1

A Study on Mesoscale Snow Band Associated with Coastal Front and Cold Air Damming along the Eastern Coast of Korean Peninsula

Jae-Gyoo Lee1 and Ming Xue2,3

1Department of Atmospheric Environmental Sciences

Kangnung National University, Kangnung, Korea

2Center for Analysis and Prediction of Storms and 3School of Meteorology University of Oklahoma, Norman OK 72072

Submitted to Mon. Wea. Rev.

September, 2007

Corresponding author’s address:

Dr. Ming Xue NWC Suite 2500, 120 David Boren Blvd

Norman OK 72073. E-mail: mxue@ou.edu.

2

Abstract

A 24-h numerical simulation with the Advanced Regional Prediction system (ARPS)

is performed and the results are used to explain the precipitation maxima at the Yeongdong

coastal region (rather than over the mountain slope and ridge) that were observed during the

heavy snowfall event of 3-4 February 1998 along the eastern coast of Korean peninsula. The

ARPS model, utilizing a 4-km horizontal grid spacing and 43 vertically stretched levels, together

with a full suite of physics, is able to simulate properly the meso- β scale characteristics of the

cold-air damming against the coastal mountain range and the associated coastal front, which is

found to control the location of the precipitation maxima in the region.

The simulated cold-air damming exhibits the following features. A very narrow wedge

of cold air is located along the eastern slope of the mountains in the coastal region, and there

exists a distinct cold dome spanning the coast to the upper part of the eastern slope of the

mountain barrier, centering roughly at the foot of the mountains. Across the frontal zone of the

cold dome, there is a strong temperature gradient of 2-3 K/10 km and directional wind shear of

about 10/40o km. The coastal front behaves like a density current, trying to propagate down the

eastern slope of the mountain range, against the onshore winds just off the coast. The simulated

cold down-slope flow is relatively shallow, with the front sloping vertically with height then

horizontally southwestward and becoming weak at a height of about 400 m. The vertical motion

between the downslope cold flow and the onshore warm air is captured well.

Due to the favorable atmosphere condition for low-level upright convection off the coast

where the air would be forced upward at coastal front, the heaviest snowfall occurs along the

cold side of the coastal front boundary, leading to consumption of most moisture there. As a

result, the precipitation directly over the mountain slope is weak. These simulated features match

available observations very well.

3

1. Introduction

When cold north or northeasterly winds blow from a well developed Siberian High over

the East Sea (Sea of Japan, see Fig. 1), the cold air mass crossing the sea becomes modified from

below by rich heat and moisture fluxes from the warm sea surface, resulting in instability and

convection in the boundary layer (e.g., Ninomiya 1968; Jhun et al. 1994). Furthermore, if the

winds over the sea are relatively strong and persistent, the convection can organize into

snowbands, leading to the heavy snowfall on the windward (facing the East Sea) side of the

Korean peninsula and the Japan Islands (Nagata et. al. 1986; Seo and Jhun 1991; Lee and Lee

1994). This often leads to heavy snowfall over short periods of time without any direct relation

with a synoptic scale trough or low system. Thus, forecasters need to pay close attention to the

development of such Siberian High.

The snowfall on the windward side of the Korean peninsula and the Japan Islands is

similar to the ocean-effect snow which has been observed in the New England coast, in the Gulf

of St. Lawrence, the Bay of Fundy, the mouths of Delaware, and Chesapeake Bays of the United

States (Waldstreicher 2002). Such snowfall is also similar to the “lake-effect snow” (Hjelmfelt

and Braham 1983) of the Great Lakes (e.g., Eichenlaub 1979; Kelly 1984; Byrd et al. 1991; Byrd

et al. 1995; Niziol et al. 1995). During cold air outbreaks, the coast line shape and sea surface

temperature pattern have profound effects on the establishment of low-level mesoscale

circulations as a result of differential heating that is dependent on both over-water path length

and water temperature (Atlas et al. 1983).

For the Korean peninsula, when cold northeasterly winds persist, the Yeongdong region,

the main windward side of the Korean peninsula, is subject to heavy snowfall. Thus, heavy

snowfall forecast in the Yeongdong region is of primary concern during the winter season.

4

Figure 1 shows the topography of the Korean peninsula and that of the Yeongdong region in an

ARPS (Xue et al. 2000) simulation domain used in this study. It is seen that the Yeongdong

region consists of narrow and steep mountains (called Taebaek mountains, with an average

height of about 1 km). The mountain range is asymmetric with its steeper slope facing the

eastern coast, and there exists a narrow coastal plain region of about 5 km in width between the

mountains and the coast. The axis of mountain range runs approximately northwest to southeast,

roughly parallel to the coastline. With such geographical and topographical conditions, the

snowfall distribution in this region can be rather complicated because of the combined effects of

ocean and topography. For these reasons, the Yeongdong region has drawn various interests

from researchers.

The possible mechanisms of orographic precipitation, as summarized by Houze (1993),

include 1) seeder-feeder mechanism, 2) upslope condensation, 3) upslope triggering of

convection, 4) upstream triggering of convection, 5) thermal triggering of convection, 6) lee-side

triggering of convection, and 7) lee-side enhancement of convection. Regarding the combined

effects of orography and ocean, Estoque and Ninomiya (1976) performed a numerical study on

snowfall distribution associated with the East Sea (Japan Sea); they showed that for their case the

occurrence of snowfall was controlled primarily by orographic lifting. Takeda et al. (1982)

examined, using radar RHI scan data, the modification of convective snow-clouds related to cold

air outbreaks from the Eurasia continent when they landed the East-Sea-facing coastal region of

Japan. In their case, most of convective radar echoes traveling from the sea showed two stages of

change in echo structure upon their landing. In the first stage, the echo center became more

intense before landing. In the second stage after landing, the intensity of the echoes increased

gradually again due to orographic lifting. Grossman and Durran (1984) showed, through 2-D

5

numerical simulations, the piling up of air mass against the windward slope of the Western Ghats

Mountain Range of southwestern India, accompanied by a deceleration of the lower-tropospheric

monsoon flow approaching the mountain range. In their case, lifting extended upstream over the

ocean as the outflow propagated down and away from the mountain slope, triggering new

convection upstream.

