• Ballard, S. P., , B. W. Golding, , and R. N. B. Smith, 1991: Mesoscale model experimental forecasts of the Haar of northeast Scotland. Mon. Wea. Rev., 119 , 21072123.

    • Search Google Scholar
    • Export Citation
  • Bao, J-W., , J. M. Wilczak, , J-K. Choi, , and L. H. Kantha, 2000: Numerical simulations of air–sea interaction under high wind conditions using a coupled model: A study of hurricane development. Mon. Wea. Rev., 128 , 21902210.

    • Search Google Scholar
    • Export Citation
  • Barker, E. H., 1992: Design of the Navy’s multivariate optimum interpolation analysis system. Wea. Forecasting, 7 , 220231.

  • Bendix, J., , B. Thies, , J. Cermak, , and T. Nauß, 2005: Ground fog detection from space based on MODIS daytime data—A feasibility study. Wea. Forecasting, 20 , 9891005.

    • Search Google Scholar
    • Export Citation
  • Bergot, T., , and D. Guedalia, 1994: Numerical forecasting of radiation fog. Part I: Numerical model and sensitivity tests. Mon. Wea. Rev., 122 , 12181230.

    • Search Google Scholar
    • Export Citation
  • Cho, Y-K., , M-O. Kim, , and B-C. Kim, 2000: Sea fog around the Korean Peninsula. J. Appl. Meteor., 39 , 24732479.

  • Cotton, W. R., , and R. A. Anthes, 1989: Storm and Cloud Dynamics. Academic Press, 833 pp.

  • Filonczuk, M., , D. Cayan, , and L. Riddle, 1995: Variability of marine fog along the California coast. Scripps Institution of Oceanography Rep. 95-2, 102 pp.

    • Search Google Scholar
    • Export Citation
  • Findlater, J., , W. T. Roach, , and B. C. McHugh, 1989: The Haar of north-east Scotland. Quart. J. Roy. Meteor. Soc., 115 , 581608.

  • Fisher, E. L., , and P. Caplan, 1963: An experiment in numerical prediction of fog and stratus. J. Atmos. Sci., 20 , 425437.

  • Fuzzi, S., and Coauthors, 1992: The Po Valley fog experiment 1989: An overview. Tellus, 44B , 448468.

  • Gao, S., , H. Lin, , B. Shen, , and G. Fu, 2007: A heavy sea fog event over the Yellow Sea in March 2005: Analysis and numerical modeling. Adv. Atmos. Sci., 24 , 6581.

    • Search Google Scholar
    • Export Citation
  • Gultepe, I., , M. D. Müller, , and Z. Boybeyi, 2006: A new visibility parameterization for warm-fog applications in numerical weather prediction models. J. Appl. Meteor. Climatol., 45 , 14691480.

    • Search Google Scholar
    • Export Citation
  • Gultepe, I., and Coauthors, 2007: Fog research: A review of past achievements and future perspectives. Pure Appl. Geophys., 164 , 11211159.

    • Search Google Scholar
    • Export Citation
  • He, R., , and R. H. Weisberg, 2002: West Florida shelf circulation and temperature budget for the 1999 spring transition. Cont. Shelf Res., 22 , 719748.

    • Search Google Scholar
    • Export Citation
  • He, R., , and R. H. Weisberg, 2003: West Florida shelf circulation and temperature budget for the 1998 fall transition. Cont. Shelf Res., 23 , 777800.

    • Search Google Scholar
    • Export Citation
  • Heo, K-Y., , and K-J. Ha, 2004: Classification of synoptic pattern associated with coastal fog around the Korean Peninsula (in Korean with English abstract). J. Korean Metor. Soc., 40 , 541556.

    • Search Google Scholar
    • Export Citation
  • Hodur, R., 1997: The Naval Research Laboratory’s Coupled Ocean–Atmosphere Mesoscale Prediction System (COAMPS). Mon. Wea. Rev., 125 , 14141430.

    • Search Google Scholar
    • Export Citation
  • Hogan, T. F., , T. E. Rosmond, , and R. Gelaro, 1991: The NOGAPS forecast model: A technical description. Naval Oceanographic and Atmospheric Research Laboratory Rep. 13, 220 pp. [Available from Naval Research Laboratory, Monterey, CA 93943–5502].

    • Search Google Scholar
    • Export Citation
  • Koračin, D., , J. Lewis, , W. T. Thompson, , C. E. Dorman, , and J. A. Businger, 2001: Transition of stratus into fog along the California coast: Observations and modeling. J. Atmos. Sci., 58 , 17141731.

    • Search Google Scholar
    • Export Citation
  • Koračin, D., , J. A. Businger, , C. E. Dorman, , and J. M. Lewis, 2005: Formation, evolution, and dissipation of coastal sea fog. Bound.-Layer Meteor., 117 , 447478.

    • Search Google Scholar
    • Export Citation
  • Kraus, H., 1993: Surface-layer fog considered using a modified Monin–Obukhov theory. Bound.-Layer Meteor., 66 , 219235.

  • Krishnamurti, T. N., , and L. Bounoua, 1996: An Introduction to Numerical Weather Prediction Techniques. CRC Press, 293 pp.

  • Kunkel, B. A., 1984: Parameterization of droplet terminal velocity and extinction coefficient in fog models. J. Climate Appl. Meteor., 23 , 3441.

    • Search Google Scholar
    • Export Citation
  • Larson, J., , R. Jacob, , and E. Ong, 2005: The model coupling toolkit: A new Fortran90 toolkit for building multiphysics parallel coupled models. Int. J. High Perform. Comput. Appl., 19 , 277292.

    • Search Google Scholar
    • Export Citation
  • Leipper, D. F., 1994: Fog on the U.S. West Coast: A review. Bull. Amer. Meteor. Soc., 75 , 229240.

  • Lewis, J., , D. Koračin, , R. Rabin, , and J. Businger, 2003: Sea fog off the California coast: Viewed in the context of transient weather systems. J. Geophys. Res., 108 , 4457. doi:10.1029/2002JD002833.

    • Search Google Scholar
    • Export Citation
  • Lewis, J., , D. Koračin, , and K. Redmond, 2004: Sea fog research in the UK and USA: Historical essay including outlook. Bull. Amer. Meteor. Soc., 85 , 395408.

    • Search Google Scholar
    • Export Citation
  • Mahrt, L., 2007: Weak-wind mesoscale meandering in the nocturnal boundary layer. Bound.-Layer Meteor., 7 , 331347.

  • Musson-Genon, L., 1987: Numerical simulation of a fog event with a one-dimensional boundary layer model. Mon. Wea. Rev., 115 , 592607.

    • Search Google Scholar
    • Export Citation
  • Oliver, D. A., , W. S. Lewellen, , and G. G. Williamson, 1978: The interaction between turbulent and radiative transport in the development of fog and low-level stratus. J. Atmos. Sci., 35 , 301316.

    • Search Google Scholar
    • Export Citation
  • Perlin, N., , and E. D. Skyllingstad, 2007: Numerical simulation of air–sea coupling during coastal upwelling. J. Phys. Oceanogr., 37 , 20812093.

    • Search Google Scholar
    • Export Citation
  • Pilié, R. J., , E. J. Mack, , C. W. Rogers, , U. Katz, , and W. C. Kocmond, 1979: The formation of marine fog and the development of fog-stratus systems along the California coast. J. Appl. Meteor., 18 , 12751286.

    • Search Google Scholar
    • Export Citation
  • Saunders, P. M., 1964: Sea smoke and steam fog. Quart. J. Roy. Meteor. Soc., 90 , 156165.

  • Shchepetkin, A. F., , and J. C. McWilliams, 2005: The Regional Ocean Modeling System (ROMS): A split-explicit, free surface, topography-following coordinate oceanic model. Ocean Modell., 9 , 347404.

    • Search Google Scholar
    • Export Citation
  • Skyllingstad, E. D., , D. Vickers, , L. Mahrt, , and R. Samelson, 2007: Effects of mesoscale sea-surface temperature fronts on the marine boundary layer. Bound.-Layer Meteor., 123 , 219237.

    • Search Google Scholar
    • Export Citation
  • Stoelinga, M. T., , and T. T. Warner, 1999: Nonhydrostatic, mesobeta-scale model simulations of cloud ceiling and visibility for an East Coast winter precipitation event. J. Appl. Meteor., 38 , 385404.

    • Search Google Scholar
    • Export Citation
  • Taylor, G., 1917: The formation of fog and mist. Quart. J. Roy. Meteor. Soc., 43 , 241268.

  • Virmani, J. I., , and R. H. Weisberg, 2003: Features of the observed annual ocean–atmosphere flux variability on the West Florida shelf. J. Climate, 16 , 734745.

    • Search Google Scholar
    • Export Citation
  • Wang, B., 1985: Sea Fog. China Ocean Press, 330 pp.

