formation, evolution, and dissipation of coastal sea fog

FORMATION, EVOLUTION, AND DISSIPATION

OF COASTAL SEA FOG

DARKO KORACIN1,5,*, JOOST A. BUSINGER2, CLIVE E. DORMAN3

and JOHN M. LEWIS1,41Desert Research Institute, Reno, Nevada, U.S.A.; 2University of Washington, Seattle,

Washington, U.S.A.; 3Scripps Institution of Oceanography and San Diego State University, SanDiego, California, U.S.A.; 4National Oceanic and Atmospheric Administration, Severe StormsLaboratory, Norman, Oklahoma, U.S.A.; 5Desert Research Institute, 2215 Raggio Parkway,

Reno, Nevada 89512, U.S.A.

(Received in final form 3 February 2005)

Abstract. Evolution of sea fog has been investigated using three-dimensional MesoscaleModel 5 (MM5) simulations. The study focused on widespread fog-cloud layers advectedalong the California coastal waters during 14–16 April 1999. According to analysis of the

simulated trajectories, the intensity of air mass modification during this advection significantlydepended on whether there were clouds along the trajectories and whether the modificationtook place over the land or ocean. The air mass, with its trajectory endpoint in the area wherethe fog was observed and simulated, gradually cooled despite the gradual increase in

sea-surface temperature along the trajectory. Modelling results identified cloud-top cooling asa major determinant of marine-layer cooling and turbulence generation along the trajectories.Scale analysis showed that the radiative cooling term in the thermodynamic equation over-

powered surface sensible and latent heat fluxes, and entrainment terms in cases of the trans-formation of marine clouds along the trajectories. Transformation of air masses along thetrajectories without clouds and associated cloud-top cooling led to fog-free conditions at the

endpoints of the trajectories over the ocean. The final impact on cloud-fog transition wasdetermined by the interaction of synoptic and boundary-layer processes. Dissipation of sea fogwas a consequence of a complex interplay between advection, synoptic evolution, anddevelopment of local circulations. Movement of the high-pressure system over land induced

weakening of the along-shore advection and synoptic-pressure gradients, and allowed devel-opment of offshore flows that facilitated fog dissipation.

Keywords: Lagrangian framework, Mesoscale model 5 (MM5), Mesoscale simulations, Off-

shore fog, Radiative cooling, U.S. West Coast.

1. Introduction

Understanding the formation and evolution of sea fog remains a researchchallenge primarily due to the lack of surface and upper-air observations overthe ocean. Satellite data are emerging as a valuable tool providing significant

*E-mail: [email protected]

Boundary-Layer Meteorology (2005) 117: 447–478 � Springer 2005DOI 10.1007/s10546-005-2772-5

weather information over the oceans; however, there is still ambiguity anduncertainty in characterizing three-dimensional cloudiness and fog usingsatellite data. Due to the existence of routine land-based observations, fogover land has received considerably more attention than fog at sea.

We focus on a sea-fog event along the U.S. West Coast where there issignificant complexity in the ocean structure, in the marine atmosphericboundary layer, and in coastal topography. As noted by Filonczuk et al.(1995), fog along the West Coast occurs over a wide range of observed windand temperature conditions, and the frequency of fog events is spatially andtemporally variable along the coast. Although there is a range of conditionsconducive to fog formation, it appears that there is no set of most favourableconditions that uniquely define the occurrence and maintenance of sea fog.As discussed in Leipper (1994), Leipper and Koracin (1998), and Lewis et al.(2004), forecasting sea fog along the California coast is a very challengingtask due to the complex topography, delicate interplay of physical processes,and the scarcity of offshore observations. Nevertheless, there have beenimportant contributions to the study of sea fog off the West Coast, notablythe early work of Byers (1930), Anderson (1931), and Petterssen (1936, 1938).In the period following World War II, Leipper (1948) and the CornellAeronautical Laboratory (CALSPAN) team (Mack et al., 1974; Pilie et al.,1979) set the stage for numerical experiments that used one-dimensionalmodels with turbulence parameterization and treatment of radiative heattransfer (e.g., see Oliver et al., 1978).

In the spirit of the early work of G.I. Taylor (Taylor, 1917) and Oliveret al. (1978), Koracin et al. (2001) (hereafter K2001) conducted a synthesizedmodelling and observational study of sea fog that accounted for the pathhistory of surface air and the turbulence-radiative processes in the marinelayer. They developed a conceptual model of the formation and evolution ofsea fog. Koracin et al. (2001) used a one-dimensional model with turbulenceclosure and radiative parameterization similar to the model used by Oliveret al. (1978); however, in contrast to Oliver et al., they emulated evolution ofthe air parcel in a Lagrangian frame of reference along multi-day trajectoriesover varying sea-surface temperatures and a strong marine inversion. In thecase study (April 1999) that was central to the K2001 investigation, sea fogformed as a result of stratus lowering along multi-day, over-water trajecto-ries. The modelling results were consistent with the observed time of fogonset and the location of fog along the coast as well as at sea, and identifiedcloud-top radiative cooling as the primary mechanism for fog formation.Radiative cooling at the top of the marine layer over-compensated forwarming at the surface in the presence of a sea surface that was severalKelvin (K) warmer than the adjoining surface air. Net cooling led to con-densation and fog. Although one-dimensional models can emulate specificadvection associated with a Lagrangian path, they inherently cannot account

DARKO KORACIN ET AL.448

for variability in the three-dimensional structure of the atmospheric bound-ary layer and advective processes.

Another approach to investigating air parcel modification within aLagrangian or quasi-Lagrangian framework was performed by Stevens et al.(2003a, b) using aircraft observations off southern California. They followedcloud parcels by means of circular aircraft patterns and focused on the rolesof cloud-top radiative cooling, entrainment, and drizzle on the evolution ofnocturnal marine stratocumulus. Stevens et al. (2003a, b) analysed the air-craft data and concluded that the observed temperature and moisture dis-continuities for a strong marine inversion over the U.S. West Coast definitelywould lead to significant entrainment and cloud drying based on cloud-topentrainment instability (Deardorff, 1980; Randall, 1980). In contrast, theobservations indicated persistent cloud layers that were even further devel-oped during the time when entrainment instability suggested cloud dissipa-tion. In addition, the consequent microphysical evolution led to significantformation of observed drizzle flux. Stevens et al. (2003a) indicated that therole of radiative processes in the maintenance and growth of clouds should beinvestigated further.

In this study, we seek to complement previous studies by examiningsimulated air parcel trajectories and investigating the roles and interplay ofadvection, cloud-top radiation, surface fluxes, and entrainment on the evo-lution of the cloudy marine atmospheric boundary layer (MABL). To betterunderstand air transformation along Lagrangian trajectories, we performedthree-dimensional numerical simulations using Mesoscale Model 5 (MM5)(Grell et al., 1994). To support the three-dimensional modelling, we relied ona recent observational study that explored the formation, maintenance, anddissipation of sea fog for this particular case (Lewis et al., 2003). The mainobjective of our study is to develop a qualitative and quantitative under-standing of the modification determinants relevant to the eventual formationof sea fog. Under certain conditions characterized by advection, strength ofthe marine inversion, and subsidence, modification of the cloudy marinelayer can lead to sea-fog formation. The influence of land-driven circulationon sea-fog dissipation is analysed and elaborated on, and implications of thestudy findings on the operational forecasting of sea fog are discussed in theconcluding remarks and epilogue.

