A Fog Event off the Coast of the Hangzhou Bay during Meiyu Period in June 2013

A dense fog with the visibility less than 100 m over 6 hours occurred around the Hangzhou Bay off the coast of the western part of the East China Sea on 24–25 June 2013 during Meiyu period. This study focuses on the physical mechanism involved in the fog process by using in-situ observations and model. The analysis indicates that the land-sea thermal contrast played an essential role in the fog episode, while a weak low-pressure wedge associated with the Meiyu front and the diurnal variation in temperature provided background conditions. Induced by the strong land-sea thermal contrast, a secondary circulation formed in the lower levels of the atmospheric boundary layer (ABL) over the coast. The southeasterly wind and subsiding motion associated with the secondary circulation contributed to the moisture supply and the lowering of the boundary layer, favoring fog formation and maintenance. The fog maintained until the weakening of the temperature gradient between land and sea, when the downward flow was replaced by upward motion controlled by the approaching of a low-pressure center. These results are helpful for improving coastal fog forecast in Meiyu period and for our understanding of mechanisms involved in coastal fog processes.


INTRODUCTION
Fog often causes losses to shipping communities and other socioeconomic activities over ocean and near coastal regions because of the low visibility in fog.The formation, maintenance and dissipation of fog are associated with physical processes, ranging from the large/synoptic scale (e.g., advection) to meso-scale (e.g., land-sea breeze) and to micro-scale (e.g., droplets) (Gultepe et al., 2007;Koračin et al., 2014).Local meteorological (e.g., wind direction and speed) and hydrological (e.g., sea surface temperature, SST) conditions are essential for fog formation near coastal regions (Huang et al., 2010;Kim and Yum, 2010).For example, influenced by the Pacific high-pressure system, the sinking of the cloud-bottom to sea surface may lead to fog along the western coast of U.S. (Koračin et al., 2001;Lewis et al., 2004).
Fog frequently occurs under the southeasterly warm/moist air advection along the Chinese coast of the South China Sea, the East China Sea and the Yellow Sea (Zhang et al., 2009;Huang et al., 2015), and due to strong tidal mixing off the western coast of the Korean Peninsula (Kim and Yum, 2010).
Meiyu is a major rainy season from Central China, the East China Sea to Japan, climatologically from June to mid-July.Meiyu rainband is accompanied by the characteristic frontal structures in the lower troposphere, i.e., Meiyu front (Sampe and Xie, 2010).Influenced by Meiyu front, low clouds, significant rainfall, fog and drizzle frequently appear in the Chinese coastal region (Wang, 1983;Wang, 2006;Zhang et al., 2008;Zhang et al., 2014).Previous studies have focused more on precipitation than on coastal fog possibly due to the lack of observations over the sea so far.
The western part of the East China Sea bears the heavy burden of fishing, shipping, oil-drilling activities and port operations.The area from the Hangzhou Bay to the open East China Sea, in particular, has the biggest deep-water port (Yangshan Port) and the largest fishery (Zhoushan fishing ground) in China (Fig. 1(c)).Fog produces low visibilities and thus heavily hampers marine activities.The lack of knowledge about the physical processes in fog formation, maintenance and dissipation has been a major obstacle in fog forecast in this marine region.
On 24-25 June 2013, a research vessel (R/V) Dongfanghong 2 anchored off the Hangzhou Bay, western part of the East China Sea and captured a fog event.A suite of rare in-situ observations from the R/V was obtained.Based on these in-situ observations, the physical processes in the fog are investigated in the present study.The reanalysis and satellite datasets, and numerical WRF modeling results are also applied.We show that the land-sea thermal contrast played an essential role in the fog formation.Data and models used in the present study are described in Section 2. Section 3 indicates the analysis based on observations.Section 4 discusses the mechanisms using model results.Section 5 presents conclusions.