  • View in gallery

    The target area and fog cases over the Yellow Sea. NASA Terra MODIS visible images and fog observation stations (filled circles) for (a) an advection fog case at 1140 LST 6 Jul 2008 and (b) a steam fog case at 1100 LST 5 Jun 2006. The stations marked by A, B, C, D, E, F, and G are the reference point used to validate the model results. The dashed lines indicate ground fog detection obtained using the method reported by Bendix et al. (2005).

  • View in gallery

    Surface weather chart at (a) 0900 LST 5 Jul 2008 and (b) 0900 LST 5 Jun 2006. The solid lines represent the isobars (4-hPa interval) and the gray solid line with hatches represents the rainy cloud region.

  • View in gallery

    (a) Surface observations and (b) simulation results of the coupled model at station A marked in Fig. 1a from 2100 LST 5 Jul to 2100 LST 6 Jul 2008. Solid line, dashed line, dash–dot line, long dashed line, and shaded area indicate the air temperature (°C), relative humidity (%), wind speed (m s−1), sea surface temperature (°C), and fog, respectively. The upper arrows indicate the wind vector. “Fog” indicates the time of visibility ≤ 1 km in (a) observation and (b) model-based visibility obtained using parameterization presented by Stoelinga and Warner (1999).

  • View in gallery

    As in Fig. 3, but at station B marked in Fig. 1b during the period between 2100 LST 4 Jun and 2100 LST 5 Jun 2006.

  • View in gallery

    (a) A map of bathymetry over the ocean adjoining the west coast of Korea. Horizontal distributions of sea surface temperature (b) on 6 Jul 2008 (advection fog case) and (c) on 5 Jun 2006 (steam fog case).

  • View in gallery

    Temporal variation in sea surface temperature of buoy and coupled model for the (a) advection fog and (b) steam fog cases. The buoys are selected near stations A and B, respectively.

  • View in gallery

    The simulated fog area (shaded, visibility ≤ 1 km), LWC at 10 m (solid line, contour interval: 0.05 g m−3), and sea surface winds (vectors) of (a) coupled and (b) uncoupled models at 0900 LST 6 Jul 2008. The filled circles represent the fog observation station with a visibility of 1 km. The box marked in Fig. 7a indicates the location where the difference between the coupled and uncoupled models was analyzed. The model-based visibility was obtained using the parameterization reported by Stoelinga and Warner (1999).

  • View in gallery

    Time series of the simulated extinction coefficient by (a) the coupled and (b) the uncoupled models at the A station during the period 2100 LST 5 Jul–2100 LST 6 Jul 2008. (c),(d) As in (a),(b), but for the center of the box squared in Fig. 7a, respectively. The threshold of fog is shown with the dashed line.

  • View in gallery

    The difference between the coupled and uncoupled simulations in the ASTD (air temperature minus SST; shaded, °C) and LWC (contour, g m−3) at (a) 0600 LST 6 Jul (starting time) and (b) 0900 LST 6 Jul 2008 (mature time). The contour interval is 0.02 g m−3.

  • View in gallery

    Time series for the simulated wind stress by using the (a) coupled and (b) uncoupled models at the A station during the period 2100 LST 5 Jul–2100 LST 6 Jul 2008. The difference in (c) latent heat flux and (d) sensible heat flux between the coupled and uncoupled models. (e),(f) As in (a),(b), but for air temperature and SST.

  • View in gallery

    Time series for the simulated wind stress by using the (a) coupled and (b) uncoupled models at the center of the box (i.e., the TA) squared in Fig. 7a during the period 2100 LST 5 Jul–2100 LST 6 Jul 2008. (c),(d) As in (a),(b), but for turbulent heat fluxes. (e),(f) As in (a),(b), but for air temperature and SST.

  • View in gallery

    Vertical profiles of (a),(b) potential temperature and (c),(d) horizontal wind speed taken from the (left) coupled and (right) uncoupled models at the center of the box identified in the insert of Fig. 7a. The times represents the formation stage (0600 LST), the mature stage (0900 LST), and the dissipation stage (1200 LST) of advection fog.

  • View in gallery

    Vertical profile of (a) TKE difference between coupled and uncoupled models in MABL and (b) TKE taken from coupled model in the ocean mixed layer at the A station during the period 2100 LST 5 Jul–2100 LST 6 Jul 2008.

  • View in gallery

    Representative surface backward trajectories where advection fogs forms in the coupled simulation. Filled circles and circles indicate fog and nonfog, respectively. The time interval is 1 h.

  • View in gallery

    Changes in SST, SHF, and LHF are shown along trajectory A marked in Fig. 14 for (a) coupled and (b) uncoupled models.

  • View in gallery

    The simulated fog area (shaded, visibility ≤ 1 km), LWC at 10 m (solid line, level = 0.1, 0.2, 0.4, and 0.6 g m−3), and sea surface winds (vectors) of the (a) coupled and (b) uncoupled models at 1100 LST 5 Jun 2006. The filled circle represents the fog observation station with a visibility of 1 km. The box marked in (a) indicates the location where the difference between the coupled and uncoupled model was analyzed. The model-based visibility was obtained using the parameterization reported by Stoelinga and Warner (1999).

  • View in gallery

    Time series for the simulated extinction coefficient by the (a) coupled and (b) uncoupled models at the B station during the period 2100 LST 4 Jun–2100 LST 5 Jun 2006. The threshold of fog is shown with a dashed line.

  • View in gallery

    Time series for the simulated wind stress by the (a) coupled and (b) uncoupled models at the B station during the period 2100 LST 4 Jun–2100 LST 5 Jun 2006. (c),(d) As in (a),(b), but for the turbulent heat fluxes. (e),(f) As in (a),(b), but for air temperature and SST, respectively.

  • View in gallery

    Vertical profiles of the potential temperature taken from the coupled (solid line) and uncoupled models (dashed line) at the B station. The times represents (a) the formation stage (0300 LST), (b) the mature stage (0600 LST), and (c) the dissipation stage (1000 LST) of steam fog.

  • View in gallery

    Representative surface backward trajectories where steam fogs forms in (a) a coupled and (b) an uncoupled simulation. Filled circles and circles indicate fog and no fog, respectively. The time interval is 1 h.

  • View in gallery

    Changes in SST, SHF, and LHF are shown along trajectory B marked in Fig. 20 for the (a) coupled and (b) uncoupled model.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 90 90 6
PDF Downloads 76 76 6

A Coupled Model Study on the Formation and Dissipation of Sea Fogs

View More View Less
  • 1 Division of Earth Environmental System, College of Natural Science, Pusan National University, Busan, South Korea
© Get Permissions
Full access

Abstract

This study examined the impact of air–sea coupling using a coupled atmosphere–ocean modeling system consisting of the Coupled Ocean–Atmosphere Mesoscale Prediction System as the atmospheric component and the Regional Ocean Modeling System as the oceanic component. Numerical experiments for advection and steam fog events were carried out to clarify the modulation of the formation and dissipation of sea fogs by the air–sea temperature difference (air temperature minus sea surface temperature) and the atmospheric stability. The coupled simulation showed that advection fog is obviously controlled by low-level atmospheric stability and downward latent heat flux with oceanic cooling through air–sea coupling. In particular, air–sea coupling stabilizes the low-level atmosphere at the dissipation stage, and then suppresses vertical mixing, which retards the dissipation of advection fog. In the case of a steam fog event, the upward turbulent heat fluxes are increased significantly from the formation time to the mature time. A decrease in sea surface temperature cools the low-level atmosphere, which increases the condensation rate and low-level atmospheric stability, eventually retarding the dissipation of steam fog.

Corresponding author address: Kyung-Ja Ha, Division of Earth Environmental System, Pusan National University, Busan 609-735, South Korea. Email: kjha@pusan.ac.kr

Abstract

This study examined the impact of air–sea coupling using a coupled atmosphere–ocean modeling system consisting of the Coupled Ocean–Atmosphere Mesoscale Prediction System as the atmospheric component and the Regional Ocean Modeling System as the oceanic component. Numerical experiments for advection and steam fog events were carried out to clarify the modulation of the formation and dissipation of sea fogs by the air–sea temperature difference (air temperature minus sea surface temperature) and the atmospheric stability. The coupled simulation showed that advection fog is obviously controlled by low-level atmospheric stability and downward latent heat flux with oceanic cooling through air–sea coupling. In particular, air–sea coupling stabilizes the low-level atmosphere at the dissipation stage, and then suppresses vertical mixing, which retards the dissipation of advection fog. In the case of a steam fog event, the upward turbulent heat fluxes are increased significantly from the formation time to the mature time. A decrease in sea surface temperature cools the low-level atmosphere, which increases the condensation rate and low-level atmospheric stability, eventually retarding the dissipation of steam fog.

