Daeun JEONG,Ki-Hong MIN,Gyuwon LEE,and Kyung-Eak KIMClimate Research Department,Asia-Pacif i c Economic Cooperation Climate Center,Busan 62-020,South Korea

2Center for Atmospheric Remote Sensing,Kyungpook National University,Daegu 702-701,South Korea

3Department of Astronomy and Atmospheric Sciences,Kyungpook National University,Daegu 702-701,South Korea

4Department of Earth,Atmospheric,and Planetary Sciences,Purdue University,West Lafayette,IN 47907-2051,U.S.A.

1.Introduction

Severe weather often causes huge loss of life and damage to property.According to the National Emergency Management Agency(NEMA,2010),more than half of the storm damages in Korea from 2001 to 2010 were caused by heavy precipitation and typhoons.In 2010 alone,there were 14 casualties and$390 million worth of property damages due to severe weather(NEMA,2010).The Korean Meteorological Administration(KMA)def i nes heavy precipitation as:(1)rainfall at an intensity of greater than 30 mm per hour,or greater than 80 mm per day,or equivalent to 10%of annual precipitation in a single day;(2)having a duration from a few tens of minutes to a few hours;and(3)having a relatively narrow areal coverage with a cell radius of 10–20 km.Heavy precipitation typically occurs during the Changma period,which is the rainy season of summer in Korea.Occasionally,heavy precipitation occurs in early autumn outside the Changma period.

Heavy rainfall caused by a mesoscale convective system(MCS)with a training-line band structure[hereafter called “mesoscale convective band”(MCB)]developed on 21 September2010,f l ooded14018housesandblackedout2706 households in Seoul and the surrounding areas with 259.5 mm of precipitation recorded in a single day.Despite the seriousness and amount of damage caused by autumn heavy precipitation events,there have been few studies of them in Korea.Existing studies on heavy precipitation have focused mainly on the formation mechanisms and development-type classif i cations that occur during the summer monsoon period(Park and Kim,1983;Park,2006;Lee,2007;Lee,2007;Ko,2008;Sun et al.,2011).

ThetriggeringmechanismbetweenthesummerChangma heavy precipitation and autumn precipitation is quite distinct.This is because the former mechanism resembles that of a tropical warm rain process,whereas the latter is that of a midlatitude baroclinic structure.Tropopause folding(TPF)is one possible mechanism that can initiate severe weather with heavy precipitation in autumn(Uccellini et al.,1985;Lee,2007;Gold and Nielsen-Gammon,2008;Lee et al.,2010).When the stratospheric air in a polar region intrudes into the mid-latitude troposphere,it injects a high potential vorticity(PV)anomaly into the upper level,which induces cyclonic vorticity at lower levels.Ko(2008)suggested that the triggering mechanism for the formation of Mesoscale Convective Complex(MCC)is the enhanced atmospheric baroclinicity over the Changma front during TPF.Lee et al.(2010)also examined the development mechanisms of heavy rainfall without a typhoon or Changma front and reported that the heavy rainfall over the Yeongdong region in the eastern part of South Korea is caused by enhanced cyclonic vorticity at the surface,which they found to be closely linked to the intrusion of dry,stable,stratospheric air with large cyclonic vorticity during the TPF.Another possible development mechanism of heavy precipitation is the presence of a low-level jet(LLJ),because air with high moisture content is situated within the atmospheric boundary layer(ABL).Heavy precipitation can be enhanced in regions where moisture is transported by LLJ(Park and Kim,1983).A statistical study on heavy precipitation during Baiu,which is the rainy season in Japan,showed that the frequency of heavy precipitation is proportional to the speed of the LLJ(Matsumoto et al.,1971;Maddox,1983).Furthermore,a recent study showed that heavy precipitation in the Middle East is associated with the location of maximum wind speed,surface cyclone,moisture f l ux convergence,and equivalent potential temperature(Hassan and Meshkatee,2011).Therefore,the presence of an upper-level tropopause fold,surface cyclone,and LLJ can beaprecursorforsevereweatherinKoreaandothercountries in Asia.