2. Offshore Cloudiness and Fog Along the California Coast During 14–15

April 1999

Extensive cloud and fog layers occurred along the California coast during14–15 April 1999 in response to a synoptic disturbance that moved on to theWest Coast on 11 April (K2001; Lewis et al., 2003). Prior to fog formation, a

COASTAL SEA FOG 449

high pressure system over the north-eastern Pacific and a low pressure systemover Arizona and California set up intense north-westerly and northerlyflows along the West Coast that are characteristic of warm season dynamicsin this region (Figure la). The high pressure system and associated subsidence(reaching 0.05 m s)1) maintained a strong marine inversion of 10 K or more(Lewis et al., 2003). As the high pressure system moved inland (Figure lb),

Figure 1. Composite mean (0000 and 1200 UTC) sea-level pressure (hPa) for 13 April (a)

and 16 April (b) 1999. Obtained from the National Oceanic and Atmospheric Administra-tion, Climate Diagnostic Center using National Centers for Environmental Prediction andNational Center for Atmospheric Research re-analysis.


the pressure gradient weakened along the coast and induced offshore flows inthe coastal zone. According to our analysis, this synoptic evolution had asignificant impact on the evolution of offshore cloudiness and fog. The initialintense northerly and north-westerly flows were favourable to the cooling ofcloudy air along over-water trajectories. Inland displacement of the highpressure system reduced horizontal pressure gradients and winds, leading toconditions favourable to the formation of sea fog. Gradual development ofoffshore flows at the end of the fog event, however, induced drying of themarine layer near the coast at night as well as drying and warming of thislayer during daytime hours. This is further explained in the following sec-tions.

3. Numerical Model

Mesoscale Model 5 is used worldwide and was developed jointly byPennsylvania State University and the National Center for AtmosphericResearch in Boulder, Colorado. Details of the model structure are describedin Grell et al. (1994). Mesoscale Model 5 has been used in a variety ofresearch and application studies focused on atmospheric dynamics andcloudiness along the California coast (Koracin and Dorman, 2001), structureand evolution of wind stress and wind-stress curl impacting ocean dynamics(Koracin et al., 2004), and as a driver for an ocean model (e.g., Powers andStoelinga, 1999; Beg-Paklar et al., 2001), among others. To account forsynoptic processes and also to resolve characteristics of mesoscale processes,coarse and nested grids were set up to cover a large portion of the U.S. WestCoast from southern Oregon to the Los Angeles region (Figure 2). Thecoarse grid was centred at 37.5� N, 122.5� W and consisted of 177 · 207 · 43points with a horizontal resolution of 6 km. The nested grid (consisting of271 · 385 · 43 points with a horizontal resolution of 2 km) extended fromthe northern California coast to Point Conception (34.4� N, 120.5� W) wherethe coastline sharply turns to the east. Most of the clouds and fog wereobserved in this area.

Table I shows the model’s vertical grid structure with average heightsconverted from 43 full-sigma levels (integer values). Horizontal wind com-ponents and thermodynamic variables are computed on half-sigma levels,while vertical velocity is computed on full-sigma levels (midpoint values). Inorder to provide high vertical resolution within the MABL, seventeen verticallevels are provided in the lowest kilometre. Topography input was extractedfrom the 30¢¢-resolution global terrain and land use files. The main physicsoptions included warm-rain microphysics; the Grell cumulus parameteriza-tion; the Gayno-Seaman second-moment, turbulence parameterization withthe prognostic turbulence kinetic energy (TKE) equation (Shafran et al.,

COASTAL SEA FOG 451

2000); a cloud-radiation algorithm; and a multi-layer soil temperature model.First guess fields and lateral boundary conditions for the coarse grid for every12 h were obtained from the National Centers for Environmental Prediction(NCEP) Global Data Assimilation System archive. Synoptic informationincluded virtual temperature, geopotential height, horizontal wind compo-nents, and relative humidity on a global grid (with a horizontal resolution of2.5� in both latitudinal and longitudinal directions). These first-guess fieldswere horizontally interpolated onto the model grid by a two-dimensional, 16-point overlapping parabolic fit. In the second step of the preprocessing thefirst guess fields were refined using observations. Similarly, the first-guess sea-surface temperature (SST) field was extracted from the U.S. Navy’s dailyvalues (with a horizontal resolution of 2.5� in both latitudinal and longitu-dinal directions), updated with buoy and coastal station data, and interpo-lated onto the model grid using a bilinear interpolation method. Simulationswere performed for the period from 12 April 1999 at 0000 UTC to 17 April1999 at 0000 UTC, corresponding to the widespread cloud and fog event thatwas described by K2001 and Lewis et al. (2003). Time steps on the coarse andnested grids were 18 and 6 s, respectively.

Figure 2. Setup of MM5 modelling domains. The outer domain (D01) consists of 177 ·207 · 35 grid points with horizontal resolution of 6 km, and the inner domain (DO2) has271 · 385 · 35 grid points with horizontal resolution of 2 km.


TABLE I

Average heights (m) of full-sigma and half-sigma vertical levels of MM5 grids.

Level

Full-sigma level Half-sigma level

height (m) height (m)

1 0

2 18 9

3 36 27

4 54 45

5 82 68

6 118 100

7 155 136

8 191 173

9 247 219

10 321 284

11 396 359

12 472 434

13 548 510

14 625 586

15 702 664

16 780 741

17 938 859

18 1180 1059

19 1429 1305

20 1701 1565

21 1995 1848

22 2294 2144

23 2607 2450

24 2931 2769

25 3262 3097

26 3610 3436

27 3973 3792

28 4344 4158

29 4738 4541

30 5149 4944

31 5574 5362

32 6028 5801

33 6506 6267

34 7005 6756

35 7543 7274

36 8118 7831

37 8726 8422

38 9393 9060

39 10121 9757

40 10911 10516

41 11805 11358

42 12820 12313

43 13983 13402

COASTAL SEA FOG 453

3.1. MODEL EVALUATION

3.1.1. Comparison with Buoy and Coastal Station DataObservations from eight buoys and two coastal land stations were used formodel evaluation. Table II shows buoy positions and summary statistics ofthe comparison between the model and observations for a five-day period(12–17 April 1999). Standard statistical parameters included the bias or meanerror (ME), the mean absolute error (MAE), the population root-mean-square error (RMSE), and the root-mean-square vector error (RMSVE).These parameters have been commonly used (Koracin et al., 2004) and alsoare defined in Koracin and Dorman (2001). The correlation coefficient forwind speed ranged from 0.43 to 0.75, which is similar to Koracin and Dor-man’s (2001) results for the comparison between model results and buoy dataalong the California coast for all of June 1996. According to standarddeviations from the model results and measurements, the model showstemporal variability similar to observations. The bias expressed by the ME isrelatively small and still should be reduced due to the difference in heightbetween the model grid points and elevation of the buoy sensors. Some of thedifferences are likely due to differences in sampling. (i.e., buoy data are an8-min average at every hour, while the model results represent grid- and shorttime-averaged values during the last timestep at each hour).

As will be shown in the next sections, accurate representation of winddirection by the model is crucial to determining the origin and fate of air-mass back-trajectories. Consequently, special consideration is given to themodel’s ability to successfully reproduce wind direction. Figure 3 shows thetime series of modelled and observed wind direction at the Point Arena andSan Francisco coastal land stations that are central to our analysis elaboratedon in the next section. Wind speeds at the Point Arena station were higherthan at San Francisco and showed persistent north-north-westerly flow in thefirst part of the period (fog development). In the second part of the period,both stations show variable wind direction including offshore flow from theeastern quadrant (prior to and during the fog dissipation period). The figureshows that MM5 was able to capture the main behaviour of the flows at thecoastal land stations.