Corresponding author address: Kyung-Ja Ha, Division of Earth Environmental System, Pusan National University, Busan 609-735, South Korea. Email: kjha@pusan.ac.kr

1. Introduction

The formation and duration of sea fog are related to thermodynamical, dynamical and physical processes as well as the sea surface conditions. Because sea fog has a spatial structure on both the microscale, mesoscale, and time scale in hours, the formation and evolution of sea fog is modulated by variety of complicated dynamic and microphysical processes (Gao et al. 2007). Gultepe et al. (2007) reported that a sea fog is a boundary layer phenomenon. Therefore, its formation and evolution are strongly affected by the surface conditions, which are determined by the state of the sea surface. The dynamical response of atmospheric models can be remarkably sensitive to temporal changes in sea surface temperature (SST) and thermodynamic flux at the air–sea interface (Bao et al. 2000).

The most common types of sea fog are advection fog and steam fog. An advection fog is formed when warm and moist air moves over colder water, and the consequent cooling of the air to below its dew point. A steam fog occurs when cold air drifts across warmer water causing condensation in the evaporating water vapors. Earlier studies of sea fogs based on observations have provided an understanding on the physical processes for the formation of sea fog. For example, turbulent heat and moisture transfer, the cooling of an air mass over colder water (Taylor 1917), and radiative cooling at the top of the fog layer (Koračin et al. 2001; Lewis et al. 2003, 2004) play an important role in the formation of sea fog. Taylor (1917) studied sea fogs along the Grand Banks and found that large horizontal gradients in SST make an advection fog. The most advection fogs occur when the air temperature in approximately 1°–2°C warmer than the SST and the wind speed ranges from 2 to 8 m s−1. In addition, weak land breeze and the difference of about 10°C between SST and the inland air temperature provided favorable conditions for the formation of steam fog (Saunders 1964). The development of sea fog is related to hydrological factors with the local sea current and SST (Wang 1985). Leipper (1994) demonstrated that advection fog forms when the air temperature heated by adiabatic heating is >5°C. He also reported that capping inversion could reach 200–300 m above the ocean because of the small-scale “water vapor driven” convection and other mixing processes. Filonczuk et al. (1995) has noted that the most sea fog along the California coast occurs at wind speeds below ∼4–6 m s−1, but the sea fog often forms in strong wind conditions between 8 and 15 m s−1. They also suggested that the sea fog forms because of the local marine environment and the diminished winds may permit the sea fog to persist once it formed. Heo and Ha (2004) showed similar results for this study in the west coast of Korea. In addition, it was reported that sea fog is formed by a stratus lowering process, which is associated with microphysical processes and entrainment (Oliver et al. 1978; Pilié et al. 1979), and steam fog is triggered by instability and mixing over warm water patches (Pilié et al. 1979).

A number of numerical studies have been performed over the last few decades since Fisher and Caplan (1963) first attempted a numerical prediction of fog. Musson-Genon (1987) placed new emphasis on the basic roles of turbulence. He reported that the appearance and suppression of a radiation fog depended on the turbulence intensity. Bergot and Guedalia (1994) illustrated the importance of the initial humidity profiles along with their role in a radiation fog formation and evolution. In addition to the numerical studies of radiation fog, sea fog was also examined using a numerical model. Oliver et al. (1978) suggested that radiative cooling at the advection fog top is an important process in the fog life cycle. Koračin et al. (2001) suggested that radiative cooling and large-scale subsidence are important factors for fog formation due to stratus lowering using a one-dimensional (1D) model. Ballard et al. (1991) attempted prediction of the advection fog event off the Scottish coast using a regional mesoscale model. They pointed out the importance of the initial conditions, vertical resolution, and the parameterization of turbulence using a 3D numerical weather prediction model to simulate sea fog. According to Findlater et al. (1989), the evolution of sensible and latent heat fluxes is one of the factors that determined the evolution of the advection fog. Gultepe et al. (2007) noted that the difference in vapor pressure between the cold air and the warm ocean leads to the formation of steam fog by evaporation and a mixing of water vapor into the cold air. Kraus (1993) has placed emphasis on the structure of steam fog as a result of divergences of turbulent fluxes of water vapor, heat, and liquid water content (LWC). A sea fog is formed under high relative humidity, which is determined, not only by the meteorological conditions, such as air temperature, humidity, wind, and stability, but also by the oceanic conditions through surface flux (Wang 1985; Filonczuk et al. 1995; He and Weisberg 2002, 2003; Virmani and Weisberg 2003; Koračin et al. 2001). The numerical study by Koračin et al. (2005) concluded that sea fog depends on the variation in SST, which affects the low-level stability as well as the surface heat and moisture flux within the near-surface layer. More recently, Skyllingstad et al. (2007) reported that a marine boundary layer is cooled by the movement of air from a warm SST over a cold SST with the construction of a cold internal boundary layer, which forms with much cooler air, stronger stratification, and weaker turbulence. These conditions often lead to the development of fog. Therefore, accurate data on the variations in SST are needed to accurately determine the effect of air–sea coupling and accurate heat flux estimations from bulk parameterizations on the formation and duration of sea fog. Although there are few reports on the effects of air–sea coupling, our current understanding of the temporal and spatial variations in the SST on the formation and dissipation of sea fog and the effect of air–sea coupling is incomplete. Sensible and latent heat fluxes modified by the variation in SST determine the evolution of sea fog together with the modification of the low-level atmospheric stability. These conditions of air and ocean set up the environment that controls the formation and dissipation of the sea fog.

To confirm this hypothesis, this study examined the effect of air–sea coupling on the formation and dissipation of sea fog. The effect of variations in SST on the formation and dissipation of advection and steam fog events were investigated by carrying out numerical experiments for an advection fog event and a steam fog event using a mesoscale coupled model and an uncoupled atmospheric model. In addition, the factors associated with the lifetime and density of sea fog were examined. To accomplish this, two sea fog case studies over the Yellow Sea were considered. Section 2 presents a brief description of the coupled model and experimental design. Section 3 describes advection and steam fog case studies with observational analyses. Section 4 discusses the effect of air–sea interaction, and section 5 reports the discussion and conclusions.

2. Model description and experimental design

A mesoscale model is needed to simulate accurately sea fog events that occur on a time scale of a few hours and spatial structure on both the microscale and mesoscale. Under weak large-scale flow, mesoscale motions lead to a meandering of the wind direction, and the generation of turbulence by unresolved mesoscale motions becomes important (Mahrt 2007). A coupled ocean–atmosphere mesoscale model, which including the Naval Research Laboratory Coupled Ocean–Atmosphere Mesoscale Prediction System (COAMPS;1 Hodur 1997) as the atmospheric component and the Regional Ocean Modeling System (ROMS; Shchepetkin and McWilliams 2005) as the ocean component, was developed for the west coast of the Korean Peninsula. Data exchange using the Model Coupling Toolkit (MCT; Larson et al. 2005; more information is available online at http://www-unix.mcs.anl.gov/mct) was performed at 20-s intervals during the data exchange sessions via the MCT. The ocean model receives the predicted wind stress and heat flux from the atmospheric model, and the atmosphere receives the SST updated from the ocean model. For a detailed description of the coupled mesoscale model for system components and model coupling strategy, please refer to the report of Perlin and Skyllingstad (2007).

Table 1 lists the numerical experiments. The model domain is 32.0°–38.5°N and 123.8°–128.0°E, with a horizontal resolution of 0.03° (about 3 km) and a vertical resolution of 30 layers for both the atmospheric and ocean models. The simulations denoted by the uncoupled COAMPS are the control run, and the coupled simulations employ the coupled COAMPS–ROMS model system. In all simulations, the lateral boundary conditions were provided by the U.S. Navy Operational Global Analysis and Prediction System (NOGAPS; Hogan et al. 1991). In addition to NOGAPS, observational data including the surface, upper-air sounding, aircraft and satellite data (from the Global Ocean Data Assimilation Experiment (GODAE); more information is available online at http://www.usgodae.org), which were quality controlled (Barker 1992), were assimilated into the initial model fields using multivariate optimum interpolation analysis (MVOI). The initial SSTs were created using the NOGAPS SST, satellite data from Geostationary Operational Environmental Satellites (GOES) and Multichannel Sea Surface Temperature (MCSST), and observation data from two buoys (from GODAE), located near the west coast of the Korean Peninsula. In the uncoupled runs, fixed SSTs were used during the integration period. In addition to the initial SSTs, the coupled simulations used the 10-yr spinup temperatures, salinity, and currents formed by cycling on the first year. Daily QuikSCAT winds were also used.