In addition, the geography and topography of the Yeongdong region may be favorable for

cold-air damming, which often occurs in eastern United States in the winter. In the latter region,

cold, stable air from a surface high over New England or Canada flows southwestward and is

channeled along the eastern slopes of the Appalachians. The mountains “dam” the air along the

eastern slopes (Richwien 1980; Forbes et al. 1987; Stauffer and Warner 1987). Furthermore, the

cold-air advection associated with the damming intensifies the thermal contrast between the

dammed air and that over the ocean, which has been warmed and moistened by the underlying

sea (Petterssen et al. 1962; Sanders and Gyakum 1980; Bosart 1981; Kuo and Low-Nam 1990).

This process enhances the baroclinic zone near the coastline, causing coastal frontogenesis

(Bosart et al. 1972; Bosart 1975; Ballentine 1980; Bosart 1981; Bosart and Lin 1984; Forbes et al.

1987; Stauffer and Warner 1987), and such frontogenesis is usually restricted to the lower

atmosphere.

If it were purely orographical lifting, the heaviest precipitation is expected near the slope

of and/or over the ridge of the mountains, instead of being located off the slope over the coastal

plain. Observations in the Yeongdong region, however, often show that the maximum

precipitation area is located over the coastal plain, the cold-air damming and associated coastal

front might have caused such a displacement. The heavy snowfall event of 3-4 February 1998

appears to be such a case. In this paper, we study this particular event by examining available

6

observations and by performing a high resolution numerical simulation. In section 2,

observational analyses are first presented, by examining surface weather maps, the distribution of

accumulated snowfall, satellite imagery, radar reflectivity, and surface wind observations. A

description of the numerical model and simulation is given in section 3 while the simulation

results are discussed in section 4. Summary and conclusions are given in section 5.

2. Observational analysis

Figure 2 shows the surface weather charts at 2100 LST (1200 UTC) 03 February (Fig. 2a)

and at 2100 LST 04, February 1998 (Fig. 2b), respectively. In Fig. 2a, a well developed Siberian

High, centered near Mongolia at 2100 LST 03 February 1998, 2-3 hours before the start of

Korean coastal snowfall, dominated most of China. A ridge extended eastward from the High

over the East Sea. On the downwind or south side of the ridge, northeasterly, quasi-geostrophic

winds of cold air blew over the warm sea surface. In particular, over the northern part of East Sea,

horizontal pressure gradient was relatively strong and the isotherms were nearly perpendicular to

the isobars (not shown), indicating strong cold advection over the sea and in the eastern coastal

region of Korea. As this continental polar air mass was advected over the warm sea surface, it

was rapidly modified by the large oceanic heat and moisture fluxes, leading to an increase in

instability similar to lake-effect cases. This rendered a favorable condition for forming

convective snow clouds, especially in the downwind stretch of the flow. Furthermore,

northeasterly, quasi-geostrophic winds normal to the axis of the mountains was favorable for

cold air damming on the eastern side of the Taebaek mountains. These mesoscale features are,

however, often difficult to identify from synoptic charts.

Figure 2b shows the surface weather chart at 2100 LST 04, or about 9 hours after the

snowfall ended. By this time, the alignment of isobars over the central and northern part of the

7

East Sea had been changed from the northeast-southwest to northwest-southeast orientation,

hence the geostrophic winds in the region backed to northwesterly. Northwesterly winds would

cause the weakening of temperature contrast in the coastal region.

The distribution of 24-h accumulated snowfall depth ending at 2400 LST 4 February

1998 is shown in Fig. 3. It is seen that toward the mountainous inland, the snowfall decreased

rapidly. The snowfall in the Yeongdong region began at close to 0000 LST and ended at around

1200 LST, 4 February. The snowfall in the region was strongest at about 0300 LST on 4

February. It can be seen from the precipitation map that the area of maximum snowfall was

located near and off the coast rather than further inland over the mountain slope. That is, a

significant amount of snowfall was located along the coastal region (e.g., 16.0 cm at Kangnung,

KN in Fig. 3) rather than over the mountain region (e.g., 6.3 cm over Taegwallyong, TG in Fig.

3) where the orographic lifting is strong. The NOAA infrared satellite image (10.5-11.5μ m) at

0251 LST 4 February 1998, when the snowfall at Kangnung was most intense, is shown in Fig. 4.

The main cloud band was clearly aligned along the east coastline, parallel to the axis of the

mountains, but the central axis of the band was located somewhat off the coast line into the sea.

More detailed features of the convective band can be identified from the PPI (Plan-

Position Indicator) scan of reflectivity, taken by a radar located at Tonghae at 0010 LST 4

February 1998 (Fig. 5). The green area of L2 (equivalent to precipitation intensity of 1~4 mm hr-

1), is aligned northwest to southeast and is located not on but 10~20 km off the coastline. This

long, narrow, banded structure suggests the existence of a coastal front. Note that the radar

reflectivity near the mountains is not available due to ground cluster.

Figure 6 shows the observed wind field at 0300 LST 4 February 1998. Northwesterly,

cross isobaric winds (c.f., Fig. 2), prevailed at this time over land and along the Yeongdong

8

coastal region (due to frictional effect), except at Kangnung which had very weak southwesterly

winds (which may be due to local orographic and/or snowfall effects). Meanwhile, northeasterly

winds, which are in the direction of geostrophic winds, were observed at the island of Ulleungdo

far off the east coast. This represents a significant change in wind direction, from northeasterly

over sea to northwesterly over the land, as the air flow approached the coastal mountains. Such

direction changes may have also contributed to the flow convergence in the coastal region,

although the orographic blocking effect should be more important.

3. Model description and numerical simulation

To better understand the physical processes associated with the coastal snow fall event, we

performed a numerical simulation using version 5.0 of the Advanced Regional Prediction system

(ARPS, Xue et al. 1995, 2000, 2001, 2003), a general-purpose, compressible, nonhydrostatic

numerical weather prediction model. The computational grid consists of 179 points in both x and

y directions with a uniform horizontal resolution of 4 km (giving a physical domain of 704 x 704

km2) and is centered at 39.3 o N, 129.0 o E (see Fig. 1). The vertically stretched grid consists of 43

levels, covering a depth of 17.2 km, with the vertical grid spacing varying from 20 m near the

ground to about 800 m near the model top. The first model level above ground is at about 10 m

AGL. This vertical grid is believed to be adequate for resolving important vertical structures of

the low-level phenomena including the cold air damming and frontal interface. The top and

bottom model boundaries are rigid with a Rayleigh damping layer located above 12 km for

controlling top boundary wave reflection. Cloud microphysics scheme used is the ice

microphysics scheme of Lin et al. (1983) which includes two liquid (cloud and rain) and three

ice categories (ice cloud, snow and hail or graupel). Subgrid-scale turbulent mixing is

represented using an anisotropic 1.5-order turbulent kinetic energy closure scheme. The surface

9

fluxes at the land surface are calculated using stability- and roughness-length-dependent drag

coefficients, and the soil temperature and volumetric water content are predicted by a two-layer

soil model. Meanwhile, the fluxes at sea surface are calculated using constant drag coefficient for

momentum (Cdm), and constant exchange coefficients for heat (Cdh) and moisture (Cdq); they

were all set to be 3102 −× following Chen et al. (1983). More details of the physics schemes in

the ARPS can be found in Xue et al. (2001).

