as weather type, horizontal visibility, ceiling ob-scuration, humidity, and sea and air temperatures. Formost anymodern study being initiated on marine fog, the first step is togather the ship observations and associated analyses initiallyorganized by Maury. We have followed this practice bymakinguse of charts from the Atlas of the Climatic Charts of the Oceans(Sections 2 and 5).

4.2. 1913 International ice patrol off Labrador

4.2.1. 1925–1927 Meteor Expedition in eastern AtlanticAs stated above, Germany sponsored the Meteor Expedi-

tion in 1925–1927 (Von Ficker, 1936; Spiess, 1985). Accuratemeasurements were made with balloon wind soundings andbox kite soundings with an attached temperature recorder.As a result, the air temperature inversion height was mappedfor the eastern Atlantic and it remains the only ocean-scaleanalysis based directly on measurements. This inversion basewas so low at a point off southwest Africa that the top of theMeteor's mast was above the air temperature inversion base.As later recognized from measurements along the Californiacoast, a similar strong, air temperature inversion capping acool, moist marine layer is a critical factor in fog formationand maintenance in that area.

4.3. 1910–1960 Oregon, California and Baja California

In addition to the Grand Banks, a well-recognized marinefog areawith different dynamics occurs over the coastal watersoff Oregon, California and Baja California. Due to the surfacepressure forces associated with the summertime subtropicalanticyclone over the Pacific Ocean and the heat low over thesouthwest U.S., the winds along the California are from thenorth–northwest and they serve to drive ocean upwelling thatbrings cold water to the surface. A result of the contrastbetween the cold ocean surface and the dry, subsiding air is anatmospheric marine layer with a cold moisture source at thebase capped by dry, hot air, which gives rise to conditionsconducive to fog.

Reports on the basic fog conditions along the California–Oregon coast began to appear in the 1910s through the 1960s.These are: the eastern side of the northeast Pacific anticyclonedominating the area with flow toward the equator; wind drivenupwelling of coldwaterwith the coldest sea surface temperatureclose to the coast; and a shallow, cold,moist atmosphericmarinelayer containing fog and stratocumulus clouds capped by astrong, sharp air temperature inversion. The earliest accounts ofthese conditions were based upon lighthouse observations forclimatology (Palmer, 1917), aircraft soundings (Willett, 1928;Anderson, 1931) and synoptic setting (Willett, 1928; Petterssen,1938). Crucial to fog in this area is the distinctive air temperatureinversion layer which is an important controlling factor, andstratus, often closely related to fog or a stage in its cycle, has been

the focus of many early field programs reviewed by Kloesel(1992).

Clarity of the atmospheric conditions accompanying fog,stratus, and cumulus in the eastern Pacific Ocean came withtropical–subtropical studies by Herbert Riehl and his studentsat the University of Chicago near the end of WWII (Riehl et al.,1951; reviewed in Lewis et al., 2012). Of particular interest inthis marine fog review is the series of upper-air soundings thatwere obtained from four stationary U. S. Navy ships thatoperated on the flight path from San Francisco to Honoluluduring the period July to October 1945. Observations fromthese ships defined the height and depth of the summertimeinversion over the subtropical latitudes of the eastern PacificOcean. Morris Neiburger and his colleagues at UCLA continuedthe line of research started by Riehl with special emphasison the stratus cloud in the eastern Pacific (Neiburger et al.,1961). The work of Neiburger and his associates has formedthe backbone for subsequent studies involving near-surfacewinds, marine conditions and marine clouds. A major resultis that the air temperature inversion is lowest (below 500 mASL) and most persistent during the extended summerbetween Cape Mendocino in northern California and PointConception in southern California. Going westward, theinversion base height rises quickly from the North andCentral California coast, and then more slowly to be above1200 m ASL at Hawaii.

4.4. 1970s, California Coast, CEWCOM, US Navy

In the 1970s, the Calspan Corporation (a component of theCornell Aeronautical Laboratory) and the Naval PostgraduateSchool,Monterey, CAparticipated in theCooperative Experimentin West Coast Oceanography and Meteorology (CEWCOM). TheNaval Air Systems Command supported the program. As part ofthis experiment, a series of single-ship cruises took place alongthe California coast from northern California to San Diego atdistances as far as 500 km from the coastline. On seven cruises,30 fog events were encountered. Especially along the northernCalifornia coastline, the fog areas were characterized by strong,low air temperature inversions and wind-driven coastal upwell-ing that occurred in patches. Using data from this experiment,Pilié et al. (1979) reported five different conditions under whichfog develops. One condition indicates that fog is triggered byinstability and mixing of colder air over warm water patches, sothat the fog layer grows along downwind trajectories. A secondcondition links fog development within a stratus layer toradiative cooling atop the cloud layer. The cooling leads tonegative buoyancy so that the stratus lowers and thickens tocreate fog. A third condition relates fog formation to low levelconvergence, growth of the entire layer, and lifting of theinversion base height. The fourth condition favors fog formationover the coastal waters in response to radiative cooling over theadjacent land at night. In wintertime, a nocturnal land breezeadvects the cool air out to sea. Finally, another instance is acomplex land–coastal interaction that occurs along southernCalifornia to form fog during the fall–winter–spring inassociation with a “Santa Ana” — a warm offshore, dry windregime. Fog in response to the Santa Ana flow regime wasoriginally proposed by Leipper (1948). Focusing on southernCalifornia, radar, acoustic sounder, aircraft, satellite, radio-sonde, ship and coastal observationswere employed to sort out

150 D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

shallow marine air movement over water with differenttemperatures (Noonkester, 1979; Pilié et al., 1979). Warm, dryoffshore flow lowers the marine inversion and passes overcoolerwaterwhere fog forms. As stated in earlier discussions, animportant factor in fog maintenance and growth is longwavecooling at the fog top.

4.5. 1980–2000s, ONR, other California coastal studies

With knowledge of the large-scale factors influencing fogclimatology, smaller scale, difficult to measure aspects of fogevents became a focus along the California coast during the lastdecades of the 20th century. Long aircraft flights west of SanFrancisco in September 1981 found that cooler air moving overwarmer water created instability and vertical mixing thatdissipated fog (Telford and Chai, 1993). Ship and coastalautomated surface observations illuminated the smaller scale,diurnal variation and higher concentrations of fog occurrencealong California (Filonczuk et al., 1995; Lundquist and Bourcy,1999). A coastally trapped atmospheric event with a shallowbore pushing poleward along the central California coastcreated a shallow stratus-coastal-cloud bore that developeddownward into fog by pushing warmer air over colder water(Dorman et al., 1998). The changing trajectories of near-surfaceair during transient weather events were linked to both theformation and dissipation of sea fog (Lewis et al., 2003; Koračinet al., 2001, 2005b).

4.6. Field programs after the 2000s

In the 2000s, fog field experiments tended to focus onsurface observations and near-surface observations that wereco-located with satellite measurements. The field work wasdesigned with an eye on strengthening operational numer-ical forecast models — model development that neededguidance on parameterization of the moist physics associatedwith sea fog.

FRAM-L was the marine phase taking place in the summerof 2006 along the Nova Scotia, Atlantic coast that was part of alarger fog program that included continental fog. Its majorstrength was the use of optical probes for detailed measure-ment of droplets and visibility at the surface combined withsatellite microwave radiometer data to examine inferred fogproperties that will be used to develop microphysical param-eterizations that could be incorporated into numerical forecastmodels (Toth et al., 2010).

5. Causative factors for marine fog

Thewide range of processes governingmarine fog have beenmentioned earlier and displayed in Fig. 1. Earlier studies havetended to focus on a limited number of these known processeswhereas recent work has often tended to dismiss significantresults from the early studies — in part due to a restrictiveviewpoint tied to the goals of funded research projects. Further,many studies have failed to be even-handed in their attention tothe phases of marine fog — often emphasizing initiation andduration without consideration of processes that maintain anddestruct fog. As is knownby thosewhohave investigated sea fog,once formed, it tends to persist. Essentially, generative processescounteract destructive processes. A negative feedback process is

at work. In this section that strives to identify most of thecausative factors related to marine fog, the dynamical/physicalprocesses are divided into the following categories: cold sea fog,warm sea fog, and elevated and land forcing. And each of thesemajor themes is further subdivided. Although we have made aneffort to avoid overlap in our review, by necessity some overlapoccurs — especially noticeable with information in Section 4(Marine fog field programs) and itemization of causative factorsfound in this section.

5.1. Advection fog — warmer air flowing over a colder sea (coldsea fog or cold fog)

The sea fog investigated by Taylor (1917) has come to belabeled advection fog — a fog that is generated through theaction of air movement over a surface with a differenttemperature. In Taylor's study, the warm/moist air initiallyover the Gulf Stream was transported northward and encoun-tered progressively colder sea surface temperatures on its pathto Newfoundland. This we label cold sea fog in contrast to theopposite case where cold air is transported over progressivelywarmerwater— the situation studied by Petterssen off the coastof southern California and labeled warm sea fog (discussed inSection 5.2).

5.1.1. Advection and air mass trajectory historyThe panels of Fig. 6 are complements to the sea fog

frequency chart shown in Fig. 3. Focusing on the high-frequencyfog area over the northwestern Atlantic, we note sea surfacetemperatures (SSTs) as low as 10 °C north of Newfoundland. Inthe other cardinal compass directions relative toNewfoundland,the SSTs are significantly warmer (the American continent tothe west and the Gulf Stream to the south and east). An airfloworiginating from any of these warmer areas can potentiallyinduce cooling of the near-surface marine air and eventuallyproduce fog (Fig. 6). Taylor concluded that all fluid states(moisture–temperature structure) in these cases must maponto a straight line in the q–Θ plane (q — total water mixingratio; Θ — potential temperature) which passes through thesurface state (q0, th0) and has a slope determined by the ratio ofthe air–seamoisture andheat fluxes. If the turbulence effects aredominant and radiation effects are negligible, Oliver et al.(1978) showed that this condition implies that temperature andmoisture fluxes are the same and that the distributions of q–q0andΘ–Θ0 and their turbulence correlationsmust be functions ofz/Lm and z/z0, where Lm is the Monin–Obukhov length, z is theheight, and z0 is the roughness length. To achieve this state,Taylor concluded that one of the most important factors wasadaptation and modification of an air mass along the longoverwater trajectory originating over the warm sea surface andcontinuing over the cold ocean current. This use of the similaritytheory complements Taylor's analysis.

5.1.2. Advection over upwelled U.S. West Coast watersAnother fog-prone area is the U.S. West Coast, which is

characterized by cold upwelled sea surface in thewarm season,a shallow atmospheric marine layer with frequent northerlyand northwesterly flows and occasional offshore flows, andsubsidence thatmaintains a strong low-level marine inversion.As itemized in the Introduction, there have been numerousstudies that have investigated West Coast fog. Byers (1930)

Fig. 6. a. Same as Fig. 3, except for temperature differences (air temp − sea surface temp). b. Same as Fig. 3, except for wind regime.

151D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

studied summertime fog over this area and hypothesized thatthe fog forms by saturation of moderately cold air moving overthe colder sea surface. Other investigators including Leipper(1948, 1994) have challenged the premise that fog formssimply as a consequence of cooling and moistening of thetransported air. According to these studies and based on manyobservations and data analysis, a primary mechanism for fogformation is as follows: In the warm season, offshore flows(linked to an anticyclone over the northwestern U.S.) descendover the coastal mountains and additionally warm by com-pression and adiabatic heating from 5 °C to as high as 24 °C(Leipper, 1994) (Fig. 7). This warm offshore flow effectivelydepresses the marine inversion almost to the surface and fogforms in a very shallow marine layer. Once fog forms, other

processes such as the longwave radiation at the fog top inducecooling and turbulent mixing that initiate fog growth. It shouldbe noted that Johnstone and Dawson (2010) also support thescenariowith positive correlations between fog occurrence andOakland upper air temperatures above the marine layer andnegative correlations within the cooler than normal marinelayer.

So, to summarize this scenario, fog does not form withinthe warm advected air, but it forms in the extremely shallowmarine layer over the cold ocean waters. The warm advectedair serves to lower the marine inversion almost to the surfaceand effectively contain a shallow and moist marine layer.Pilié et al. (1979) used observations and conceptual modelingand also concluded that one of the main mechanisms for the

Fig. 7. A schematic diagram of Leipper's conceptual model of sea fogformation during a hot-spell event. The upper portion of the diagram showsthe surface airflows associated with the position of the Pacific anticyclone.The inset shows the relative position of clear air, fog, and stratus relative tothe coastline. See text for further details.The plot is reproduced from Lewis et al. (2004).

152 D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

fog formation in the warm season over theWest Coast is that awarm advected offshore flow erodes the marine inversionalmost to the surface and allows saturation of this near-surfacemarine layer. Koračin et al. (2005b) simulated processes in thisscenario of fog formation in the very shallowmarine inversioncapped by the strong low-level inversion. The importance oflongwave cooling in rapid fog growth was discussed as well asthe necessity of high vertical resolution in resolving radiationfluxes that are crucial for fog evolution.

5.1.3. Fog during coastally-trapped disturbancesInterestingly, some authors have proposed physical mech-

anisms different from the classical mechanisms for cold sea fogmentioned above. One example is related to fog over the coldupwelled West Coast water that appears during occasionalcoastally-trapped disturbances [CTD, also called coastally-trapped wind reversals (Nuss et al., 2000; Thompson et al.,2005)]. The CTD events usually occur several times each yearover the West Coast during the warm season synoptic setup(anticyclone in the northern Pacific and low over the U.S.southwest). In these areas, persistent northerly and northwest-erly flow regimes are interrupted by a period of opposingsoutherly flow near the coast where clouds and fog are spreadover several hundred kilometers of the coast and rapidly

propagate northward (Nuss et al., 2000). These conditionsappear to represent a characteristic case of warm and moist airpropagating northward over cold upwelled water [documentedby Dorman et al. (1998)]. However, Thompson et al. (2005)argued that according to their sensitivity tests from 3Dsimulations using the COAMPS model (Coupled Ocean–atmo-sphere Mesoscale Prediction System; Hodur, 1997; Hodur et al.,2002), the main driver for the fog maintenance during thenorthward propagation was due to wind convergence andassociated rising motions at the leading edge of the cloud/fog.The convergence and rising motions change the stability of themarine layer from strongly stable to a shallow and cool mixedlayer with relative humidity increasing upward. Future studiesincluding observations and advanced modeling are needed tofurther investigate these hypotheses based on testing particularmodel sensitivity to someof the processes such as surface fluxes.

5.1.4. Haar off British IslesThe haar is a well-known sea fog that occurs most

frequently along the east coast of Scotland. The unusual namecomes from the speech of the inhabitants of LincolnshireCounty on the east coast of England. This fog has been traced tothe cooling of warm air masses by the relatively cold waters ofthe North Sea (east of Scotland). British meteorologists HubertLamb and C.K.M. Douglas are credited with an early investiga-tion of haar that gave evidence of both classic cold sea fogprocess augmented by radiative cooling atop the fog layer(Douglas, 1930; Lamb, 1943). Lamb's study made it clear thatlocal land–sea circulations were often active in haar wherefoggy air at sea was brought to land by a sea breeze. The poorvisibility is often extreme with haar (visibility b25 m). Furtherinsight into haar is shown by Taylor (1987) and Findlater et al.(1989), among others.

5.1.5. Merging synoptic and local effects — Yellow Sea fogThe classic concepts of cold sea fog are in evidence from

studies of fog over the Yellow Sea and the East and South ChinaSea. There are a significant number of observational andmodeling studies aimed at understanding the active physicalprocesses governing fog in this area. Themain backdrop includessynoptic and local effects. InMarch–April, heating of the air overcentral-eastern China usually induces a shallow anticyclone overthe cool Yellow Sea and the northern East China Sea thatincludes a prominent inversion at 100 to 350 m altitude. On thewest flank of the anticyclone, this setup generates persistentsoutherlies which advect warm and humid air over the coolwaters of the Yellow Sea and East China Sea that leads to fogoccurrence (Zhang et al., 2009). In July–August, the East Asian–western Pacific monsoon initiates a strong shift from southerliesto easterlies over those two areas that brings drier air to theregion that erodes the marine inversion and leads to disappear-ance of the fog.

Additionally to mechanisms for fog along the U.S. WestCoast due to offshore flow and lowering of the marine inver-sion (Leipper, 1948, 1994; Koračin et al., 2005a), Zhang et al.(2009) indicate that the offshore flow from the heatedcontinent (at about 925 hPa) strengthens themarine inversionand traps the moisture in a shallowmarine layer. Compared tothe central Yellow Sea, the sea surface approaching the Koreancoast is even colder due to intense tidal mixing that promotesmore fog events (Cho et al., 2000; Zhang et al., 2009).

153D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

Numerical simulations of sea fog over the Yellow Sea havebeen conducted by Gao et al. (2007). Their results usingMesoscale Model 5 (MM5; Grell et al., 1994) showed thatformation of a surface-based inversionwas a critically importantelement in the formation of sea fog. The inversion led to devel-opment of a shallow, moist marine layer and formation of athermal internal boundary layer at the base. Sea fog formed inresponse to gradual cooling and moistening by turbulentmixing. Dissipation of fog was caused by the shift in advectionfrom warm and moist southerlies to dry and cold northerlies.

5.1.6. Frequency and persistence of fog over the Yellow SeaKim and Yum (2010) analyzed the climatologies of sea fog

along the Korean coast and open ocean and categoricallyseparated events into coastal and open ocean fog–sea fog.Although there were significantly more cases of coastal fog,the sea fog (cold sea fog being more frequent than warm seafog) was characterized by longer durations. Based on thesynoptic setup discussed above, the occurrence of cold fogduring the warm season is in agreement with results fromZhang et al. (2009) while the occurrence of warm sea fog wasusually from January to May. Kim and Yum (2010) noted thatthe dew point preceding the cold fog was greater than theSST. On sea fog days, this dew point was approximately 5 °Cgreater than on fog-free days. Consequently, fog was formedby cooling of already sufficiently humid air.