3.1.2. Comparison with Satellite DataIn addition to the point comparison described in the previous subsection, wealso compared model results with satellite data. Figure 4 shows Geosta-tionary Operational Environmental Satellite 10 (GOES-10) visible imagesover the West Coast at hourly intervals from 1600 to 2300 UTC on 13 April1999. An extensive cloud and fog layer was observed over the entire Cali-fornia coast at the beginning of the period, and the layer cleared from thenorth toward the end of the period. Figure 5 shows horizontal cross-sections


TABLE

II

StatisticalparametersofthecomparisonbetweenMM5resultsandbuoyandcoastallandstationdata

fortheperiod12–17April1999.

BuoyID

N.Lat.

(�)

W.Long.

(�)

Heightbuoy

HeightMM5

NMean

Std

dev

Corr

ME

(ms)

1)

MAE

(ms)

1)

RMSE

(ms)

1)

RMSVE

(ms)

1)

(mMSL)

(mMSL)

MM5

OBS

MM5

OBS

(ms)

1)

(ms)

1)

(ms)

1)

(ms)

1)

Windspeed

Santa

Maria

46011

34.88

120.87

59

121

4.52

4.02

2.53

2.15

0.52

0.50

1.86

2.35

3.40

BodegaBay

46013

38.23

123.30

59

121

6.68

4.82

2.77

3.72

0.75

1.86

2.64

3.09

4.06

Pt.Arena

46014

39.22

123.97

59

121

5.73

5.22

3.03

3.56

0.63

0.51

2.28

2.89

3.55

Pt.Arguello

46023

34.71

120.97

10

9121

4.70

5.26

2.53

2.68

0.54

)0.55

2.07

2.54

3.99

SanFrancisco

46026

37.75

122.82

59

97

5.57

3.66

2.38

2.26

0.43

1.91

2.63

3.12

5.01

SanMartin

46028

35.74

121.88

59

121

4.87

4.31

2.18

2.50

0.70

0.56

1.52

1.90

2.79

Santa

Barbara

46053

34.24

119.85

59

121

5.55

3.59

3.34

2.35

0.70

1.96

2.46

3.09

5.32

Pt.Conception

46063

34.25

120.66

59

121

6.50

5.81

3.18

3.21

0.71

0.68

1.98

2.53

3.76

Pt.Arena

(land)

PTAC1

38.96

123.74

30

26

121

4.48

4.08

1.74

2.22

0.59

0.37

1.56

1.88

3.06

SanFran.

Airport

SFO

37.60

122.40

39

120

3.31

2.97

2.39

2.67

0.64

0.35

1.70

2.17

2.96

COASTAL SEA FOG 455

Table

IIContinued.

BuoyID

N.Lat.

(�)

W.Long.

(�)

Heightbuoy

HeightMM5

NMean

Std

dev

Corr

ME

(�C)

MAE

(�C)

RMSE

(�C)

RMSVE

(�C)

(mMSL)

(mMSL)

MM5

OBS

MM5

OBS

(�C)

(�C)

(�C)

(�C)

Tem

perature

Santa

Maria

46011

34.88

120.87

49

121

11.51

9.34

1.81

2.81

0.63

2.17

2.51

3.06

BodegaBay

46013

38.23

123.30

49

121

11.07

10.59

1.18

3.34

0.72

0.47

2.20

2.68

Pt.Arena

46014

39.22

123.97

49

121

10.95

9.18

1.15

3.36

0.77

1.78

2.57

3.18

Pt.Arguello

46023

34.71

120.97

10

9121

11.12

9.36

1.75

2.87

0.70

1.76

2.20

2.71

SanFrancisco

46026

37.75

122.82

49

97

11.31

11.73

1.82

3.73

0.76

0.42

2.06

2.65

SanMartin

46028

35.74

121.88

49

121

11.14

10.67

2.00

2.92

0.65

0.47

1.89

2.26

Santa

Barbara

46053

34.24

119.85

49

121

11.22

11.21

1.42

2.14

0.75

0.01

1.12

1.42

Pt.Conception

46063

34.25

120.66

49

121

10.99

9.66

1.55

2.63

0.66

1.33

1.85

2.34

P.Arena(land)

PTAC1

38.96

123.74

24

26

121

9.70

11.31

4.67

1.63

0.73

)1.66

3.51

3.97

SanFran.Airport

SFO

37.60

122.40

39

120

15.21

14.81

5.39

5.04

0.91

0.42

1.87

2.32


of the colour-filled, simulated total-cloud liquid water mixing ratio at 90 mnear the beginning of the period (1200 UTC on 13 April), near the middle ofthe period (1800 UTC on 13 April), and at the end of the period (0000 UTCon 14 April). These periods correspond to the times shown in the satelliteimage (Figure 4). The simulations accurately reproduced advection of thecloud-fog edge from north-west to south-east along the coast. Figure 5 andresults of the statistical comparison between the model and measurementsshow that the model was able to predict the general behaviour of cloud andfog-layer evolution. Consequently, model results can be used for investigat-ing the formation, maintenance, and dissipation of sea fog as discussed in thefollowing sections.

4. Modification of the Marine air Along Trajectories

Using buoy data from along the West Coast, K2001 inferred that coastal fogin the southern part of the domain occurred in the early hours of 14 April1999. Analysis of the one-dimensional modelling and a limited set of

12 12.5 13 13.5 14 14.5 15 15.5 16 16.5 170

50

100

150

200

250

300

350

Day of April 1999

Win

d di

rect

ion

(°)

(a)

Sim Meas

12 12.5 13 13.5 14 14.5 15 15.5 16 16.5 170

50

100

150

200

250

300

350

Day of April 1999

Win

d di

rect

ion

(°)

(b)

Sim Meas

Figure 3. Time series of measured and simulated wind direction at the Point Arena (a) andSan Francisco (b) coastal land stations from 12 to 17 April 1999.

COASTAL SEA FOG 457

Figure 5. Horizontal, colour-filled cross-sections of the simulated liquid water mixing ratio(kg kg)1) at 90 m on 13 April at 1200 UTC (a), on 13 April at 1800 UTC (b), and on 14April 1999 at 0000 UTC (c).

Figure 4. GOES-10 visible images in hourly intervals from 1601 to 2301 UTC on 13 April1999.


observations described in K2001 led to the development of and conceptualmodel for conditions of north-westerly advection of the cloudy marine airthat defined modification of the MABL. Since the SST was lower off thesouthern Oregon and northern California coasts and higher off the centraland southern California coasts (Figure 6), surface sensible and latent heatfluxes were increasing southward and consequently warming and moisteningthe marine layer. Figure 6 confirms this characteristic SST structure at thetime of sea-fog formation. At the same time, cloud-top cooling caused

Figure 6. Colour-filled, sea-surface temperature contours (�C) off the southern Oregon andCalifornia coasts at the time of sea-fog formation over southern California coastal waterson 14 April 1999 at 0200 UTC. Contour interval is 0.5 K.

COASTAL SEA FOG 459

buoyant instability and overall cooling of the MABL, which propagateddownward. As shown by K2001, cloud-top cooling can overpower warmingfrom the surface and, together with increased moisture, can lead to fog underconditions of strong subsidence and marine inversion. Spatial and temporalvariations of advection, however, strongly modify the temperature andhumidity properties of marine air. In particular, it is extremely important todetermine whether trajectories at different levels originate over the land orocean, as well as whether a trajectory originally over the ocean passes overthe land and undergoes air mass transformation (mainly warming anddrying).