The soil model that predicts the soil skin temperature and soil moisture content is a two-

layer force-restore model based on Noilhan and Planton (1989), where offsets of top soil layer

temperature and bottom soil layer from surface air temperature, as defined in Ren and Xue

(2004), are set to be 1.0 K and 5.0 K, respectively, following the observations at sites in

Yeongdong region. The terrain data used was derived from a 30-second terrain data base. Initial

and boundary conditions came from the 40-km resolution RDAPS (Regional Data Assimilation

and Prediction System of Korea Meteorological Administration, KMA 1996). The ARPS was

run for 24 hours, starting at 2100 LST 3 February, about 3 hours before the onset of heavy

snowfall.

4. Results of model simulation

As mentioned earlier, the ARPS model was run for a 24-h period starting at 2100 LST 3

February 1998. The results at 6 h model time will be described mainly, a time of most intense

precipitation. The convergence zone associated with the coastal front remained quasi-stationary

for about 3 hours in its mature stage. At this stage, a balance exists between the dammed cold air

over the eastern slope and the coastal onshore flow (more details on model results later).

a. Predicted temperature fields

10

Figure 7 shows 6-h predicted potential temperature θ field at the first model level or

about 10 m above ground, valid at 0300 LST, 4 February 1998. A very narrow wedge of cold air

(thermal trough), as indicated by the darkened 268 K contour, is located along the eastern slope

of the Taebaek mountains and the coastal region, indicating the presence of cold air damming.

Meanwhile a broader thermal ridge just off the coast is quasi-parallel to the coast. Along line A-

B shown in Fig. 1 that is roughly perpendicular to the mountain axis, the vertical cross sections

of temperature T, potential temperature θ, and equivalent potential temperature eθ are shown in

Fig. 8 for the same time. They show clearly that there exists a distinct cold dome extending from

the coast to the upper part of the eastern slope of the mountain barrier, centering nearly at the

foot of the mountains. The damming formed because the cold air flowing from the northeast as

part of the Siberian High circulation was blocked at low levels by the mountains and trapped or

dammed to the east of the mountains. The cold air supplied by the Siberian High was channeled

southeastward along the eastern slopes of the mountains, reducing the temperature east of the

mountain through thermal advection. In the sea-level pressure chart of the simulation (not

shown), wedge-shaped pressure ridge is clearly identified along the eastern slopes of the

mountains, which is the result of hydrostatic pressure effect of the dammed cold-air.

Cold-air damming is not unique to the eastern part of Korea or the eastern United States; it is

also observed on the eastern slope of the Rockies (Church and Stephens 1941; Murakami and Ho

1981) and on other mountain slopes of the world (Smith 1982). Baker (1970) showed that

shallow cold air is channeled southward establishing an inverted surface pressure ridge to the

east of the Appalachian mountains. Stauffer and Warner (1987) stated that the wedge ridge

appeared to be hydrostatically consistent with the low-level dome of cold air, and the axis of the

pressure ridge was generally located where the low-level air was coldest and the deepest.

11

Following Forbes et al. (1987) and Bell and Bosart (1988), the cold air wedge is maintained by a

combination of frictional and isallobaric effects, evaporation, adiabatic ascent, and orographic

channeling. Furthermore, Nielsen (1989) commented that the effect of cold air damming on the

coastal front is to accumulate the cold air against the mountains, leading to a greater tendency for

coastal fronts to remain stationary, and it may become the dominant influence on frontal motion

after the cold dome has been established. These processes appear to also apply to the case

studied here.

Due to the modification of the boundary layer air off the coast by sensible and latent heat

fluxes from the ocean, relatively warm air is established and located extensively just off the coast.

This distribution creates the broader thermal ridge and leads to a strong thermal gradient or a

coastal front. The thick dotted line in Fig. 8a indicates the coastal front, which is analyzed based

on both temperature and wind fields from the simulation. Across the front at the surface, there is

a strong temperature gradient of approximately 0.2 - 0.3 K km-1 and wind shear is about 4

degree km-1 (Figs. 8a and 9a). This temperature contrast is nearly twice as large as the 10-15 K

per 100 km reported by Bosart (1981) based on buoy data for a case of coastal frontogenesis in

the Carolinas, but is about half as large as the 5-10 K over 5-10 km across the New England

coastal front studied in Bosart et al. (1972).

Referring to the definition of the coastal front in Glossary of Meteorology (Glickman

2000), "a coastal front is a shallow mesoscale frontal zone marked by a distinct cyclonic wind

shift in a region of enhanced thermal contrast" and " ..... In-situ coastal front developments can

occur near mountain barriers where upslope flow results in differential air-mass cooling and

stabilization and where offshore troughs form due to differential heating across oceanic thermal

boundaries. ... ". Thus, the coastal front of this study satisfies this definition.

12

Furthermore, this front appears as an “extended coastal front” (Forbes et al. 1987) - the

sloping stable layer or inversion separating the trapped cold dome from the warm onshore flow

above. Kocin and Uccellini (1990) commented that in general the coastal fronts that form during

damming episodes are easily found by an inverted sea level pressure trough near the coast or just

offshore of the coast, and are distinguished by a remarkable temperature contrast and wind

direction changes. Several studies (Bosart et al. 1972; Bosart 1975; Marks et al. 1979; Bosart

1981) have suggested that the convergence accompanied by coastal frontogenesis is probably

responsible for the increased precipitation rate along the cold side of the frontal boundary, where

the heaviest snowfall is often observed.

In our case, the cold air inland and the warm maritime air off the coast are divided by the

shallow, extended coastal front, which slopes vertically with height first and then more

horizontally with its maximum height being at about 400 m. Due to its shallowness, high enough

vertical resolution, as used in this study, will be necessary to forecast the essential mesoscale

features.