3. Experimental cases

a. Advection fog (5–6 July 2008)

An extensive cloud and sea fog area was observed over the Yellow Sea on 5 and 6 July 2008. Figure 1a shows the Moderate Resolution Imaging Spectroradiometer (MODIS) visible image of the National Aeronautics and Space Administration (NASA) Terra satellite, which was obtained at 1140 LST2 on 6 July 2008. A large patch of sea fog appeared in the Yellow Sea, which can be identified as fog according to the method reported by Bendix et al. (2005) and low visibility (<1 km) based on surface observations. The heights at the top of the fog over the west coast of Korea were estimated to be approximately 237 and 351 m according to sounding data obtained from the A station (47102, 38.0°N, 124.6°E) and B station (47169, 34.7°N, 125.5°E) marked in Fig. 1a, respectively. The depth of the fog is defined as a dew point depression <3°C below the inversion layer. The height at the top of the fog at the stations was lower, and was estimated to be approximately 250 m at 0900 LST 6 July 2008.

A synoptic condition of this advection fog event was characterized by a light-to-moderate southerly wind that supplied warm, moist air over the west coast (Fig. 2a). A high pressure system over the northwest Pacific region persisted from 2100 LST 5 July to 0900 LST 6 July 2008. A synoptic low propagated eastward in the northern area of the Korean Peninsula and promoted an inflow of warm, moist air from the south. As a result, the SST of the Yellow Sea was much cooler than the air temperature. Therefore, the advection fog originated from this area. Prior to the formation of the advection fog, the west coast was dominated by a stationary high pressure system with a stable atmosphere and warm-moisture advection. This synoptic condition was one of the important factors that induce the development of advection fog over the Yellow Sea (Heo and Ha 2004).

Surface observation at the B station indicated that this advection fog could be characterized by durations longer than 48 h. During the formation of an advection fog, strong tidal mixing produced cold water (18°–19°C) in the southwest coastal region. The surface temperature in the coastal region is lower by 2°C than that in the offshore area because of tidal mixing (cf. Fig. 7 of Cho et al. 2000). An analysis of the sea level observation data at the B’ station from the National Oceanographic Research Institute indicated strong tidal mixing, in the range from 26 to 373 cm during the period between 2022 LST 5 July and 2156 LST 6 July 2008. Strong tidal mixing is one of the main factors associated with the formation of advection fog over the Yellow Sea in summer (Cho et al. 2000).

To validate the simulation result of the coupled model, Fig. 3 shows the temporal variation for the observed and simulated variables including wind direction, wind speed, relative humidity, air temperature, and fog at A station during the advection fog event between 2100 LST 5 July and 2100 LST 6 July 2008. The formation and dissipation of advection fog over the ocean was validated by selecting the observational data at the A station only instead of the B and C stations because the fog was maintained at the B station during the experimental period and the C station was located near land. The fog showed a persistent time longer than 7 h between 0700 and 1300 LST 6 July (Fig. 3a). From 1300 LST 6 July, the relative humidity decreased to 65% with increasing air temperature to 26°C and decreasing southerly wind speed to 2 m s−1. As shown in Fig. 3b, a positive air–sea temperature difference (ASTD) was maintained during the fog event. The relative humidity decreased, even though the maximum ASTD was observed at the A station at 1500 LST 6 July 2008. That may be the result of an increase in air temperature from 21.0° to 23.5°C. Although the simulated fog dissipated 1 h later than the observed fog, the coupled model provided meaningful results for the formation and duration of advection fog. Table 2 shows the root-mean-square error (RMSE) for the simulated 10-m air temperatures, relative humidity, wind direction, and wind speed at stations A, B, and C. The temporal changes in the simulated wind, temperature, and relative humidity are in good agreement with those for the observed variables, even though the magnitude of the simulated wind speed and air temperature was lower than that of the observed values.

b. Steam fog (4–5 June 2006)

An extensive sea fog occurred over the Yellow Sea between 4 and 5 June 2006. Figure 1b shows the NASA Terra MODIS visible images at 1100 LST 5 June 2006. The sea fog detected using the method reported by Bendix et al. (2005) and the low visibility ≤ 1 km from the surface observations can be clearly seen in the middle (near 38°N) and southwest coast (near 34°N) of Korea.

In the surface weather maps on 4–5 June 2006, a high pressure center was located in the western part of the Korean Peninsula before the formation of the sea fog, which induced a southwesterly wind on the middle of the west coast and a northeasterly wind on the southwest coast of Korea (Fig. 2b). At 1200 LST 5 June, a high pressure system was persistent, and sea fog was reported in those regions. A northerly wind was dominant over the southwest coast during the negative ASTD (air temperature < SST). However, a southwesterly wind dominated during the positive ASTD in the middle of the west coast. Therefore, steam and advection fog was located on the southwest coast and the middle of the west coast, respectively. The steam fog remained for 9 h in the nighttime at the B station marked in Fig. 1b and the advection fog persisted for 26 h at the A station. The prevailing wind directions were from the north-northwest to the northeast during the steam fog event, and the prevailing wind speed was 2–4 m s−1 at the B station (Fig. 4a). The height at the top of the fog at the B station was estimated to be approximately 80 m according to sounding data. A shallow surface-based inversion, which caps the saturated layer from 168 to 236 m, existed at the B station.

To validate the results of the coupled simulation, Fig. 4 shows the temporal variation for the observed and simulated variables, such as the wind direction, wind speed, relative humidity, air temperature, and fog at the B station during the steam fog event between 2100 LST 4 June and 2100 LST 5 June 2006. The observational data at the A station was used for validation because the fog event around the station was advection fog. The D and E stations were also omitted because these stations are automatic weather stations that do not produce visibility data. In Fig. 4a, as compared with the advection fog, the steam fog event is characterized by a northerly wind and a decrease in air temperature prior to fog formation. A change in wind direction from the south to north was observed at the time of steam fog formation. The steam fog dissipated when the wind speed increased to approximately 5 m s−1 and the air temperature increased above 15.5°C. Figure 4b shows that the simulated air temperature and relative humidity are similar to the observed values. The variation in the simulated 10-m air temperature correlated with this observation but the maximum air temperature was underestimated. The variation in relative humidity appeared to follow the trend in the observations quite closely. Both the simulated wind speed and wind direction by the coupled model were slightly different from observed values. In particular, the wind speed was smaller than the observed wind speed during the fog event. The simulated fog dissipated 2 h earlier than the observed data because of an increase in air temperature above the SST from 0900 LST 5 June 2006. Table 3 shows the RMSE statistics for the coupled model output at stations B, D, and E denoted in Fig. 1b. For the air temperature, the model-data RMSE for the three stations was 1.57°–2.09°C, indicating that the model performance for temperature is quite reasonable. The model reproduced the variation in relative humidity well, as illustrated by the RMSE values, which range from 3.04% to 19.74%. Overall, the model performance for relative humidity was not as good as those for the air temperature. For the wind speed, the simulated winds had an RMSE of 0.83–2.06 m s−1, suggesting good model performance at the stations except for the B station. However, in the case of wind direction, there were high RMSE values up to 40° at the stations except for the D station because the stations are under the influence of local surface roughness due to the weak wind conditions.

c. Validation of sea surface temperature

The numerical simulations using the coupled model have been performed over the eastern part of the Yellow Sea, which is a typical shallow and semienclosed sea (Fig. 5a). The horizontal distribution of cold water is corresponded with that of shallow depth of the west coast of Korea. Figure 6 shows the validation of simulated SST with observed SST at the buoy stations for the advection fog event, even though the simulated as a 3-km grid mean and the observed SST is a station point value. The F and G buoy stations are marked in Fig. 1. In the ocean adjoining the west coast of Korea, there are only two buoys. The RMSEs between buoy and model SST for advection and steam fog events are 0.69° and 0.56°C, respectively. These RMSEs may partly come from the grid versus station values difference. This station point value and grid mean value difference has to be considered. The temporal variations in SSTs of buoy and model during the integration time are good agreement although the model tends to overestimate the SST for the advection fog event. In addition, the model does not simulate a rapid decreasing of SST about 1°C during 3 h. Nevertheless, variable SST in a coupled simulation provides meaningful results rather than fixed SST in an uncoupled simulation.

4. Identification of sea fog

The fog intensity, which is measured by the visibility, is inversely dependent on the LWC. Many studies on the relationship between LWC and fog have been reported. Rapid temporal variations in the LWC from near zero up to 0.5 g m−3 have been observed during a fog event (Fuzzi et al. 1992). According to Cotton and Anthes (1989), fog typically exhibits an LWC from 0.05 to 0.2 g kg−1. The visibility is usually calculated by the relationship between visibility and the LWC (Gultepe et al. 2006). The LWC must be converted to an extinction coefficient (β) or visibility in order to determine the practice of fog (Kunkel 1984). The relationship between visibility and extinction coefficient is
i1520-0493-138-4-1186-e1
where η is the threshold of contrast (normally equal to 0.02). Here β is the extinction coefficient given by
i1520-0493-138-4-1186-e2
where Qext is the normalized extinction cross section, and ni is the number density for droplets of radius ri. Because the droplet size distribution is unknown, an empirical relationship between β and LWC proposed by Kunkel (1984) is used in this study:
i1520-0493-138-4-1186-e3

According to Kunkel (1984), an extinction coefficient β for fog is correlated with the LWC, and a visibility < 1 km is equivalent to a LWC > 0.017 g m−1 and β > 4.01 km−1. In this study, the threshold value of β (4.01 km−1) has been chosen to determine the practice and absence of sea fog.