Mesoscale heavy precipitation systems(HPSs)in Korea are classif i ed based on phenomenological analysis of convective systems(Lee and Kim,2007).The four major types are:isolated thunderstorm(IS),convective band(CB),squall line(SL),andcloudcluster(CC).Themajorcriteriafor the classifi cation process are shape,size,and movement of HPSs.The convective band among these four types is typically 2 km to 20 km in width and 100 km in length,and their lifespan ranges from approximately tens of minutes to a few hours.Lee and Kim(2007)reported that CBs normally cause the strongest and most concentrated localized precipitation in the shortest time among the four types of HPSs.Sun and Lee(2002)conducted a numerical study of an intense quasistationary CB over Korea and found that converging air fl ow between the mid-latitude cyclone to the north and the western Paci fi c subtropical high to the south is important.In addition,heavy rain is due to several long-lived precipitation cells along the band initiated by a convectively unstable environment.Despite recent studies,it is still dif fi cult to forecast heavy rainfall caused by CBs because the development mechanism is not completely understood.

Bluestein and Jain(1985)identi fi ed four types of severe,mesoscale convective-line developments in Oklahoma,U.S.A,that occur during spring based on an 11-yr radar ref l ectivity dataset:(1)broken-line;(2)back-building;(3)broken-areal;and(4)embedded-areal.Rain bands can form with or without fronts.However,these four types do not accompany a stationary front.The convective line or band is an important mode of organized convection and frequently occurs in many regions around the world,including East Asia.Unlike the fast-moving squall lines,these CBs are generally slow-moving or quasi-stationary(Sun and Lee,2002).When deep convective cells are organized in a way that they move repeatedly over a given area,it is referred to as radar“echo training”and produces large local rainfall totals(Doswell et al.,1996).Parker and Johnson(2000)identif i ed the primary precipitation dynamics of three modes of linear MCSs that are common in the U.S.central plains;namely,trailing-stratiform(TS),leading-stratiform(LS),andparallelstratiform(PS)precipitation,respectively.They found that LS MCSs typically move more slowly than the other modes,and may produce more extreme rainfall and f l ash f l ooding.

The development mechanisms and characteristics of the various MCBs in East Asia are still under study.This is because theoretical models,observations,and the mechanisms,which vary from storm to storm,have limitations and are incomplete.In the present study,we examined the mechanisms for the formation and development of the initial convection and subsequent evolution of an MCB on a quasi-stationary front using TPF and PV dynamics.In addition,the thermodynamic and dynamic characteristics of the MCB along the quasi-stationary front were investigated and analyzed to explain the synoptic,thermodynamic,and dynamic characteristics of atmospheric f l ow,where the convection was embedded.

2.Data and analysis method

2.1.Selected case and data

The selected case for this study was based on the aforementioned KMA criteria for heavy precipitation.There was 71.0 mm of rainfall recorded in a single hour at 0600 UTC 20 September 2010,with a daily total precipitation of 259.5 mm at Seoul weather station(37.57°N,126.9°E)from 1500 UTC 20 to 1400 UTC 21 September 2010(Fig.1).This amount broke the daily maximum precipitation record in September and ranked second since records began in 1908 at the Seoul weather station(KMA,2010).The study case met all of the KMA criteria for heavy precipitation,as discussed in the introduction.

The data used for the study were surface observations,radar imagery,satellite data and images,surface and upperlevel weather charts and soundings.The radar data were used to classify the HPS by analyzing the shape,size and movement of convective cells through the radar Constant Altitude Plan Position Indicator(CAPPI)imagery.The location of the initial convection,height of cloud,and cloud top temperature were analyzed using Multifunctional Transport Satellite-2(MTSAT-2)enhanced infrared and water vapor imagery.The soundings used in this study were taken from Qingdao,China(36.06°N,120.33°E)and Baengyeongdo,South Korea(37.97°N,124.63°E).NCEP Global Final Analysis data were used to further understand and analyze the thermodynamic and dynamic characteristics in detail.The spatial resolution of these data is 1°×1°in both latitude and longitude,with 21 mandatory levels,and a temporal resolution of 6 h.The dataset is composed of seven variables:absolute vorticity,geopotential height,relative humidity,temperature,horizontal wind components(u,v),and vertical wind component(ω).