From Fig. 8b, the isentropes in the frontal zone at the foot o the mountain are nearly

vertical up to ~300 m AGL; they then turn to be more horizontal, implying that isentropic

upward motion was occurring mainly at the leading edge of the front, along the almost vertical

slope located just off the coast. The wind fields show that the location of the vertical isentropes

also coincide with the wind shift/convergence line (Fig. 9), another evidence of strong ascending

motion at the front.

A dry adiabatic lapse rate is found in the mixed boundary layer below 900 m over the

warm sea (Fig. 8b). Furthermore, negative vertical eθ gradient is found east of the front over the

warm water (Fig. 8c), indicating convective instability. An upward bulge of equivalent potential

13

temperature is found ahead of the frontal zone (Fig. 8c), which is a sign of forced upward motion

in the zone and of a high potential for developing convection in this region. Meanwhile, the

stratification over the eastern slope of the mountains is much more stable as warm air from the

sea is forced up over the shallow cold dome.

Comparing the coastal front of this study which occurred along the eastern coast of the

Korean Peninsula with those occurring in New England of the United States (Bosart et al. 1972),

there exist some differences. The development of the coastal front of this study is not related

directly to synoptic-scale cyclone genesis, and the front does not mark an axis of maximum

cyclonic vorticity. Thus, the coastal front of this case did not serve as a boundary along which

intensifying synoptic-scale cyclones move northward. The New England coastal fronts studied

by Bosart et al. (1972) dissipated when a cyclonic disturbance offshore reached the latitude of

southern New England. Meanwhile, the front of this study developed as the Siberian High

expands toward the northern part of the East Sea, resulting in northeasterly onshore winds along

the eastern coast of Korean Peninsula. The dissipation of the front occurs with the cession of the

onshore winds when the Siberian High shrinks, thus, the dissipation is not directly related to a

cyclone movement, an aspect that is different from the classic New England cases.

b. Predicted wind fields

The winds at the model initialization time were nearly northerly near the coast (not

shown). By 3 h, a low-level convergence zone had developed about 10 km off the coast as the

winds west of the convergence zone backed with time from northerly to northwesterly (to be

roughly parallel to the mountains) due to the effect of mountain blocking to the northerly flow,

while the winds east of the convergence zone remained northerly (not shown).

14

Figure 9a shows the model predicted surface wind vector field at 6 h. In this figure,

northerly to northeasterly winds are prevalent east of the convergence zone over the East Sea,

meanwhile, cold, northwesterly winds, which are nearly parallel to the mountains, dominate from

the eastern side of the mountains to about 10 km off the coastline. As discussed earlier, these

northwesterly winds are the result of mountain blocking and orographic channeling, which

causes the cold air damming against the mountain slope. The low-level northwesterly flow of

cold air over the eastern slope and the coastal plain contrasts sharply with the northerly to

northeasterly flow of maritime air over the East Sea. This zone of strong low-level temperature

contrast or baroclinicity is called a "coastal front" (Bosart et al. 1972; Bosart 1981; Bluestein

1993). The northeasterly winds, in particular, already have a long fetch over the warm sea

surface to absorb ample heat and moisture. Thus, a distinct moisture convergence zone is well

developed where the northwesterly winds and the northerly to northeasterly winds meet, and the

zone is aligned nearly parallel to the axis of the mountains (Fig. 9b) but off the coast and is well

matched with the mesoscale snowbands seen in the NOAA satellite image in Fig. 4. It is

estimated that the maximum intensity of moisture convergence is about 3107.4 −× g kg 1− s 1− .

Compared to the location of the L2 banded radar echo in Fig. 5, this simulated maximum

moisture convergence zone is well positioned, indicating a proper simulation of the frontal zone

by the ARPS model. After this hour, the coastal front or convergence zone remained nearly

stationary for about 6 hours in the model. After 15 h of model time, the front begins to weaken,

apparently due to the weakening of the cold air damming and its seaward movement due to the

weakening of easterly flow ahead of it. The front eventually disappears in the model by 23 h (not

shown).

Figure 10 shows the wind vectors projected into the plane of a vertical cross-section, the

15

horizontal wind barbs in the same cross section, and the vertical velocity contours, at 6 h of

model time. The cross section is the same as in Fig. 8 and runs through Kangnung and is normal

to the axis of mountains. In the region of cold dome, cold air is seen running down along the

eastern slope, analogous to a density current. It meets the onshore winds just off the coast then

turns upward, forming a closed vertical circulation. The warm moist air is seen to rise over the

cold dome and its circulation, and eventually flows over the mountain ridge (Fig. 10a).

In Fig. 10b, it is clear that the northwesterly, which is roughly parallel to the mountains, is

dominant over the mountain slope, and northerly winds, on the contrary, prevail off the coast.

This leads a rather strong directional shear, at about /40o 10 km across the coastal front. The

wind shear becomes very weak above the 500 m height.

From Fig. 10c, a maximum zone of upward motion of over 0.35 m s 1− just off the coast is

found above the low-level coastal convergence zone, while the weaker zone of upward motion

over the upper part of the eastern slope of the mountains are due to upslope motion over the cold

dome. The lower part of the maximum vertical velocity zone is located nearly along the vertical

slope of the coastal front. Meanwhile, a shallow layer of downward motion is found along the

eastern slope of the mountains, due to the cold air running down the slope. This feature lasts for

about 4 hours, while precipitation is going on off the coast, as was observed. The intensity of

upward motion just off the coast increases steadily over time, reaching a maximum of about 0.42

m s 1− by 8 h. After that time, the intensity steadily decreases. The shape of the vertical velocity

profile just off the coast (not shown) resembles those documented by Ballentine (1980) and

Sanders (1955) for New England coastal surface fronts.

Figure 11 shows the east-west and north-south velocity components in the same vertical

cross section. They show a well-developed, low-level, northwesterly jet located east of the

16

Taebaek mountains at this time (6 h). The strongest northwesterly wind, the jet core, of about 12

m s 1− remains below the ridge of the mountains, just below the extended coastal front inversion;

this is called the mountain barrier jet. For the Appalachian ice-storm of 13-14 January 1980, a

12-14 m s 1− northeasterly jet, identified from surface observations, appeared near the base of the

extended coastal front inversion (Forbes et al. 1987). Unfortunately there is no wind data

available above the surface at Kangnung to compare the simulated winds with. Stauffer and

Warner (1987) carried out a 24-h numerical simulation to investigate the mesoscale structure of

these phenomena associated with the Appalachian ice-storm of 13-14 January 1980 and showed

a similar, low-level northerly jet east of the Appalachians. In addition, low-level, mountain-

parallel jets are known to be frequently observed along the Brooks range in Alaska

(Schwerdtfeger 1975a) and along the Antarctic Peninsula and the Transantarctic Mountains in

Antarctica (Schwerdtfeger 1975b). Also, such barrier winds along the western slope of the Sierra

Nevada Range are common wintertime features (Parish, 1982). Xu (1990) shows through an

idealized model that the barrier jet develops as a result of geostrophic adjustment when the on-

slope flow is slowed down by the mountain barrier.

c. Moisture and precipitation fields

Figure 12 shows the vertical cross-sections of specific humidity at 6 h. Strong

northeasterly winds from the sea, combined with upward motion just off the coast and over the

eastern slopes of the mountains, result in a tongue of moist air extending from the sea surface

inland up over the mound of cold, drier air on the slope. The largest horizontal gradient of

specific humidity is located at the coastal region below the moist tongue.