Using numerical simulations, Kim and Yum (2012) inves-tigated a case of cold fog over the Yellow Sea which they foundto occur more frequently than warm sea fog. They simulatedthis fog event with both a 1D model (Bott and Trautmann,2002) and a 3Dmodel— theWeather Research and Forecasting(WRF) model (Skamarock et al., 2008). They indicated thatsome processes relevant to cold sea fog begin with cooling ofthe overlaying warmer air by radiative cooling and turbulentmixing which in turn leads to a thermally stable internalboundary layer. The moisture loss due to the downwardmoisture flux near the surface is balanced by the continuousadvection of the moisture — this keeps the dew pointquasi-constant and, due to net cooling, the air mass eventuallyreaches saturation.

5.1.7. Importance of air–sea interaction for fog evolutionHeo and Ha (2010) used coupled and uncoupled versions

of the COAMPS model (Hodur, 1997; Hodur et al., 2002) andthe ROMS ocean model (Shchepetkin and McWilliams, 2004)to investigate cold sea fog cases and a steam fog case over theYellow Sea. Regarding the cold sea fog, there was advection ofwarm and moist air over the cold coastal waters that wereadditionally cooled compared to offshore regions by tidalmixing (Cho et al., 2000). Simulated trajectories showed thatthewarm airmass coming from the southwas gradually cooledover the SST that decreased northward. Condensation oc-curred, thus inducing negative latent heat flux and increasingstability near the sea surface causing weakening of the windsand restricting downward mixing of drier air. Decreasing SSTalong the advection trajectory appears to be a significantpre-cursor for fog formation allowing cooling and moisteningof the near-surface air. This also maintained the fog until thesoutherly advection was reduced. Heo and Ha (2010) notedthat uncoupled simulations showed underestimation of

stability, and they were unable to represent these processesrelevant to the formation of fog.

Huang et al. (2011) studied observations during a cold seafog event over southern China coastal waters with warm airadvection over the cooler waters. They suggested that themarine boundary layer consisted of two major sub-layerscharacterized by mechanical turbulence near the sea surfaceand thermal radiation turbulence effects due to longwavecooling from the stratus/fog top. They indicated that a thermalturbulence interface was observed between 180 and 350 maltitude that separates these sub-layers. Fog dissipation or/andlifting to stratus was attributed to reduction of moisturetransport, rapid lifting of the fog top and associated entrain-ment. Results from this study indicate that the thermalturbulence interface persisted even when the fog lifted toform stratus. In short, the thermal turbulence interface has asignificant role in maintaining stratus. This could motivatefuture observational and modeling studies to further under-stand these observed characteristics.

5.2. Advection fog — colder air flowing over a warmer sea(warm sea fog or warm fog)

Another concept of fog formation stems from advection ofcolder air over the warm sea where saturation occurs inresponse to mixing of the cold and sufficiently moist air withwarm/moist air. Taylor (1917) also gave some attention tothis process of warm fog formation.

5.2.1. Petterssen's and Pilié's viewpointsLewis et al. (2004) reviewed the work by Petterssen

regarding fog formation due to advection of colder air overwarmwater (Petterssen, 1936, 1938, 1939). Petterssendiscussedwarm sea fog in the context of mixing — i.e., mixing of twounsaturated air masses, one the incoming cool and unsaturatedair and the other the near-surface warm and saturated ornear-saturated air. The verticalmixing is a response to buoyancy.Sufficient mixing and radiative cooling then lead to saturation ofthe air mixture.

Some of the main processes during this type of fog arediscussed in detail by Pilié et al. (1979). They collectedshipboard measurements in a situation characterized by fogat the site of the ship but fog-free upwindof the ship. Therewasa surface inversion layer on the upwind side that effectivelytrappedmoist near-saturated air at the air–sea interface.Whenthis layer traversed warm sea, the stability was eroded. Mixingand saturation then took place between the warm and moistair at the sea surface and the advancing cool, moist air (Fig. 8).The temperature decreased in the fog layer due to mixing andradiative cooling and a superadiabatic layer was maintainednear the surface. The cold fog layer also represents a sink formoisture that consequently enhances evaporation from thewarm sea surface.

Another study of warm fog evolution was shown by Kimand Yum (2010). They analyzed data from sea fog (cold andwarm) and also coastal fog over the west coast of Korea. Theirmeasurements indicated that the air temperature slightlydropped when the fog was formed over the warm waters.They concluded that the dew point increase, with thedifference between the air temperature and the dew pointduring the fog mature phase ranging from 0.5 to 1.9 °C, is due

Fig. 8. (Left) Profile sketch of the upwind edge of a fog patch formed over warm water, 30 Aug 1972; (right) Selected vertical profiles at indicated positionsrelative to fog edge, 30 Aug 1972.From Pilié et al. (1979).

154 D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

to moisture transport from the surface and is the dominantprocess for fog formation. They indicated that this is possiblydue to longwave cooling of the fog top which overpowerswarming from the sea surface. Notably, Pilié et al. (1979) alsoobserved the effect of fog cooling over the warmwaters (Fig. 8;their Figs. 2 and 3). Others also demonstrated pre-conditioningcooling and/or fog cooling over warm waters. Koračin et al.(2001) used 1D simulations emulating cooling of a Lagrangianair mass along a long overwater trajectory due to longwavecooling from the cloud/fog that can eventually overpowerwarming from the surface. With a continuous supply from thesurface and cooling-generated instability and turbulence, thisprocess can lead to fog formation. These effects are furtherelaborated on in Section 5.3.2.

5.2.2. “Steam fog” — advected air much colder than the seasurface

When a stream of cold, dry air (typically originating overland) traverses a much warmer sea surface, a “steam fog” canoccur. Such events also occur with regularity over lakes inwintertime. The large heat capacity of water leads to slowercooling of the lake water relative to the land mass (especiallynotable in late fall and earlywinter). In both cases, i.e., over lakesor oceans, the latent and sensible heat fluxes are extreme (oftenreaching values of 100s ormore ofW m−2 for both sensible andlatent heat fluxes). The steam fog is frequently observed in theArctic (Saunders, 1964; Økland and Gotaas, 1995; Gultepe et al.,2003). Further observations and simulations, especially for casesof larger air–sea temperature differences, are needed to revealthe role of SST in steam fog.

5.3. Air mass transformation leading to fog — processes at anelevated level

The earlier discussion of sea fog clearly indicates that muchof the attention has focused on the relationship between thesurface atmospheric layer and the ocean surface. However,there are certainly other pathways that can bring the surfacelayer to near saturation over the sea. Some of the otherpossibilities occur in response to elevated processes above thesurface layer. These include convergence and divergence above

the surface layer that can change the subsidence and inversiondynamics on the top of the surface layer. Above this surfacelayer, the presence of clouds, haze, and particulate matter alterthe upward and downward radiation. Precipitation can alsoaffect surface layer processes. Slight changes in these factors cantip the balance for fog formation. However, a near-saturationcondition can also persist without fog formation where the sumof elevated processes is a significant factor. Once formed, thereare tremendous differences between cloudy and cloud-freemarine layers with respect to thermal, radiative, and turbulenceprocesses that favor maintenance of a cloudy layer (Tjernströmand Koračin, 1995).

5.3.1. Inversion and cloud forcing leading to sea fogKoračin et al. (2001) studied a transient-season (April) sea

fog over the U.S. West Coast in the presence of an along-coastnortherly and northwesterly flow. They emphasized a need toconsider the sea fog formation and evolution in a Lagrangianframework that allows for understanding air mass modificationalong a long over-water trajectory. In essence, these researchersfollowed Taylor's (1917) lead on considering air mass transfor-mation in a Lagrangian framework. The modification wascharacterized by the sea being warmer than the air withcontinuous cooling of the marine air along the trajectory.According to their analysis of observations and a high-resolution 1D model, they concluded that the main mechanismfor fog formationwas cloud-top cooling and associated turbulentmixing, supplying moisture due to a positive heat flux, and all inthe presence of intense subsidence, maintained strong marineinversion, and shrinking of the marine layer along the advectiontrajectory (Fig. 9). The 1-Dmodel was run in a Lagrangian modewhere the advection was emulated by varying the SST over aspatial trajectory estimated from the weather analysis. Thedownward propagation of cloud cooling can be seen clearly inthis time–height cross section of air temperature (Fig. 9). Theonset of cooling first occurs at higher elevations and graduallydescends to lower levels. For example, the net cooling beginsafter 12 h of simulation at 200 m,while it takes 2more hours forthe cooling to reach the 100-m level. Significant cooling occurredafter 39 h when the elevated air driven by cloud-top coolingmerged with the near-surface layer. Cloud-top cooling and

Fig. 9. Contours of ambient temperature (°C) for the baseline 1D Lagrangian simulation. Contour interval is 0.5 °C. See the text for further details.From Koračin et al. (2001).

155D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

subsidence appear to be dominant factors for fog formation andevolution during cloud lowering along the advection trajectory.Their sensitivity tests showed that there is an optimum strengthof the inversion and that moisture above the inversion controlsthe longwave cooling and entrainment processes.

5.3.2. Warm advection depressing the inversion — triggeringmechanisms

A triggering mechanism for saturation within the shallowmarine layer is still under investigation and debate. As men-tioned earlier, Koračin et al. (2005b) used a high-resolution 1Dmodel to simulate depression of the marine inversion andconsequent fog formation. In this case, differential advection ispresent — warm air advection above the inversion and cold airbelow the inversion. Their hypothesis is that the increasedlongwave flux divergence at the top of a shallow marine layerwith sufficiently uniform moisture leads to condensation at thetop of this shallow layer, which then rapidly propagates to thesurface due to further increases in longwave cooling at the fogtop. Their simulations show that longwave cooling at theinterface of the warm, dry air and the moist, cool air at the verytop of the depressedmarine layer is the dominant process for fog

formation. Sufficient mechanical and radiatively-driven turbu-lence initiate saturation at the very top of the marine layer,which rapidly propagates downward to the sea surface withina very shallowmarine layer. Consequently, longwave cooling atthe fog top is the main driver of the fog growth that is limitedby the strength of the subsidence and gradual warming anddrying while growing and mixing vertically. Their sensitivitystudy (Fig. 10) clearly indicates that the model setup (verticalresolution and physics options), and initial conditions (lowerboundary— SST, humidity and temperature profiles within themarine layer, and properties above the inversion such ashumidity, temperature, and subsidence) significantly influencefog initiation and evolution.

These tests clearly indicate a need for sophisticatedensemble modeling providing probabilities for fog occurrence,evolution, and dissipation (further discussion in the Epilog).

5.3.3. Synoptic forcing for fog formationLewis et al. (2003) re-analyzed the case studied by

Koračin et al. (2001) focusing on the larger-scale influenceson fog formation and dissipation. They contrasted synopticconditions for fog and fog-free events and found that the

0 20 40 60 80 100 120 140 1600

50

100

150

200

250

Hour

Hei

ght (

m)

5 10 15 20 25 30 35 40 45−60

−50

−40

−30

−20

−10

0

10

20

Hour

Hea

ting

(K/d

ay)

Fig. 10. (Left) Evolution of the simulated fog top for the baseline run (solid line), a case with colder sea-surface temperature by 2 °C (○), a case with warmer sea-surface temperature by 2 °C as compared to the baseline run(*), a case with larger near-surface inversion (×), and a case with a drier hot-air layer (+), as compared with the baseline run; (right) time series of the simulated average net heating of the fog layer (°C day−1) due tolongwave and shortwave radiative heat transfer for the baseline simulation with 180 vertical points (solid), the test run with 90 vertical points (dash–dot), and the test run with 45 vertical points (dashed line) within1200 m.From Koračin et al. (2005b).

156D.K

oračinet

al./Atm

osphericResearch

143(2014)

142–175

157D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

strength and evolution of the synoptic-scale subsidence,strength, and height of the marine inversion, air–sea tempera-ture difference, cloudiness at the top of the marine layer as wellas air-mass transformation along trajectories were criticallyimportant factors in sea fog formation. Their back-trajectoryanalysis showed that trajectory residence time over the landsignificantly reduced fog formation over the sea. Koračin et al.(2005a) performed 3D simulations of the same case fromKoračin et al. (2001). The simulations were able to demonstratethat the intensity of air mass modification during this advectionsignificantly depended on whether there were clouds along thetrajectories and whether the trajectories had residence timeover the land or ocean. The model was able to simulate overallcooling of the marine layer (due to longwave cooling at thecloud top) in spite of a gradual increase in the SST along thetrajectory. A scale analysis of the model results showed thatlongwave cooling at the cloud and fog top overpowered surfacesensible and latent heat fluxes and entrainment in cases oftransformation of cloudy marine layer over long over-watertrajectories. Fog dissipation was significantly influenced by thedevelopment of land-driven circulations as a result of thecomplex interplay among advection, synoptic evolution, andspecifics of local circulations. Displacement and weakening ofthe horizontal synoptic pressure gradients and the consequentdecrease of the marine winds allowed for the development ofoffshore flows that induced warming and drying and conse-quent fog dissipation.

5.3.4. Local circulations and advection of aerosolsWhile categorizing fog by advection properties, an impor-

tant component is advection of aerosols and fog condensationnuclei that bears importance for fog both over the sea and land.The advection can be due to local circulations or synopticpatterns. Transport of sea aerosols and water vapor onto theland during sea breeze circulations can trigger fog over the landwhere the condensation nuclei came from the sea (Bartok et al.,2012). Obviously, fog can form over the land and then drift tothe sea during a land breeze and vice versa during a sea breeze(Pilié et al., 1979). This further complicates investigation of theorigin of fog and physical processes relevant to fog formation.These issues are comprehensively addressed in Section 6(microphysics of fog).

5.3.5. Fog and precipitationSome of the early studies indicated that cloud precipitation

cools and moistens the subcloud layer and lowers the conden-sation level. Pilié et al. (1979) describe this process as cloudthickening. It should be noted that in many cases, this process isinaccurately called “cloud lowering”. The term “cloud lowering”should be reserved for the whole entity of cloud lowering(including the cloud top) and this is actually explained in thestudy by Koračin et al. (2001) where subsidence overpoweredmarine turbulence and the whole inversion lowered along thetrajectory of the air mass.

Asmentioned earlier, Tardif and Rasmussen (2008) showedthat the effect of precipitation evaporating and moistening thesubcloud layer can occur for a wide range of precipitationintensities and can strongly influence fog formation, evolutionand dissipation. Although the majority of the cases fall intocategories of light precipitation, rain showers and snow aspre-condition forcing can also lead to fog formation. They claim

that the moistening is a more dominant process compared toevaporative cooling leading to fog formation. In a subsequentstudy, Tardif and Rasmussen (2010) used a Lagrangian-framesimulation of a microphysical column model and showed thatevaporation from raindrops departing from equilibrium, i. e., thelatent heat loss from droplet evaporation does not balancesensible heat flux from the ambient air, facilitates fog formation.Consequently, they indicate that the non-equilibrium state of thefalling raindrops needs to be taken into account while consid-ering precipitation fog events.

5.4. Environmental impact on terrestrial systems

Advecting marine fog transports water and materials ontoadjacent coastal and terrestrial systems, profoundly altering theenvironmental conditions of these systems. The transport of fogwater droplets and their associated soluble ions, dust particles,microorganisms, andmixtures of organic and inorganic reactionproducts changes the hydrologic (Bruijnzeel et al., 2005;Dawson, 1998), thermodynamic (Madej et al., 2006; Iacobellisand Cayan, 2013), nutrient (Weathers and Likens, 1996; Ewinget al., 2009; González et al., 2011), and toxicological (Wurl andObbard, 2004) regimes of ecosystems along the coast. Althoughthe interactions are variable and complex, several prominentecological mechanisms associated with each type of flux(water, energy, and microscopic particulates both inorganicand organic) have been elucidated.

The strong ecological response to the spatial and temporalvariability of marine fog results in discernible impacts on diurnalto seasonal (Fischer et al., 2009; Carbone et al., 2013), inter-annual to decadal (Cereceda et al., 2007; Garreaud et al., 2008),and longer scales (Gutiérrez et al., 2008; Williams et al., 2008).The strong biogeographic climate signature of transportedmarine fog water and minerals has led to terrestrial proxies forpaleoclimate change in coastal upwelling. Fog associated speciessuch as coastal redwoods [Sequoia sempervirens ((D. Don)Endl.)], coastal fog opportunists such as Bishop pines (Pinusmuricata (D. Don)), and endemic fog-dependent species such asTorrey pine (Pinus torreyana ssp. insularis (Haller)) and ChileanTique [Aextoxicon punctatum (Ruiz and Pav)] have providedpollen (Heusser, 1998; Barron and Anderson, 2011) andtree-ring based chronologies (Roden et al., 2009; Williams etal., 2008; Gutiérrez et al., 2008; Díaz et al., 2001) that are used aspaleoclimate proxies (Poole and van Bergen, 2006) in theinterpretation of atmosphere–ocean linkages associated withmajor climate events. Johnstone and Dawson (2010) relatedhow, for example, over the last 8 millennia, expansion andcontraction of fog associated vegetation is interpreted as asynchronous response to marine fog-conducive upwellingconditions that are an expression of ENSO/PDO variations(Barron and Anderson, 2011). In the context of longerpaleoecological timescales, fog imbued coastal regions nowact as refugia for forest ecosystems that were once distributedacross North America but do not tolerate current combinationsof arid and freezing conditions such as redwood forests that arenow restricted to a narrow 50 km belt along the Californiacoast (Ahuja, 2009; LePage et al., 2005; Johnstone and Dawson,2010). Reported decline of summer fog along the Californiacoast (Johnstone and Dawson, 2010) and Hokkaido Island,Japan (Sugimoto et al., 2013) could result in a cascade ofecosystem responses should the trend continue.