2.2.Analysis method

2.2.1.Thermodynamic and dynamic analyses

The equivalent potential temperature was used to determine the potential/convective instability,stability index,and location of the front(Petty,2008).The instability in the HPS anditsvicinitywereanalyzedbytheequivalentpotentialtemperature,which was calculated by Bolton’s(1980)formula:

whereθeis the equivalent potential temperature(units:K),θ is the potential temperature(units:K),w is the mixing ratio(unitless),and TLis the temperature at the lifting condensation level(LCL).

The dynamic characteristics of a convective system,such as PV,wind,relative vorticity and divergence,are analyzed to examine the formation and development of HPSs.PV is used to analyze the dynamic tropopause because PV is conserved under adiabatic and frictionless f l ow conditions(Ertel,1942;Hoskins et al.,1985).Large anomalies in PV are associated with atmospheric disturbances,and such anomalies could be advected rapidly under pseudo-adiabatic conditions in the upper troposphere(Bluestein,1993).PV shows a sharp gradient between the isentropic surfaces when making a transition from the troposphere to stratosphere(Davis and Emanuel,1991;Morgan and Nielsen-Gammon,1998).The 1–3 PVU[PV units(10?6K kg?1m?2s?1)]range lies within the transition zone(Morgan and Nielsen-Gammon,1998).A PV surface is selected within the transition zone to def i ne the dynamic tropopause,in which the value varies from study to study.In this study,1 PVU,as suggested by Bithell et al.(1999),was chosen.In the low and middle troposphere,the average PV ranges from approximately 0.3 PVU to 0.5 PVU(Santurette and Georgiev,2005),so anything greater than this PVU range is an anomaly in the troposphere.The PV was calculated using the following equation:

where g is the gravitational acceleration,f is the Coriolis factor,?k istheverticalunitvector,θisthepotentialtemperature,V is the horizontal wind vector,and p is the atmospheric pressure.

In addition,divergence,relative vorticity,vertical motion and their time evolution were analyzed to identify the dominant mechanisms in the genesis of mesoscale convective HPSs and to determine the degree of convection.Anomalies of cyclonic relative vorticity would suggest strong severe weatherwith lowpressure and convergencethatenhanceconvection.

2.2.2.Isentropic analysis

Isentropic surfaces act as material surfaces over synoptic scales(Rossby and Collaborators,1937;Hoskins et al.,1985).That is,the air parcels are thermodynamically bound to their isentropic surfaces in the absence of diabatic process because the potential temperature is conserved(Moore,1993a).On the isentropic surface,adiabatic f l ow is conf i ned and slopes upward in the area of strong convection,showing the nature of horizontal and vertical three-dimensional f l ow(Moore,1993b).In addition,air parcels on the isentropic surfaces conserve their mixing ratio or specif i c humidity.Owing to these features,dry lines and moist tongues,which represent the degree of moisture,can be forecasted using the component of the wind perpendicular to the isohumes(Wilson et al.,1980).

Isentropic surfaces,which are chosen individually,are dependent on the time of year(Namias and Stone,1940;Namias,1983;Hoskins et al.,1985).The case for the present study was late September,meaning the low-level isentropic surface fell under 300–305 K according to Namias and Stone(1940).Due to the inf l uence of the North Pacif i c suptropical high,typhoons,and tropical depressions during this period,the isentropic surface of 310 K(low value for summer)was chosen for this study.

3.Results

3.1.Surface observational data analysis

The time series of precipitation rate,temperature,wind,and mixing ratio are shown in Fig.1.The rectangular boxed area represents the period in which there were signif i cant changes in these f i ve meteorological variables.The precipitation rate was less than 5 mm h?1before 0400 UTC but increased rapidly from 0400 UTC reaching a maximum of 75 mm h?1at 0530 UTC(Fig.1a).The amount of precipitation was 164 mm from 0500 to 0700 UTC,amounting to 63%of the daily total.This suggests that the precipitation was quite concentrated during that time.The solid line in the upper panel(Fig.1a)shows the temperature,which indicates a steady temperature of 21°C until 0300 UTC,after which it increased rapidly to approximately 25°C due to a wind shift from the northeasterly to southwesterly direction(Fig.1b).From 0400 UTC,the temperature then decreased to approximately 22°C with heavy precipitation.Both the wind direction and speed changed signif i cantly during the observation period,as shown in the middle panel of Fig.1.The wind speed was less than 2 m s?1before 0300 UTC,but increased with the change in wind direction,reaching 9.6 m s?1at 0400 UTC.The mixing ratio was approximately 15 g kg?1before 0240 UTC,and increased rapidly from that time,reaching more than 18 g kg?1.From 0400 UTC,the mixing ratio decreased and became steady at around 16 g kg?1(Fig.1c).This suggests that the northeastward warm moist air advection toward Seoul contributed signif i cantly to the development of the HPS during the period.