5. Effects of air–sea interaction

a. Advection fog

The simulation period for the advection fog event was from 2100 LST 5 July to 2100 LST 6 July 2008. Figure 7 shows the area of the advection fog, LWC, and sea surface wind vector simulated by the coupled and uncoupled models. A comparison of Figs. 7a,b shows two differences between the two simulations. First, the distribution of the LWC in the southwest coastal region (33°–35°N), which indicates the fog intensity, is clearly different. Second, there is an obvious difference in the simulated fog in the target area (TA) squared in Fig. 7a and in the region near the A station around 38°N, 124.5°E. The effect of air–sea coupling on the formation and duration of the advection fog will be analyzed at the A station and the TA.

The difference in simulated advection fog between the coupled and uncoupled models was compared. Figure 8 shows the temporal variations for the simulated β at the 10-m level. In the coupled simulation, the temporal variation in β at the A station was above the threshold value of 4.01 km−1 from 0600 to 1430 LST 6 July, and had two peaks at 17.0 and 27.3 km−1 at 0800 and 1300 LST 6 July 2008, respectively (Fig. 8a). However, the uncoupled model simulated β less than the threshold value at the A station between 0600 and 1430 LST 6 July 2008 (Fig. 8b). Furthermore, at the center of the TA, the temporal variation in β in Figs. 8c,d show a similar result to that in Figs. 8a,b. The uncoupled model did not explicitly produce the fog between 0600 and 1230 LST 6 July 2008.

To examine the causes of the difference in the fog area around the TA, the role of air–sea coupling was analyzed by comparing the simulation results from the coupled and uncoupled models. There was no remarkable difference in the temporal variations of the wind, air temperature, and relative humidity between the two models except for visibility (or β) and SST. There was a significant difference in the change in SST. To evaluate the effect of ASTD, Fig. 9 shows the difference in the ASTD and LWC between the coupled and uncoupled simulations. A positive difference in ASTD was dominant over the Yellow Sea at 0600 and 0900 LST 6 July 2008, which results from a decrease in SST. The distribution of the difference in LWC was similar to that for the difference in ASTD. This should be analyzed by two factors. Cooling of the low-level atmosphere over the cooler water, which promotes the condensation of water vapor at low-levels, is an underlying cause. Another factor is a decrease in wind speed. The coupled model simulates a weaker wind speed than the uncoupled model, which suggests that the increase in ASTD strengthens the stable stratification at the low level resulting in a decrease in wind speed. Compared with the simulation results with the unchanged SST in the uncoupled model at the center of the TA, it is obvious that the decrease in SST from 22.0° to 20.6°C in the coupled simulation plays a important role in the development of advection fog through the approximate 0.4°–0.6°C increase in ASTD. The increase in LWC is related to an increase in ASTD, which is also found in the area of decreasing ASTD in the coupled simulation. To explain this result, the advection of the LWC was calculated in the area. The LWC in this area increased in response to the LWC advection ranging from 0.01 to 0.03 g m−3 h−1 at 0600 and 0900 LST 6 July 2008. The advection of the LWC resulted in an increase in the LWC, which has a positive impact on the extension of fog area in the TA, even though the ASTD partly decreases in the TA.

Figures 10a,b shows the time variation of wind stress during an advection fog event at the A station. In the coupled simulation, the strong wind stress results from an increase in wind speed 3 h before the fog. Fog formation was restricted by the strong wind stress, which induces downward mixing of the upper dryer air (Fig. 10b), while the relative humidity was >90% (cf. Fig. 3b). Prior to the advection fog formation, the wind stress decreased to 0.16 N m−2 in the coupled simulation, while it increased from 0.11 to 0.22 N m−2 in the uncoupled simulation. Figures 10c,d show the temporal variation of difference in surface heat fluxes between the coupled and uncoupled simulations and its association with the evolution of advection fog. The air–sea energy exchange is determined mainly by the latent heat flux rather than by sensible heat flux during an advection fog event. Prior to the formation of advection fog, the difference in latent heat flux between the coupled and uncoupled simulations reached 20 W m−2, while the difference in sensible heat flux between the coupled and uncoupled simulations was <5 W m−2. Regarding the effect of air–sea coupling, an important feature of the surface heat flux is a negative latent heat flux before the fog formation, which is apparently due to condensation in the stable surface layer. The negative latent heat flux ranging from −20 to 0 W m−2 is maintained during the fog event, which suggests that southerly wind supplies moisture to initiate and maintain the fog event, and cooling of the SST then contributes to condensation. In the coupled simulation, the negative sensible heat flux showed stable stratification, while the vertical momentum flux accompanied by strong wind speed disturbs the formation of fog. The fog begins to dissipate when latent heat flux changes to a positive value. After the advection fog event, although the ASTD increased (Fig. 10e), the advection fog was dissipated by the decrease in relative humidity due to the increasing air temperature. Therefore, the downward mixing of drier air and the warm and moist advection play important roles in the formation of advection fog, and the diurnal variations in air temperature are concerned with its dissipation. In Fig. 11, at the center of the TA squared in Fig. 7a, the difference in latent heat flux between the two simulations can be clearly seen both before and during the fog event, even though the wind stresses are similar. The negative latent heat flux ranging from −20 to −5 W m−2 was maintained both before and during the fog event in a coupled simulation. However, the latent heat flux has a positive value below 5 W m−2 in an uncoupled simulation (Figs. 11c,d). A decrease in SST from 22.0° to 20.7°C in a coupled simulation results in the condensation of advected moist air in the stable surface layer due to a 0.7°C increase in ASTD from 3 h before fog formation (Fig. 11e). Therefore, the formation of advection fog is determined not only by the cooling of SST, but also by the indirect effect of a negative latent heat flux as the condensation of advected moist air.

To examine the effect of a decrease in SST, Fig. 12 shows the vertical profiles of the potential temperature and horizontal wind speed. The potential temperature profiles between the two simulations suggest that the potential temperature in the coupled simulation is noticeably cooler than that in the uncoupled simulation. During the fog event, 0.5°–0.8°C cooling at the surface increases the low-level stability, which restricts vertical mixing with air aloft. As a result, the horizontal wind speed at the sea surface is significantly weaker than that in the upper level below a 100-m height so that the wind speed increases with increasing height above the ground. Air–sea coupling induces a decrease in the sea surface wind and an increase in wind speed aloft during an advection fog event. In contrast, over warmer ocean water in the uncoupled simulation, the air rises easily and mixes with the upper air moving at higher wind speed. As a result, the difference in wind speed between the sea surface and upper level in an uncoupled simulation is lower than that in the coupled simulation. This suggests that vertical mixing contributes to the vertical exchange of winds and accelerates the wind speed at the sea surface. Weaker vertical mixing induced by cooling of the SST and the resulting decrease in wind speed (down to 1.5 m s−1 at the mature time) produce favorable conditions for the formation and development of advection fog. In the dissipation stage, although the low-level atmospheric stability is strengthened, the advection fog is dissipated by the increase in air temperature and latent heat flux resulting from the weakened advection of moist air as a result of a 3 m s−1 decrease in southerly low-level wind.

To examine the stabilizing effect of the decrease in SST, the time–height cross section of the difference in turbulent kinetic energy (TKE) between the coupled and uncoupled simulations at the A station is displayed in Fig. 13a, which allows us to provide a definitive comparison of the effect of the air–sea coupling. The distribution of the TKE difference can explain the difference in the formation of the advection fog between the coupled and uncoupled simulations presented in Fig. 8. Weak turbulence in the marine boundary layer (MABL) restricts the upward transport of low-level fog water, which confines the fog water to within the low level. It is shown that the magnitude of TKE difference before the fog formation is much larger than that after the fog formation, and the maximum TKE difference occurs at the 300-m height. Therefore, the decrease in SST stabilizes the low-level atmosphere, which promotes the condensation of water vapor confined within the low level. The temporal variation in the upper-ocean TKE profile for the advection fog event is shown in Fig. 13b. The temporal variation in the upper-ocean TKE shows a significant mixing from shear, which results in the variation in SST. The southerly wind averaging 6–8 m s−1 was dominant over the Yellow Sea between a synoptic low north of Korea and a North Pacific high positioned south of Japan (Fig. 2a). The atmospheric forcing of the sea surface wind induces the horizontal and vertical mixing in the upper-ocean mixed layer, which is supported by the fact that the temporal variation in the ocean TKE corresponds with that in the sea surface wind speed shown in Fig. 3b.