4.1 BACK-TRAJECTORIES WITH END POINTS IN FOG-COVERED AND CLEAR-SKY

AREAS

To better understand how marine air is modified while being advected alongthe coast, we constructed a series of backward trajectories computed fromMM5-simulated wind fields and examined the simultaneous effects of cloud-top cooling, surface heating, and moistening along these trajectories. Itshould be noted that the specified height of a backward trajectory refers tothe height at its end point. While tracing a trajectory backward in time, theheight of the trajectory will vary in time as a consequence of vertical motions.We focus on two regions: one at which fog occurrence was inferred frombuoy measurements (Point Arguello, hereafter PA), and one at which fog wasabsent according to the analysis of buoy observations (San Francisco,hereafter SF). The effects of cloud-top cooling can be inferred by tracingsimulation results of air temperature at various levels and differences withrespect to the SST. Trends in dew-point temperature as well as surface sen-sible and latent heat fluxes can be used to infer effects of the ocean surface onmodification of the MABL as it is advected along the coast. Analysis of thesimulations indicated fog formation upwind of PA in the early hours of 14April 1999. At 0300 UTC on 14 April, model results show fog formation atthe Point Arguello buoy location and no-fog conditions at the SF buoylocation – predictions in agreement with saturation conditions at the buoys.Figure 7 shows back-trajectories from these two sites at 10 m (near thesurface), 90 m (the lower part of the marine layer), 270 m (the upper part ofthe marine layer), and 1000 m (above the marine layer) with an ending timeof 0300 UTC on 14 April. The typical depth of the marine layer (300–500 m)is indicated in the observational (e.g., Rogers et al., 1998) and modellingstudies (e.g., Koracin and Dorman, 2001). The surface back-trajectory fromPA (Figure 7a) was significantly offshore, while the back-trajectory at 270 mwas originally located closer to the coast, approached the coastline south ofthe Monterey Bay area for a short time, and then turned offshore afterwards.


The flow above the marine layer was fully offshore. Back-trajectories with theendpoint at the SF buoy location (Figure 7b) were offshore whileapproaching the buoy, but the time history indicated an origin over landwhere the back-trajectories encountered warming and drying. Figure 7indicates that the origin and modification of the marine layer throughwarming and drying over land can be a significant determinant of cloud-freeand fog-free conditions.

4.2 MODIFICATION OF AIR ALONG BACK-TRAJECTORIES RELEVANT TO FOG

OCCURRENCE

In order to examine simulated atmospheric conditions along back-trajecto-ries, we constructed time series of air temperature, dew-point temperature,SST, and cloud occurrence in terms of the vertically integrated liquid waterpath (ILWP) (Stull, 1988) as well as surface sensible and latent heat fluxes.Figure 8 shows a time series of these parameters simulated along a surfaceback-trajectory with the end point at PA where fog was inferred frommeasurements. The SST gradually increased by about 2.5 K along the pathduring the first 12 h, and sensible and latent heat fluxes increased by about

(a) (b)

Figure 7. Horizontal projection of simulated back-trajectories at 10 m (black), 90 m (blue),270 m (red), and 1000 m (green) with endpoints at the Point Arguello (a) and SanFrancisco (b) buoys on 14 April at 0300 UTC. Cloud and fog conditions at the Point

Arguello buoy and cloud and fog-free conditions at the San Francisco buoy were inferredfrom buoy and coastal station measurements.

COASTAL SEA FOG 461

30 35 40 45 504

5

6

7

8

9

10

11

12pa – surface t(b–o) td(g–x) sst(r– –)

Hour after simulation start

Tem

pera

ture

(C

)

30 35 40 45 500

50

100

150

200

250

300

350

400pa – lwp (b–o)


Inte

grat

ed li

quid

wat

er p

ath

(g m

–2)

30 35 40 45 5020

0

20

40

60

80

100pa – shflx (b–o) & lhflx (g–x)


Hea

t flu

x (W

m–2

)

Figure 8. Time series of simulated air temperature (solid line with circles), sea-surface tem-perature (dashed line), and dew-point temperature (solid line with x) (a); integrated liquid

water path (b); and surface heat (solid line with circles) and latent heat (solid with x) fluxes(c) along the surface back-trajectory at 10 m with the end point at the Point Arguello buoylocation on 14 April at 0300 UTC.


70 W m)2. Despite this significant increase in the SST, as well as in thesurface sensible and latent heat fluxes, the air temperature gradually de-creased about 4 K during the same time. Since the surface back-trajectoryposition was entirely offshore, we infer that the cooling mechanism origi-nated from cloud-top and fog-top cooling. The importance of cloud- and fog-top cooling on the evolution of the MABL has been emphasized in bothobservational (Nicholls, 1984) and modelling studies (Koracin and Rogers,1990; Rogers and Koracin, 1992; K2001). The mechanism and main deter-minants for cooling of the air as it is advected offshore are discussed furtherin the next section.

Examination of the present simulation results (Figure 8) shows that fogformed at the beginning of the surface back-trajectory period and wasmaintained throughout the travel to PA, except for a brief break-up andlifting into the low-level cloud between hours 18 and 21 from the back-trajectory start. It should be noted that the main increase in temperatureoccurred after a significant drop in the ILWP (after hour 13). Air tempera-ture was increasing and recovering toward the SST and, consequently, sur-face fluxes were reduced. After hour 20, the gradual decrease in the SST andincrease in the ILWP led to a decrease in air temperature and saturation nearthe surface while approaching PA. Since the back-trajectory was graduallyapproaching the coast, the simulated boundary-layer height was greater inthe starting area (about 500 m) than in the end-point area (about 300 m).Consequently, even the presence of relatively high air temperatures did notprevent fog formation and maintenance due to the moisture confined in theshallow marine layer and low surface fluxes.

Figure 9 shows a time series of simulated atmospheric conditions alongthe back-trajectory ending at the SF buoy location where fog was notpresent. Evolution of the simulated atmospheric conditions was significantlydifferent from conditions simulated along the back-trajectory ending at PA.Near-surface air temperature for the SF back-trajectory gradually increased(for the most part) due to the increase in the SST. There was a brief occur-rence of fog in the far upwind side in the beginning period (first four hours)during which the air temperature initially dropped significantly and then keptrelatively steady during the occurrence of fog. After the fog cleared, the airtemperature adjusted toward the SST, reducing the humidity. Sensible heatflux decreased as the air temperature converged toward the SST, while thelatent heat flux increased somewhat due to decreasing humidity of the near-surface air. Since there were no significant clouds and associated cloud-topcooling present, these conditions led to a fog-free state as the flowapproached SF.