Fig.1.Time series of(a)precipitation rate(mm h?1)and temperature(°C),(b)wind direction(°)and wind speed(m s?1),and(c)mixing ratio(g kg?1)over Seoul,Korea.The temporal resolution is 10 min from 0000 UTC to 1000 UTC 21 September 2010.

3.2.Description of the mesoscale convective band(MCB)

3.2.1.Synoptic overview

The distribution of pressure,formation and movement of the front,and the synoptic conditions of the heavy rainfall were investigated with surface and upper-level weather charts(Fig.2).The distribution of the subtropical high was similar to that of a typical summer weather pattern in Korea,even though the weather maps were typical of those in September.A cold continental air mass(cP)was situated in Mongolia and a warm moist subtropical air mass(mT)was located to the southeast of Korea.A quasi-stationary front formed across the Yellow Sea at 0000 UTC 21 September,which extended from the Shandong region(36.5°N,119°E)in China(Fig.2a).Another mature cyclone with a central pressure of 990 hPa was located at Sakhalin,Russia(51°N,143°E),northeast of Korea.Such a distribution of pressure creates deformation,one of the kinematic properties of wind f i elds,and often plays an important role in frontogenesis(Bluestein,1993).The front moved southeastward with time as the center of low pressure gradually deepened from 1009 to 1006 hPa(Fig.2b).A surface front can form along the trough of low pressure if the distribution of isotherms is oriented in such a way that air f l ows perpendicular to the isotherms near the area of the “saddle-point”,which is the point of intersection of a trough and a ridge in the 925 hPa chart(Fig.3).In this situation,there are two characteristic axes:an axis of dilatation(stretching)toward which streamlines converge asymptotically,andanaxisofcontractionfromwhichstreamlines diverge asymptotically in a rectangular hyperbolic curve(Bluestein,1993).If the angle between the axis of dilatation and the isotherms is less than 45°,the condition for frontogenesis is satisf i ed(Bluestein,1993).The angle was approximately 30°–35°at 1800 UTC 20 September,which was a favorableconditionforafronttoform,wellbefore6hoftheactual surface front(Fig.3a).As the angle became smaller and the temperature gradient became sharper(Fig.3b),a quasistationary front formed in the middle of the Korean peninsula(Figs.2a and b).At 0000 UTC 21 September,the center of low pressure was located at Baengyeongdo and the front extended toward Seoul and Gyeonggi-do.Afterwards,the low pressure and front moved slowly southeastward from 0600 to 1200 UTC 21 September,and rather quickly after 1200 UTC 21 September(Fig.2b).They eventually moved out of the Korean peninsula by 0000 UTC 22 September.

Similarly,at 850 hPa,the cP air mass to the northwest of Korea and the warm sector to the southwest created an area of dilatation that developed a high meridional temperature gradient of 3.6°C(1000 m)?1(Fig.2c).Bluestein(1993)suggested that a surface front develops near an area parallel to and concentrated in isotherms and lies 2°–3°south of this area.In our case,the concentrated isotherms existed around(39°–41°N,122°E)and the quasi-stationary front developed about 2°–3°south of this region(Fig.2d).The f i gure also suggests that moisture was provided by strong southwesterly fl ow of 10 m s?1or more from the Yellow Sea.At 200 hPa,a straight-line jet stream was located to the north of the Korean peninsula with a core of 63 m s?1or more near Heilongjiang,China(48°N,129°E)(Fig.2e).Uccellini and Kocin(1987)described the interaction of transverse vertical circulations associated with two separate jet streak/trough systems.They identi fi ed that a direct vertical circulation is located on the warm side(south)of the upstream con fl uent entrance region of an upper-level jet streak,whereas an indirect vertical circulation is located on the cold side(north)of the downstream dif fl uent exit region.Although the crest and trough of the jet stream were not well de fi ned at 1200 UTC 21 September,an MCB was located in the right-entrance region(south)when it was mature and the vertical circulation was enhanced in this region(Fig.2f).Further analysis of a vertical cross section of the divergence and convergence was made to investigate the development of the HPS,as reported in section 3.4.