To determine the spatial pattern and source of the air parcel connected to the advection fog, low-level back trajectories are calculated based on the sea surface wind. The trajectory calculation followed the methodology outlined by Krishnamurti and Bounoua (1996). Figure 14 shows that the trajectories of the surface marine boundary layer air that is tracked backward from station A were calculated for the past 12 h from the fog onset. The trajectory A shown in Fig. 14a reflects the persistent southerly wind and allows cooling and moistening of the near-surface air, which can occur with a sufficiently long fetch over the cooler ocean. Note that the fog was maintained in the coupled simulation while the fog dissipated in the uncoupled simulation after 5 h from the start along the trajectory A. The changes of SST, the surface heat flux (SHF), and the latent heat flux (LHF) in the coupled and uncoupled simulations are presented at many points along the trajectory A (Fig. 15). The SST change along the trajectory decreases from 22.3° to 19.9°C in the coupled simulation. The SHFs related to the decreasing SST along the trajectory indicate that heat is transported from the air to the ocean (Fig. 15a). It is obvious that the dissipation of the advection fog along trajectory A follows the change of the SST in the uncoupled simulation (Fig. 15b). The fog dissipation is associated with the positive SHFs because the surface marine boundary layer air is warming during about 5 h. In the uncoupled simulation, there is an oscillation of 0.3°C during a 12-h period along trajectory A. Little change in the SHFs results in the weak SST gradient.

b. Steam fog

The simulation period for the steam fog event was 2100 LST 4 June to 2100 LST 5 June 2006. Figure 16 shows the steam fog area (represented by the low visibility), LWC, and sea surface wind vector simulated in the coupled and uncoupled models. The fog area and LWC distribution showed little difference between the two simulations. In Figs. 16a,b, a weak anticyclonic circulation occurred around the D station and the wind speed increased with increasing distance from the center. Under the wind fields, a weak northwesterly wind stretched out toward the southwest coast, and the southwesterly wind was dominant in the middle of the west coast. However, there is a difference in the fog areas around the B and D stations and the density of the LWC in the fog patches. The effect of air–sea coupling was analyzed quantitatively by comparing the simulation results from the coupled and uncoupled models at the B station.

The simulated steam fog between the coupled and uncoupled models was compared. Figure 17 shows the temporal variation for β at a 10-m level simulated by the coupled and uncoupled models. The temporal variation in β at the B station in the coupled simulation is similar to that obtained in the uncoupled simulation but the fog dissipation was retarded slightly. The β begins to increase above a threshold value of 4.01 km−1 from 0300 LST 5 June and began to decrease below the threshold value from 1040 LST 5 June 2006 in the coupled simulation (Fig. 17a). The temporal variation in β simulated from the uncoupled model showed that the fog had dissipated at 1000 LST 5 June 2006 (Fig. 17b).

Figure 18 shows the temporal changes in wind stress, turbulent heat flux, air temperature, and SST during the steam fog event at the B station in the coupled and uncoupled simulations. During the steam fog event, the wind stresses were almost zero, and there was no remarkable difference in wind stress magnitudes between the two simulations, as shown in Figs. 18a,b. The magnitudes of the sensible and latent heat fluxes increased significantly from 2.8 and 4.5 W m−2 at the formation time to 9.4 and 17.7 W m−2 at the mature time of the steam fog, respectively. These were induced by an increase in the absolute value of the ASTD and evaporation over a warmer ocean. A comparison of the turbulent heat flux between the two simulations showed that less turbulent heat fluxes appear in the coupled simulation before and during the fog event. Because the SST was fixed in the uncoupled simulation, the unchanged SST overestimates the turbulent heat fluxes by 2–W m−2 because of the larger temperature gradient between the sea surface and low-level atmosphere. Although positive turbulent heat fluxes during the steam fog are factors that regulate the duration of a steam fog, Figs. 18c,d suggest that the magnitude of sensible and latent heat fluxes does not govern the formation and intensity of a steam fog. Because a decrease in the SST (Fig. 18e) tends to decrease the low-level air temperature, which may reduce the saturation vapor pressure, an increase in condensation rate and low-level stability might retard a steam fog. However, the change of SST and its impact on the air are small. Figure 19 shows the vertical profile of the potential temperature in the formation/mature/dissipation stage of the steam fog event. In the coupled simulation, the cooling of air below a 200-m altitude was simulated in the formation and mature stage. Air–sea coupling decreased the water vapor in the air because of a deficit of evaporation from the ocean during the formation and mature time. The difference in evaporation rates between the two simulations reached 2.9–4.4 g m−2 h−1 during the steam fog event (not shown). In the dissipation stage, although the amount of water vapor in the coupled simulation was weakened before and during the steam fog event, the dissipation time was slightly retarded, as mentioned above (Fig. 17). Generally, in order to dissipate steam fog, the fog liquid water must be reduced by evaporation, requiring an increase in air temperature. As a result, the water vapor formed by the evaporation of liquid water can be accommodated for in the atmosphere, which would result in a decrease in evaporation from the ocean. However, the lower SST in the coupled simulation, which results in an increase in low-level stability, affected the air temperature within 90-m height at 1000 LST 5 June 2006, as shown in Fig. 19c. The increase in stability and the decrease in air temperature restrict the evaporation of liquid water, which retards the dissipation time of the steam fog.

We have calculated the trajectories to show the modification of the cold air parcel tracked from the land to the ocean. With regard to the back trajectories, the air parcels generally reach station B from the land (Fig. 20). Cold surface air from nighttime cooling over land traveled over the warm ocean before reaching station B over the ocean and this advection allows the favorable condition for the formation of steam fog. Figure 21 shows the changes of SST, SHF, and LHF in the coupled and uncoupled simulations at various points along trajectory B. The ASTD reaches about −2°C due to a diurnal decrease in the air temperature from 0300 LST, while SST variations are weak along trajectory B. As the result, SHF and LHF have increased about 7 and 11 W m−2, respectively. The decrease in ASTD after 0300 LST 5 June 2006 induces evaporation over the ocean and a mixing of water vapor into the cold air, which gives rise to the steam fog. Therefore, it appears that the modification of cold air in conjunction with the diurnal variation in air temperature play a major role in the formation of steam fog rather than local SST gradient or SST variation.

6. Discussion and conclusions

This study focused on the effect of temporal variations of SST in modulating the formation and duration of sea fog using the coupled atmosphere–ocean model. An improved prescription of the SST simulated in the ocean model during the atmospheric model integration leads to a better simulation of reasonable turbulent fluxes and boundary layer structures. This study examined the effect of air–sea coupling to the show factors associated with the formation and dissipation of sea fog.

The results of air–sea coupling indicate that the mechanisms of the formation and duration are different for advection and steam fog. In the results using the coupled model, the reproduced meteorological variables (i.e., air temperature, relative humidity, wind direction, wind speed, and fog) were reasonably consistent with the observed data. Enhanced cooling of the air in the MABL over a cooler ocean surface increases the condensation of advected moist air and strengthens the stable stratification. Stable stratification causes a decrease in the sea surface wind speed and restricts the downward mixing of drier air. In addition, LWC advection from an abundant LWC region over the cooler ocean promotes the formation and extension of advection fog. Before fog formation, a southerly wind supplies moisture to initiate and maintain the fog, and the cooling of the SST supports the condensation of advected moist air. This results in negative latent heat flux ranging from −20 to 5 W m−2 before and during the fog. These processes for the formation of advection fog failed in the uncoupled simulation. In the dissipation stage of the advection fog, the air temperature increased to a maximum, and moist advection was weakened by an approximately 3 m s−1 decrease in southerly low-level wind. In addition, the advection fog is characterized by the persistent southerly wind along the over a long water trajectory. Comparing the simulation results from the coupled and uncoupled models, the decreasing SST along the trajectory allows cooling and moistening of the near-surface air, which maintains the advection fog.

On the other hand, in the case of steam fog, the effect of a temporal variation of SST has little impact on the formation process, but affects the dissipation process. The upward turbulent heat flux increased significantly between the time of formation to the mature phase, but its magnitude does not govern the formation and intensity of a steam fog. However, a decrease in SST tends to cool the low-level atmosphere, which retards fog dissipation by virtue of an increase in condensation rate and low-level atmospheric stability while the evaporative flux from the ocean decreases. This retarding process for fog dissipation was not reproduced in the uncoupled simulation. The steam fog event is characterized by cold surface air from nighttime cooling over land with weak wind conditions. The cold and dry air traveled over the warm ocean before the steam fog formation. Therefore, the ASTD can be reached −2°C by a diurnal decrease in air temperature and almost unchanged SST.