The distribution of 24-h accumulated precipitation from the simulation ending at 2400

17

LST 4 February 1998 is shown in Fig. 13. It is clear that starting from mountainous inland to off

the coast, the snowfall amount increases remarkably, agreeing well with the observation shown

in Fig. 3. The predicted maximum precipitation band extends quasi-parallel to the coastline from

Sokcho (SC) south through Tonghae, just off the coast. The precipitation on the eastern mountain

slope is much weaker than off the coast, where strong lifting at the coastal front is present. By

the time that onshore air reaches the mountain slope, most of the moisture has been exhausted by

the convection. The remaining precipitation is delivered over the ridge following a relatively

horizontal path along the upper part of the vertical circulation. This precipitation distribution

matches very well the radar reflectivity observations taken at Tonghae at 0010 LST 4 February

1998 (Fig. 5) when the snowfall intensity was strong. These results suggest that the mesoscale

precipitation, especially its spatial distribution, within this event is reproduced well by the model.

The actual predicted precipitation amounts over land at Kangnung and Tonghae are lower than

observed. The quantitative aspect of precipitation forecast is always a bigger challenge, and

many aspects of the model, including the often insufficient resolution and inadequate model

microphysics can cause error in quantitative precipitation forecast.

d. Froude number and similarity to density current

The effect of mountain blocking can be quantified by the Froude number, NHUFr /=

(e.g., Forbes 1987), where U is the speed of the component of the undisturbed wind normal to

the mountain, and N is the Brunt-Väsälä frequency upstream of the mountain, and H is the

mountain height, which is 900 m in this study. The square of Fr is proportional to the ratio of

kinetic energy of air to the potential energy that must be overcome for the air parcel to reach the

mountain top. The air can flow over the mountain without major deflection or blocking at large

18

enough Fr. At low Fr, ranging from 0.5 to 2.3 (e.g., Kitabayashi 1977; Mason and Skyes 1978;

Baines 1979), the flow becomes blocked at low levels by the terrain. Forbes et al. (1987) also

showed in their study of an Appalachian ice storm associated with cold air damming that at small

Fr (about 0.3 to 0.4), the flow was blocked and was flowing mainly in the mountain-parallel

direction. Similarly, the calculated Froude number for this case is about 0.3 to 0.4 in the

beginning stage at 1 h of simulation. The blocking effect by the terrain is therefore dominant so

that the air flows are deflected leftward to become parallel to the mountains. Another reason for

the leftward deflection is the large-scale pressure gradient force, which drives the flow in the low

pressure direction when the Coriolis force associated with the decelerated flow is reduced (Xu

1990). At 6 h of simulation or the mature stage of cold air damming, Fr is increased to about 0.8

to 0.9, indicating that the strength of the cold-air damming is lessened compared to the beginning

stage. The flow is still blocked at low levels by the terrain. For this case, the layer of blocked

flow does not reach the mountain top, consistent with the low Froude number.

The dense, cold air that flows down the slope of the mountains and rises when

encountering the warm air from the sea (see Fig. 10a) is very similar to a density current.

Bluestein (1993) illustrated that topography can “trap” a density current and showed an example

of the evolution of a “coastal surge” on the west coast of the United States.

To see how much the mesoscale front of our case behaves like a density current, the

front propagation speed is compared to that of a density current, as given by

C = kvc

vcvw

TTT

gH−

+ 0.62u , (1)

where subscripts w and c refer to the warm and cold air, vwT and vcT are the average virtual

temperatures on the warm and cold sides, respectively, g is gravitational acceleration, k is the

19

density current constant and u is the vertical average of the wind parallel to the density current

motion in the warm air ahead of the front over the depth of the density current, and u is positive

in the direction of current motion (Bluestein 1993). There has been a wide range of k values,

between 0.7 and 1.0, obtained for thunderstorm-induced density currents (Bluestein 1993). Note

that this equation considers an ambient wind, u , following Simpson and Britter (1980).

For our case, vwT = 273 K, vcT = 270 K, H = 300 m, u is about - 6 ms 1− , g = 9.8 ms 1− ,

k is set to be 0.77 following Wakimoto (1982). The calculated speed, C, of the density current

according to (1) is about 0.6 m s-1, which is nearly identical to the propagation speed (quasi-

stationary) of the simulated front in the model. This indicates that the front in this case has the

characteristics of an atmospheric density current, and there is a balance between the density

current and the environmental wind at the mature stage. Because the onshore flow originates

from the well mixed marine boundary layer and it does not contain much shear, the updraft

forced at the leading edge of the front tends to be tilted rearward instead of rising high (Rotunno

et al. 1988), the precipitation system tends to be shallow. Vertical stratification is another factor

limited the depth of the precipitation system, as Fig. 8c shows that the minimum eθ layer is at

about 2 km above the sea surface and the layer above is rather stable.

5. Summary and conclusions

In summary, a 24-h numerical simulation performed using a high-resolution numerical

model has revealed the processes and mechanisms of a heavy snowfall event that occurred in the

Yeongdong region in association with a coastal front that formed along the eastern coast of the

Korean peninsula due to cold air damming against the coastal mountains. Specific evolution

20

from the initial stage to decaying stage involves the following:

1) With the expansion of the Siberian High, cold air outbreak occurred over the northern part of

the East Sea. This provided the initial trigger for the heavy snowfall near the coast. The cold

air, being on the east side of the High, was advected southwestward over the warm sea

surface, collecting a large amount of heat and moisture.

2) Scattered snow-cloud bands formed over the sea due to the release of convective instability

resulting from the heat and moisture fluxes from the warm sea surface, and the clouds were

advected downwind toward the eastern coastal region of the Korean peninsula.