Fig. 11. Atmospheric processing of a 0.2 μm organic marine aerosol.From Ellison et al. (1999).

158 D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

5.4.1. Marine fog — hydrologic impacts on ecosystem dynamicsOne of the earliest records of hydrologic transport from

marine fogs into coastal ecosystems is the “sacred” fountaintree, attributed to Pliny the Elder. In 1764, GeorgeGlas describedit as a laurel [Ocotea foetens (Aiton) (Benth. & Hook.f)] on theCanary island of El Hierro used by indigenous people until itsuprooting during a hurricane in 1610 (Gioda et al., 1995). Inrecent times, coastal drylands adjacent to eastern ocean basinssuch as California, Chile, and Namibia have been used as aproductive laboratories to understand the mechanisms ofhydrologic input frommarine fog at scales frommicrobiologicalto landscape. In the Mediterranean climate system of California,precipitation from the strong cyclonic activity occurs during thewinter with little to no precipitation falling during theanticyclonic dry summer period. By late summer the waterdeficit experienced by plants as they transpire more water thanis available to maintain tissue turgor is at its most extreme, andif unabated will cause a plant to wilt and die.

Two mechanisms effectively transfer liquid water frommarine fogs into ecosystems. The first is fog drip, where fogwater accumulates on surfaces such as leaves, needles, or stemsthen drips or is channeled to the soil surface providingmoisture to shallow roots (Means, 1927; Oberlander, 1956;Azevedo and Morgan, 1974; Ingraham and Matthews, 1995;Corbin et al., 2005; Simonin et al., 2009; Roth-Nebelsick et al.,2012) or to collection containers for human use (Schemenauerand Cereceda, 1994). The second is foliar uptake whereleaf-wetting fog events increase foliar hydration when watermoves directly through leaf pores and surfaces into internaltissues along a water potential gradient (Limm et al., 2009;Burgess and Dawson, 2004; Vasey et al., 2012; Eller et al.,2013). Critically important features for this mechanism are theabsolute liquid water content of the marine fog event anddeposition rate (Lovett, 1984; Slinn, 1982; Joslin et al., 1990;Katata et al., 2010; Hiatt et al., 2012). The first is predominantlya function of cloud condensation nuclei (CCN) composition andpath history while the second is influenced by wind speed,wind direction, and dew-point depression. Liquid water inputsbased on annual averages from flat screen fog collectors and foggauges range from 0.2 to 15 l/m2/day (Schemenauer andCereceda, 1991; Larrain et al., 2002; Juvik et al., 2011; Klemmet al., 2012).

Marine fog also increases the relative humidity, suppress-ing transpiration and effectively reducing plant water deficit(Ritter et al., 2009; Fischer et al., 2009; Williams et al., 2008).This effect is also notable in regions with a predominantnocturnal marine fog layer that acts to reduce nighttimetranspiration rates (Dawson et al., 2007; Alvarado-Barrientoset al., 2013). The increase in available water and reduction ofevaporative stress has marked impact on overall ecosystemproductivity. At the canopy level, forest photosynthesis andcarbon uptake is substantially higher during fog events thanon clear, sunny days — a result of diffuse radiation on canopyCO2 (Carbone et al., 2013). At the soil level, increasedmicrobial activity measured as soil respiration also increasesavailable nutrients leading to an overall increase in ecosys-tem productivity (Carbone et al., 2011).

5.4.2. Marine fog — chemical and microbiological impactsFogs have long been observed to carry more than water

resulting in both deleterious and beneficial impacts. The

same mechanisms that can lead to fog formation, such asanticyclone mediated temperature inversion, can trap andconcentrate urban, industrial, and agricultural aerosols to toxiclevels. Despite an extensive literature on nutrient and pollutantdeposition flux from fog (Weathers et al., 1986; Collett et al.,2002) much is still poorly understood about the impact ofmarine fog chemistry and microbiota on humans and ecosys-tems. Concerns over impacts from deposition nitrogen oxidesand anthropogenic sulfur dioxide (which are also found inmarine fogs) have resulted in regulations on industrial andurban emissions (Lynch et al., 2000). Other impacts such asasthmatic or allergic reactions to constituents in marine fogshave not been adequately studied or regulated. In part, thisresults from the complexity of marine chemical compositionand the reactions that occur onmultiple temporal scales aswellas the difficulty of ascribing causal linkage within an epidemi-ological context.

The marine fog air mass is a complex and reaction richaqueous environment that reflects its ocean origin and pathtrajectory. The droplet “core” varies in size depending onchemical composition and can have a variety of surface films(Fig. 11) that are altered over time (Donaldson and Vaida,2006). The interstitial space between fog droplets can containinactivated or hydrophobic dry aerosol particles and aerosolprecursor gases. These include, in varying quantities and forvarious durations, ionic ocean salts, anthropogenic and natu-rally occurring sulfate and organic compounds, atmosphericparticles entrained from above such as wind-blown mineraldust and trace metals, and bioaerosols.

Large differences in reported marine fog water chemistry(Gundel et al., 1994, Northern California; Klemm et al., 1994,New England; Watanabe et al., 2001, Japan; Yue et al., 2012,South China Sea) suggest a strong geographic and temporalresponse of marine fog to ambient conditions. This is alsomirrored in the differences between terrestrial and marine fogchemistry, the latter tending to have higher concentrations ofinorganic ions, lower pH, and lower primary sulfates (Kimballet al., 1988). In contrast, marine fog of longer exposure hashigher concentrations of secondary ammonium sulfate andnitrate (Chow et al., 1996). The bidirectional coastal fluxbetween marine fog and anthropogenic sulfur dioxide (theprecursor of sulfuric acid and sulfate aerosols) from coppersmelters in Chile is an example of inland-sourced aerosolsaffecting marine fog formation and evolution. The reactionsthat occur in the fog airmass as itmoves from the ocean to landsurfaces and toward a dissipationor deposition locationdefinesthe material composition and size and, as a consequence, thetype and level of impact on the receiving body, either organismor landscape unit.

159D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

High chloride ion concentrations from sea spray-originatingCCN result in amuchmore corrosive atmospheric environmentcausing cars to rust faster in seaside cities (Ma et al., 2009). Theimpacts of other ions including nitrogen and sulfur compoundsare beneficial for nutrient limited ecosystems but deleteriouswhen in excess of ecosystem needs. When marine fog scav-enges excess ammonium sulfate, ammonium nitrate, and otherair pollutants, air quality improves however the subsequentdeposition (acid rain) results in deleterious impacts on vegeta-tion and water bodies. Essential plant nutrients transported bymarine fog such as nitrogen, calcium, and to some extent sulfurhave measurable beneficial impacts on the growth of forest,shrub, lichen, and fungal communities (Weathers et al., 2000;Ewing et al., 2009; Maphangwa et al., 2012).

The highly curved surface microlayer of bursting oceanbubbles is an important interface of hydration microphysics.Toxicologically significant pollutants such as chlorinated hydro-carbons, organotin compounds, petroleum hydrocarbons, poly-cyclic aromatic hydrocarbons (PAH), and heavy metals are up to500 timesmore concentrated in this sea-surface microlayer (top1–1000 μm) than in the underlying bulk water column (Wurland Obbard 2004). The flux of these materials into marine fogaerosols and their epidemiological impact is not yet wellquantified.

Airborne particles of biological origin such as fungal spores,bacteria, algae, viruses, pollen, excretions, andbiotic fragments ofvarying size represent a significant fraction of air particles(Després et al., 2012). Experiments in 1769 by Spallanzani todisprove spontaneous generation initiated the field of aerobiol-ogy. Fog temperature, chemical composition, and acidity affectthe concentration of bacteria and yeasts but not of molds (Fuzziet al., 1997). Metabolic activity of aerosol microbes cantransform the chemical constituents of fog aerosols affectingthe microstructure, wettability, and hydration of the aerosols(Deguillaume et al., 2008). The sources and impacts ofmicrobial aerosols are not yet well understood. Urbano et al.(2011) used both culture dependent and independent tech-niques such as DNA clone libraries to examine 55 sequencesobtained from a coastal pier in southern California. Themajorityof these were fungi and bacteria taxa commonly found fromterrestrial sources suggesting a beach or intertidal source.Dueker et al. (2012) used 16S rRNA sequencing on bacterialcultures from ocean and onshore coastal Maine collectionsduring foggy and clear days and found that cultures from foggydays were dominated by ocean surface microbial aerosols andmicrobial viability was enhanced when fog was present.

The epidemiology of airborne material has a rich literature(Fernstrom and Goldblatt, 2013) but the specific flux andconsequence from marine fog is sparse. Many different atmos-pheric aerosols that are also constituents of marine fogs haveknown human health effects such as viruses, cyanobacteria(Genitsaris et al., 2011), dust (Sandstrom and Forsberg, 2008),and heavy metals such as mercury (Tchounwou et al., 2003).Pandemics that can affect large portions of the global populationthat also have an aerosol/droplet component, such as the1918–1920 Spanish Flu, are increasingly investigated using anatmospheric aerosol conceptual model (Fuhrmann, 2010). Theepidemiological requirement to ascertain unambiguous causallinkagemakes the challenge of identifying impacts ofmarine fogon human health more challenging. The inter-connectedness ofthe air–ocean–land–anthropic system at global, mesoscale, and

local scales increases the potential for an environmental burdenassociatedwithmarine fogs. Efforts are underway to investigatethe environmental burden inmarine fog of several categories oftoxicologically damaging constituents such as mercury (WeissPenzias et al., 2012), nitrosamines (Cornell et al., 2003), andother organic compounds (Herckes et al., 2013).

5.4.3. Marine fog — thermodynamic ecosystem impactsMarine fog-related reductions in daytime temperature and

increases innighttime temperaturemoderate climatic conditionsfor coastal ecosystems and reduce extremes in climate variability(Lebassi-Habtezion et al., 2011; Iacobellis and Cayan, 2013). Theimpact of the thermal moderation has wide ranging effects bothfor human communities and natural resources. Marine fogreduction of thermal extremes decreases emergency responseto extreme heat events and human mortality (Gershunov andJohnston, 2011). Energy consumption is reduced in severalsectors such as irrigation pumping and air conditioning therebylowering energy infrastructure costs (Miller et al., 2008;Lebassi et al., 2010). Impacts on coastal wildlife from marinefog are also widespread. Sessile organisms such as intertidalmarine invertebrates experience high mortality during lowtide events coinciding with anomalously low summertime fogconditions (Helmuth et al., 2006).Marine fog blocks shortwaveradiation during California summer months when coastalstream flow is low helping to reduce salmon mortality fromthermal stress. Fisheries managers in the eastern Pacific aregreatly concerned that the 20th century trend (Johnstone andDawson, 2010) of 33% declines in coastal fog could continue.

5.4.4. ConclusionMarine fogs provide many valuable ecosystem services with

most not yet fully quantified. Once considered primarily for thehazard it posed to ship, aviation, and other transportation sec-tors, it is now also recognized for beneficial properties (Bendix etal., 2011). Climatological projections of future marine fogconditions (O'Brien et al., 2013; Haensler et al., 2011; Tseng etal., 2012) are of extreme interest to many social and manage-ment sectors including energy (Gonzalez-Cruz et al., 2013),emergency response (Gershunov and Johnston, 2011), coastalfisheries (King et al., 2011), and biodiversity (Ackerly et al.,2012). Interdisciplinary inquiry will help advance our under-standing of the environmental impacts of marine fog andquantify the ecosystem services provided by marine fog.

6. Microphysics of fog

Most of what is known about fog microphysics pertains toland fog. Nevertheless, it is suspected that results obtainedfrom land-based fog observations are applicable in principleto sea fogs and especially to fogs near the coastline.

Probably the greatest difference between fog and cloudmicrophysics is the greater perceived importance of smallerdroplets in fog compared to most aloft cloud studies. This ispartly due to the greater prevalence of smaller droplets in morepolluted environments that are more frequent at the surfacewhere most of the pollution originates. Another reason is thegreater importance of the 2nd moment of droplet sizes com-pared to higher moments that are more important for manycloud studies; i.e., precipitation, radarmeasurements. This lowermoment, which is directly related to the prime motivation of

160 D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

most fog studies, visibility, attaches more importance to smallerdroplets.

One of the earliest investigations of fog microphysics wasconducted by Pedersen and Todsen (1960). They were thefirst to note bimodal drop size distributions, which were alsooccasionally observed by May (1961) at 1–2 and 15–25 μmdiameters. Garland (1971) found bimodality (Fig. 12) in halfof a more extensive investigation of 25 rural England fogs,where small droplet concentrations exceeded 500 cm−3.

He attributed the bimodality to the difference between thelarger activated droplets and the smaller unactivated droplets.Activated droplet sizes exceeded their critical sizes (at the peakof the Kohler curve of each particle) because the ambientsupersaturation (S) exceeded the critical S (Sc) of the particlesupon which these droplets had formed. Unactivated dropletshad grown on particles with Sc higher than ambient S. Garlandnoted that although the unactivated (haze) droplets did notcontribute much to the liquid water content of the fogs, theirdisproportionate surface area made a greater relative contri-bution to visibility reduction because in nine of the twenty fivefogs these haze droplets contributed 30% of the atmosphericextinction and in five of the fogs they produced 60% of theextinction. Roach et al. (1976) also noted that the activateddroplet mode contributed greatly to the liquid water content(LWC) but little to visibility reduction. Elias et al. (2009) alsofound that small unactivated haze droplets produced themajority of the extinction in polluted Paris fogs. Fig. 13 showslarger fog droplet size distributions in rather unpolluted fogs atAlbany, New York (Fuzzi et al., 1984).

Roach et al. (1976) were the first to point out the importanceof gravitational collection of droplets at the ground, whichconsiderably reduced LWC from 1–2 g m−3 to b0.3 g m−3.

Fig. 12. Droplet concentrationsFrom Garland (1971).

Gultepe et al. (2007b) noted that most fogs have LWC 0.01–0.4 g m−3. Roach et al. (1976) also placedmore importance onmicrophysics by noting that the droplets themselves contrib-ute to the radiative cooling that sustains the fog. AlthoughRoach (1976) predicted fog S of a few hundredths of a percent,estimates of fog S by matching activated mode fog dropletconcentrations (Nc) with CCN concentrations (NCCN)measuredat various S, inferred 0.8% S in rural England fogs (Roach et al.,1976). This type of inference of fog or cloud S by matchingNCCN(S) (CCN spectra) with nearby measured Nc is dubbedeffective S (Seff). With CCN measurements at much lower S byusing an isothermal haze chamber (IHC), Fitzgerald (1978)inferred Seff of 0.055–0.79% in sea fogs off Nova Scotia. Withsimilar equipment, Hudson (1980; Fig. 14) found fog Seff of0.06–0.11% at four different locations along theU.S.West Coast,including one on a ship at sea (Fig. 14b).

These similar Seff in spite of very different NCCN indicated thevalidity of the proportionality between NCCN and measured Nc.Another measurement at San Diego, California in more pollutedconditions showed Seff ≪ 0.04% where the haze droplets alonereduced visibility well below 1 km (Fig. 14). Gultepe et al.(2007b) noted that the distinction between activated and un-activated droplets is much less in more polluted environments.On the other hand, measurements inmountain impacted stratusby Hudson and Rogers (1986) showed higher Seff in the cleanerSan Marcos Pass, California (Seff N 0.1%) than at the morepolluted Henninger Flats (0.02–0.05% Seff), which was immedi-ately downwind of Los Angeles. This seemed to confirmpredictions of Twomey (1959) that Seff should be inverselyrelated to NCCN. Similar aircraft measurements in stratus cloudsoff the US west coast indicated slightly higher Seff of ~0.2%(Hudson, 1983). This was confirmed by more extensive aircraft

in various size intervals.

Fig. 13. Volume percent distribution of fog droplets taken at different times during the October 1, 1982 fog event. The solid line represents the integral percentvolume as a function of droplet size. The data from the Particle Measuring Systems — Forward Scattering Spectrometer Probe 100 are integrated on 12-minperiods. The droplet size interval taken into account is 4–47 μm diameter.From Fuzzi et al. (1984).

161D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

measurements of stratus off the US west coast by Hudson et al.(2010) and Hudson and Noble (2014) where a wider range ofNCCN and Nc definitively showed the expected decrease of Seffwith NCCN predicted by Twomey (1959), with clean stratusshowing Seff N 1%. This was quite different from fog Seff withsimilarly low NCCN and Nc on a ship off the west coast where Seff

Fig. 14. Data showing ambient Nc (×) and simultaneous cumulative drop distributionsthe IHC curves andmove the sizes from r100 (the size in the IHC, which is at 100% RH ordrop distribution but is only used to determine Seff, from the r at the intersection of th≪0.04% (c). Visibility calculated from the ambient fog drop distributionwas 6000 m (a1977; (b) on board the R.V. Wecoma, 30 km off Oregon coast, 0730–0740 PST 29 July 1From Hudson (1980).

was 0.06% (Hudson, 1980). Further measurements with thesame CCN instruments and similar droplet instruments inradiation fog at Albany New York showed Seff of 0.026–0.20%whereas similar measurements in impacted stratus atWhitefaceMountain, New York showed Seff of 0.14–0.35% (Hudson, 1984).Thus, in all comparisons fog seemed to show a lower Seff than

(+)within the isothermal haze chamber (IHC). The third distributions (○) take0% S) to rc (critical radius; rc = 30.5 r100). This curve does not represent any reale × and ○ lines so that Sc = 7.1 × 10−2 / rc, which was 0.06% (a) and (b) and), 1900 m (b), and 670 m (c). (a) YaquinaHead, Oregon, 0121–0125 PST 26 July977; (c) San Diego, CA 0030–0040 PST 27 December 1975.