Data
As shown in Fig. 1 Väisälä ceilometer (CL31), surface air temperature (SAT), relative humidity (RH), sea level pressure (SLP), visibility, wind speed and directions from an automatic weather station (ATWS); vertical profiles of temperature, humidity and wind information were from GPS soundings.Sampling intervals of the ceilometer and ATWS measurement were 16 s and 1 min, respectively.
Routine weather data are provided by the Meteorological Information Comprehensive Analysis and Process System (MICAPS) from the China Meteorological Administration (CMA).The reanalyzed data are obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF, Dee et al., 2011), which are available four times daily at 0000, 0600, 1200, and 1800 UTC with a horizontal resolution 0.125° × 0.125° and 37 vertical levels from 1000 hPa to 10 hPa.
The simulated fog patches is identified as the region with the cloud water mixing ratio at the lowest level of model being greater than 0.016 g kg -1 (Zhang et al., 2012) and the atmospheric horizontal visibility (hereafter visibility) in fog is calculated with Koschmieder's formula (Koschmieder, 1924) as employed by Stoelinga and Warner (1999) where X VIS is the visibility (m) and β is the extinction coefficient that is determined by cloud water (β cw ), cloud ice (β ci ), snow (β snow ) and rain water (β rain ).In summer in the East China Sea, there is no snow and cloud ice near surface.The rain water in the CTL run is below 0.000008 kg kg -1 , which is much less than cloud water (above 0.000016 kg kg -1 ) and thus has little influence on visibility.Therefore the cloud water is used to calculate the visibility: in which q is the cloud water mixing ratio (g kg -1 ), ρ the air density (kg m -3 ).All of these measurements indicated the formation and maintenance of fog.Fog dissipated after 0200 LST 25 June 2013 when the visibility increased remarkably and drizzle was recorded at 0400 LST on the R/V (We distinguish between fog and drizzle mainly according to manual observations, along with the measurements of atmospheric horizontal visibility and backscatter coefficient.When the backscatter coefficient intensity is larger than 10000 near the surface, the present weather is heavy fog or drizzle (Young and Whiteman, 2015).When the visibility is lower than 1 km the present weather is very likely fog).Low clouds and precipitation were reported by the CMA weather stations nearby (Fig. 1(c)); some cloud ceilings were captured by the CL31 (Fig. 2(c)).These were typical weather phenomena during Meiyu period (Ninomiya, 1984;Sampe and Xie, 2010).

Observations
A GPS sounding launched onboard indicated a stable atmospheric boundary layer with an average temperature inversion from 30 m to 390 m (Fig. 2(b)), which was favorable for the formation and maintenance of fog.The warm advection associated with the southwesterly wind at 950 hPa might contribute to the temperature inversion (Fig. 3(b)), together with the clockwise shift in wind direction from southeasterly to south-southwesterly from surface to 500 m (Fig. 2(b)).The fog layer was below 100 m, consistent with the measurement from the CL31 (Fig. 2(c)).

Synoptic Situations
As shown in Fig. 3(a), a weak low-pressure wedge associated with the Meiyu front stretched eastwards from the southeast of China to the East China Sea along 30°N, and Hangzhou Bay was within the low-pressure wedge at 2000 LST 24 June 2013 when the R/V was anchored there (triangle in Fig. 3(a)).Since the fog mainly occurred around the Hangzhou Bay (Fig. 1(c)), it seemed unreasonable to attribute the cause of fog simply to synoptic situation and diurnal variations.
Generally, fog generates under high-pressure system and descending motion (Koračin et al., 2001;Zhang et al., 2012) rather than under low-pressure system.Indeed, low clouds (stratocumulus and cumulus) and showers were reported in

Model Simulation
By comparing Figs.2(a) and 3(a), the southeasterly wind recorded by the ATWS cannot be well represented in the synoptic map produced from the reanalyzed data possibly due to the relative coarse resolution of the data.We use WRF model to simulate the process and to investigate the mechanism in the fog process, especially the influence of meso-scale systems.
The results from the CTL show that the model can well reproduce the low-pressure wedge (Fig. 4(a)), the horizontal wind directions and the air temperature distribution (Figs. 4(b) and 4(c)).The simulated fog (cloud water mixing ratio, qc > 0.016 g kg -1 ) starts after 2100 LST 24 June and ends at around 0300 LST 25 June near the position of the R/V (Fig. 4(b)) in agreement with the measurements onboard.The simulated fog area is mainly located in and off the Hangzhou Bay, the western part of the East China Sea (refer to Fig. 7(a)).The southeasterly-northeasterly winds in the