3) Cold air flowing near the coast was dammed against the eastern slope of the mountain range

parallel to the coast, due to the blocking effect of the mountain barrier. It led to strong

temperature gradient and directional wind shear between the cold air dome and the onshore

winds, which arrived from the northeast after a long fetch over the warm water.

4) A coastal front developed in the convergence zone off the coast, where strong temperature

gradient was established. Due to its mesoscale nature, the coastal front and cold air behind

behaved like a density current system.

5) The development and organization of a snow-cloud band along the coastal front led to the

heavy snowfall along the coast, with its maximum located about 10 km off the coast. The

cold air damming and the formation of the coastal front explain why the heaviest snowfall

was not found over the mountain slope, where the orographic lifting is expected to be the

strongest. Detailed studies on such processes and mechanism had not been done before for

the coastal region of Korea.

6) A synoptic-scale pressure pattern change, which in this case was the retreat of the Siberian

High away from the northern East Sea, led to the shift in the wind direction from north-

21

northeasterly to northwesterly in the coastal region and off coast at later times. This resulted

in weakening of the wind direction shear and of temperature gradient across the front and the

seaward propagation of the front moves away from the coast. The precipitation eventually

ceased in the region.

Due to limited observational data available in the coastal region, and the lack of in-situ

data over the ocean, much of our understanding summarized above were obtained through

numerical simulations. A simulation, using a 4-km horizontal grid spacing and 43 vertically

stretched levels, reproduced many physically consistent meso-β scale features associated with

the above sequence of events. The simulated coastal front, with the cold dome lying behind,

behaved like a density current. The simulation did under-predict the amount of precipitation in

the region, and model resolution, physics, as well as initial condition deficiencies can all

contribute to such quantitative errors.

Acknowledgements. This work was primarily supported by the Korea Meteorological

Administration Research and Development program under Grant CATER 2007-2309. Ming Xue

was supported by NSF grants ATM-ATM-0530814, EEC-0313747, ATM-0331594 and ATM-

0331756 and ATM-0608168.

22

References

Atlas, D., S-H. Chou and W. P. Byerly, 1983: The influence of coastal shape on winter meso-

scale air-sea interaction. Mon. Wea. Rev., 111, 245-252.

Baines, P. G., 1979: Observations of stratified flow past three-dimensional barriers. J. Geophys.

Res., 84, 7834-7838.

Baker, D. G., 1970: A study of high pressure ridges to the east of the Appalachian mountains. Ph.

D. Dissertation, Massachusetts Institute of Technology, 127 pp.

Bell, D. G., and L. F. Bosart, 1988: Appalachian cold-air damming. Mon. Wea. Rev., 116, 137-

162.

Ballentine, R. J., 1980: A numerical investigation of New England coastal frontogenesis. Mon.

Wea. Rev., 108, 1479-1497.

Bluestein, H. B., 1993: Synoptic-Dynamic Meteorology in Midlatitudes. Vol. 2: Observations

and Theory of Weather Systems. Oxford University Press, 594 pp.

Bosart, L. F., D. J. Vaudo and J. H. Heldson, Jr., 1972: Coastal frontogenesis. J. Appl. Meteor.,

11, 1236-1258.

Bosart, L. F., 1975: New England coastal frontogenesis. Quart. J. Roy. Meteor. Soc., 101, 957-

978.

Bosart, L. F., 1981: The President's Day snowstorm of 18-19 February 1979: A subsynoptic-

scale event. Mon. Wea. Rev., 109, 1542-1566.

Bosart, L. F., and S. C. Lin, 1984: A diagnostic analysis of the Presidents' day storm of February

1979. Mon. Wea. Rev., 112, 2148-2177.

Byrd, G. P., R A Anstett, J. E. Heim and D. M. Usinski, 1991: Mobile sounding observations of

lake-effect snowbands in western and central New York. Mon. Wea. Rev., 119, 2323-

23

2332.

Byrd, G. P., D. E. Bikos, D. L. Schleede and R. J. Ballentine, 1995: The influence of upwind

lakes on snowfall to the lee of Lake Ontario. Preprints, 14th Conf. Wea. Analysis

Forecasting, Dallas, TX, Amer. Meteor. Soc., 204-206.

Chen, T. C., C. B. Chang and D. J. Perkey, 1983: Numerical study of an AMTEX 75 oceanic

cyclone. Mon. Wea. Rev., 111, 1818-1829.

Church, P. E., and T. E. Stephens, 1941: Influence of the Cascade and Rocky Mountains on the

temperature during the westward spread of polar air. Bull. Amer. Meteor. Soc., 22, 25-30.

Eichenlaub, V. L., 1979: Weather and Climate of the Great Lakes Region, University of Notre

Dame Press, 335 pp.

Estoque, M. A., and K. Ninomiya, 1976: Numerical simulation of Japan Sea effect snowfall.

Tellus, 28, 243-253.

Forbes, G. S., R. A. Anthes and D. W. Thomson, 1987: Synoptic and mesoscale aspects of an

Appalachian ice storm associated with cold-air damming. Mon. Wea. Rev., 115, 564-591.

Glickman, T. S., Ed., 2000: Glossary of Meteorology, 2nd Edition. American Meteorological

Society, 855 pp.

Grossman, R. L., and D. R. Durran, 1984: Interaction of low-level flow with the western Ghat

Mountains and offshore convection in the summer monsoon. Mon. Wea. Rev., 112, 652-

672.

Hjelmfelt, M. R., and R. R. Braham, Jr., 1983: Numerical simulation of the air flow over Lake

Michigan for a major lake-effect snow event. Mon. Wea. Rev., 111, 205-219.

Houze, R. A., Jr., 1993: Cloud Dynamics. Academic Press, 573 pp.

Jhun, J.-G, D.-K. Lee and H.-A. Lee, 1994: A study on the heavy snowfalls occurred in South

24

Korea. J. of Korean Meteor. Soc., 30, 97-117 (in Korean with English abstract).

Kelly, R. D., 1984: Horizontal roll and boundary layer interrelationships observed over Lake

Michigan, J. Atmos. Sci., 41, 1816-1826.

Kitabayshi, K., 1977: Wind tunnel and field studies of stagnant flow upstream of a ridge. J.

Meteor. Soc. Japan, 55, 193-203.

KMA, 1996: A new numerical weather prediction system at KMA. Technical Report 96-2, 147

pp (in Korean).

Kocin, P. J., and L. W. Uccellini, 1990: Snowstorms along the Northeastern Coast of the United

States: 1955 to 1985. Meteor. Monogr., No. 44, Amer. Meteor. Soc.