162 D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

stratus thatwas either detached from the surface or impacted onmountains. This Seff difference between fog and stratus isprobably partly due to the lower vertical velocities of fog due tothe inhibition of motion due to the proximity of the surface.

Meyer et al. (1980) differentiated haze from fog conditionsin that visual range (V) which depends strictly on particleconcentrations in unsaturated haze conditions but changesabruptly to dependence on particle/droplet size at V ~ 1–2 kmwhere activation to fog/cloud conditions occurs (Fig. 15).

In haze conditions, where there are no activated droplets, Vdepended only on particle concentrations and very little onparticle sizes, but in activated fog conditions particle/dropletsize first became equal to particle concentration effects andthen overtook particle concentration dependence (Fig. 15b).The larger sized activated mode tends to persist more than thesmaller sized unactivated mode during fog dissipation stages.

Precipitation often accompanies fog, especially fogs associ-ated with frontal passages. Tardif and Rasmussen (2007) showprecipitation fog as one of the four major fog types along withradiation, advection, and cloud-base-lowering fogs. Fig. 16ashows a fog case with precipitation throughout the fog whileFig. 16b shows precipitation only after the fog.

Fig. 16c shows relationships between fog visibility andprecipitation, which shows that at high precipitation ratesvisibility ismuch lower than estimated from earlier studies andthat visibility decreases much more gradually with precipita-tion rate than the earlier studies. Furthermore, drizzle seemedto have more effect than rain on visibility. The huge scatteringof the data shows that more detail of the drop and droplet sizesare needed to better relate to visibility. Haeffelin et al. (2005)reported that light precipitation contributed to the reduction invisibility in three quarters of non-precipitation types of fogs.

Recent research that has revisited Sean Twomey's inter-esting conjectures of over 50 years ago (Twomey, 1959)gives evidence of the interaction of vertical motion fluctua-tions in stratus (turbulence) and cloud microphysics. Thoughclouds are generally caused by rising air (positive vertical

Fig. 15. Visual range versus (a) cumulative aerosol concentrations and (b) rms diamone standard deviation.From Meyer et al. (1980).

velocity [W]), unlike cumulus clouds, stratus clouds cannothave net positive W because they are usually verticallyconfined. Thus, over sufficient distances mean W is usuallyzero in stratus. This is why the fluctuations of W rather thanmeanW are generally considered in stratus (Peng et al., 2005).Hudson and Noble (2014) showed that droplet concentrations(Nc) were proportional to standard deviations of W (σw) inpolluted stratus clouds where Nc was not correlated with CCNconcentrations (NCCN) as was the case in cleaner air masses.Hudson et al. (2010) demonstrated the suppression of cloudsupersaturation (S) by higher NCCN that had been predicted byTwomey (1959). At these lower values of S, cumulative CCNspectra are generally steeper (i.e., higher k; i.e., greater NCCN

differences per S difference). Twomey (1959) pointed out thatas k increases, Nc switches from predominant dependence onNCCN to predominant dependence onW. So at the higher k thatis more relevant as S is depressed, variations of W rather thanvariations of NCCN become more important for determining Nc

and there is a higher correlation between Nc and W thanbetween Nc and NCCN (Hudson and Noble, 2014). In the case offog, this is σw rather than W. Twomey (1959) said that high kmakes W more important than NCCN for determining Nc, but itis the high NCCN and high k of the low S portion of the CCNspectrum that makes W or σw variations more important thanNCCN variations for determining Nc.

The studies described here affirm the contention in theintroductory paragraphs of the relative importance of smallerdroplets in fogs compared to clouds. Smaller droplets havegreater relative significance for visibility and chemistry,which are more important for fogs than for most cloudstudies. Thus, even unactivated haze droplets take on muchmore significance for fogs compared to clouds. Furthermore,this is exacerbated by the greater losses due to fallout andimpaction of the larger droplets, which is more prevalentcloser to the surface, where there are also obstacles such asvegetation (mainly trees). Fogs can even occur withoutsupersaturation, especially in more polluted environments,

eter squared (mean surface diameter squared). Solid vertical lines indicate

Fig. 16. (a, b) Visibility, RH, T, and PRR time series from observations of two fogs and the results for a proposed parameterization (Vis-RHw), green line isparameterization. (c) Vis-vs-PR relationships for rain and drizzle, shown with red solid lines and red dashed lines, respectively, overlaid on all data points (red dots).From Gultepe et al. (2009).

163D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

where supersaturations are suppressed by competition amongdroplets. The impact of pollution (higher CCN concentrations)seems to be more recognizable in fogs where the smallerdroplets take on more importance and the larger droplets alsofall out.

7. Marine fog forecasting

Two-way communication is essential forweather forecasts—timely receipt of observations at the operational predictioncenter anddelivery of the forecast product to clients. This process

164 D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

was initiated by land telegraph that connected stations acrosscontinents by 1870; undersea cables linked the world by 1900.At sea, long-distance communications began with early radio,which received a big expansion in 1912 with the sinking of theTitanic in fog, followed by a steady stream of technicalimprovements that later included satellites and ultimately theWorld Wide Web.

Weather observations at sea were taken with basicinstruments read by eye and transmitted by radio. This evolvedto automated instruments on ships that recorded informationdigitally and transmitted electronically. The era of satellitemeasurements has made it possible to sample the entire worldocean, beginningwith sea surface temperatures, later includingestimates of winds based on cloud tracking, and verticalthermal structure from radiance soundings, and finally surfacewind based on sea-state structure.

Over the past hundred years, fog forecasts have steadilyimproved. In the early 20th century, only the ship-basedclimatology was available. With the development of aviationduring these same decades, land forecast centers were esta-blished and issued forecasts based onhand-plotted and analyzedmaps, fog-prediction diagrams, and subjective experience of thesynoptic meteorologist. By the beginning of WWII, weatherforecasting was viewed in terms of the synoptic situation orso-called synoptic typing. The evolution of a weather pattern forperiods up to a week was based on the analog method —

matching current weather with a similar pattern in the past andusing the historical pattern as a guide for the current forecast. Asmentioned in the history of sea fog research (Lewis et al., 2004),C. K. M. Douglas had an encyclopedic memory of historicalweather patterns and the D-Day forecast benefitted greatly fromhis memory of past weather. After WWII, the advent of thedigital computer led to computer-based numerical weathermapanalyses that fed into the dynamical prediction models —

essentially objective analyses and forecasts based on the physicsof the atmosphere and the supporting numerical methodsrequired to solve the governing equations. With knowledgethat came from numerical prediction experiments/simulationson the global scale in the 1960s and beyond, operational forecastmodels began to produce products over the oceanic areas by the1970s. Nevertheless, the coarse resolution of the early globalmodels made it impossible to capture the small-scale processesrelated to sea fog initiation and maintenance.

Fig. 17.Marine fog envelops theWest Quoddy Head Lighthouse and a Coast Guard veLabrador Current/Gulf Stream complex. Coastal fog climatology often depends upoPhotos from http://www.photolib.noaa.gov/index.html and http://www.snodoglog

The focus of this section is forecasting fog at sea includingthose situations where the marine influence laps over thecoast. Forecasting is intrinsically perishable — it ceases to beuseful to the operational forecaster after a given timepoint. The“nowcast” is a short-period extrapolation of analyses based oncurrent observations and background information such as aforecast from an earlier time and/or climatology. The longer-range forecast relies on an accurate initial condition and adynamical model. Themerit of a forecast is judged on its abilityto improve on the climatological background state.

In the following, we supply details on some of the majorissues that have faced sea fog forecasting.

7.1. Initial climatology

Basic climatology of marine fog over the world sea wasdeveloped through the international collection of weatherobservations from ships and lighthouses (Fig. 17) starting inthe mid-1800s (see Section 4, Marine fog field programs). Anexample of a California coastal fog climatology based uponship and coastal stations was shown in Fig. 4.

Over time, the number of stations and data volumetransmission expanded — critically important for weatherforecasts on the synoptic scale. Significant weather data trans-mission by telegraph was transcontinental by the later 1800s;transoceanic undersea cableswerewell established by 1900, andship radio was expanding by the 1920s.

Early prediction was based upon fog-prediction diagrams(Petterssen, 1956). Themain client was a coastal airport wherethe forecast relied on surface observations such as airtemperature and dew point depression along with soundingdata that delivered wind shear used to predict the chance offog, its hour of onset, severity (dense/moderate/light fog), andtime of breakup. While Taylor (1917) was among the first touse this technique at sea, forecasts based on these subjectiveprinciples remained in use well into late 20th century untilnumerical fog prediction replaced this approach.

7.2. Surge expansion with aviation

Dramatic expansion of operational weather surface obser-vations so important for the understanding of fog andshort-term forecasts occurred with the initial development of

ssel in Penobscot Bay, both on the Maine coast and under the influence of then lighthouse and ship observations..com/09Summer-Pg5.html.

165D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

commercial aviation, especially air mail and passenger serviceover land in the late 1920s and 1930s (George, 1960, http://celebrating200years.noaa.gov/foundations/aviation_weather/#get). Interest inmarine fog forecasting tended to be restrictedto coastal airports subject to fog that included bays. In theseearly days, there were a disproportionate number of floatplaneoperations due to themajor travel junctions at coastal bays andthe limited number and quality of land runways (Johnson,2009). The forecasting technique tended to be based on acombination of climatology, fog-prediction diagrams and asubjective estimate of the synopticweather trend. The latter forsome stations was a subjective analysis from a central agencyrelying on experience, climatology, surface weather observa-tions and soundings, performed by skilled forecasters such asC.K.M. Douglas and H.H. Lamb in Great Britain (Lewis et al.,2004) and the National Weather Service in the U.S. (http://www.noaa.gov/features/protecting_1208/weatherservice.html). Some forecasters maintained a reputation of superi-ority over machines in forecasting fog long into the era ofcentral, computer-based forecasts, due to the difficultyof such systems in accurately expressing the subsynopticscale, never mind the microscale, in the early stages of theirdevelopment.

More systematic collection of surface observations andballoon soundings at fixed sea locations began with thedevelopment of Ocean Weather Ships (in part the responseto a Pan-American aircraft accident over the Pacific Ocean in1938) and the heightened trans-North Atlantic ship and airtraffic with the escalation of WWII. A dozen Ocean WeatherShip stations were in the north Atlantic and three were in thenorth Pacific starting about 1940 and ending in the 1970s(Adams, 2010).

7.3. After WWII

Following WWII, the weather forecasting industry in bothgovernment and private sectors expanded and flourished. Cen-tral to these expansions were the world wide communicationcircuits that morphed from land, undersea cable and radioteletype to theWorld WideWeb at century's end. In the 1950s–1960s, most surface observations were made by human obser-vers at airports, encoded onto paper forms and transmitted byteletype. By 2000, most of these observations were digitized byautomated weather devices and entered into the World WideWeb. Satellite detection of marine fog lagged, but has improvedwith use of visual and IR bands as well as satellite-basedsoundings for enhanced measurement of the synoptic structureoverwater. By the late 20th century, the expanded coverage thatcame with numerical weather prediction included all of theworld's temperate latitudes (see Section 7.4). The forecast of lowcloudandproperties of themarine layer (including fog) still needconsiderable development.

Although numerical weather prediction's developmentwasinitially slow following WWII, advances in both computingpower and theories of atmospheric processes led to improvedmodels and faster execution times by the mid-1960s. By the1980s, satellite-based measurement of SST, clouds and tem-perature structure through inversion of radiance measure-ments led to improved initial conditions for the numericalmodels — especially noticeable over the data-sparse oceans.

7.4. Growth of operational numerical models

The main approach to marine fog forecasting starts withthe larger scale motions. Output from the large-scale modelsgenerally provides boundary conditions for the subsynopticand mesoscale models that cover a coherent geographical areasuch as a specific coast or a small section of an ocean and theadjacent area. The horizontal resolution of these local modelsapproaches 10 km or better, although upper-air observationscannot resolve structures at this resolution. The benefits of fogforecasts that use output from these small-scale models inconcert with climatology have received some attention (Wells,2007), but the sample size is limited.

Many coastal stations, such as U.S. Gulf Coast stations,forecast fog based upon a conceptual decision-tree processthat is a mix of climatology (large scale and station specifics),observations (local surface, local sounding, satellite imagery),centrally produced numerical guidance and analyses, localmodel diagnostic software, and qualitative assessment ofcloud condensation nuclei (Croft et al., 1997). This combina-tion of factors is used to forecast fog initiation, visibility, andbreakup of fog.

Bartok et al. (2012) present a recent example of the use of ahigh resolution, 3-dimensional numericalmodel (WRF) coupledwith a one-dimensional fogmodel that employs boundary layerprocesses and parameterized microphysics to forecast theoccurrence of fog for road traffic on the north coast of theUnited Arab Emirates with significant skill. The fogs are formedat night from Persian Gulf marine air advected over land.Comparisons of a forecast with corresponding satellite imageare shown in Fig. 18, where a correct forecast is when fog isforecasted and occurs (27 out of 84 cases), a correct negativeforecast is when fog is forecasted to not occur and does not (16out of 84 cases), and a false alarm is when fog is forecasted butdoes not (19 out of 84 cases). Not shown is the case when fogoccurs when not forecasted, which happened only in a smallproportion of the cases studied (5 out of 84 cases).

It has been suggested that improvements could be madefrom better integration of operationally available surface basedmeasurements, remotemeasurements (satellite, ground basedprofilers), and numerical model outputs (Ellrod, 1995; Isaac etal., 2006). Another example is Zhou et al. (2007) who proposetwo different approaches that use an operational numericalmodel — one improves the probability of fog occurrence whilethe other also improves liquid water content which is requiredfor visibility forecasts.

A variation of working on a numerical weather predictionsystem itself is to base fog prediction on multi-mesoscalemodels related in subgroupings to form systems (Zhou et al.,2007; Zhou and Du, 2010). An advantage is that this allows arelative check on how individual models and various groupingsperform on deterministic and probabilistic forecasts. Ensemble-based forecasts are significantly better than those based onindividual models.

Another alternative to sea fog forecasting comes with thepattern recognition method (a statistical methodology asopposed to a dynamic prediction methodology). Among thesestatistical methods is the classification and regression tree(CART) method — a decision tree-building technique (Lewis,2000). The advantage of this approach is that when theavailable data sources, including numerical models, are not

Fig. 18. Comparison pairs of fog forecast (left, WRF model, relative humidity in units of percent, gray scale with white = 100%; arrows are 10-m flow streamlines)with corresponding satellite image (right, EUMET satellite) for north coast of the United Arab Emirates at time and date posted in upper left. White areas denotefoggy regions. See text for discussion.Figure adapted from Bartok et al. (2012).

166 D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

strongly related to fog formation, combinations of indirectfactors can sometimes improve the forecast. Lewis (2000)applied this technique to fog forecasting at the Kunsan Air Basein Korea. His model input included sea surface temperatures,land-based surface observations, upper-air soundings, andoutput from a numerical weather prediction model. Forecast-ing for fog has been also conducted using artificial neuralnetworks (e.g., Fabbian et al., 2007) and for cloud ceiling andvisibility using fuzzy logic (e.g., Hansen, 2007) and neuralnetworks (Marzban et al., 2007).

7.5. Satellite development and expansion

The analysis of current observations is the essential compo-nent of the “nowcast” — the prediction on the order of severalhours. As expected, satellite observations are crucial for thenowcast over the oceanic areas. The visible imagery cangenerally detect thick cloud, but thin cloudy areas includingfoggy areas are difficult to detect — essentially transparentduring the daylight hours. The infrared radiance measurementsat night generally fail to differentiate between the temperature ofthe fog top and the temperature of the sea surface. Nevertheless,work is underway to differential these temperature based onradiance differences in the shorter- and longer-waves in theinfrared spectrum (Eyre et al., 1984; Ellrod, 1995). A filteringprocess helps identify those image pixels with partial coverageby fog. Operational algorithms for daytime detection of fog andlow stratus have been proposed by Bendix et al. (2006) (TerraMODIS) and Cermak and Bendix (2007) (Meteosat SEVIRI).

Satellite-based fog detection over a range of oceanic scalesis paramount to improving the fog forecasts at sea (Ellrodand Gultepe, 2007). Bendix et al. (2006) have proposed a fogdetection algorithm for the MODIS instrument that includes

channels in the near infrared that reportedly can detect fog to500 m resolution.

7.6. The elusive sub-mesoscale and microscale

A major obstacle to operational numerical prediction of fogis the inability to directly incorporate fogmicrophysics into themodel. In part, this is due to increased computational demandsthat come with the microphysics parameterization (Müller etal., 2007; Gultepe et al., 2006c; Tardif and Rasmussen, 2010).Further, as presented and discussed in Section 6 (Microphysicsof marine fog), equations governing the microphysics of fogdroplets are not easily linked with the equations that govern amesoscale model. That is, the parameterization is not straight-forward and generally involves serious assumptions. A lot ofuncertainty is also present in unknown initial and boundaryconditions of aerosols. Another challenge is to optimally couplethe background climatology with the observations. Thus, onboth the large- and small-scale, factors critical to fog and itsevolution are not easily incorporated into a forecast system(Stoelinga and Warner, 1999; Gultepe et al., 2006a).

There has been considerable work on the relationshipsbetween the sub-mesoscale or microscale and fog. To give animpression of the complexities missed by operational models,some of thework on treatment of physical processes and needsfor their parameterizations in models follows.