Another determinant for the occurrence or absence of fog is verticalstructure and turbulent mixing within the boundary layer. As discussedearlier and shown in Figure 7, the back-trajectories at various vertical levels

COASTAL SEA FOG 463

have different pathways and origins. Figure 8 shows that the air temperaturealong the PA back-trajectory gradually decreased within the first 10 h or soand then gradually increased in approximately the middle of its duration

40 42 44 46 48 504

5

6

7

8

9

10

11

12sf – surface t(b–o) td(g–x) sst(r– –)


Tem

pera

ture

(C

)

40 42 44 46 48 500

20

40

60

80

100sf – lwp (b–o)


Inte

grat

ed li

quid

wat

er p

ath

(g m

–2)

40 42 44 46 48 50100

80

60

40

20

0

20

40sf – shflx (b–o) & lhflx (g–x)


Hea

t flu

x (W

m–2

)

Figure 9. Same as Figure 8, except with the end point at the San Francisco buoy location.


while approaching the coastline under intermittent clear-sky conditions.After hour 17, however, the back-trajectory turned offshore and undercloudy-fog conditions cooled down almost 4 K and significantly increased inhumidity while approaching PA. In contrast to the PA back-trajectory, thetemperature along the SF back-trajectory near the surface (Figure 9) grad-ually increased, with low humidity throughout the travel time. These hightemperatures and low humidities at higher levels, in conjunction withincreasing surface temperature and relatively low near-surface humidity(Figure 9), led to an absence of fog at the SF location. Our present three-dimensional simulations confirm the results from the one-dimensional sim-ulations of K2001, that there was noticeable TKE generated by surface fluxesand longwave cooling at the cloud top. Simulated mean TKE along the PAtrajectories was 0.26 and 0.13 m2 s)2 at the surface and 270 m, respectively,and was less along the SF trajectories (0.17 and 0.02 m2 s)2 at the surface and270 m, respectively) without significant cloud-fog top cooling. Reduced TKEin the latter case was a consequence of small surface fluxes and the absence ofcloud-generated turbulence, mainly through cloud-top cooling, as shown inprevious studies such as Koracin and Rogers (1990), Koracin and Tjernstrom(1992), and Tjernstrom and Koracin (1995).

5. Fog Formation in Response to the Interplay of Surface Forcing, Cloud-Top

Cooling and Entrainment

On the basis of our modelling results, we investigated the roles of the maincomponents of heat transfer that can lead to net cooling or heating of theMABL. As a first step towards understanding the importance of the majorcomponents, we performed a scale analysis of the simplified thermodynamicequation from a mixed-layer model (Stage and Businger, 198la,b):

dhdt¼ w0h0e

zbþ Rc � Rb

zbþ weDhe

zb: ð1Þ

The term on the left-hand side represents the total change of the potentialtemperature due to the following terms on the right-hand side: divergence ofsensible heat flux, divergence of net radiative flux, and entrainment processes.In this equation, w0h0e is the surface kinematic heat flux; zb is the height of theMABL (equal to the height of the inversion base); Rc, and Rb are the nor-malized net radiative fluxes at the cloud top and base, respectively; we is theentrainment rate velocity; and Dhe is the jump in equivalent potential tem-perature within the inversion.

COASTAL SEA FOG 465

5.1. SCALE ANALYSIS OF THE TERMS IN THE EQUATION FOR POTENTIAL

TEMPERATURE

Simulated surface fluxes were computed based on the similarity theorydescribed by Grell et al. (1994) and Beg-Paklar et al. (2001), while the heightof the MABL was estimated by jointly examining vertical gradients of thepotential temperature and TKE, as well as the top of the cloud layer. Heatingand cooling rates due to radiative heat transfer were simulated for each levelusing the parameterization by Dudhia (1989). Entrainment velocity (we) wascomputed using the simplified prognostic equation for evolution of theMABL height (zb):

dzbdt¼ we þ wm ð2Þ

where wm is the mean model vertical velocity at the top of the MABL. Fromthe model results we computed the change of the boundary-layer height (left-hand side of Equation (2)). Then we computed we from Equation (3) usingthese wm and dzb

dt estimates.Since modelled radiation tendency is a computed value for each grid

point, we converted radiation tendency from a particular point into theassociated effect for the entire MABL. Radiation tendency for each point isassumed to be representative of the half-grid vertical interval above andbelow the considered point. The magnitude of the radiation tendency was

0 5 10 15 201.5

1

0.5

0

0.5

1

1.5


Pot

entia

l tem

pera

ture

tend

ency

(10

–3 K

s–1

)

Figure 10. Time series of simulated terms on the right-hand side of Equation (1) contribut-

ing to the MABL heating and cooling processes: surface heat flux (dashed), radiation ten-dency (solid), and entrainment (dashed-dotted line with circles) for back-trajectory withendpoint at the Point Arguello buoy location on 14 April at 0300 UTC.


then multiplied by the ratio of this vertical separation to the MABL depth;then these values for each layer are summed up to the top of the MABL. Inour specific case, we performed this procedure for the vertical profile forevery hour at the point where the trajectory was at that hour. By followingthis process, we were able to extrapolate the value of the radiation tendencyderived for each particular point and its corresponding z into its particularcontribution to heating or cooling of the entire MABL. As discussed in theprevious section, turbulence was sufficient to provide mixing and redistri-bution of this heating and cooling.

Figure 10 shows a time series of the main components contributing to thepotential temperature tendency. Use of the simplified thermodynamicequation (Equation (1)) imposes a residual with respect to the simulated totalchange of the potential temperature that includes treatment of full three-dimensional processes. The average difference between the left-hand side ofEquation (1) and the simulated total change of the potential temperature is26%. Relative magnitudes among the components shown in Figure 10,however, remain the same as the initial estimate of individual importance.Figure 10 clearly indicates that longwave cooling is the main contributor tothe cooling of the MABL in the first part of the period when the overallcooling of the marine layer was simulated along the back-trajectory with theend point at Point Arguello (Figure 8). Cloud–and fog-induced cooling iscounteracted by heating due to the surface heat flux and entrainment.According to this scale analysis, cloud- and fog-top cooling is definitely aprocess that can dominate surface forcing and entrainment and, in con-junction with increased moisture, can lead to condensation. The effect wasprominent during the first 7 h and at the final stage of the trajectory when thesurface heat flux was low. Note that the moisture flux increased during theperiod of increased heat flux and that apparent cooling of the air along thetrajectory during the last several hours prior to fog formation was sufficientto produce fog.

6. Dissipation of Sea Fog

In accordance with the observations in K2001, the sea fog analysed in theprevious section dissipated in the area north of Point Conception after 1600UTC on 15 April. In order to examine the effect of advection and surfacefluxes on fog dissipation, we constructed back-trajectories with the end pointin the PA area at the time after fog clearing (0000 UTC on 16 April) andextending 18 h backward in time, while the trajectories were in the domain.Figure 11 shows the position of the back-trajectories at the surface, 90 m,270 m, and 1000 m. In contrast to Figure 7, which shows the offshore originof the back-trajectories, Figure 11 clearly indicates that trajectories both

COASTAL SEA FOG 467

within and above the MABL originated over land and entered the coastalwaters mainly as an easterly flow. Figure 12 shows a time series of thetemperature, dew-point temperature, ILWP, and surface sensible and latentheat fluxes along the surface back-trajectory shown in Figure 11. In contrastto the properties of the air mass along the back-trajectory relevant to fogformation (Figure 8), the air mass at the origin of the trajectory over landwas warm and dry and encountered cooling and moistening while mixingwith coastal air over the ocean. The air temperature gradually decreased,approaching the SST, as shown by the reduction in sensible heat flux. Themain temperature decrease was from hour 6 to 11 when the air massencountered thin cloud, as indicated in Figure 12. ILWP was more than 10times lower in this case than in the case of the offshore trajectory relevant tocloud and fog formation (Figure 8). During the short-term cooling fromhours 6 to 11, surface fluxes were small and did not contribute to verticalmoisture transport. In spite of the gradual increase in the dew-point tem-

Figure 11. Horizontal projection of simulated back-trajectories at 10 m (black), 90 m (blue),

270 m (red), and 1000 m (green) that are relevant to fog dissipation. The back-trajectorieshave their end point at the Point Arguello buoy location on 16 April at 0000 UTC.