The temporal changes in SST have direct and indirect effects on the formation and duration of an advection fog event, but may have little effect on a steam fog event. It should be noted that the effect of air–sea coupling is ineffective as a factor for the formation and duration of a steam fog event because there is very little change in the SST in a steam fog event. Because of the restrictive case study for a steam fog event, further study will be needed to determine how the change in SST affects the formation and duration of a steam fog event with a variable of steam fog cases. The change of sea surface wind out of North Pacific high, which affects the wind stress, controls upper-oceanic turbulent mixing and causes a decrease in SST during the advection fog event. The stress is strongly affected by the SST–wind stress feedback. Over cooler water, low-level air is cooled and thus does not mix with the upper air, which results in a weakening of turbulence and wind stress. The weakened wind stress might limit the decrease in turbulent mixing in the ocean, leading to an increase in SST and the dissipation of advection fog. To determine if this result is indeed a negative feedback mechanism, further studies will examine the feedback mechanism between the change in SST and the duration of sea fogs.

Acknowledgments

This study was supported by a grant from the “Eco-Technopia 21 Project” by the Korean Ministry of Environment and the Brain Korea 21 Project. The authors are very grateful to Prof. Kyung-Eak Kim of the Department of Astronomy and Atmospheric Sciences, Kyungpook National University, for his constructive criticism and helpful discussions during the preparation of this manuscript. The authors thank Natalie Perlin of the College of Oceanic and Atmospheric Sciences, Oregon State University, for her assistance in the model coupling. The manuscript was further improved by the criticism from two reviewers: J. Lewis and an anonymous one.

REFERENCES

  • Ballard, S. P., , B. W. Golding, , and R. N. B. Smith, 1991: Mesoscale model experimental forecasts of the Haar of northeast Scotland. Mon. Wea. Rev., 119 , 21072123.

    • Search Google Scholar
    • Export Citation
  • Bao, J-W., , J. M. Wilczak, , J-K. Choi, , and L. H. Kantha, 2000: Numerical simulations of air–sea interaction under high wind conditions using a coupled model: A study of hurricane development. Mon. Wea. Rev., 128 , 21902210.

    • Search Google Scholar
    • Export Citation
  • Barker, E. H., 1992: Design of the Navy’s multivariate optimum interpolation analysis system. Wea. Forecasting, 7 , 220231.

  • Bendix, J., , B. Thies, , J. Cermak, , and T. Nauß, 2005: Ground fog detection from space based on MODIS daytime data—A feasibility study. Wea. Forecasting, 20 , 9891005.

    • Search Google Scholar
    • Export Citation
  • Bergot, T., , and D. Guedalia, 1994: Numerical forecasting of radiation fog. Part I: Numerical model and sensitivity tests. Mon. Wea. Rev., 122 , 12181230.

    • Search Google Scholar
    • Export Citation
  • Cho, Y-K., , M-O. Kim, , and B-C. Kim, 2000: Sea fog around the Korean Peninsula. J. Appl. Meteor., 39 , 24732479.

  • Cotton, W. R., , and R. A. Anthes, 1989: Storm and Cloud Dynamics. Academic Press, 833 pp.

  • Filonczuk, M., , D. Cayan, , and L. Riddle, 1995: Variability of marine fog along the California coast. Scripps Institution of Oceanography Rep. 95-2, 102 pp.

    • Search Google Scholar
    • Export Citation
  • Findlater, J., , W. T. Roach, , and B. C. McHugh, 1989: The Haar of north-east Scotland. Quart. J. Roy. Meteor. Soc., 115 , 581608.

  • Fisher, E. L., , and P. Caplan, 1963: An experiment in numerical prediction of fog and stratus. J. Atmos. Sci., 20 , 425437.

  • Fuzzi, S., and Coauthors, 1992: The Po Valley fog experiment 1989: An overview. Tellus, 44B , 448468.

  • Gao, S., , H. Lin, , B. Shen, , and G. Fu, 2007: A heavy sea fog event over the Yellow Sea in March 2005: Analysis and numerical modeling. Adv. Atmos. Sci., 24 , 6581.

    • Search Google Scholar
    • Export Citation
  • Gultepe, I., , M. D. Müller, , and Z. Boybeyi, 2006: A new visibility parameterization for warm-fog applications in numerical weather prediction models. J. Appl. Meteor. Climatol., 45 , 14691480.

    • Search Google Scholar
    • Export Citation
  • Gultepe, I., and Coauthors, 2007: Fog research: A review of past achievements and future perspectives. Pure Appl. Geophys., 164 , 11211159.

    • Search Google Scholar
    • Export Citation
  • He, R., , and R. H. Weisberg, 2002: West Florida shelf circulation and temperature budget for the 1999 spring transition. Cont. Shelf Res., 22 , 719748.

    • Search Google Scholar
    • Export Citation
  • He, R., , and R. H. Weisberg, 2003: West Florida shelf circulation and temperature budget for the 1998 fall transition. Cont. Shelf Res., 23 , 777800.

    • Search Google Scholar
    • Export Citation
  • Heo, K-Y., , and K-J. Ha, 2004: Classification of synoptic pattern associated with coastal fog around the Korean Peninsula (in Korean with English abstract). J. Korean Metor. Soc., 40 , 541556.

    • Search Google Scholar
    • Export Citation
  • Hodur, R., 1997: The Naval Research Laboratory’s Coupled Ocean–Atmosphere Mesoscale Prediction System (COAMPS). Mon. Wea. Rev., 125 , 14141430.

    • Search Google Scholar
    • Export Citation
  • Hogan, T. F., , T. E. Rosmond, , and R. Gelaro, 1991: The NOGAPS forecast model: A technical description. Naval Oceanographic and Atmospheric Research Laboratory Rep. 13, 220 pp. [Available from Naval Research Laboratory, Monterey, CA 93943–5502].

    • Search Google Scholar
    • Export Citation
  • Koračin, D., , J. Lewis, , W. T. Thompson, , C. E. Dorman, , and J. A. Businger, 2001: Transition of stratus into fog along the California coast: Observations and modeling. J. Atmos. Sci., 58 , 17141731.

    • Search Google Scholar
    • Export Citation
  • Koračin, D., , J. A. Businger, , C. E. Dorman, , and J. M. Lewis, 2005: Formation, evolution, and dissipation of coastal sea fog. Bound.-Layer Meteor., 117 , 447478.

    • Search Google Scholar
    • Export Citation
  • Kraus, H., 1993: Surface-layer fog considered using a modified Monin–Obukhov theory. Bound.-Layer Meteor., 66 , 219235.

  • Krishnamurti, T. N., , and L. Bounoua, 1996: An Introduction to Numerical Weather Prediction Techniques. CRC Press, 293 pp.

  • Kunkel, B. A., 1984: Parameterization of droplet terminal velocity and extinction coefficient in fog models. J. Climate Appl. Meteor., 23 , 3441.

    • Search Google Scholar
    • Export Citation
  • Larson, J., , R. Jacob, , and E. Ong, 2005: The model coupling toolkit: A new Fortran90 toolkit for building multiphysics parallel coupled models. Int. J. High Perform. Comput. Appl., 19 , 277292.

    • Search Google Scholar
    • Export Citation
  • Leipper, D. F., 1994: Fog on the U.S. West Coast: A review. Bull. Amer. Meteor. Soc., 75 , 229240.

  • Lewis, J., , D. Koračin, , R. Rabin, , and J. Businger, 2003: Sea fog off the California coast: Viewed in the context of transient weather systems. J. Geophys. Res., 108 , 4457. doi:10.1029/2002JD002833.

    • Search Google Scholar
    • Export Citation
  • Lewis, J., , D. Koračin, , and K. Redmond, 2004: Sea fog research in the UK and USA: Historical essay including outlook. Bull. Amer. Meteor. Soc., 85 , 395408.

    • Search Google Scholar
    • Export Citation
  • Mahrt, L., 2007: Weak-wind mesoscale meandering in the nocturnal boundary layer. Bound.-Layer Meteor., 7 , 331347.

  • Musson-Genon, L., 1987: Numerical simulation of a fog event with a one-dimensional boundary layer model. Mon. Wea. Rev., 115 , 592607.

    • Search Google Scholar
    • Export Citation
  • Oliver, D. A., , W. S. Lewellen, , and G. G. Williamson, 1978: The interaction between turbulent and radiative transport in the development of fog and low-level stratus. J. Atmos. Sci., 35 , 301316.

    • Search Google Scholar
    • Export Citation
  • Perlin, N., , and E. D. Skyllingstad, 2007: Numerical simulation of air–sea coupling during coastal upwelling. J. Phys. Oceanogr., 37 , 20812093.

    • Search Google Scholar
    • Export Citation
  • Pilié, R. J., , E. J. Mack, , C. W. Rogers, , U. Katz, , and W. C. Kocmond, 1979: The formation of marine fog and the development of fog-stratus systems along the California coast. J. Appl. Meteor., 18 , 12751286.