Fig.2.Surface(a,b)and upper-level weather charts of 850 hPa(c,d)and 200 hPa(e,f)for 0000 UTC 21 and 1200 UTC 21 September 2010,respectively.The weather symbols and contours represented in the charts correspond to those of WMO standards.

Enhanced infrared imagery from satellite MTSAT-2 were used to estimate the size,direction and movement speed,as well as the location of a convective system with time.Figure 4 presents the satellite imagery from 2000 UTC 20 to 1500 UTC 21 September 2010.A convective cell“A”formed over Shandong at 2000 UTC 20 September and moved eastward(Fig.4a).A new convective cell“B”then formed to the east of“A”at 2200 UTC 20 September(Fig.4b)and merged to become“C”at 0033 UTC over the Yellow Sea of the Korean peninsula(Fig.4c).The cell“C”continued to develop with time,and a new cell“D”formed over the Yellow Sea at 0300 UTC(Fig.4d),which merged with “C”to become“E”(Fig.4e).The cloud “C”in Fig.4d was a tapering cloud,which is a brush-form(or carrot-form)cloud area that gradually thins toward the windward direction in the upper and middle levels.According to the Meteorological Satellite Center of Japan Meteorological Agency(1991),the forma-tion of a tapering cloud is favorable under the condition in which the equivalent potential temperature is higher than 320 K at 850 hPa.This type of cloud tends to form in the warm sector of low pressure,frequently in summer,and causes heavy precipitation over a narrow area.The narrowness of the area makes the amount of precipitation much more intense and localized.There is also a back-building rainband signature with cloud“E”in Fig.4e,which is known to cause narrow-band of heavy precipitation(Schumacher and Johnson,2005).During 0633 to 0700 UTC 21 September,the precipitation rate in Seoul recorded its maximum amount of 71 mm h?1.After this time,the typical shape of the convective cells changed and convective cell“E”moved southeastward out of the Korean peninsula(Fig.4f).

Fig.3.Relationship between the axis of contraction and dilatation,and the isotherms at 925 hPa for(a)1800 UTC 20 and(b)0000 UTC 21 September 2010.The thick dashed(L)and dotted(H)arrows represent the cyclonic and anticyclonic streamlines,respectively.The small individual arrows represent wind vectors(m s?1).

The height of the mature cloud top was calculated by satellite imagery and soundings.The sounding data from Osan,South Korea(37.08°N,127.03°E)at 1200 UTC 21(not shown)were used because it is the nearest radiosonde site from the mature cloud.The height of the cloud top can be estimated by matching the temperatures represented by the enhanced infrared imagery and sounding data.The lowest temperature in the cloud was about?50°C to ?55°C,and thistemperaturerangeexistedatapproximately12km,which corresponds to the cloud top height.

The horizontal dimension,shape and movement of the HPS,as well as the precipitation rate were analyzed using radar CAPPI imagery(Fig.5).Small convective cells were scattered over the Gulf of Gyeonggi at 2000 UTC 20 September,and moved closer together with time(Figs.5a and b).The region of the radar-estimated precipitation rate of 40–50 mm h?1could be found near Baengyeongdo at 0000 UTC 21 September,and its shape became almost linear after 1 h(Fig.5b).The axis of the echoes lay northeast–southwest and moved east-southeastward.At 0400 UTC 21 September,when the precipitation rate increased rapidly,the HPS had a narrow width and began to elongate zonally with slight bowing.The width and length of the HPS was approximately 20 km and 240 km,respectively,which met the shape criteria of the CB,as classif i ed by Lee and Kim(2007).