Kuo, Y.-H., and S. Low-Nam, 1990: Prediction of nine explosive cyclones over the western

atlantic ocean with a regional model. Mon. Wea. Rev., 118, 3-25.

Lee, H., and T.-Y. Lee, 1994: The governing factors for heavy snowfalls in Youngdong area. J.

of Korean Meteor. Soc., 30, 197-218 (in Korean with English abstract).

Lin, Y.-L., R. D. Farley and H. D. Orville, 1983: Bulk parameterization of the snow field in a

cloud model. J. Climate Appl. Meteor., 22, 1065-1092.

Marks, F. D., Jr., and P. M. Austin, 1979: Effects of the New England coastal front on the

distribution of precipitation. Mon. Wea. Rev., 107, 53-67.

Mason, P. J., and R. I. Skyes, 1978: On the interaction of topography and Ekman boundary layer

pumping in a stratified atmosphere. Quart. J. Roy. Meteor. Soc., 104, 475-490.

Murakami, T., and L. Y. C. Ho, 1981: Orographic influence of the Rocky Mountains on the

winter circulation over the contiguous United States. Part II. J. Meteor. Soc. Japan, 59,

683-708.

Nagata, M., M. Ikawa, S. Yoshizumi and T. Yoshida, 1986: On the formation of a convergent

25

cloud band over the Japan sea in winter; Numerical experiments. J. Meteor. Soc. Japan,

64, 841-855.

Nielsen, J. W., 1989: The formation of New England coastal fronts. Mon. Wea. Rev., 117, 1380-

1401.

Ninomiya, K., 1968: Heat and water budget over the Japan sea and the Japan islands in winter

season. J. Meteor. Soc. Japan, 46, 343-372.

Niziol, T. A., W. R. Snyder and J. S. Waldstreicher, 1995: Winter weather forecasting througout

the eastern United States. Part IV: Lake effect snow. Wea Forecasting, 10, 61-77.

Noilhan, J. and S. Planton, 1989: A simple parameterization of land surface processes for

meteorological models. Mon. Wea. Rev., 117, 536-549.

Parish, T. R., 1982: Surface airflow over East Antarctica. Mon. Wea. Rev., 110, 84-90.

Petterssen, S., D. L. Bradbury and K. Pedersen, 1962: The Norwegian cyclone models in relation

to heat and cold sources. Geofys. Publ., Norske Viderskaps-Akad. Oslo, 24, 243-280.

Ren, D. and M. Xue, 2004: A revised force-restore model for land-surface modeling. J. App.

Meteor., 43, 1768-1782.

Richwien, B. A., 1980: The damming effect of the southern Appalachians. Nat. Wea. Dig., 5, 2-

12.

Rotunno, R., J. B. Klemp, and M. L. Weisman, 1988: A theory for strong long-lived squall lines.

J. Atmos. Sci., 45, 463-485.

Sanders, F., 1955: An investigation of the structure and dynamics of an intense surface frontal

zone. J. Meteor., 122, 542-552.

Sanders, F., and J. R. Gyakum, 1980: Synoptic dynamic climatology of the bomb. Mon. Wea.

Rev., 108, 1589-1606.

26

Schwerdtfeger, W., 1975a: Mountain barrier effect on the flow of stable air north of the Brooks

Range. Climate of the Arctic. G. Weller and S. Bowling, Eds., Geophysical Institute

University of Alaska, Fairbanks, 204-208.

Schwerdtfeger, W., 1975b: The effect of the Antarctic Peninsula on the temperature regime of

the Weddell Sea. Mon. Wea. Rev., 103, 45-51.

Seo, E.-K., and J.-G. Jhun, 1991: A case study of the heavy snowfalls occurred in the Korean

peninsula from 29 January to 1 February 1990. J. of Korean Meteor. Soc., 27, 165-179

(in Korean with English abstract).

Simpson, J. E., and R. E. Britter, 1980: A laboratory model of an atmospheric mesofront. Quart.

J. Roy. Meteor. Soc., 106, 485-500.

Smith, R. B., 1982: Synoptic observations and theory of orographically disturbed wind and

pressure. J. Atmos. Sci., 39, 60-70.

Stauffer, D. R., and T. T. Warner, 1987: A numerical study of Appalachian cold-air damming

and coastal frontogenesis. Mon. Wea. Rev., 115, 799-821.

Takeda, T., K. Isono, M. Wada, Y. Ishizaka, K. Okada, Y. Fujiyoshi, M. Maruyama, Y. Izawa

and K. Nagaya, 1982: Modification of convective snow-clouds in landing the Japan Sea

coastal region. J. Meteor. Soc. Japan, 60, 967-977.

Wakimoto, R. M., 1982: The life cycle of the thunderstorm gust fronts as viewed with Doppler

radar and rawinsonde data. Mon. Wea. Rev., 110, 1060-1082.

Waldstreicher, J. S., 2002: A foot of snow from a 3000-foot cloud - The ocean-effect snowstorm

of 14 January 1999. Bull. Amer. Meteor. Soc., 83, 19-22.

Xu, Q., 1990: A theoretical study of cold-air damming. J. Atmos. Sci., 47, 2969-2985.

Xue, M., K. K. Droegemeier, V. Wong, A. Shapiro and K. Brewster, 1995: ARPS Version 4.0

27

User's Guide. [Available at http://www.caps.ou.edu/ARPS], 380 pp.

Xue, M., K. K. Droegemeier, and V. Wong, 2000: The Advanced Regional Prediction System

(ARPS) - A multiscale nonhydrostatic atmospheric simulation and prediction tool. Part I:

Model dynamics and verification. Meteor. Atmos. Physics, 75, 161-193.

Xue, M., K. K. Droegemeier, V. Wong, A. Shapiro, K. Brewster, F. Carr, D. Weber, Y. Liu and

D. Wang, 2001: The Advanced Regional Prediction System (ARPS) - A multi-scale

nonhydrostatic atmospheric simulation and prediction tool. Part II: Model physics and

applications. Meteor. Atmos. Phys., 76, 143-166.

Xue, M., D.-H. Wang, J.-D. Gao, K. Brewster and K. K. Droegemeier, 2003: The Advanced

Regional Prediction System (ARPS), storm-scale numerical weather prediction and data

assimilation. Meteor. Atmos. Physics, 82, 139-170.