Variations in Surface Temperature
Cooling and moistening can lead to saturation and condensation.As shown in Fig. 2(a), both the SAT and T d (dew-point temperature) decreased remarkably as the visibility onboard dropped from over 5 km to less than 1 km from 2000 LST to 2200 LST, denoting the formation of fog.It was assumed that the cooling played a leading role in fog formation because the SAT decreased faster than the T d did.The SAT was gradually close to the T d , thus resulting in the decrease in difference between SAT and T d and the relative humidity close to 100%.The relative humidity is defined as the ratio of actual water vapor pressure (e a ) to saturation water vapor pressure (e s ): The former is positive related with T d and the latter is the function of T (SAT).In the formation phase of the fog, the relative humidity increased as the difference between SAT and T d decreased (Fig. 2(a)), which suggests the coolingcontrol mechanism in agreement with fog in spring in the Yellow Sea (Zhang and Lewis, 2017).
One may expect the diurnal variation for the decrease in SAT.Indeed, the drop in temperature in inland city of Hangzhou was more remarkable than on the R/V from 1700 LST to 2000 LST, indicating stronger diurnal variation over land (Fig. 5(a)).The diurnal variation possibly contributed to the cooling in SAT both on land and over sea, favoring fog formation.Yet this process alone cannot explain the foggy area that was limited merely in and off the Hangzhou Bay.Note that the temperature decreased about 1.5°C onboard vs. 0.6°C in Hangzhou from 2000 LST to 2300 LST when the fog maintained (Figs.2(a) and 5(a)).This faster cooling onboard was very likely due to the feedback between the longwave radiation from fog top and the temperature in fog layer.The longwave radiation cooling from fog top has been documented as a significantly important process in the formation and maintenance of fog in the previous literatures (Koračin et al., 2001;Lewis et al., 2004;Zhang et al., 2012).We will also confirm this aspect from CTL and EXP in the subsequent text.

Variations in Humidity
By examining Fig. 2(a), one can find that the visibility decreased as the wind direction shifted from northeasterly to southeasterly around 2100 LST.The southeasterly wind blew from warmer water and carried moisture from open sea toward the cooler sea surface near the Hangzhou Bay (Fig. 6).The negative SST-SAT (Fig. 3(a)) there provided stable air-sea interface.All of these conditions were conducive to cold coastal (sea) fog (Koračin et al., 2014).
Note that the southeasterly wind appeared in a limited area over the southern area off the Hangzhou Bay where the R/V was anchored (Fig. 6).Thus the moisture supply was local, with a limited extent and not sufficient to support Fig. 6.The SST (shaded/°C), specific humidity (q; red contours at the interval of 1 g kg -1 ) and horizontal wind (arrows) at 10 m in CTL.▲ is the location of R/V.The SST is from the Real Time Global Dataset (RTG High Res 0.083, http://polar.ncep.noaa.gov/sst/).the rise in T d during fog formation.Indeed, the T d decreased in the formation phase of this coast fog (Fig. 2(a)).A similar situation is found in the Yellow Sea with the decreased T d in fog formation phase in spring when the southeasterly moist air comes from local effects, whereas opposite condition (T d keeps increasing in fog formation phase) is measured in summer fog due to the huge amount of moisture supply from the tropical/subtropical oceans by a deep large-scale summer monsoon (Zhang et al., 2012).

Effect of Local Land-Sea Thermal Contrast
From the analysis above, it seems that the southeasterly wind may be the main cause for this coastal fog in terms of the spatial temporal scales.But what made the southeasterly wind?For this question, we turn to local land-sea thermal contrast.In this present case, the near surface air temperature in the CTL indicated a sharp temperature gradient over the southern part of the Hangzhou Bay between warmer land and cooler sea surface at 2000 LST (Fig. 7(a)).The variations in air temperature at Hangzhou weather station and on the R/V showed that the intensity of the cross-coastal temperature gradient peaked (~3.22°C/200 km) at 1700 LST (Fig. 5(a)).We also found apparent cyclonic shear in wind directions along the southern coastline of Hangzhou Bay (Fig. 6).The vertical structure was a narrow transition zone of relatively sharp air temperature gradient that rose steeply from surface to a height of 50-500 m along the southern coast of the Hangzhou Bay (Fig. 8(a)).The difference in air pressure between land and sea became large and up to the maximum of 2.3 hPa/200 km at 2000 LST, and decreased to 0.2 hPa/ 200 km near the end of this fog episode (Fig. 5(b)), which confirmed the existence of the local land-sea thermal contrast effect.The local land-sea thermal structure was just opposite to coastal front which is derived from a strong thermal Fig. 7.The air temperature at 15 m above sea level (contours at the interval of 24.5, 25 and 25.5°C) and fog area (yellow shaded when qc > 0.016 g kg -1 ) at 2000 LST 24 June from CTL (a) and EXP (b) and these at 0200 LST 25 June (c).Gray curves represent the location of strong thermal-contrast area; A-B line denotes the position of cross-front vertical section in Fig. 8; ▲ is the location of R/V; the black curve denoting the coastal line in model data.
contrast between warmer open sea and colder coastal land (Riordan, 1990).Our observed temperature gradient was sharper than the Carolina coastal fronts (~3°C/280 km, Appel et al., 2005), but its vertical structure was weaker (shallower) and scale was smaller than coastal fronts in previous studies (Riordan, 1990;Holt and Raman, 1992;Riordan et al., 1995).