From 0400 to 1000 UTC 21 September,the band-like shape of the echo became more linear.New cells formed in the upwind part of the band(Figs.5c and d).This was the case of “back-building”,which is one of the development types of linear precipitation systems(Bluestein and Jain,1985).The band itself remained quasi-stationary during this period and later moved in the direction parallel to the line.Afterwards,the direction of HPS movement was again east-southeastward(Fig.5e).This HPS met the movement criteria of the CB,as suggested by Lee and Kim(2007).They reported that cells in CBs move along the band following the mean f l ow,and the band itself moves in the direction parallel to the band.Therefore,this HPS can be referred to as an MCB.Upon leaving Seoul and Gyeonggi-do,the MCB began to move faster and the precipitation intensity over Seoul dropped sharply to 5 mm h?1.The movement of the MCB was somewhat slow from 0300 to 1000 UTC,when a quasistationary front formed at the surface.

3.2.2.Evolution of MCB

To assist in weather forecasting and provide guidance to the forecasters in this region,we identif i ed the periods and development stage of the MCB based on its movement and shape observed from radar and satellite imagery,change in the precipitation rate,and movement of the center of low pressure and quasi-stationary front.A composite of satellite imagery,surface front analysis and radar was used to identify the evolution of the MCB.

Fig.4.Enhanced infrared imagery from MTSAT-2 showing the evolution of the convective system for(a)2000 UTC 20,(b)2200 UTC 20,(c)0033 UTC 21,(d)0300 UTC 21,(e)0500 UTC 21,and(f)1500 UTC 21 September 2010.

The f i rst period was the “cell-forming”period when a convective cell emerged near Shandong,China(37.5°N,122°E),asshowninFig.4a.Thesecondperiodwasthe“frontogenetic period”.This is because the center of low pressure and a quasi-stationary front formed near Baengyeongdo at 0000 UTC 21 September(Fig.4b).A convective cell produced by the merger of two convective cells formed to the south of the quasi-stationary front(Fig.4b).The main precipitation area,with a radar-estimated rainfall of approximately 45–50 mm h?1,was located to the south of the quasistationary front,as shown in Fig.5b.The third period was the“stationary period”from 0400 to 1000 UTC 21 September,identif i able because the quasi-stationary front and radar echo band moved southward very slowly(Figs.5c and d).The location of the radar echo band was consistent with that of the quasi-stationary front,and the precipitation rate near Seoul was approximately 70 mm h?1.The fourth period was the“mature period”at 1200 UTC 21 September,because the size of the convective cell and the extent of the low temperature of the cloud top were at their largest(Fig.5e).At 1100 UTC,the precipitation rate reached its maximum of 96 mm h?1at Icheon,Gyeonggi-do(37.15°N,127.29°E).The precipitation area,with its maximum,was located to the north of the quasistationary front.The last period was the“dissipating period”,which began at 1800 UTC 21 September(Fig.5f).During this period,the convective cell no longer existed over the Ko-rean peninsula.Only a light precipitation area was present along the quasi-stationary front,which was located to the southwest of the low pressure center(36°N,128°E).

Fig.5.Radar CAPPI imagery from the KMA showing the evolution of the convective band for(a)2000 UTC 20,(b)0000 UTC 20,(c)0400 UTC 21,(d)1000 UTC 21,(e)1200 UTC 21,and(f)1800 UTC 21 September 2010.