Fog occurs in aerosol-laden surface air with high relativehumidity, ranging from undersaturated to slightly supersatu-rated (Pruppacher and Klett, 1997); it is a mixture of micron-size haze (unactivated) particles and activated particlesreaching 10s of microns in size (Pinnick et al., 1978; Hudson,1980; Gerber, 1981). Fog'smicrostructure and life cycle dependon the properties of aerosols (Bott, 1991), as does superstation(Pilié et al., 1975; Gerber, 1991). Fog droplets are generally

167D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

smaller than cloud droplets; fog liquid water content isgenerally small, and most fog liquid water content rangesfrom 0.01 to 0.4 g m−3 (Gultepe et al., 2007b).

Liquid water content is related to droplet concentration(Gerber, 1981, 1991; Garcia-Garcia et al., 2002; Fuzzi et al.,1992), gravitational settling of larger drops (Bott et al., 1990),and droplet size (Jiusto, 1981). Liquid water content has beenempirically related to extinction coefficients (Eldridge, 1971;Tomasi and Tampieri, 1976; Kunkel, 1984; Gultepe et al.,2006a) while liquid water content times the droplet numberconcentration has been related to visibility (Gultepe et al.,2006a), an important forecasting characteristic.

The role of small-scale turbulence in fog has been inves-tigated by Zdunkowski and Barr (1972), Turton and Brown(1987), and Musson-Genon (1987). Turbulence exchange coef-ficients in the nocturnal boundary layer with fog have beenexplored in Turton and Brown (1987) and with a second-orderturbulence closure in Nakanishi and Niino (2004, 2006).

The cooling of moist air by radiative flux divergence hasbeen analyzed by Duynkerke (1991). Another complexity isthat clouds above the surface layer increase the downwardlongwave radiation, reducing the longware radiation loss at thetop of the fog which is important to fog dynamics (Gultepe etal., 2007a).

The role and importance of advection terms and their role infog formation and evolution have been shown by Guédalia andBergot (1994). Related is that large-eddy simulations showdistinct flow regimes in different stages of fog layer evolution(Nakanishi, 2000).

Fog has been explored using models of different dimen-sions. Local surface measurements were assimilated into a1-dimensional fog model (Bergot et al., 2005). Oliver et al.(1978) used a second-order closure model to investigateturbulence radiation properties in fog. Koračin et al. (2001,2005b) used a 1Dmodel in a Lagrangian framework and also a3-dimensional model (Koračin et al., 2005a) to simulateinversion and cloud forcing leading to fog. Bott et al. (1990)employed a 2-dimensional fog model to examine the effects offog microphysics and radiation processes. Ballard et al. (1991)and Pagowski et al. (2004) used 3-dimensional models toexplore fog variations. On the other hand, Bretherton et al.(1999), Duynkerke et al. (1999) and Teixeira (1999) usedsingle-column versions of existing 3-D models that have beenused for fog studies. Muller (2006) developed a 1-D variationassimilation scheme for surface observations and coupled two1-Dmodels with several operational 3-Dmodels to produce anensemble forecast. Zhou and Du (2010) have developed amultimodel mesoscale ensemble prediction system for fog andshowed that the ensemble-based forecasts are in generalsuperior to the single control forecasts.

As noted in the previous paragraph, a major restriction forfog forecasting improvement was that forecasting is donelargely through operational numerical models that do not dealdirectly with the fog physics but employ coefficients andparameterizations. To improve this, the field project FRAMincluded data from coastal and continental sites (Gultepe et al.,2006a; Gultepe and Milbrandt, 2009; Toth et al., 2010). Themarine phase took place in the summer of 2006 along the NovaScotia Atlantic coast. Extensivemeasurementsweremade of thelower atmosphere, including variables known to be importantfor fog but not operationally available such as aerosol size and

concentration, cloud-base height, droplet size, droplet numberconcentration, liquid water content, liquid water path, radiativefluxes, vapor mixing ratio, precipitation type and intensity, 3Dwind speed and turbulence, and wind profiler with RASS whilesatellite observations were included from GOES, MODIS Terra.The data can be used to develop and improve microphysicalparameterizations which will be incorporated into numericalforecast models. An example parameterization is that ofvisibility versus the inverse of the liquid water content timesthe cloudwater drop number concentration (Nd) (Gultepe et al.,2006b). Another example is the concentrationweighted particleterminal velocity times liquid water content versus a functionbased upon the liquid water content and Nd.

Future improvement of world-wide forecasting of marinefog will be based upon better observations of fog that will mostlikely be through technical developments of satellite based,remotely sensed systems such as hyperspectral channels toprovide a basis to improve detection algorithms (Ellrod andGultepe, 2007). Better remotely sensed observations will, inturn, provide a basis for comparison with large scale opera-tional numerical models to deal with their lack of sensitivity inthe smaller scale, near sea surface conditions to better handlefog forecasting.

It is expected that fog forecasting will become at leastsomewhat more realistic via algorithm development to makeup for unmeasured variables that are essential throughout thelife of fog (i.e., drop size distribution, condensation nuclei,sub-grid scale motions and structure) as well as to have a basisto forecast useful but previously unapproachable variables thatcharacterize fog conditions such as visibility and fog depth.

In recent years there has been strong interest in the evolutionof open sea and coastal zone fog from a climatological per-spective. While global models are still operating with coarsehorizontal and vertical resolutions, regional climatemodels (e.g.,O'Brien et al., 2013) are showing some promise in addressingthis issue.

8. Remote sensing of marine fog

Satellite imagery is used to identify low clouds and fog bothin conjunction with in-situ field studies (Gultepe et al., 2009)and operationally by meteorological services (Molenar et al.,2000). The standard approach to detecting fog relies ondetection of passive infrared emission or scattering from thecloud-top. This is achieved by evaluating the difference betweenthe narrow-band brightness temperature observed in theinfraredwindow region of the spectrum (typically awavelengthin the 10–12 μm range) and a brightness temperature observedin the near-infrared (typically in the 3–4 μm range). Theemissivity of both increaseswith cloud depth, but the differenceis relatively uniform for clouds thicker than approximately100 m (Ellrod, 1995). For low, liquid water clouds, the infraredwindow channel emissivity is greater than the near-infraredemissivity, leading to a positive difference (window channelminus near-infrared channel) in the observed brightnesstemperatures that increases with cloud thickness (Ellrod,1995). At night, the difference is small (generally less than2 °C) for clear-sky conditions and increases to as much asapproximately 5 °C for thick stratus or stratocumulus clouds.Thin cirrus clouds yield a negative brightness temperaturedifference that distinguishes them from lower liquid water

168 D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

stratus clouds. Fig. 19 shows an operational application of thetechnique applied to GOES 15 geostationary satellite imagerydepicting nighttime low clouds and fog intruding on the SanFrancisco Bay Area. In addition to the presence of fog, depth ofthe fog based upon the brightness temperature difference(Ellrod, 1995) is also estimated operationally.

During the day, the signal of reflected near-infraredradiation from the sun overwhelms the emission signal.However, the near-infrared reflectance of low liquid watercloud is substantially brighter than the reflectance of surface orcirrus clouds, which generally yields a large negative bright-ness temperature difference useful for daytime detection of fogand low stratus clouds (Fig. 20).

The infrared brightness temperature difference techniqueis applicable over the ocean and a wide variety of land surfacetypes. It has been applied in a range of fog studies (Ellrod,1995; Lee et al., 1997; Bendix, 2002).

One fundamental limitation of this approach is that it doesnot precisely discriminate cloud near the surface from higher-level stratocumulus or altostratus clouds. Higher-level cloudsmay either obscure fog below or be mistaken for low-levelclouds. Furthermore, even for low clouds, the infrared emissionsignatures of the cloud measured from above yield littleinformation regarding the proximity of the cloud base to thesurface. One solution to this problem is combining satellite datawith information about surface conditions. Gultepe et al.(2007a) report that successful fog detection rates using satelliteimagery compared against surface monitoring stations inCanada are only between 0.26 and 0.32, largely owing to thepresence of mid- or high-level clouds. However, they improve

Fig. 19. Low clouds and fog (yellow pixels) intruding on the San Francisco Bay Ar10.7 μm brightness temperature measured by GOES 15 geostationary satellite. Cirrtemperature difference.Image obtained from the NOAA Satellite and Information Service (http://www.star

the rate to between 0.55 and 1.00 by using numerical weatherprediction model-derived near-surface temperature estimates.They report a false alarm rate of only 0.10. Similarly, Zhang andYi (2013) use a climatology of sea surface temperature todiscriminate fog from low stratus clouds over seas adjacent toChina. They find that the difference between the temperature atthe fog top and the sea surface temperature differs betweencases of fog from cases of low stratus cloud. Hence themonthly-mean climatology of SST is used to determine a dynamicthreshold on infrared brightness temperature useful for detect-ing fog.

Daytime application of the brightness temperature differ-ence technique is also complicated by large variations innear-infrared illumination from the sun through the day. Datafrom Lee et al. (1997) documents the large daytime variation inthe brightness temperature difference through the day, anddemonstrate that the daytime detection of fog and low cloud ismade more reliable by first estimating the near-infraredreflectance, which is a considerably more stable quantitythrough the day, and using the high reflectance of low stratusclouds to distinguish fog and low cloud from less reflectivecirrus, ocean, or land surface. Nevertheless, further ambiguity inthe brightness temperature difference is present in the dawnand dusk hours, when theweak signal in reflected near-infraredradiation renders fog indistinguishable from the surface in thebrightness temperature difference. Lee et al. (2011)make use ofthe cloud-free visible reflectance derived from 15 days of priorsatellite imagery to identify scenes that are brighter than thecloud-free case and distinguish brighter fog in the visible bandfrom darker surface. Based on the daytime and nighttime

ea at 0300 LST July 22, 2013 determined from the difference of 3.9 μm andus clouds (blue and black pixels) exhibit the opposite sign of the brightness

.nesdis.noaa.gov/smcd/opdb/aviation/fog.html).

Fig. 20. Infrared window channel brightness temperature (“longwave”) andnear-infrared channel brightness temperature (“shortwave”) from GOES-9satellite imagery illustrating the change in sign of the brightness temper-ature difference at sunset.From Lee et al. (1997).

169D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

brightness temperature difference and the additional dawn/dusk criterion, Lee et al. (2011) present a smooth 24-hour fogdetection algorithm.

The limited visibility caused by fog is a consequence of themicrophysical aspects of the cloud. Detection of the potentialpresence of fog using satellites is valuable, but furtherinterpretation of the potential impact on visibility relies onretrieval of the microphysical properties of the clouds. Thereflectance of near-infrared radiation, which has so far beendiscussed as a means of detecting daytime fog, is alsosubstantially dependent upon the size of the cloud droplets. Acloud composed of smaller drops enhances the reflectance ofnear-infrared radiation relative to a cloud with larger drops.Near-infrared reflectance, when combined with a measure ofvisible reflectance (for example at 0.64 μm), has been used toperform a combined retrieval of cloud optical thickness andcloud droplet effective radius (Nakajima and King, 1990).These data are now produced routinely using imager data suchas that from the MODIS instrument (Platnick et al., 2003).Wetzel et al. (1996) compare retrievals of effective radius and

optical thickness for a fog case over California against in situprofiles of fog droplet sizes. Based on a favorable agreement,they discuss the potential to estimate the surface visual rangein the presence of fog based on some assumptions about thevertical profile of the drop sizes. MODIS retrievals for anotherfog case are presented in Gultepe et al. (2009). Retrievals ofcloud drop number concentration are also made by satellitewith some success (Rausch et al., 2010), which could beapplied to translating satellite observations to surface visibilityestimates. This retrieval, however, also relies on an assumptionfor the vertical structure — in this case the adiabatic model forthe vertical variability of LWC and drop effective radius.

Active remote sensing of fog can eliminate some of theambiguity of satellite remote sensing, but at present is limitedto select surface sites, and hence does not offer the globalcoverage of satellite data. Visible light is strongly attenuated byclouds, therefore lidar technology deployed from the surface isuseful for detecting cloud base, such as with operationalceilometers. However, such technology cannot profile a foglayer. Suborbital aircraft and satellite lidar systems can identifycloud top heights with substantially greater precision thanpassive infrared imaging techniques, but are similarly unable todetermine whether the cloud base reaches the surface.

Radar systems have been used to study the structure andevolution of fog layers. Operational weather radars operating atcentimeter wavelengths are designed to detect precipitationsized cloud drops, but the smaller drops typical of fog layers donot effectively scatter radar signals at these wavelengths.Experimental cloud radars, however, have been deployed at35 GHz and 95 GHz frequencies to study the layer thicknessand vertical structure of cloud layers. Hamazu et al. (2003)describe a scanning Doppler radar system at 35 GHz and use itto describe the variability of reflectivity within a sea fogcase. Gultepe et al. (2009) and Boers et al. (2013) evaluate theprospects for determining surface visibility from radar reflec-tivity. While relationships between the two quantities areapparent (Fig. 21), Boers et al. (2013) conclude that thevisibility–reflectivity relationship varies as the fog layer evolves.

A 94 GHz satellite cloud radar, CloudSat, is in orbit,offering the prospect of global fog studies. However, weaksensitivity and clutter attributable to the surface returncomplicates the observation of cloud properties near thesurface by space-borne radar.

Operational techniques for satellite remote sensing of seafog based on passive visible and infrared imagery are nowroutinely applied to monitoring fog events. These tech-niques work best at night and are limited in their ability todistinguish fog at the surface from ordinary low stratusclouds. Active remote sensing of fog by radar showspotential for retrieving visibility estimates in fog layers, butis so far limited to experimental cases rather than wide-spread monitoring.

9. Epilog

Viewed historically, we have been investigating fog at seafor exactly 100 years if we label G.I. Taylor's monumental workof 1913 as the initial point in time.What dowe know about thelife cycle of sea fog commencing with this early investigation?First of all, the factors identified by Taylor have stood the test oftime — Lagrangian trajectories of initially warm/moist air that

Fig. 21. Visibility against cloud radar reflectivity (“Z”) for samples from theFRAM field campaign based in Ontario, Canada.From Gultepe et al. (2009).

170 D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

is cooled from below by colder ocean temperatures and mixedvertically by velocity-shear turbulence. Since then, we alsoknow that turbulence in an unstably stratified air mass (warmocean temperatures relative to the overlying air) very effec-tively mixes the air upward and entrains air from atop themixed layer. Further, wemore clearly understand thewarmingand cooling of the fog layer in response to the processes of bothlong- and short-wave radiation; and the “capping lid”—marineinversion atop the fog layer is generally a response tosubsidence associated with the semi-permanent anticyclonesover the oceans. These are among the processes that have beenthoroughly investigated and we have some confidence thattheir action is understood.

What do not we know about the cycle of sea fog?Certainly we are ignorant of the complex interactions of thevarious processes, some of which were mentioned above.And part of the dilemma rests on our inability to observethese processes individually, let alone collectively. And thisshould come as no surprise since the phenomenon exhibitsextreme scales — a range of ~1013.

Among the issues that are poorly understood are thefollowing:

– Accurate estimates/predictions of subsidence elude us.– Models cannot maintain a strong inversion.– Many model parameterizations have been calibrated for

different conditions (frequently over the land).– Models with coarse resolution cannot fully represent

coastal local circulations.– Ocean input to atmospheric models is generally too

coarse.– Over-ocean measurements are sparse and non-existent in

many areas.– Model coupling (atmosphere–ocean) and surface fluxes

are frequently approximated.– Initial state, composition, and history of air masses

including fog condensation nuclei are generally unknown.– Condensation triggering for various subsaturation and

supersaturation scenarios is difficult to represent in modelswith limited condensation parameterizations.

– Detailed microphysics and precipitation are usually sim-plified in model parameterizations.

– Physical processes relevant to marine fog in Lagrangian andEulerian frameworks need to be more closely examined.

– Fog is an elusive target for airborne field programs;airborne measurements in fog are limited by the FederalAviation Authority (FAA) and other agencies.

– Climatology of marine air characteristics is poorly knownover many regions of the world sea.

Improvement in our knowledge of sea fog will undoubtedlycome from detection of sea fog through satellite observations.While limitations persist in the quantitative informationattainable about fog from satellites, the vastly improvedspatio-temporal sampling of the oceans afforded by satellitesand the demonstrated ability of satellite detection of fog eventssuggests that many more fog events can be identified andstudied with the aid of satellites.

Of course, we will gain much frommeasurements of sea fogin field programs that incorporate observations from ship andcoastal stations. The future trend will emphasize programs thatamalgamate observations in the hope of clarifying the interac-tions of processes. Field measurements will be combined withmodel output to obtain the most accurate representation of asystem state that stretches across the scales from synoptic tomicrophysical.

From the many modeling studies of sea fog, essentiallynumerical experiments/simulations/forecasting that started inthe immediate post WWII period, it becomes clear thatdeterministic forecasting of sea fog onset and its duration hasgenerally been unsuccessful. The extreme sensitivity of modeloutput to elements of control [initial conditions, boundaryconditions, and forcing (empirical/physical parameterization)],in concert with the chaotic nature of dynamic prediction, is at theheart of prediction inaccuracy. Ensemble prediction is a possibil-ity, but it comes with complications when applied to sea fog. Thecomplications arise because the phenomenon is discontinuouswith an impulsive start and an abrupt end. The fundamentals ofensemble prediction applied todiscontinuousdynamical systemssuch as this one are in their infancy.Without doubt, however, thisarea of investigation is needed and promising.

When will we be able to make accurate forecasts of sea fogonset and its duration? At this time, we are unable to answerthis question with any degree of certainty. However, we knowfrom the history of geophysical science that improvements inunderstanding and forecasting come incrementally withdependence on better observations (both temporal andspatial) that lead to improved four-dimensional analyseswhich in turn lead to improved dynamical forecasts. And notonly from models, but advances will come from the individualscientists (observationalists, analysts, and theoreticians) withtheir phenomenological viewpoints — viewpoints that lead toconjectures or hypotheses thatwhen followed to their terminalpoints contribute to the incremental advance.