Secondary Circulation
A secondary circulation existed in the boundary layer, with its sinking (rising) branch on the cooler (warmer) sea (land) surface (Fig. 8(a)).Induced by the secondary circulation, southeasterly wind appeared at near the surface level, which was compatible with the observations from the ATWS and the soundings (Figs.2(a) and 2(b)).The subsiding motion may help lower the top of boundary layer (Fig. 4(b)).Similar process was documented by Koračin et al. (2001) for coastal fog along the California Coast, but the subsidence motion is induced from synoptic high-pressure system rather than secondary circulation.The fog droplets, once formed, would be confined in a thinner layer near surface, favoring fog formation and maintenance.
The air parcels kept sinking while along their historical paths from southeast to northwest as shown by a suite of backward trajectories tracing from foggy area (Fig. 9(a)).The decrease in air temperature and specific humidity accompanied the increase in cloud water mixing ratio and relative humidity, suggesting the formation of fog (Fig. 9(b)).Hence the secondary circulation and the associated southeasterly wind should be the dominant cause for the coastal fog episode.
After 0200 LST 25 June, the low-pressure centre moved easterly approaching the Hangzhou Bay (Fig. 4(a)), and the land-sea thermal contrast weakened (Fig. 7(c)).The secondary circulation and hence the southeasterly wind and the downward air motion over the cooler sea surface disappeared.Instead, the horizontal wind was westerly~southwesterly and the updrafts prevailed (Fig. 8(c)) induced by the lowpressure system.Fog dissipated under such circumstances and drizzles appeared (Fig. 1(c)).

Numerical Experiment
It is assumed that the secondary circulation was induced from the local land-sea temperature contrast.The EXP was performed by changing a part of land into sea in the WRF to investigate the effect of land-sea thermal contrast on the secondary circulation and surface winds.As a warm center over 27°C was on the land near the coast (Fig. 3(a)), we change this part of land into sea based on 27°C isotherm (the green frame in Fig. 3(a)).In this way, the temperature gradient would be smoothed along the original coast.
The land-sea surface thermal contrast was much weaker in the EXP (~0.5°C/200 km, Fig. 7

DISCUSSION AND SUMMARY
Based on observations and numerical simulations and experiments of WRF model, mechanism in the formation of a coastal fog event in Meiyu period was investigated.The major summaries including some discussions are as follows: (1) This coastal fog occurred under a weaker synoptic low-pressure wedge associated with the Meiyu front.The heavy humid surroundings and the diurnal variation in temperature provided favorable background conditions for the formation and maintenance of fog.
(2) Meso-scale conditions can explain the spatial and temporal extent of the coastal fog.A strong land-sea thermal contrast existed near the southern part of the Hangzhou Bay, which induced a secondary circulation appeared in the boundary layer.Subsiding motion occurred on the cold sea surface though under the influence of a synoptic lowpressure system.The southeasterly wind and subsiding motion associated with the secondary circulation contributed to the moisture supply and the lowering of the boundary layer, favoring fog formation and maintenance.The fog dissipated when a low-pressure center moved to the Hangzhou Bay and the local land-sea thermal contrast weakened.
(3) Note that the cloud ceiling lowered in the initial stage of this fog from the CL31 measurement (Fig. 2