0 2 4 6 8 10 12 14 16 180

5

10

15

20

Hour after backtrajectory start

Tem

pera

ture

(C

)

0 2 4 6 8 10 12 14 16 180

5

10

15

20

25

30

35

40


Inte

grat

ed li

quid

wat

er p

ath

(g m

)

0 2 4 6 8 10 12 14 16 18100

50

0

50


Hea

t flu

x (W

m2 )

(a)

(b)

(c)

Figure 12. Time series of simulated air temperature (solid line with circles), sea-surface tem-

perature (dashed line), and dew-point temperature (dashed dotted line with x) (a); inte-grated liquid water path (b); and surface heat (solid line with circles) and latent heat(dashed line with x) fluxes (c) along the surface back-trajectory that are relevant to fog dis-

sipation. The end point of the back-trajectory is at the Point Arguello buoy location on 16April 1999 at 0000 UTC.

COASTAL SEA FOG 469

perature until the last few hours of the back-trajectory, there was no sig-nificant cooling of the surface air when the trajectory was approaching PA.All these processes led to clear-sky conditions at the end of the consideredback-trajectory.

Another complexity in the formation and dissipation of sea fog is theinteraction of synoptic processes and locally induced coastal circulations(Lewis et al., 2003). A process of gradual cooling of the cloudy marine airoccurred along the offshore advection path with sufficient moisture, turbu-lence, and saturation. In the case of synoptic pressure centre displacementsand weakening of the synoptic pressure gradients, land-breeze flows developduring nighttime and, by mixing with alongshore marine flows, induce gen-eral drying of the MABL. During daytime, sea breezes develop and in thereturn flow, they bring warm and dry air that originated over land. Dryingand warming are enhanced by daytime shortwave radiational heating of thecloud and fog layers (Nicholls, 1984; Koracin and Rogers, 1990). All theseeffects promote fog dissipation. The effects of radiative processes on thediurnal evolution of coastal clouds are discussed by Betts (1990) and Koracinand Dorman (2001).

Figure 13a shows surface vector winds at the time (0200 UTC on 14 April)when fog onset was simulated in the PA area. This area was under the

MAXIMUM VECTOR: 13.9 m s-1 MAXIMUM VECTOR: 21.6 m s-1

(a) (b)

Figure 13. Simulated vector winds and wind speed contours on 14 April at 0200 UTC (13

April at 1800 LST) during fog formation, at the surface (a) and at 500 m (b). Contourinterval for the wind speed is 2 m s)1. For clarity, wind vectors are plotted at every tenthgrid point.


influence of alongshore marine flows with strong winds near San Franciscoand Monterey Bay that weakened south of Monterey Bay. We simulatedsimilar flow characteristics, although with greater wind speeds, at 500 m(Figure 13b). Simulated vector winds at the surface and at 500 m (prior tofog dissipation) are shown in Figure 14a, b, respectively. As indicated inFigure 1 and explained further in this section (see Figures 15 and 16), thehigh pressure system moved inland and synoptic pressure gradients weresubstantially weaker. This produced a significant decrease in winds within themarine layer and allowed for the development of offshore flows. Offshoreflows were simulated throughout the marine layer, as shown by the vectorwinds at the surface and 500 m (Figure 14a, b). The main impact of theoffshore flows was to warm and dry the MABL near the coast. This is clearlyseen in Figure 15, which shows colour-filled, surface-temperature contoursduring fog formation (Figure 15a) and 48 h later (Figure 15b). The figureshows gradual warming and a decrease of the pressure gradients over theocean in the southern and south-western regions of the domain. Gradualheating of the air over land associated with the high pressure system also canbe seen in Figure 15. Another important feature of the offshore flows is theconsequent drying of the MABL near the coast. Figure 16 shows gradual

MAXIMUM VECTOR: 13.5 m s-1

(a)


(b)

Figure 14. Simulated vector winds and wind speed contours on 15 April at 1300 UTC(0500 LST) prior to fog dissipation, at the surface (a) and at 500 m (b). Contour intervalfor the wind speed is 2 m s)1. For clarity, wind vectors are plotted at every tenth grid

point.

COASTAL SEA FOG 471

drying of the surface air in colour-filled, surface relative humidity contoursfor the same time intervals shown in Figure 15. During the fog period,humidity was high over the whole offshore region (Figure 16a), with dryingpropagating from the north-west to the south-east as the offshore flowsdeveloped (Figure 16b). The combined effect of the MABL warming anddrying led to dissipation of clouds and fog.

7. Summary and Conclusions

Using MM5, we simulated a case of widespread offshore cloud and foglayers along the California coast during 14–16 April 1999. Our mainobjective was to investigate and quantify the effects of advection, radia-tion, surface fluxes, and entrainment relevant to the formation, evolution,and dissipation of sea fog. Areas where fog was observed (PA) and wherefog was not present (SF) were considered for analysis using buoy and landstation observations. Our results emphasize that it is crucial to investigatethe formation, evolution, and eventual dissipation of sea fog in aLagrangian framework (i.e., along long over-water trajectories and tra-


(a)


(b)

Figure 15. Colour-filled, simulated surface air temperature (�C) overlaid with sea-level pres-

sure (hPa) and surface winds (knots) at 0200 UTC at a 48-h interval: 14 April 1999 (a) and16 April 1999 (b). Contour interval for the sea-level pressure is 0.5 hPa. For clarity, windvectors are plotted at every tenth grid point.


jectories that originate over land). The study also shows that modificationof the MABL is significantly dependent on advection processes over theland and ocean and whether clouds are present during the transformation.

During the time when the fog formed, a fog layer and cloud layer werepresent with significant ILWP (0.2–0.4 kg m)2). Cloud-top cooling gener-ated net cooling of the marine layer by about 4 K despite the gradualincrease in the SST by 2–3 K along the trajectory. Within 12 h this trig-gered an increase of the surface heat and latent heat fluxes of about70–80 W m)2 along the trajectory. This surface heating counteractedcloud-top cooling but not sufficiently to overcome the effect of cloud-generated cooling and fog formation that was simulated at the end of theback-trajectories. Scale analysis of the major factors (radiative cloud-topcooling, surface fluxes, and entrainment) determining the evolution andfate of sea fog showed that in this case cloud-top cooling was a dominantprocess creating net cooling of the MABL, and leading to the formationand maintenance of the cloud and fog layers. When the cloud and foglayers were absent with a greater increase of the SST along the trajectory,surface fluxes continuously increased air and dew-point temperatures.


(a)

MAXIMUM VECTOR: 8.4 m s -1

(b)

Figure 16. Colour-filled, simulated surface relative humidity (%) overlaid with sea-levelpressure (hPa) and surface vector winds at 0200 UTC at a 48-h interval: 14 April 1999 (a)and 16 April 1999 (b). Contour interval for the sea-level pressure is 0.5 hPa. For clarity,

wind vectors are plotted at every tenth grid point.

COASTAL SEA FOG 473

Consequently, these conditions did not yield fog conditions at the endpoints of the trajectories.

Our model results show that dissipation of sea fog along the West Coast issignificantly influenced by the development of land-driven circulations.Dissipation of sea fog is governed by the complex interplay between advec-tion, synoptic evolution, and development of local circulations. Displacementand weakening of horizontal synoptic pressure gradients and the consequentdecrease in marine winds allows for development of offshore flows. Theseoffshore flows merge with weak marine flows and cause drying of the MABL.During daytime, the offshore flows can induce warming and consequent fogdissipation.