Acknowledgments

Two of the authors (Koračin and Lewis) acknowledgesupport from the Office of Naval Research grant N00014-

171D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

00-1-0524. Funding for this project was partially provided byNASA UCSD contract 20053743 (Dorman), DOE ASR DE-SC0009162 (Hudson), and (Torregrosa) US Geological SurveyClimate and Landuse Change Program, Gordon and Betty MooreFoundation Grant #2861, and California Landscape ConservationCollaborative. We also want to express our gratitude to the lateProfessor Dale Leipper, a protégé of Harald Sverdrup at Scripps inthe late 1940s–early 1950s,whowas an inspiration to those of uswho had the privilege of working with him. All authors aregrateful toMr. Travis McCord of the Desert Research Institute forhis thorough editorial efforts.

References

Ackerly, D.D., Ryals, R.A., Cornwell, W.K., Loarie, S.R., Veloz, S., Higgason, K.D.,Silver, W.L., Dawson, T.E., 2012. Potential Impacts of Climate Change onBiodiversity and Ecosystem Services in the San Francisco Bay Area.CaliforniaEnergy Commission (Publication number: CEC-500-2012-037).

Adams, Michael R., 2010. Ocean Station: Operations of the U.S. Coast Guard,1940–1977.Nor'Easter Press, Eastpoint, Maine978-0-9779200-1-3.

Ahuja, M.R., 2009. Genetic constitution and diversity in four narrow endemicredwoods from the family Cupressaceae. Euphytica 165 (1), 5–19.

Alvarado-Barrientos, M.S., Asbjornsen, H., Holwerda, F., 2013. Is nighttimetranspiration enhanced after fog events? In IX International Workshopon Sap Flow 991, pp. 133–139.

Anderson, J., 1931. Observations from airplanes of cloud and fog conditionsalong the southern California coast. Mon. Weather Rev. 59, 264–270.

Ångström, A., 1920. Applications of heat radiation measurements to problemsof the evaporation from lakes and heat convection at their surfaces. Georg.Ann. 2, 237–252.

Azevedo, J., Morgan, D.L., 1974. Fog precipitation in coastal California forests.Ecology 55 (5), 1135–1141.

Ballard, S., Golding, B., Smith, R., 1991.Mesoscale model experimental forecastsof the Haar of northeast Scotland. Mon. Weather Rev. 191, 2107–2123.

Barron, J.A., Anderson, L., 2011. Enhanced Late Holocene ENSO/PDO expressionalong the margins of the eastern North Pacific. Quat. Int. 235 (1), 3–12.

Bartok, B., Bott, A., Gera, M., 2012. Fog prediction for road traffic safety in acoastal desert region. Bound.-Layer Meteorol. 145 (3), 485–506.

Bendix, J., 2002. A satellite-based climatology of fog and low-level stratus inGermany and adjacent areas. Atmos. Res. 64, 3–18.

Bendix, J., Thies, B., Nauss, T., Cermak, J., 2006. A feasibility study of daytimefog and low stratus detection with TERRA/AQUA-MODIS over land.Meteorol. Appl. 13, 111–125.

Bendix, J., Eugster, W., Klemm, O., 2011. Fog—boon or bane? Erdkunde 65(3), 229–232.

Bergot, T., Carrer, D., Noilhan, J., Bougeault, P., 2005. Improved sit-specificnumerical prediction of fog and low clouds: a feasibility study. WeatherForecast. 20, 627–646.

Bergthorsson, P., 2000. The Wineland Millennium.Mal og menning,Reykjavik, Iceland (304 pp.).

Boers, R., Baltink, H.K., Hemink, H.J., Bosveld, F.C., Moerman, M., 2013. Ground-based observations andmodeling of the visibility and radar reflectivity in aradiation fog layer. J. Atmos. Ocean. Technol. 30, 288–300.

Bott, A., 1991. On the influence of the physico-chemical properties of aerosolson the life cycle of radiation fogs. Bound.-Layer Meteorol. 56, 1–31.

Bott, A., Trautmann, T., 2002. PAFOG—a new efficient forecast model of radiationfog and low-level stratiform clouds. Atmos. Res. 64 (1–4), 191–203.

Bott, A., Sievers, U., Zdunkowski, W., 1990. A radiation fog model with adetailed treatment of the interaction between radiative transfer and fogmicrophysics. J. Atmos. Sci. 47, 2153–2166.

Bowen, I.S., 1926. The ratio of heat losses by conduction and evaporationfrom water surfaces. Phys. Rev. 27, 779–787.

Bretherton, C.S., Krueger, S.K., Wyant, M.C., Bechtold, P., Van Meijgaard, E.,Stevens, B., Teixeira, J., 1999. A GCSS boundary-layer cloud modelintercomparison study of the first ASTEX Lagrangian experiment.Bound.-Layer Meteorol. 93 (3), 341–380.

Bruijnzeel, L.A., Eugster, W., Burkard, R., 2005. Fog as a hydrologic input. In:Anderson, M.G., McDon nell, J.J. (Eds.), Encyclopedia of HydrologicalSciences, 1. John Wiley & Sons, Inc, Chichester, pp. 559–582.

Burgess, S.S.O., Dawson, T.E., 2004. The contribution of fog to the waterrelations of Sequoia sempervirens (D. Don): foliar uptake and preventionof dehydration. Plant Cell Environ. 27 (8), 1023–1034.

Byers, H.R., 1930. Summer sea fogs of the central California coast. Univ. Calif.Publ. Geogr. 3 (5), 291–338.

Carbone, M.S., Still, C.J., Ambrose, A.R., Dawson, T.E., Williams, A.P., Boot,C.M., Schimel, J.P., 2011. Seasonal and episodic moisture controls onplant and microbial contributions to soil respiration. Oecologia 167(1), 265–278.

Carbone, M.S., Park Williams, A., Ambrose, A.R., Boot, C.M., Bradley, E.S.,Dawson, T.E., Schaeffer, S.M., Schimel, J.P., Still, C.J., 2013. Cloud shadingand fog drip influence the metabolism of a coastal pine ecosystem. Glob.Chang. Biol. 19, 484–497. http://dx.doi.org/10.1111/gcb.12054.

Cereceda, P., Larrain, H., Osses, P., Farías, M., Egaña, I., 2007. The spatial andtemporal variability of fog and its relation to fog oases in the AtacamaDesert, Chile. Atmos. Res. 87 (3–4), 312–323. http://dx.doi.org/10.1016/jatmosres2007.11.012.

Cermak, J., 2012. Low clouds and fog along the south-western African coast—satellite-based retrieval and spatial patterns. Atmos. Res. 116, 15–21.

Cermak, J., Bendix, J., 2007. Dynamical nighttime fog/low stratus detectionbased on Meteosat SEVIRI data: a feasibility study. Pure Appl. Geophys.164, 1179–1192.

Cho, Y.-K., Kim, M.-O., Kim, B.-C., 2000. Sea fog around the Korean Peninsula.J. Appl. Meteorol. 39, 2473–2479.

Chow, J.C., Watson, J.G., Lowenthal, D.H., Countess, R.J., 1996. Sources andchemistry of PM10 aerosol in Santa Barbara County, CA. Atmos. Environ.30 (9), 1489–1499.

Collett Jr., J.L., Bator, A., Sherman, D.E., Moore, K.F., Hoag, K.J., Demoz, B.B.,Rao, X., Reilly, J.E., 2002. The chemical composition of fogs andintercepted clouds in the United States. Atmos. Res. 64, 29–40.

Corbin, J.D., Thomsen, M.A., Dawson, T.E., D'Antonio, C.M., 2005. Summerwater use by California coastal prairie grasses: fog, drought, andcommunity composition. Oecologia 145 (4), 511–521.

Cornell, S.E., Jickells, T.D., Cape, J.N., Rowland, A.P., Duce, R.A., 2003. Organicnitrogen deposition on land and coastal environments: a review ofmethods and data. Atmos. Environ. 37 (16), 2173–2191.

Croft, P.J., Darbe, D.L., Garmon, J.F., 1995. Forecasting significant fog insouthern Alabama. Natl Weather Dig. 19, 10–16.

Croft, P.J., Pfost, R.L., Medlin, J.M., Johnson, G.A., 1997. Fog forecasting forthe southern region: a conceptual model approach. Weather Forecast.12, 545–556.

Dawson, T.E., 1998. Fog in the California redwood forest: ecosystem inputsand use by plants. Oecologia 117 (4), 476–485.

Dawson, T.E., Burgess, S.S., Tu, K.P., Oliveira, R.S., Santiago, L.S., Fisher, J.B.,Simonin, K.A., Ambrose, A.R., 2007. Nighttime transpiration in woodyplants from contrasting ecosystems. Tree Physiol. 27 (4), 561–575.

Deguillaume, L., Leriche, M., Amato, P., Ariya, P.A., Delort, A.M., Pöschl, U., Morris,C.E., 2008. Microbiology and atmospheric processes: chemical interactions ofprimary biological aerosols. Biogeosci. Discuss. 5 (1), 841–870.

Després, V.R., Huffman, J.A., Burrows, S.M., Hoose, C., Safatov, A.S., Buryak, G.,Jaenicke, R., 2012. Primary biological aerosol particles in the atmo-sphere: a review. Tellus B 64.

Díaz, S.C., Touchan, R., Swetnam, T.W., 2001. A tree-ring reconstruction ofpast precipitation for Baja California Sur, Mexico. Int. J. Climatol. 21,1007–1019. http://dx.doi.org/10.1002/joc.664.

Donaldson, D.J., Vaida, V., 2006. The influence of organic films at the air–aqueous boundary on atmospheric processes. Chem. Rev. 106 (4),1445–1461.

Dorman, C.E., Armi, L., Bane, J.M., Rogers, D.P., 1998. Sea surface mixed layerduring the 10–11 June 1994 California coastally trapped event. Mon.Weather Rev. 126, 600–619.

Douglas, C., 1930. Cold fogs over the sea. Meteorol. Mag. 65, 133–135.Dueker, M.E., O'Mullan, G.D., Weathers, K.C., Juhl, A.R., Uriarte, M., 2012.

Coupling of fog and marine microbial content in the near-shore coastalenvironment. Biogeosciences 9 (2), 803–813.

Duynkerke, P.G., 1991. Radiation fog: a comparison of model simulationswith detailed observations. Mon. Weather Rev. 119, 324–341.

Duynkerke, P.G., Jonker, P.J., Chlond, A., Van Zanten, M.C., Cuxart, J., Clark, P.,Sanchez, E., Martin, G., Lenderink, G., Teixeira, J., 1999. Intercomparisonof three- and one-dimensional model simulations and aircraft observa-tions of stratocumulus. Bound.-Layer Meteorol. 92, 453–487.

Eldridge, R.G., 1971. The relationship between visibility and liquid watercontent of fog. J. Atmos. Sci. 28, 1183–1186.

Elias, T., Haeffelin,M., Drobinski, P., Gomes, L., Rangognio, J., Bergot, T., Chazette,P., Raut, J.-C., Colomb, M., 2009. Particulate contribution to extinction ofvisible radiation: pollution, haze, and fog. Atmos. Res. 92, 443–454.

Eller, C.B., Lima, A.L., Oliveira, R.S., 2013. Foliar uptake of fog water andtransport belowground alleviates drought effects in the cloud foresttree species, Drimys brasiliensis (Winteraceae). New Phytol. 199,151–162.

Ellison, G.B., Tuck, A.F., Vaida, V., 1999. Atmospheric processing of organicaerosols. J. Geophys. Res. 104 (D9), 11633-11.

Ellrod, G.P., 1995. Advances in the detection and analysis of fog at nightusing GOES multi-spectral infrared imagery. Weather Forecast. 10,606–619.

172 D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

Ellrod, G.P., Gultepe, I., 2007. Inferring low cloud base heights at night foraviation using satellite infrared and surface temperature data. In: Fogand boundary layer clouds: fog visibility and forecasting. BirkhäuserBasel, pp. 1193–1205.

Emmons, G., Montgomery, R., 1947. Note on the physics of fog formation.J. Meteorol. 4, 206.

Ewing, H.A., et al., 2009. Fog water and ecosystem function: heterogeneity ina California redwood forest. Ecosystems 12, 417–433.

Eyre, J.R., Brownscombe, J.L., Allan, R.J., 1984. Detection of fog at night usingAdvanced High Resolution Radiometer (AVHRR) imagery. Meteorol.Mag. 113, 66–271.

Fabbian, D., deDear, R., Lellyett, S., 2007. Application of artificial neural networkforecasts to predict fog at Canberra International Airport. WeatherForecast. 22, 372–381.

Fernstrom, A., Goldblatt, M., 2013. Aerobiology and its role in the transmissionof infectious diseases. J. Pathog. 2013, 13. http://dx.doi.org/10.1155/2013/493960.

Filonczuk, M.K., Cayan, D.l.R., Riddle, L.G., 1995. Variability of marine fogalong the California coast. SIO-Reference, No 95-2. Climate ResearchDivision, Scripps Institution of Oceanography, University of California,San Diego (http://www.escholarship.org/uc/item/2kc7x97f#page-1).

Findlater, J., Roach, W., McHugh, B., 1989. The haar of north-east Scotland. Q.J. R. Meteorol. Soc. 115, 581–608.

Fischer, D.T., Still, C.J., Williams, A.P., 2009. Significance of summer fog andovercast for drought stress and ecological functioning of coastalCalifornia endemic plant species. J. Biogeogr. 36, 783–799.

Fitzgerald, J.W., 1978. A numerical model of the formation of droplet spectrain advection fogs at sea and its applicability to fogs off Nova Scotia.J. Atmos. Sci. 35, 1522–1535.

Ford, K.W., 1968. Basic Physics.Blaisdell Publishing Company (968 pp.).Fuhrmann, C., 2010. The effects of weather and climate on the seasonality of

influenza: what we know and what we need to know. Geogr. Compass 4(7), 718–730.

Fuzzi, S., Castillo, R.A., Jiusto, J.E., Lala, G., 1984. Chemical composition ofradiation fog water at Albany, New York, and its relationship to fogmicrophysics. J. Geophys. Res. 89, 7159–7164.

Fuzzi, S., Facchini, M.C., Orsi, G., Lind, J.A., Wobrock, W., Kessel, M., Maser, R.,Jaeschke, W., Enderle, K.H., Arends, B.G., Berner, A., Solly, I., Kruisz, C.,Reischl, G., Pahl, S., Kaminski, U., Winkler, P., Ogren, J.A., Noone, K.J.,Hallberg, A., Fierlinger-Oberlinninger, H., Puxbaum, H., Marzorati, A.,Hansson, H.-C., Wiedensohler, A., Svenningsson, I.B., Martinsson, B.G.,Schell, D., Georgii, H.W., 1992. The Po Valley fog experiment 1989. Tellus B44, 448–468. http://dx.doi.org/10.1034/j.1600-0889.1992.t01-4-00002.

Fuzzi, S., Mandrioli, P., Perfetto, A., 1997. Fog droplets—an atmosphericsource of secondary biological aerosol particles. Atmos. Environ. 31 (2),287–290.

Gao, S., Lin, H., Shen, B., Fu, G., 2007. A heavy sea fog event over the YellowSea in March 2005: analysis and numerical modeling. Adv. Atmos. Sci.24, 65–81.

Garcia-Garcia, F., Virafuentes, U., Montero-Martinez, G., 2002. Fine-scalemeasurements of fog-droplet concentrations: a preliminary assessment.Atmos. Res. 64, 179–189.

Garland, J.A., 1971. Some fog droplet size distributions obtained by animpaction method. Q. J. R. Meteorol. Soc. 97, 483–494.

Garmon, J.F., Darbe, D.L., Croft, P.J., 1996. Forecasting significant fog on theAlabama coast: impact climatology and forecast checklist development.NWS Technical Memorandum NWS SR-176 (16 pp.).

Garreaud, R., Barichivich, J., Christie, D.A., Maldonado, A., 2008. Interannualvariability of the coastal fog at Fray Jorge relict forests in semiarid Chile.J. Geophys. Res. Biogeosci. 113 (G4) (2005–2012).

Genitsaris, S., Kormas, K.A., Moustaka-Gouni, M., 2011. Airborne algae andcyanobacteria: occurrence and related health effects. Front. Biosci. 3,772–787.

George, J.J., 1960. Weather Forecasting for Aeronautics.Academic Press297–372.

Gerber, H.E., 1981. Microstructure of a radiation fog. J. Atmos. Sci. 38,454–458.

Gerber, H.E., 1991. Supersaturation and droplet spectral evolution in fog.J. Atmos. Sci. 48, 2569–2588.

Gershunov, A., Johnston, Z., 2011. The California heat wave 2006 with impactson statewide medical emergency: a space–time analysis. GeographyResearch Forum, vol. 31, pp. 6–31.

Gioda, A., Maley, J., Guasp, R.E., Baladón, A.A., 1995. Some low elevation fogforests of dry environments: applications to African paleoenvironments.Tropical Montane Cloud Forests. Springer, US, pp. 156–164.

González, A.L., Fariña, J.M., Pinto, R., Pérez, C., Weathers, K.C., Armesto, J.J.,Marquet, P.A., 2011. Bromeliad growth and stoichiometry: responses toatmospheric nutrient supply in fog-dependent ecosystems of the hyper-arid Atacama Desert, Chile. Oecologia 167 (3), 835–845.

Gonzalez-Cruz, J., Sequera, P., Molina, Y., Picon, R., Pillich, J., Ghebreegziabhe,A.T., Bornstein, B., 2013. 3.01— Climate and energy vulnerability in coastalregions: the case for US Pacific and northeast corridor coastal regions. In:Pielke, Roger (Ed.), Climate Vulnerability. Academic Press, Oxford,pp. 3–35. http://dx.doi.org/10.1016/B978-0-12-384703-4.00302-6.