28

List of Figures

Fig. 1. Topography of the Korean peninsula in the ARPS domain. Contour interval is 400 m. The

inner box with bold, solid lines indicates the Yeongdong region which is the focus region

for the study. Solid line A-B indicates the location of the cross sections shown in later

figures, which is roughly normal to the axis of the costal mountains passing through

Kangnung. The outer box shows more detail topography in the Yeongdong area. Contour

interval in the box is 100 m. The locations marked by black inverse triangles include

surface observation sites SC (Sokcho), KN (Kangnung), TG (Taegwallyong), and radar

site TH (Tonghae). The numbers above the triangles indicate the height of the sites above

sea level.

Fig. 2. Synoptic surface weather charts at (a) 2100 LST (1200 UTC) 3 February and (b) at 2100

LST 4 February 1998.

Fig. 3. Distribution of 24-h accumulated snowfall depth (5 cm contour intervals) ending at 2400

LST 4 February 1998. SC, KN, TG and TH represent the locations of Sokcho, Kangnung,

Taegwallyong and Tonghae, respectively.

Fig. 4. The NOAA satellite image at 0251 LST 4 February 1998 when the snowfall intensity at

Kangnung was the greatest.

Fig. 5. Observed radar echoes from Tonghae radar, at 0010 LST 4 February 1998. Color scales

for precipitation intensities in mm hr-1 for the radar echo are given at lower right corner

of the figure.

Fig. 6. Observed surface wind field at 0030 LST 4 February 1998. One full barb represents 5 m

s-1.

Fig. 7. The 6-hour predicted horizontal potential temperature field at 10 m, the first model level

29

above ground, valid at 0300 LST 4 February 1998. The contour interval is 2 K.

Fig. 8. The vertical cross-sections along line A-B in Fig. 1, of (a) temperature, (b) potential

temperature and (c) equivalent potential temperature at 6 h of forecast time. The cross

section roughly passes through Kangnung and is about perpendicular to the coastal

mountain range. The horizontal axis labels represent the distances in km from point A,

following the path of the cross section. The dot line in Fig. 8a indicates the coastal front.

Fig. 9. As in Fig. 7, but for wind (a) and moisture convergence (b) fields. In Fig. 9a, the length of

each vector is proportional to wind speed and the length scales for the components of the

wind vector are given in the lower left corner. The interval of the moisture convergence

in Fig. 9b is 3 1 11 10 g kg s− − −× .

Fig. 10. As in Fig. 8 but for (a) the wind vectors that present the vertical velocity and the

horizontal velocity projected to the cross-section, (b) the horizontal wind barbs, and (c)

vertical velocity w contours. In (a), the length of each vector is proportional to wind

speed, and the vector key is given at the lower left corner of the plot. In (b), one full barb

represents 5 1ms− . The interval in (c) is 2 15 10 m s− −× .

Fig. 11. As Fig. 8 but for (a) u-component (west-east), and (b) v-component (south-north) of

wind. The contour intervals are 1 m s-1.

Fig. 12. As Fig. 8 but for specific humidity. The contour interval is 0.25 g kg-1.

Fig. 13. The 24-h accumulated precipitation for a subdomain with 2 mm contour intervals.

30

A

EAST SEA (JAPAN SEA)

Yeongdong region

AB

B

ATGTG

IJIJ

SCSC

HGHGKGKG

THTH

WOWO

Fig. 1. Topography of the Korean peninsula in the ARPS domain. Contour interval is 400 m. The inner box with bold, solid lines indicates the Yeongdong region which is the focus region for the study. Solid line A-B indicates the location of the cross sections shown in later figures, which is roughly normal to the axis of the costal mountains passing through Kangnung. The outer box shows more detail topography in the Yeongdong area. Contour interval in the box is 100 m. The locations marked by black inverse triangles include surface observation sites SC (Sokcho), KN (Kangnung), TG (Taegwallyong), and radar site TH (Tonghae). The numbers above the triangles indicate the height of the sites above sea level.

31

Fig. 2. Synoptic surface weather charts at (a) 2100 LST (1200 UTC) 3 February and (b) at 2100 LST 4 February 1998.

(a)

(b)

32

Fig. 3. Distribution of 24-h accumulated snowfall depth (5 cm contour intervals) ending at 2400 LST 4 February 1998. SC, KN, TG and TH represent the locations of Sokcho, Kangnung, Taegwallyong and Tonghae, respectively.

33

Fig. 4. The NOAA satellite image at 0251 LST 4 February 1998 when the snowfall intensity at Kangnung was the greatest.

34

Fig. 5. Observed radar echoes from Tonghae radar, at 0010 LST 4 February 1998. Color scales for precipitation intensities in mm hr-1 for the radar echo are given at lower right corner of the figure.

35

Fig. 6. Observed surface wind field at 0030 LST 4 February 1998. One full barb represents 5 m s-1.

36

Fig. 7. The 6-hour predicted horizontal potential temperature field at 10 m, the first model level above ground, valid at 0300 LST 4 February 1998. The contour interval is 2 K.

37

(b)

(c)

(a)

Fig. 8. The vertical cross-sections along line A-B in Fig. 1, of (a) temperature, (b) potential temperature and (c) equivalent potential temperature at 6 h of forecast time. The cross section roughly passes through Kangnung and is about perpendicular to the coastal mountain range. The horizontal axis labels represent the distances in km from point A, following the path of the cross section. The dot line in Fig. 8a indicates the coastal front.

38

(a)

(b)

Fig. 9. As in Fig. 7, but for wind (a) and moisture convergence (b) fields. In Fig. 9a, the length of each vector is proportional to wind speed and the length scales for the components of the wind vector are given in the lower left corner. The

39

interval of the moisture convergence in Fig. 9b is 3 1 11 10 g kg s− − −× .

40

(a)

(c)

(b)

Fig. 10. As in Fig. 8 but for (a) the wind vectors that present the vertical velocity and the horizontal velocity projected to the cross-section, (b) the horizontal wind barbs, and (c) vertical velocity w contours. In (a), the length of each vector is proportional to wind speed, and the vector key is given at the lower left corner of the plot. In (b), one full barb represents 5

1ms− . The interval in (c) is 2 15 10 m s− −× .

41

(b)

(a)

Fig. 11. As Fig. 8 but for (a) u-component (west-east), and (b) v-component (south-north) of wind. The contour intervals are 1 m s-1.

42

Fig. 12. As Fig. 8 but for specific humidity. The contour interval is 0.25 g kg-1.

43

Fig. 13. The 24-h accumulated precipitation for a subdomain with 2 mm contour intervals.