In conclusion, we would like to emphasize that we have provided atheoretical background for the formation, evolution, and dissipation of seafog and indicated the major components determining sea-fog characteris-tics. This methodology should be applied to other cases and types of seafog and further evaluated using observations from future field programs tobe conducted in Lagrangian and Eulerian frameworks of reference overthe ocean.

8. Epilogue on Sea-Fog Research and Implications for Operational

Forecasting

In addition to the summary and conclusions section, we believe that it isimportant to provide a general picture of sea-fog research, its challenges, andassociated implications for operational forecasting.

We have completed a suite of observational andmodelling studies related tocoastal sea fog (K2001; Lewis et al., 2003, 2004; and the current contribution).The observational studies of the sea-fog event of April 1999 (part ofK2001 andLewis et al., 2003) and the view of sea fog in the context of synoptic processesover the West Coast (Lewis et al., 2003) showed that widespread sea fog iscontrolled by large- and regional-scale processes: subsidence, long over-watertrajectories and land-based trajectories, and the consequent structure of themarine layer. It appears that cooling of the marine layer in the presence of awarm ocean is due to radiative cooling associated with stratus. The value ofthese precise simulations of radiative fluxes stems from the model’s very finevertical resolution (approximately 10 m, K2001).

Our three-dimensionalmodelling results indicate that delineation of fog andfog-free areas can be ascertained through careful analysis of trajectories withinand above the MABL. These modelling results also provide reasonable valuesof the liquid water content of the fog layer. Further, results indicate that thedissipation of fog is linked to changes in the mesoscale wind field. Radiativecloud-top cooling generates turbulence that can induce fog maintenance and


evolution; however, due to a general lack of turbulence measurements in seafog, the role of turbulence transfer is not yetwell understood (Nakanishi, 2000).

What do these observational and modelling results portend for the successof operational prediction? First, our studies indicate that the description oflarge-scale subsidence, as well as the presence and placement of stratusclouds, are an essential part of the background information. Subsidencestrength should be predicted to within 0.01 m s)1. Timing of frontal passagesin the coastal region (time between successive passages) must be accurate(probably within ±6 h based on results in Lewis et al., 2003). The posi-tioning and strength of the cyclone-anticyclone couplets also must be precisein order to accurately determine trajectories of the air within and above theMABL. Can synoptic-scale numeric models faithfully produce these features?On the optimistic side, the ability of the model to predict the movement ofsynoptic systems (strength and positioning) is generally very good. On thepessimistic side, the ability of the model to address associated subsidence,and in particular subsidence just above the MABL, is problematic (not somuch in position but in strength).

Given good background information, can mesoscale features – includingthe boundary-layer physics – be predicted with sufficient accuracy to deter-mine the occurrence or absence of fog? Again, on the positive side, the abilityof the model to predict radiative and turbulence processes shows promise inbringing the marine layer to a saturated state (in the foggy region), andindicating that the time history of the trajectories is key to determiningwhether fog is likely or not.

Thus, there is promise for reliable operational prediction of fog, yet thesensitivity of parameter estimation (air-sea flux, radiative cooling, initial stateof the MABL and overlying dry layer) leads us to believe that a prudentapproach would involve ensemble forecasting (i.e., prediction of the meancondition as well as variances). We also realize that models will have greatdifficulty addressing the persistence of sea fog. Once sea fog forms, it tends topersist, especially if the synoptic systems are more steady than transient.There is negative feedback, in effect. That is, if one process (e.g., heating fromthe sea surface) tends to dissipate the fog, the induced turbulence couldpossibly lead to mixing with radiatively cooled air at the stratus base andmaintenance of the fog. To parameterize physical processes within the marinelayer so precisely as to properly account for these negative feedbacks is aformidable challenge, yet to be achieved.

While we strive to improve our modelling capabilities, we also need topromote the development of measurement tools for the ocean environment.The dew-point sensors that exist on a limited number of buoys have provedcrucial in our studies, giving us valuable data on temporal changes in vapournear the surface. Cloud-fog structure and cloud-top temperature estimatesover the ocean from satellites also have been of great value. Missing, and of

COASTAL SEA FOG 475

critical importance, are the temperature and dew-point structure in theMABL and immediately above it, as well as the spatial and temporal evo-lution of wind fields and turbulence over the ocean, especially near the coast.At present, we are unable to estimate these features of the MABL usingsatellite data.

Based on these recent studies on sea fog off the California coast, physicalprocesses associated with sea fog have been clarified. Yet, operational pre-diction must await the availability of measurements over the sea that will morefaithfully define the initial state and evolutionary process. A certain form ofensemble forecasting, a synthesis of dynamic and probabilistic predictions,will be most meaningful in the next step of solving this challenging problem.

Acknowledgements

Two of the authors (Koracin and Lewis) acknowledge support from theOffice of Naval Research grant N00014-00-1-0524. All authors would like toexpress their gratitude to the late Dr. Dale Leipper for his great motivationand inspiration in coastal sea-fog research. Dr. Jordan Powers of theNational Center for Atmospheric Research is acknowledged for technicalassistance and helpful comments. One of the authors (Koracin) thanks Dr.Michael Tjernstrom of the University of Stockholm in Sweden for helpfulcomments and also Dr. Ragothaman Sundararajan for help in initial analysisduring his appointment at the Desert Research Institute. Mr. TravisMcCord, Mr. Domagoj Podnar, and Mr. Adam Kochanski of the DesertResearch Institute are acknowledged for technical preparation of the man-uscript. All authors are grateful to Mr. Roger Kreidberg of the DesertResearch Institute for his thorough editorial efforts. Implemented sugges-tions from three anonymous reviewers contributed significantly as well.

References

Anderson, J. B.: 1931, ‘Observations from Airplanes of Cloud and Fog Conditions Along theSouthern California coast’, Mon. Wea. Rev. 59, 264–270.

Beg-Paklar, G., Isakov, V., Koracin, D., Kourafalou, V. and Orlic, M.: 2001, ‘A Case Study ofBora-Driven Flow and Density Changes on the Adriatic shelf (January 1987)’, Cont. Shelf

Res. 21, 1751–1783.Betts, A.: 1990, ‘Diurnal Variation of California Coastal Stratocumulus from Two Days of

Boundary Layer Soundings’, Tellus 42A, 302–304.

Byers, H.: 1930, ‘Summer Sea Fogs of the Central California Coast’, Publ. Geogr. 3, 291–338.Deardorff, J. W.: 1980, ‘Cloud Top Entrainment Instability’, J. Atmos. Sci. 37, 131–147.Dudhia, J.: 1989, ‘Numerical Study of Convection Observed During Winter Monsoon

Experiment using a mesoscale two-dimensional model’, J. Atmos. Sci. 46, 3077–3107.


Filonczuk, M. K., Cayan, D. R. and Riddle, L. G.: 1995, Variability of Marine Fog Along the

California Coast, Report 95-2, Scripps Institution of Oceanography, La Jolla, CA, 92093-0224, 91 pp.

Grell, G. A., Dudhia, J. and Stauffer, D. R.: 1994, A Description of the Fifth-Generation Penn

State/NCAR Mesoscale Model (MM5), Techn. Note TN-398, National Center forAtmospheric Research, 122 pp.