(c)).
There could be relations between the low clouds and the fog similar to the situations along the California coast (Koračin et al., 2001).Such relations also existed between low clouds in the northern part of the East China Sea and fog in the southern part of the Yellow Sea, and were documented by Zhang et al. (2014), Li and Zhang (2013).Based on those previous studies, the vapor coming from the clouds may also contribute to the fog formation and maintenance forced by the sinking motion in this present fog case.
(4) Coastal fog and drizzle often occurs in the East China Sea in the Meiyu period.Such kind of phenomenon may be due to a secondary circulation overlaid with a weaker synoptic low-pressure system.It becomes clear that the thermal contrast between land and sea surface should be considered in the forecast of coastal fog in the relatively long lasting Meiyu period.
(a), the quasi-stationary Meiyu front and associated cloud band extended from the southeast of China, the East China Sea to the south of Japan at 1315 LST (Local Standard Time: UTC+8 hours) 24 June 2013.Meanwhile the research vessel (R/V) Dongfanghong 2 (DFH2) of the Ocean University of China captured the coastal fog episode when it anchored off the mouth of the Hangzhou Bay (Fig. 1(c)).The observations onboard used in the present study were the height of cloud base measured by a
On 24 June 2013 the R/V Dongfanghong 2 anchored off the mouth of the Hangzhou Bay (Fig. 1(c)) when fog occurred.The visibility dropped significantly after 2000 LST 24 June 2013 (Fig. 2(a)).The cloud base lowered from about 400 m to near sea surface measured by the CL31 (Fig. 2(c)).The RH increased over 98% observed by the ATWS (Fig. 2(a)).

Fig. 3 .
Fig. 3. (a) Sea level pressure (SLP; black contours at the interval of 1 hPa), horizontal wind (arrows), surface air temperature (SAT ; red contours at the interval of 2°C) and SST-SAT (shaded in °C) at 2000 LST 24 June; (b) temperature advection (advT; shaded), horizontal wind (arrows) and ascending motion (-ω; black contours at the interval of 0.05 Pa s -1 ; only positive value is represented) on 950 hPa at 2000 LST 24 June.The sign ▲ represents the location of the R/V Dongfanghong 2.

Fig. 4 .
Fig. 4. The sea level pressure (SLP; contours at the interval of 1 hPa) at 10 m in CTL at 2000 LST, the low-pressure centers (red L) at 1400 LST 24 and 0200 LST 25 being marked, the green curve denoting the coastal line.(b) Temporal variations of the cloud water mixing ratio (qc; contours at the interval of 0.05 g kg -1 and shaded when qc > 0.016 g kg -1 ), horizontal winds in CTL (red vectors) and in EXP (blue vectors) at 10m, the surface air temperature in CTL (tmp; red line) and in EXP (tmp_EXP; blue line) at 10m and planetary boundary layer height (pblh; black dashed line) in CTL from 1930 LST 24 to 0500 LST 25 at (122.2°E, 30.05°N) near the R/V.▲ is the location of the R/V Dongfanghong 2. (c) The air temperature (shaded/°C) and horizontal wind (arrows) at 10m in CTL.The location of weather stations (•) and the air temperature (numbers/°C) and horizontal wind (vectors) at stations are marked.

Fig. 5 .
Fig. 5. Observed air temperature (a) and sea level pressure (b) at Hangzhou weather station (red lines) and on the R/V (blue lines) every 3 hours from 1400 LST 24 to 0200 LST on 25 June.The bar graphs are the differences between Hangzhou and R/V (Hangzhou-R/V).

Fig. 8 .
Fig. 8.The air temperature (θ; red contours at the interval of 0.2°C), the secondary circulation (arrows; horizontal component in m s -1 and vertical component in 10 -2 m s -1 ) across A-B line in Fig. 7(a) at 2000 LST 24 June, in (a) CTL and (b) EXP, and these in CTL at 0200 LST 25 June (c).▲ is the location of the strong thermal-contrast area.Yellow shaded denotes fog area.

Fig. 9 .
Fig. 9. (a) The backward trajectory of air parcels at 6 m (starting points) from 2300 LST 24 JUN for 7 hours in CTL.The black lines are the SST from the Real Time Global Dataset, the solid circles denote the starting spots of tracing and the colors represent the variations in height of air parcels in their paths.(b) The time series of variables averaged from all the trajectories, height (hgt), air temperature (tmp), relative humidity (rh), cloud-water-mixing-ratio (qc) and specific humidity (q).
(b)) compared with that in the CTL (~2.5°C/200 km, Fig. 7(a)), and the secondary circulation disappeared (Fig. 8(b)).In the EXP, the horizontal wind was westerly-southwesterly (Fig. 8(b)), which suggested that the southeasterly wind was very likely derived from the strong coastal land-sea surface thermal contrast.Without the southeasterly moisture supply the fog might not form (Figs. 8(b) and 8(c)).These results confirmed the vital importance of the local land-sea thermal effect for the coastal fog.Without fog, the decrease extent in air temperature would be much less than those in the CTL from 2100 LST 24 June to 0100 LST 25 June (Fig. 4(b)), indicating the longwave radiation cooling effect from fog top.

Table 1 .
Specifications of WRF modeling.