Grell, G.A., Dudhia, J., Stauffer, D.R., 1994. A description of the fifth-generationPenn State/NCAR mesoscale model (MM5). NCAR Tech. Note NCAR/TN-398 + STR (122 pp.).

Guédalia, D., Bergot, T., 1994. Numerical forecasting of radiation fog. Part II: acomparison of model simulations and several observed fog events. Mon.Weather Rev. 122, 1231–1246.

Gultepe, I., Milbrandt, J.A., 2009. Probabilistic parameterizations of visibilityusing observations of rain precipitation rate, relative humidity, andvisibility. J. Appl. Meteorol. 49, 36–46.

Gultepe, I., Isaac, G., MacPherson, I., Marcotte, D., Strawbridge, K., 2003.Characteristics of moisture and heat fluxes over leads and polynyas, andtheir effect on Arctic clouds during FIRE. ACE Atmos. Ocean 41, 15–34.

Gultepe, I., Muller, M.D., Boybeyi, Z., 2006a. A new warm fog parameteriza-tion scheme for numerical weather prediction models. J. Appl. Meteorol.45, 1469–1480.

Gultepe, I., Müller, M.D., Boybeyi, Z., 2006b. A new visibility parameteriza-tion for warm fog applications in numerical weather prediction models.J. Appl. Meteorol. 45, 1469–1480.

Gultepe, I., Cober, S.G., King, P., Isaac, G., Taylor, P., Hansen, B., 2006c. The fogremote sensing and modeling (FRAM) field project and preliminaryresults. AMS 12th Cloud Physics Conference, July 9–14, 2006, Madison,Wisconsin, USA (Print in CD, P4.3).

Gultepe, I., Pagowski, M., Reid, J., 2007a. A satellite-based fog detectionscheme using screen air temperature. Weather Forecast. 22, 444–456.

Gultepe, I., Tardif, R., Michaelides, S.C., Cermak, J., Bott, A., Bendix, J., Muller,M.D., Pagowski, M., Hansen, B., Ellrod, G., Jacobs, W., Toth, G., Cober, S.G.,2007b. Fog research: a review of past achievements and futureperspectives. Pure Appl. Geophys. 164, 1121–1159.

Gultepe, I., Pearson, G., Milbrandt, J.A., Hansen, B., Platnick, S., Taylor, P.,Gordon, M., Oakley, J.P., Cober, S.G., 2009. The fog remote sensing andmodeling field project. Bull. Am. Meteorol. Soc. 90, 341–359.

Gundel, L.A., Benner, W.H., Hansen, A.D.A., 1994. Chemical composition offog water and interstitial aerosol in Berkeley, California. Atmos. Environ.28 (16), 2715–2725.

Gutiérrez, A.G., Barbosa, O., Christie, D.A., del-Val, E., Ewing, H.A., Jones, C.G.,Marquet, P.A., Weathers, K.C., Armesto, J.J., 2008. Regeneration patternsand persistence of the fog dependent Fray Jorge forest in semiarid Chileduring the past two centuries. Glob. Change Biol. 14, 161–176. http://dx.doi.org/10.1111/j.1365-2486.2007.01482.x.

Haeffelin, M., Barthès, L., Bock, O., Boitel, C., Bony, S., Bouniol, D., Chepfer, H.,Chiriaco, M., Cuesta, J., Delanoe, J., Drobinski, P., Dufresne, J.L., Flamant, C.,Grall, M., Hodzic, A., Hourdin, F., Lapouge, F., Lemaitre, Y.,Mathieu, A.,Morille,Y., Naud, C., Noel, V., OH'irok, B., Pelon, J., Pietras, C., Protat, A., Romand, B.,Scialom, G., Vautard, R., 2005. SIRTA, a ground-based atmospheric observa-tory for cloud and aerosol research. Ann. Geophys. 23, 253–275.

Haensler, A., Cermak, J., Hagemann, S., Jacob, D., 2011.Will the southern Africanwest coast fog be affected by future climate change? Results of an initial fogprojection using a regional climate model. Erdkunde. 261–275.

Hamazu, K., Hashiguchi, H., Wakayama, T., Matsuda, T., Doviak, R.J., Fukao, S.,2003. A 35-GHz scanning Doppler radar for fog observations. J. Atmos.Ocean. Technol. 20, 972–986.

Hansen, 2007. A fuzzy logic based analog forecasting system for ceiling andvisibility. Weather Forecast. 22, 1319–1330.

Helmuth, B., Broitman, B.R., Blanchette, C.A., Gilman, S., Halpin, P., Harley,C.D., O'Donnell, M.J., Hofmann, G.E., Menge, B., Strickland, D., 2006.Mosaic patterns of thermal stress in the rocky intertidal zone:implications for climate change. Ecol. Monogr. 76 (4), 461–479.

Heo, K.-Y., Ha, K.-J., 2010. A coupled model study on the formation anddissipation of sea fogs. Mon. Weather Rev. 138, 1186–1205.

Herckes, P., Valsaraj, K.T., Collett Jr., J.L., 2013. A review of observations oforganic matter in fogs and clouds: origin, processing and fate. Atmos.Res. 132, 434–449.

Heusser, L., 1998. Direct correlation of millenial-scale changes in westernNorth American vegetation and climate with changes in theCalifornia Current system over the past 60 kyr. Paleoceanography13 (3), 252–262.

Hiatt, C., Fernandez, D., Potter, C., 2012. Measurements of fog waterdeposition on the California central coast. Atmos. Clim. Sci. 2, 525–531.

Hodur, R.M., 1997. The Naval Research Laboratory's Coupled Ocean/Atmo-sphere Mesoscale Prediction System (COAMPS). Mon. Weather Rev. 125,1414–1430.

Hodur, R.M., Hong, X., Doyle, J.D., Pullen, J., Cummings, J., Martin, P., Rennic,M.A., 2002. The Coupled Ocean/Atmosphere Mesoscale PredictionSystem (COAMPS). Oceanography 15, 88–98 (Wash DC).

173D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

Huang, H., Liu, H., Jiang, W., Huang, J., Mao, W., 2011. Characteristics of theboundary layer structure of sea fog on the coast of Southern China. Adv.Atmos. Sci. 28 (6), 1377–1389.

Hudson, J.G., 1980. Relationship between fog condensation nuclei and fogmicrostructure. J. Atmos. Sci. 37, 1854–1867.

Hudson, J.G., 1983. Effects of CCN on stratus clouds. J. Atmos. Sci. 40 (480),486.

Hudson, J.G., 1984. CCNmeasurements within clouds. J. Clim. Appl. Meteorol.23 (42), 51.

Hudson, J.G., Noble, S., 2014. CCN and vertical velocity influences on dropletconcentrations and supersaturations in clean and polluted stratusclouds. J. Atmos. Sci. 71, 312–331.

Hudson, J.G., Rogers, C.F., 1986. Relationship between critical supersatura-tion and cloud droplet size: implications for cloud mixing processes.J. Atmos. Sci. 43 (2341), 2359.

Hudson, J.G., Noble, S., Jha, V., 2010. Stratus cloud supersaturations. Geophys.Res. Lett. 37, L21813. http://dx.doi.org/10.1029/2010GL045197.

Iacobellis, S.F., Cayan, D.R., 2013. The variability of California summertimemarine stratus: impacts on surface air temperatures. J. Geophys. Res.http://dx.doi.org/10.1002/jgrd.50652 (Online Access published July, 19,2013).

Ingraham, N.L., Matthews, R.A., 1995. The importance of fog-drip water tovegetation—Point Reyes peninsula, California. J. Hydrol. 164 (1–4), 269–285.

Isaac, G.A., Bailey,M., Cober, S.G., Donaldson,N., Driedger, N., Glazer, A., Gultepe,I., Hudak, D., Korolev, A., Reid, J., Rodriguez, P., Strapp, J.W., Fabry, F., 2006.Airport vicinity icing and snow advisor. AIAA 44th Aerospace Sci. Meetingand Exhibit, Reno Nevada, 9–12 January 2006, AIAA-2006-1219.

Jeffreys, H., 1918. Some problems of evaporation. Phil. Mag. 35 (6th series),270–280.

Jiusto, J.E., 1981. Fog structure. In: Hobbs, P.V., Deepak, A. (Eds.), Clouds,Their Formation, Optical Properties and Effect. Academic Press (495 pp.).

Johnson, E.R., 2009. American Flying Boats and Amphibious Aircraft: AnIllustrated History.McFarland, Jefferson, North Carolina (376 pp.).

Johnson, G.A., Graschel, J., 1992. Sea fog and stratus: a major aviation andmarine hazard in the northern Gulf of Mexico. Symp. on WeatherForecasting. Amer. Meteor. Soc., Atlanta, GA, pp. 55–60.

Johnstone, J.A., Dawson, T.E., 2010. Climatic context and ecologicalimplications of summer fog decline in the coast redwood region. PNAS2010. http://dx.doi.org/10.1073/pnas.0915062107.

Joslin, J.D., Mueller, S.F., Wolfe, M.H., 1990. Tests of models of cloudwaterdeposition to forest canopies using artificial and living collectors. Atmos.Environ. 24A, 3007–3019 (journal.pone.0034603).

Juvik, J.O., DeLay, J.K., Kinney, K.M., Hansen, E.W., 2011. A 50th anniversaryreassessment of the seminal ‘Lana'i fog drip study’ in Hawai'i. Hydrol.Process. 25 (3), 402–410.

Katata, G., Nagai, H., Kajino, M., Ueda, H., Hozumi, Y., 2010. Numerical studyof fog deposition on vegetation for atmosphere–land interactions insemi-arid and arid regions. Agric. For. Meteorol. 150 (3), 340–353.

Kim, C.-K., Yum, S.-S., 2010. Local meteorological and synoptic characteristicsof fogs formed over Incheon international airport in the west coast ofKorea. Adv. Atmos. Sci. 27, 761–776.

Kim, C.-K., Yum, S.-S., 2012. A numerical study of sea-fog formation over coldsea surface using a one-dimensional turbulence model coupled with theweather research and forecasting model. Bound.-Layer Meteorol. 143,481–505.

Kimball, K.D., Jagels, R., Gordon, G.A., Weathers, K.C., Carlisle, J., 1988.Differences between New England coastal fog andmountain cloud waterchemistry. Water Air Soil Pollut. 39 (3–4), 383–393.

King, J.R., Agostini, V.N., Harvey, C.J., McFarlane, G.A., Foreman, M.G.G.,Overland, J.E., Di Lorenzo, E., Bond, N.A., Aydin, K.Y., 2011. Climate forcingand the California Current ecosystem. ICES J. Mar. Sci. 68, 1199–1216.

Klemm, O., Bachmeier, A.S., Talbot, R.W., Klemm, K.I., 1994. Fog chemistry atthe New England coast: influence of air mass history. Atmos. Environ. 28(6), 1181–1188.

Klemm, O., Schemenauer, R.S., Lummerich, A., Cereceda, P., Marzol, V., Corell,D., Fessehaye, G.M., 2012. Fog as a fresh-water resource: overview andperspectives. Ambio 41 (3), 221–234.

Kloesel, K.A., 1992. A 70-year history of marine stratocumulus cloud fieldexperiments off the coast of California. Bull. Am. Meteorol. Soc. 73,1581–1585.

Koračin, D., Lewis, J., Thompson, W.T., Dorman, C.E., Businger, J.A., 2001.Transition of stratus into fog along the California coast: observations andmodeling. J. Atmos. Sci. 58, 1714–1731.

Koračin, D., Businger, J.A., Dorman, C.E., Lewis, J.M., 2005a. Formation,evolution, and dissipation of coastal sea fog. Bound.-Layer Meteorol. 117,447–478.

Koračin, D., Leipper, D.F., Lewis, J.M., 2005b. Modeling sea fog on the U.S.California coast during a hot spell event. Geofizika 22, 59–82.

Kunkel, B.A., 1984. Parameterization of droplet terminal velocity andextinction coefficient in fog models. J. Clim. Appl. Meteorol. 23, 34–41.

Lamb, H., 1943. Haars or North Sea Fogs on the Coasts of GreatBritain.Meteorology Office Publication, M. O. 504.

Larrain, H., Velasquez, F., Cereceda, P., Espejo, R., Pinto, R., Osses, P., Schemenauer,R.S., 2002. Fog measurements at the site ‘Falda Verde’ north of Chan˜aralcompared with other fog stations of Chile. Atmos. Res. 64, 273–284.

Lebassi, B., González, J.E., Fabris, D., Bornstein, R., 2010. Impacts of climatechange in degree days and energy demand in coastal California. J. Sol.Energy Eng. 132 (3) (031005-1).

Lebassi-Habtezion, B., González, J., Bornstein, R., 2011. Modeled large scalewarming impacts on summer California coastal cooling trends.J. Geophys. Res. 116, D20114. http://dx.doi.org/10.1029/2011JD015759.

Lee, T.F., Turk, F.J., Richardson, K., 1997. Stratus and fog products using GOES-8-9 3.9-μm data. Weather Forecast. 12, 664–677. http://dx.doi.org/10.1175/1520-0434(1997)012b0664:SAFPUGN2.0.CO;2.

Lee, J.-R., Chung, C.-Y., Ou, M.L., 2011. Fog detection using geostationarysatellite data: temporally continuous algorithm. Asia–Pac. J. Atmos. Sci.47, 113–122.

Leem, H.H., Kim, D.H., Song, K.W., 2005. The importance of fog prediction at theIncheon Int'l Airport through flight delay and cancel statistics. J. Korean Soc.Aeronaut. Space Sci. 13, 1–10 (in Korean with English abstract).

Leipper, D., 1948. Fog development at San Diego, California. J. Mar. Res. 7,337–346.

Leipper, D.F., 1994. Fog on the U.S. West Coast, a review. Bull. Am. Meteorol.Soc. 72, 229–240.

LePage, B.A., Williams, C.J., Yang, H. (Eds.), 2005. The Geobiology and Ecologyof Metasequoia, vol. 22. Springer.

Lewis, J.M., 1996. Winds over the world sea: Maury and Köppen. Bull. Am.Meteorol. Soc. 77, 935–952. http://dx.doi.org/10.1175/1520-0477.

Lewis, R.J., 2000. An Introduction to Classification and Regression Tree(CART) Analysis.Department of Emergency Medicine, Harbor-UCLAMedical Center, Torrance, CA.

Lewis, J.M., Koračin, D., Rabin, R., Businger, J., 2003. Sea fog off the Californiacoast: viewed in the context of transient weather systems. J. Geophys.Res. Atmos. 108 (D15), 4457. http://dx.doi.org/10.1029/2002JD002833.

Lewis, J.M., Koračin, D., Redmond, K.T., 2004. Sea fog research in the UnitedKingdom and United States: a historical essay including outlook. Bull.Am. Meteorol. Soc. 85, 395–408.

Lewis, J.M., Fearon, M.G., Klieforth, H.E., 2012. Herbert Riehl: intrepid andenigmatic scholar. Bull. Am. Meteorol. Soc. 93, 963–985. http://dx.doi.org/10.1175/BAMS-D-11-00224.1.

Limm, E.B., Simonin, K.S., Bothman, A.G., Dawson, T.E., 2009. Foliar wateruptake: a common water acquisition strategy for plants of the redwoodforest. Oecologia 161, 449–459.

Lovett, G.M., 1984. Rates and mechanisms of cloud water deposition to asubalpine balsam fir forest. Atmos. Environ. 18, 361–371.

Lundquist, J., Bourcy, T.B., 1999. California and Oregon Humidity and CoastalFog SIO, Reference No. 99-17.

Lynch, J.A., Bowersox, V.C., Grimm, J.W., 2000. Changes in sulfate depositionin eastern USA following implementation of Phase I of Title IV of theClean Air Act Amendments of 1990. Atmos. Environ. 34, 1665–1680.

Ma, Y., Li, Y., Wang, F., 2009. Corrosion of low carbon steel in atmosphericenvironments of different chloride content. Corros. Sci. 51 (5), 997–1006.

Madej, M.A., Currens, C., Ozaki, V., Yee, J., Anderson, D.G., 2006. Assessingpossible thermal rearing restrictions for juvenile coho salmon (Onco-rhynchus kisutch) through thermal infrared imaging and in-streammonitoring, Redwood Creek, California. Can. J. Fish. Aquat. Sci. 63,1384–1396. http://dx.doi.org/10.1139/F06-043.

Maphangwa, K.W., Musil, C.F., Raitt, L., Zedda, L., 2012. Differential interceptionand evaporation of fog, dewandwater vapour and elemental accumulationby lichens explain their relative abundance in a coastal desert. J. AridEnviron. 82, 71–80.

Marzban, C., Leyton, S., Colman, B., 2007. Ceiling and visibility forecasts vianeural nets. Weather Forecast. 22 (3), 466–479.

Maury, M.F., 1861. Physical Geography of the Sea and Its Meteorology.May, K.R., 1961. Fog droplet sampling using a modified impactor technique.

Q. J. R. Meteorol. Soc. 87, 534–548.Means, H.R., 1927. Fog precipitated by trees. Science 66, 402–403.Meyer, M.B., Jiusto, J.E., Lala, G.G., 1980. Measurements of visual range and

radiation-fog (haze) microphysics. J. Atmos. Sci. 37, 622–629.Miller, N.L., Hayhoe, K., Jin, J., Auffhammer,M., 2008. Climate, extremeheat, and

electricity demand in California. J. Appl. Meteorol. Climatol. 47, 1834–1844.Molenar, D.A., Schrab, K.J., Purdom, J.F.W., 2000. RAMSDIS contributions to

NOAA satellite data utilization. Bull. Am. Meteorol. Soc. 81, 1019–1030.Muller, M.D., 2006. Numerical Simulation of Fog and Radiation in Complex

Terrain. (Ph.D. Thesis) Stratus, 12. University of Basel (103 pp.).Müller, M.D., Schmutz, C., Parlow, E., 2007. A one-dimensional ensemble

forecast and assimilation system for fog prediction. J. Pure Appl. Geophys.164, 1241–1264 (Special issue).