Koracin, D. and Rogers, D. P.: 1990, ‘Numerical Simulations of the Response of the Marine

Atmosphere to Ocean Forcing’, J. Atmos. Sci. 47, 592–611.Koracin, D. and Tjernstrom, M.: 1992, ‘Impact of Clouds on a Boundary Layer Structure’, in

preprint, Tenth Symposium on Turbulence and Diffusion, Portland, OR, September 29-

October 2, 1992, American Meteorological Society, 45 Beacon St., Boston, MA, pp. 19–21.Koracin, D. and Dorman, C. E.: 2001, ‘Marine Atmospheric Boundary Layer Divergence and

Clouds along California in June 1996’, Mon. Wea. Rev. 129, 2040–2055.Koracin, D., Dorman, C. E. and Dever, E. P.: 2004, ‘Coastal Perturbations of Marine Layer

Winds, Wind Stress, and Wind Stress Curl along California and Baja California in June1999’, J. Phys. Oceanogr. 34, 1152–1173.

Koracin, D., Lewis, J., Thompson, W. T., Dorman, C. E. and Businger, J. A.: 2001, ‘Tran-

sition of Stratus into Fog along the California Coast: Observations and Modeling’,J. Atmos. Sci. 58, 1714–1731.

Leipper, D. F.: 1948, ‘Fog Development at San Diego, California’, J. Mar. Res. 7, 337–346.

Leipper, D. F.: 1994, ‘Fog on the United States West Coast: a review’, Bull. Amer. Meteorol.Soc. 75, 229–240.

Leipper, D. F. and Koracin, D.: 1998, ‘Hot Spells and their Role in Forecasting Weather

Events on the U.S. West Coast’, in preprint, Second Conference on Coastal Atmospheric andOceanic Prediction and Processes, Phoenix, AZ, January 11–16, 1998, American Meteo-rological Society, 45 Beacon St., Boston, MA, pp. 127–132.

Lewis, J., Koracin, D., Rabin, R. and Businger, J.: 2003, ‘Sea Fog off the California Coast:

Viewed in the Context of Transient Weather Systems’, J. Geoph. Res. (Atmos)., 108, No.D15, 4457, doi 10.1029/2002JD002833.

Lewis, J., Koracin, D. and Redmond, K.: 2004, ‘Sea Fog Research in the UK and USA:

Historical Essay Including Outlook’, Bull. Amer. Meteorol. Soc. 85, 395–408.Mack, E. J., Katz, U., Rogers, C. and Pilie, R.: 1974, The Microstructure of California

Coastal Stratus and Fog at Sea, Rep. CJ-5405-M-1, Calspan Corp., 74 pp.

Nakanishi, M.: 2000, ‘Large Eddy Simulation of Fog Radiation’, Boundary-Layer Meteorol.94, 461–493.

Nicholls, S.: 1984, ‘The Dynamics of Stratocumulus: Aircraft Observations and Comparisonwith a Mixed Layer Model’, Quart. J. Roy. Meteorol. Soc. 110, 783–820.

Oliver, D., Lewellen, W. and Williamson, G.: 1978, ‘The Interaction Between Turbulent andRadiative Transport in the Development of Fog and Low-level Stratus’, J. Atmos. Sci. 35,301–316.

Petterssen, S.: 1936, ‘On the Causes and the Forecasting of the California Fog’, J. Aerosp. Sci.3, 305–309.

Petterssen, S.: 1938, ‘On the Causes and the Forecasting of the California Fog’, Bull. Amer.

Meteorol. Soc. 19, 49–55.Pilie, R. J., Mack, E. J., Rogers, C. W., Katz, U. and Kocmond, W. C.: 1979, ‘The Formation

of Marine Fog and the Development of Fog-stratus Systems along the California Coast’,

J. Appl. Meteorol. 18, 1275–1286.Powers, J. G. and Stoelinga, M. T.: 1999, ‘A Coupled Air-sea Mesoscale Model: Experiments

in Atmospheric Sensitivity to Marine Roughness’, Mon. Wea. Rev. 128, 208–228.

COASTAL SEA FOG 477

Randall, D.: 1980, ‘Conditional Instability of the First Kind Upside Down’, J. Atmos. Sci. 37,

125–130.Rogers, D. P. and Koracin, D.: 1992, ‘Radiative Transfer and Turbulence in the Cloud-

Topped Marine Atmospheric Boundary Layer’, J. Atmos. Sci. 49, 1473–1486.

Rogers, D. P., Dorman, C. E., Edwards, K. A., Brooks, I. M., Melville, W. K., Burk, S. D.,Thompson, W. T., Holt, T., Strom, L. M., Grisogono, B., Bane, J. M., Nuss, W. A.,Morley, B. M. and Schanot, A. J.: 1998, ‘Highlights of Coastal Waves 1996’, Bull. Amer.

Meteorol. Soc. 79, 1307–1326.Stage, S. A. and Businger, J. A.: 1981a, ‘A Model for Entrainment into a Cloud-Topped

Marine Boundary-Layer. Part I: Model Description and Application to a Cold-air Out-

break Episode’, J. Atmos. Sci. 38, 2213–2229.Stage, S. A. and Businger, J. A.: 1981b, ‘A Model for Entrainment into a Cloud-Topped

Marine Boundary-Layer. Part II: Discussion of Model Behaviour and Comparison withOther Models’, J. Atmos. Sci. 38, 2230–2242.

Shafran, P. C., Seaman, N. L. and Gayno, G. A.: 2000, ‘Evaluation of Numerical Predictionsof Boundary Layer Structure during the Lake-Michigan Ozone Study’, J. Appl. Meteorol.39, 412–426.

Stevens B., Lenschow, D. H., Vali, G., Gerber, H., Bandy, A., Blomquist, B., Brenguier, J.-L.,Bretherton, C. S., Burnet, F., Campos, T., Chai, S., Faloona, L., Friensen, D., Haimov, S.,Laursen, K., Lilly, D. K., Loehrer, S. M., Malinowski, S. P., Morely, B., Peners, M. D.,

Rogers, D. C., Russell, L., Savic-Jovcic, V., Snider, J. R., Straub, D., Szumowski, M. J.,Takagi, H., Thorton, D. C., Tshudi, M., Twohy, C., Wetzel, M. and Van Zanten, M. C.,2003a, �Dynamics and Chemistry of Marine Stratocumulus–DYCOMS-II�. Bull. Amer.

Meteorol. Soc., 84, 579–593.Stevens, B., Lenschow, D. H., Faloona, I., Moeng, C. H., Lilly, D. K., Blomquist, B., Vali, G.,

Bandy, A., Campos, T., Gerber, H., Haimov, S., Morley, B., Thornton, D. C., 2003b: �OnEntrainment Rates in Nocturnal Marine Stratocumulus�. Quart. J. Roy. Meteorol. Soc.

129, 3469–3493.Stull, R. B.: 1988, An Introduction to Boundary Layer Meteorology, Kluwer Academic

Publishers, The Netherlands, 666.

Taylor, G. I.: 1917, ‘The Formation of Fog and Mist’, Quart. J. Roy. Meteorol. Soc. 43,241–268.

Tjernstrom, M. and Koracin, D.: 1995, ‘Modeling the Impact of Marine Stratocumulus on the

Boundary-Layer Structure’, J. Atmos. Sci. 52, 863–878.


formation, evolution, and dissipation of coastal sea fog

Documents

evolution of sea fog

fog dissipation

seafog event

cloudfog transition

oshore fog

fog usingsatellite data

fogfree conditions

area wherethe fog