3.3.Thermodynamic conditions

3.3.1.Moisture conditions

Moisture advection in the lower,mid and upper troposphere was analyzed using equivalent potential temperature with wind,water vapor satellite imagery,and sounding data.Figure 6 shows the distribution of the equivalent potential temperature(θθ)at and near the MCB along with the wind vector at 850 hPa.The solid lines and black arrows represent the isopleths ofθθand wind vectors,respectively.Figure 6a corresponds to the cell-forming period,in which the values ofθθover Shandong were higher than 345 K.This relatively highθθarea extended to Baengyeongdo with a southwesterly f l ow bringing warm moist air to western mid-Korea(Fig.6b).Owing to the shape of the 345 K contour,it is often referred to as a “warm air advection tongue”(Hassan and Meshkatee,2011).After 6 h,this relatively highθθarea covered the mid-and southern part of the Korean peninsula(Fig.6c)and moved southward with a northerly wind allowing a strong temperature gradient to form over 37°N(Fig.6d).This horizontal temperature gradient is one measure of the frontogenetic tendency,and the process can be described quantitatively in terms of the frontogenesis function,where?θis the rate of change of horizontal potential temperature gradient and d/dt is a total derivative(Keyser et al.,1988).Frontogenesis can be understood by expanding Eq.(3),which is described by four terms:(1)the diabatic term;(2)the horizontal deformation term;(3)the vertical shear term;and(4)the divergence term in the direction/presence of the existing temperature gradient.Many active fronts are associated with a deformation f i eld,which lead to an intensif ication of the horizontal temperature gradient(Sawyer,1956).Figure 6 also suggests that moisture was provided by strong southwesterly f l ow of 20 m s?1or more from the Yellow Sea.As mentioned previously,the inf l ow of warm moist southwesterly wind at 850 hPa supports MCB development.

Fig.6.Equivalent potential temperature(K)at 850 hPa for(a)1800 UTC 20,(b)0000 UTC 21,(c)0600 UTC 21,and(d)1200 UTC 21 September 2010.The solid lines and black arrows represent the equivalent potential temperature and wind vectors,respectively.

Figure 7 shows a series of wind,isobars and mixing ratios on an isentropic surface of 310 K.The level of the 310 K isentropic surface lies at approximately 800 hPa.The solid black lines and blue colored region represent the isobars and mixing ratio,respectively.The mixing ratio is blue-shaded at 1 g kg?1intervals and the color bar is light-coded with increasing mixing ratio.The orange contours depict the mixing ratio of more than 10 g kg?1.During the cell-forming period(Fig.7a),ascending motion with a mixing ratio of 10.5 g kg?1existed,which is marked by the arrows crossing the 750-hPa level over Shandong,and extending toward the southwest coast of Korea.In the quasi-stationary period(Fig.7b),air ascended near Baengyeongdo with mixing ratio of10.5gkg?1crossingthe750-hPalevel.Themixingratioat Baengyeongdo was less than 10 g kg?1for the cell-forming period,but increased to more than 11 g kg?1over the middle and southern parts of the Korean peninsula during the quasistationary period(Fig.7c).For the mature period(Fig.7d),mixing ratios of more than 12 g kg?1existed over the mid-

Fig.7.Mixing ratio(g kg?1),wind vector,and isobar(hPa)at the surface of 310 K for(a)1800 UTC 20,(b)0000 UTC 21,(c)0600 UTC 21,and(d)1200 UTC 21 September 2010.The blue coloring represents the distribution of the mixing ratio.The black solid lines and orange solid lines represent the isobars and isohumes(mixing ratio),respectively.

western part of the Korean peninsula,with southwesterly and westerly winds converging and ascending across the 750-hPa level.Ascending air with a mixing ratio of more than 10 g kg?1helped the MCB to develop,and the MCB developed further with increasing mixing ratio.

3.3.2.Atmospheric instability

The sounding data at Qingdao and Baengyeongdo were analyzed to determine the vertical intrusion of dry or moist air and its atmospheric instability.The intrusion of dry air at the upper level is a classical signature of an atmospheric destabilizing mechanism,which can assist in the development of convective cells and heavy precipitation(Browning,1997;James and Clark,2003).Prior to 8 h of the convective cell initiation,SSI and K-index were 2.03 and 23.3,respectively(Fig.8a).The atmospheric condition was neutral to stable at this time.These values become 0.1 and 37.3,respectively,at 0000 UTC 21 September,i.e.,the atmosphere became more unstable with time.In addition,dry air intrusion occurred between 500 and 600 hPa,suggesting that there could have been some upper-level dynamical support in the convection development(Fig.8d).The dew-point depression was 8 K at 550 hPa,whereas it was 20 K immediately above and below due to entrainment of dry air.This created a sharp contrast in moisture between the low levels and mid to upper levels.Overall,the atmosphere became unstable due to warm and moist air advection in the lower troposphere and dry air intrusion in the mid and upper troposphere.The instability of the atmosphere further enhanced the development of precipitation.