Musson-Genon, L., 1987. Numerical simulation of a fog event with a one-dimensional boundary layer model. Mon. Weather Rev. 115, 592–607.

174 D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

http://dx.doi.org/10.1175/1520-0493(1987)115b0592:NSOAFEN2.0.CO;2.

Nakajima, T., King, M.D., 1990. Determination of the optical thickness andeffective particle radius of clouds from reflected solar radiationmeasurements. Part I: theory. J. Atmos. Sci. 47, 1878–1893.

Nakanishi, M., 2000. Large-eddy simulation of radiation fog. Boundary LayerMeteorol. 94, 461–493.

Nakanishi, M., Niino, H., 2004. An improved Mellor–Yamada level-3 modelwith condensation physics: its design and verification. Boundary LayerMeteorol. 112, 1–31.

Nakanishi, M., Niino, H., 2006. An improved Mellor–Yamada level-3 model:its numerical stability and application to a regional prediction ofadvection fog. Boundary Layer Meteorol. 119, 397–407.

Neiburger, M., Johnson, D.S., Chien, C.W., 1961. Studies of the Structure of theAtmosphere Over the Eastern Pacific Ocean in Summer.University ofCalifornia Press (94 pp.).

Noonkester, V.R., 1979. Coastal marine fog in southern California. Mon.Weather Rev. 107, 830–851.

Nuss, W.A., et al., 2000. Coastally trapped wind reversals: progress towardunderstanding. Bull. Am. Meteorol. Soc. 81, 719–743. http://dx.doi.org/10.1175/1520-0477(2000)081b0719:CTWRPTN2.3.CO;2.

O'Brien, T.A., Sloan, L.C., Chuang, P.Y., Faloona, I.C., Johnstone, J.A., 2013.Multidecadal simulation of coastal fog with a regional climate model.Clim. Dyn. 40, 2801–2812.

Oberlander, G.T., 1956. Summer fog precipitation on the San Franciscopeninsula. Ecology 37, 851–852.

Økland, H., Gotaas, Y., 1995. Modelling and prediction of steam fog. Beitr.Phys. Atmos. 68, 121–131.

Oliver, D., Lewellen, W., Williamson, G., 1978. The interaction betweenturbulent and radiative transport in the development of fog and low-level stratus. J. Atmos. Sci. 35, 301–316.

Pagowski, M., Gultepe, I., King, P., 2004. Analysis andmodeling of an extremelydense fog event in Southern Ontario. J. Appl. Meteorol. 43, 3–16. http://dx.doi.org/10.1175/1520-0450(2004)043b0003:AAMOAEN2.0.CO;2.

Palmer, A.K., 1917. Fog along the California Coast. Mon. Weather Rev. 45, 496.Pedersen, K., Todsen, M., 1960. Some measurements of the microstructure of

fog and stratus in the Oslo area. Geophys. Pub., 21. Norske VidenskapsAkad, Oslo, pp. 1–15.

Peng, Y., Lohmann, U., Leaitch, W.R., 2005. Importance of vertical velocityvariations in the cloud droplet nucleation process of marine stratus clouds.J. Geophys. Res. D21213. http://dx.doi.org/10.1029/2004JD004922.

Petterssen, S.V., 1936. On the causes and forecasting of the California fog.J. Aerosp. Sci. 3, 305–309.

Petterssen, S.V., 1938. On the causes and forecasting of the California fog.Bull. Am. Meteorol. Soc. 19, 49–55.

Petterssen, S., 1939. Some aspects of formation and dissipation of fog.Geofys. Publ. 12, 15–22.

Petterssen, S., 1956. Weather analysis and forecasting. Weather andWeatherSystems, vol. II. McGraw-Hill, New York (266 pp.).

Pilié, R.J., Mack, E.J., Kocmond, W.C., Rogers, C.W., Eadie, W.J., 1975. The lifecycle of valley fog. Part I: micrometeorological characteristics. J. Appl.Meteorol. 14, 347–363.

Pilié, R.J., Mack, E.J., Rogers, C.W., Katz, U., Kocmond, W.C., 1979. Theformation of marine fog and the development of fog-stratus systemsalong the California coast. J. Appl. Meteorol. 18, 1275–1286.

Pinnick, R.G., Hoihjelle, D.L., Fernandez, G., Stenmark, E.B., Lindberg, J.D.,Hoidale, G.B., 1978. Vertical structure in atmospheric fog and hazeand its effects on visible and infrared extinction. J. Atmos. Sci. 35,2020–2032.

Platnick, S., King, M.D., Ackerman, A., Menzel, W.P., Baum, B.A., Riédi, J.C.,Frey, R.A., 2003. The MODIS cloud products: algorithms and examplesfrom Terra. IEEE Trans. Geosci. Remote Sens. 41, 459–473.

Poole, I., van Bergen, P.F., 2006. Physiognomic and chemical characters inwood as palaeoclimate proxies. Plants and Climate Change. Springer,Netherlands, pp. 175–196.

Pruppacher, H.R., Klett, J.D., 1997. Microphysics of Clouds and Precipitation2nded. Kluwer Pub. Inc., Boston (954 pp.).

Rausch, J., Heidinger, A., Bennartz, R., 2010. Regional assessment of microphysicalproperties of marine boundary layer cloud using the PATMOS-x dataset.J. Geophys. Res. 115, D23. http://dx.doi.org/10.1029/2010JD014468.

Riehl, H., Yeh, T.C., Malkus, J., La Seur, N., 1951. The northeast trade of thePacific Ocean. Q. J. R. Meteorol. Soc. 77, 598–626.

Ritter, A., Regalado, C.M., Aschan, G., 2009. Fog reduces transpiration in treespecies of the Canarian relict heath-laurel cloud forest (GarajonayNational Park, Spain). Tree Physiol. 29 (4), 517–528.

Roach, W.T., 1976. On the effect of radiative exchange on the growth bycondensation of a cloud or fog droplet. Q. J. R. Meteorol. Soc. 102,361–372.

Roach, W.T., 1995. Back to basics: fog: part 3—the formation and dissipationof sea fog. Weather 50, 80–84.

Roach, W., Brown, R., Caughey, S.J., Garland, J.A., Readings, C.J., 1976. Thephysics of radiation fog: I—a field study. Q. J. R. Meteorol. Soc. 102,313–333.

Roden, J.S., Johnstone, J.A., Dawson, T.E., 2009. Intra-annual variation in thestable oxygen and carbon isotope ratios of cellulose in tree rings of coastredwood (Sequoia sempervirens). The Holocene 19 (2), 189–197.

Roth-Nebelsick, A., Ebner, M., Miranda, T., Gottschalk, V., Voigt, D., Gorb, S.,Stegmaier, T., Sarsour, M., Linke, J., Konrad, W., 2012. Leaf surfacestructures enable the endemic Namib desert grass Stipagrostis sabulicolato irrigate itself with fog water. J. R. Soc. Interface. 9, 1965–1974. http://dx.doi.org/10.1098/rsif. 2011.0847.

Sandstrom, T., Forsberg, B., 2008. Desert dust: an unrecognized source ofdangerous air pollution? Epidemiology 19 (6), 808–809.

Saunders, P.M., 1964. Sea smoke and steam fog. Q. J. R. Meteorol. Soc. 90,156–165.

Schemenauer, R.S., Cereceda, P., 1991. Fog-water collection in arid coastallocations. Ambio 303–308.

Schemenauer, R.S., Cereceda, P., 1994. Fog collection's role in water planningfor developing countries. Natural Resources Forum, 18. Wiley, p. 2.

Shchepetkin, A.F., McWilliams, J.C., 2004. The regional oceanic modelingsystem: a split-explicit, free-surface, topography-following-coordinateocean model. Ocean Modell. 9, 347–404.

Simonin, K.S., Santiago, L.S., Dawson, T.E., 2009. Fog interception by Sequoiasempervirens (D. Don) crowns decouples physiology from soil waterdeficit. Plant Cell Environ. 32, 882–892.

Skamarock, W., Klemp, J.B., Dudhia, J., Gill, D.O., Barker, D., Duda, M.G.,Huang, X.-Y., Wang, W., 2008. A description of the advanced researchWRF version 3. NCAR Technical Note NCAR/TN-475 + STR. http://dx.doi.org/10.5065/D68S4MVH.

Slinn, W.G.N., 1982. Predictions for particle deposition to vegetativecanopies. Atmos. Environ. 16, 1785–1794.

Spiess, F., 1985. In: Emery, W.J. (Ed.), The Meteor Expedition: ScientificResults of the German Atlantic Expedition. Amerind Pub. Co (429 pp.).

Stoelinga, M.T., Warner, T.T., 1999. Nonhydrostatic, mesobeta-scale modelsimulations of cloud ceiling and visibility for an east coast winterprecipitation event. J. Appl. Meteorol. 38, 385–404.

Sugimoto, S., Sato, T., Nakamura, K., 2013. Effects of synoptic-scale control onlong-term declining trends of summer fog frequency over the Pacificside of Hokkaido Island. J. Appl. Meteorol. Climatol. 52 (10), 2226–2242.

Taljaard, J.J., Schumann, T.E.W., 1940. Upper air temperatures and humiditiesat Walvis Bay, southwest Africa. Bull. Am. Meteorol. Soc. 21, 293–296.

Tardif, R., Rasmussen, R.M., 2007. Event-based climatology and typology of fogin the NewYork City region. J. Appl. Meteorol. Climatol. 46 (8), 1141–1168.

Tardif, R., Rasmussen, R.M., 2008. Process-oriented analysis of environmen-tal conditions associated with precipitation fog events in the New YorkCity region. J. Appl. Meteorol. Climatol. 47, 1681–1703.

Tardif, R., Rasmussen, R.M., 2010. Evaporation of non-equilibrium raindropsas a fog formation mechanism. J. Atmos. Sci. 67 (2), 345–364.

Taylor, G.I., 1915. Eddymotion in the atmosphere. Philos. Trans. R. Soc. Lond. Ser. A215, 1–26 (Containing Papers of a Mathematical or Physical Character).

Taylor, G.I., 1917. The formation of fog andmist. Q. J. R.Meteorol. Soc. 43, 241–268.Taylor, W., 1987. Mesoscale case study: project haar. Met Office Met. O.11

Tech. Note 239.Tchounwou, P.B., Ayensu, W.K., Ninashvili, N., Sutton, D., 2003. Review:

environmental exposure to mercury and its toxicopathologic implica-tions for public health. Environ. Toxicol. 18 (3), 149–175.

Teixeira, J., 1999. Simulation of fog with the ECMWF prognostic cloudscheme. Q. J. R. Meteorol. Soc. 125, 529–553.

Telford, J.W., Chai, S.K., 1993. Marine fog and its dissipation over warmwater. J. Atmos. Sci. 50, 3336–3349.

Thompson, W.T., Burk, S.D., Lewis, J., 2005. Fog and low clouds in a coastallytrapped disturbance. J. Geophys. Res. 110, D18.

Tjernström, M., Koračin, D., 1995. Modeling the impact of marine stratocumu-lus on the boundary-layer structure. J. Atmos. Sci. 52, 863–878.

Tomasi, C., Tampieri, F., 1976. Features of the proportionality coefficient inthe relationship between visibility and liquid water content in haze andfog. Atmosphere 14, 61–76.

Toth, G., Gultepe, I., Milbrandt, J., Hansen, B., Pearson, G., Fogarty, C.,Burrows, W., 2010. The environment Canada handbook on fog and fogforecasting. Her Majesty the Queen in Right of Canada, Represented bythe Minister of the Environment. ISBN: 978-1-100-52518-1 (93 pp., Cat.no. En56-231/2010).

Tremant, M., 1987. La prévision du brouillard enmer. Météorologie Maritimeet Activities Océanographique Connexes. Rapport No. 20. TD no. 211.World Meteorological Organization, Geneva, Switzerland.

Tseng, Y.-H., Chien, S.-H., Jin, J., Miller, N.L., 2012. Modeling air–land–seainteractions using the integrated regional model system in MontereyBay, California. Mon. Weather Rev. 140, 1285–1306.

Turton, J.D., Brown, R., 1987. A comparison of a numerical model of radiationfog with detailed observations. Q. J. R. Meteorol. Soc. 113, 37–54.

175D. Koračin et al. / Atmospheric Research 143 (2014) 142–175

Twomey, S., 1959. The nuclei of natural cloud formation, II; the supersat-uration in natural clouds and the variation of cloud droplet concentra-tion. Geophys. Pure Appl. 43, 243–249.

U.S. Department of Agriculture, 1938. Atlas of the Climatic Charts of theOceans. Publication No. 1247, Prepared under the supervision of W. F.McDonald, 130 charts. U.S. Weather Bureau.

U.S. Weather Bureau, 1930. International Radio Weather Code for Use onUnited States Selected Ships. W.B No. 1005.U.S. Government PrintingOffice, Washington, DC, USA (14 pp.).

Urbano, R., Palenik, B., Gaston, C.J., Prather, K.A., 2011. Detection andphylogenetic analysis of coastal bioaerosols using culture dependentand independent techniques. Biogeosciences 8 (2), 301–309.

US Navy, 1955. Marine Climatic Atlas of the World.Chief of Naval Operations,Washington, District of Columbia, U.S.A.

van Schalkwyk, L., Dyson, L., 2013. Climatological characteristics of fogat Cape Town International Airport. Weather Forecast. 28 (3),631–646.

Vasey, M.C., Loik, M.E., Parker, V.T., 2012. Influence of summer marine fogand low cloud stratus on water relations of evergreen woody shrubs(Arctostaphylos: Ericaceae) in the chaparral of central California.Oecologia 170 (2), 325–337.

Von Ficker, H., 1936. Die Passatinversion. Veroffentl. D. Meteorol. Inst. D.Univ. Berlin, Bd., I, Heft, 4.

Wang, B., 1985. Sea Fog.China Ocean Press (330 pp.).Watanabe, K., Ishizaka, Y., Takenaka, C., 2001. Chemical characteristics of

cloud water over the Japan Sea and the Northwestern Pacific Ocean nearthe central part of Japan: airborne measurements. Atmos. Environ. 35(4), 645–655.

Weathers, K.C., Likens, G.E., 1996. Clouds in southern Chile: an importantsource of nitrogen to nitrogen-limited ecosystems? Environ. Sci.Technol. 31 (1), 210–213.

Weathers, K.C., Likens, G.E., Bormann, F.H., Eaton, J.S., Bowden, W.B., Andersen,J.L., Smiley, D., 1986. A Regional Acidic Cloud/Fog Water Event in theEastern United States.

Weathers, K.C., Lovett, G.M., Likens, G.E., Caraco, N.F., 2000. Cloudwaterinputs of nitrogen to forest ecosystems in southern Chile: forms, fluxes,and sources. Ecosystems 3 (6), 590–595.

Weiss Penzias, P.S., Ortiz, C., Acosta, R.P., Heim, W., Ryan, J.P., Fernandez, D.,Flegal, A.R., 2012. Total and monomethyl mercury in fog water from thecentral California coast. Geophys. Res. Lett. 39 (3).

Wells, L.A., 2007. Marine Layer Stratus Study. Fog and Boundary LayerClouds: Fog Visibility and Forecasting Pageoph Topical Volumes 2007.1397–1421.

Wetzel, M.A., Borys, R.D., Xu, L.E., 1996. Satellite microphysical retrievals forland-based fog with validation by balloon profiling. J. Appl. Meteorol. 35,810–829.

Wheeler, J.A., Ford, K.W., 1998. Geons, Black Holes, and Quantum Foam (ALife in Physics).W.W. Norton & Co.(380 pp.).

Willett, H.C., 1928. Fog and haze, their causes, distribution and forecasting.Mon. Weather Rev. 56, 435–468.

Williams, A.P., Still, C.J., Fischer, D.T., Leavitt, S.L., 2008. The influence ofsummertime fog and overcast clouds on the growth of a coastalCalifornia pine: a tree-ring study. Oecologia 156, 601–611.

Wurl, O., Obbard, J.P., 2004. A review of pollutants in the sea-surfacemicrolayer (SML): a unique habitat for marine organisms. Mar. Pollut.Bull. 48 (11), 1016–1030.

Yang, D., Ritchie, H., Desjardins, S., Pearson, G., MacAfee, A., 2010. High-resolution GEM-LAM application in marine fog prediction: evaluationand diagnosis. Weather Forecast. 25, 727–748.

Yue, Y., Niu, S., Zhao, L., Zhang, Y., Xu, F., 2012. Chemical composition of sea fogwater along the South China Sea. Pure Appl. Geophys. 169 (12), 2231–2249.

Zdunkowski, W.G., Barr, A.E., 1972. A radiative conductive model for theprediction of radiation fog. Bound. Layer Meteor. 2, 152–177.

Zhang, S., Yi, L., 2013. A comprehensive dynamic threshold algorithm fordaytime sea fog retrieval over the Chinese adjacent seas. Pure Appl.Geophys. 170, 1931–1944.

Zhang, S.-P., Xie, S.-P., Liu, Q.-L., Yang, Y.-Q., Wang, X.-G., Zhao-Peng Ren, Z.-P., 2009. Seasonal variations of Yellow Sea fog: observations andmechanisms. J. Clim. 22 (24), 6758–6772.

Zhou, B., Du, J., 2010. Fog prediction from a multimodel mesoscale ensembleprediction system. Weather Forecast. 25, 303–322.

Zhou, B., Du, J., Ferrier, B., McQueen, J., Dimego, G., 2007. Numerical forecastof fog — central solutions. 18th Conference on Numerical WeatherPrediction, AMS, 25–29 June 2007, Park City, UT.