    • Search Google Scholar
    • Export Citation
  • Saunders, P. M., 1964: Sea smoke and steam fog. Quart. J. Roy. Meteor. Soc., 90 , 156165.

  • Shchepetkin, A. F., , and J. C. McWilliams, 2005: The Regional Ocean Modeling System (ROMS): A split-explicit, free surface, topography-following coordinate oceanic model. Ocean Modell., 9 , 347404.

    • Search Google Scholar
    • Export Citation
  • Skyllingstad, E. D., , D. Vickers, , L. Mahrt, , and R. Samelson, 2007: Effects of mesoscale sea-surface temperature fronts on the marine boundary layer. Bound.-Layer Meteor., 123 , 219237.

    • Search Google Scholar
    • Export Citation
  • Stoelinga, M. T., , and T. T. Warner, 1999: Nonhydrostatic, mesobeta-scale model simulations of cloud ceiling and visibility for an East Coast winter precipitation event. J. Appl. Meteor., 38 , 385404.

    • Search Google Scholar
    • Export Citation
  • Taylor, G., 1917: The formation of fog and mist. Quart. J. Roy. Meteor. Soc., 43 , 241268.

  • Virmani, J. I., , and R. H. Weisberg, 2003: Features of the observed annual ocean–atmosphere flux variability on the West Florida shelf. J. Climate, 16 , 734745.

    • Search Google Scholar
    • Export Citation
  • Wang, B., 1985: Sea Fog. China Ocean Press, 330 pp.

Fig. 1.
Fig. 1.

The target area and fog cases over the Yellow Sea. NASA Terra MODIS visible images and fog observation stations (filled circles) for (a) an advection fog case at 1140 LST 6 Jul 2008 and (b) a steam fog case at 1100 LST 5 Jun 2006. The stations marked by A, B, C, D, E, F, and G are the reference point used to validate the model results. The dashed lines indicate ground fog detection obtained using the method reported by Bendix et al. (2005).

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 2.
Fig. 2.

Surface weather chart at (a) 0900 LST 5 Jul 2008 and (b) 0900 LST 5 Jun 2006. The solid lines represent the isobars (4-hPa interval) and the gray solid line with hatches represents the rainy cloud region.

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 3.
Fig. 3.

(a) Surface observations and (b) simulation results of the coupled model at station A marked in Fig. 1a from 2100 LST 5 Jul to 2100 LST 6 Jul 2008. Solid line, dashed line, dash–dot line, long dashed line, and shaded area indicate the air temperature (°C), relative humidity (%), wind speed (m s−1), sea surface temperature (°C), and fog, respectively. The upper arrows indicate the wind vector. “Fog” indicates the time of visibility ≤ 1 km in (a) observation and (b) model-based visibility obtained using parameterization presented by Stoelinga and Warner (1999).

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 4.
Fig. 4.

As in Fig. 3, but at station B marked in Fig. 1b during the period between 2100 LST 4 Jun and 2100 LST 5 Jun 2006.

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 5.
Fig. 5.

(a) A map of bathymetry over the ocean adjoining the west coast of Korea. Horizontal distributions of sea surface temperature (b) on 6 Jul 2008 (advection fog case) and (c) on 5 Jun 2006 (steam fog case).

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 6.
Fig. 6.

Temporal variation in sea surface temperature of buoy and coupled model for the (a) advection fog and (b) steam fog cases. The buoys are selected near stations A and B, respectively.

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 7.
Fig. 7.

The simulated fog area (shaded, visibility ≤ 1 km), LWC at 10 m (solid line, contour interval: 0.05 g m−3), and sea surface winds (vectors) of (a) coupled and (b) uncoupled models at 0900 LST 6 Jul 2008. The filled circles represent the fog observation station with a visibility of 1 km. The box marked in Fig. 7a indicates the location where the difference between the coupled and uncoupled models was analyzed. The model-based visibility was obtained using the parameterization reported by Stoelinga and Warner (1999).

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 8.
Fig. 8.

Time series of the simulated extinction coefficient by (a) the coupled and (b) the uncoupled models at the A station during the period 2100 LST 5 Jul–2100 LST 6 Jul 2008. (c),(d) As in (a),(b), but for the center of the box squared in Fig. 7a, respectively. The threshold of fog is shown with the dashed line.

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 9.
Fig. 9.

The difference between the coupled and uncoupled simulations in the ASTD (air temperature minus SST; shaded, °C) and LWC (contour, g m−3) at (a) 0600 LST 6 Jul (starting time) and (b) 0900 LST 6 Jul 2008 (mature time). The contour interval is 0.02 g m−3.

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 10.
Fig. 10.

Time series for the simulated wind stress by using the (a) coupled and (b) uncoupled models at the A station during the period 2100 LST 5 Jul–2100 LST 6 Jul 2008. The difference in (c) latent heat flux and (d) sensible heat flux between the coupled and uncoupled models. (e),(f) As in (a),(b), but for air temperature and SST.

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 11.
Fig. 11.

Time series for the simulated wind stress by using the (a) coupled and (b) uncoupled models at the center of the box (i.e., the TA) squared in Fig. 7a during the period 2100 LST 5 Jul–2100 LST 6 Jul 2008. (c),(d) As in (a),(b), but for turbulent heat fluxes. (e),(f) As in (a),(b), but for air temperature and SST.

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 12.
Fig. 12.

Vertical profiles of (a),(b) potential temperature and (c),(d) horizontal wind speed taken from the (left) coupled and (right) uncoupled models at the center of the box identified in the insert of Fig. 7a. The times represents the formation stage (0600 LST), the mature stage (0900 LST), and the dissipation stage (1200 LST) of advection fog.

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 13.
Fig. 13.

Vertical profile of (a) TKE difference between coupled and uncoupled models in MABL and (b) TKE taken from coupled model in the ocean mixed layer at the A station during the period 2100 LST 5 Jul–2100 LST 6 Jul 2008.

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 14.
Fig. 14.

Representative surface backward trajectories where advection fogs forms in the coupled simulation. Filled circles and circles indicate fog and nonfog, respectively. The time interval is 1 h.

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 15.
Fig. 15.

Changes in SST, SHF, and LHF are shown along trajectory A marked in Fig. 14 for (a) coupled and (b) uncoupled models.

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 16.
Fig. 16.

The simulated fog area (shaded, visibility ≤ 1 km), LWC at 10 m (solid line, level = 0.1, 0.2, 0.4, and 0.6 g m−3), and sea surface winds (vectors) of the (a) coupled and (b) uncoupled models at 1100 LST 5 Jun 2006. The filled circle represents the fog observation station with a visibility of 1 km. The box marked in (a) indicates the location where the difference between the coupled and uncoupled model was analyzed. The model-based visibility was obtained using the parameterization reported by Stoelinga and Warner (1999).

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 17.
Fig. 17.

Time series for the simulated extinction coefficient by the (a) coupled and (b) uncoupled models at the B station during the period 2100 LST 4 Jun–2100 LST 5 Jun 2006. The threshold of fog is shown with a dashed line.

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 18.
Fig. 18.

Time series for the simulated wind stress by the (a) coupled and (b) uncoupled models at the B station during the period 2100 LST 4 Jun–2100 LST 5 Jun 2006. (c),(d) As in (a),(b), but for the turbulent heat fluxes. (e),(f) As in (a),(b), but for air temperature and SST, respectively.

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 19.
Fig. 19.

Vertical profiles of the potential temperature taken from the coupled (solid line) and uncoupled models (dashed line) at the B station. The times represents (a) the formation stage (0300 LST), (b) the mature stage (0600 LST), and (c) the dissipation stage (1000 LST) of steam fog.

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 20.
Fig. 20.

Representative surface backward trajectories where steam fogs forms in (a) a coupled and (b) an uncoupled simulation. Filled circles and circles indicate fog and no fog, respectively. The time interval is 1 h.

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Fig. 21.
Fig. 21.

Changes in SST, SHF, and LHF are shown along trajectory B marked in Fig. 20 for the (a) coupled and (b) uncoupled model.

Citation: Monthly Weather Review 138, 4; 10.1175/2009MWR3100.1

Table 1.

Summary of the experimental design.

Table 1.
Table 2.

Statistics of the RMSE for the simulated wind direction, wind speed, air temperature, and relative humidity at a 10-m level using the coupled model during 2100 LST 5 Jul–2100 LST 6 Jul 2008 at island stations over the Yellow Sea.

Table 2.
Table 3.

As in Table 2, but for the period 2100 LST 4 Jun–2100 LST 5 Jun 2006.

Table 3.

1

COAMPS is a registered trademark of the Naval Research Laboratory.

2

The conversion from UTC to local standard time (LST) requires the addition of 9 h (e.g., 0900 LST should be understood as 0000 UTC).

Save