3.4.Dynamic conditions and lifting

Fig.8.Sounding data at Qingdao(36.06°N,120.33°E)for(a)1200 UTC 20,(b)0000 UTC 21,and at Baengyeongdo(37.97°N,124.63°E)for(c)1200 UTC 20,(d)0000 UTC 21 September 2010.The wind barb f l ags have units of knots.

Fig.9.(a,b)Zonal and(c,d)meridional winds(m s?1)with relative vorticity(×10?5s?1)and(e,f)vertical winds(Pa s?1)with divergence(?1×10?5s?1)along(a,c,e)122°E at 1800 UTC 20 and(b,d,f)127°E at 0600 UTC 21 September 2010.The bold solid line represents 1 PVU;the heavy dashed line represents the axis of strong westerly wind;the black box represents the area of cyclonic rotation;and shading indicates the degree of relative vorticity and convergence.The black arrow in(a,c,e)indicates the location where the cell formed during the cell-forming period,and in(b,d,f)the center of low pressure in the stationary period.

Thezonal,meridional,andverticalwindswithwindshear were analyzed together to understand the 3-D effect of wind on convective cell development.Two of the f i ve periods that showed clear evolution were selected for the presentation:the cell-forming period and the quasi-stationary period.Figures 9a and b show the vertical cross section of the zonal wind for the cell-forming and quasi-stationary period,respectively.The solid and dotted lines represent westerly(W)and easterly(E)winds,respectively.The black arrow indicates the location where a convective cell formed.The bold solid line represents 1 PVU,which corresponds to the dynamic tropopause.A PV anomaly with TPF formed,making an intrusionchannelofstratosphericairintothemidandlowertroposphere,as shown in Figure 9a.The PV anomaly extended toward the surface of 39°N and induced cyclonic vorticity at the surface.A strong westerly wind greater than 80 m s?1was located at 200 hPa of 48°N and its axis extended toward the surface at 35°N,marked by the heavy-black dashed line.Along this axis,the wind speed was mostly higher than that of other latitudes at the same level.There was easterly wind from the surface to approximately 800 hPa from 36°N to 42°N with a maximum speed of 9 m s?1.The isotachs show a strong gradient at the boundary between the easterly and westerly wind near the surface,indicated by the black box.This zone corresponds to the center of cyclonic rotation according to Byun and Kim(1993).The relative vorticity was dominant along the PV anomaly located at 40°N with its surface maximum of 18×10?5s?1.This cyclonic vorticity assisted in the formation of a convective cell after 2 h.

The axis of the strong westerly in the quasi-stationary period became more distinct than in the cell-forming period(Fig.9b).The black arrow indicates the location of the center of low pressure.This axis extended toward the surface of 30°N and a strong westerly wind faster than 80 m s?1existed at 250 hPa of 45°N.This region descended to a lower level and moved southward compared to that of the cell-forming period.The level of the easterly area was approximately 800 hPa.The gradient of the wind speed was signif i cant at the axis of cyclonic rotation,similar to the cell-forming period.This discontinuity of the wind direction suggests the existence of a strong front from the surface to approximately 800 hPa.The extent of the relative vorticity above the stationary front in the quasi-stationary period was much greater than that of the cell-forming period.The strong cyclonic vorticity of approximately 12×10?5s?1and more than 8×10?5s?1existed close to the surface at 37°N and along the PV anomaly.

A discontinuity also existed between southerly and northerly winds(Fig.9c).The solid and dotted lines represent the southerly and northerly winds,respectively,with units in m s?1.A relatively strong southerly wind of 6 to 8 m s?1blew along the PV anomaly.The boundary between the southerly and northerly coincided with the axis of cyclonic rotation(black box).The gradient of the wind speed was sharp at this boundary,as in the case of zonal wind.The speed of southerly and northerly wind in the quasi-stationary period became greater than that in the cell-forming period(Fig.9d).

The aforementioned evidence strongly supports the notion that the PV anomaly near the surface induced cyclonic vorticity at Shandong(38°N,122°E)for the cell-forming period.The southwesterly ascended along this region of the PV anomaly.Under this region blew a northeasterly wind toward 800 hPa.For the quasi-stationary period,the axis of the westerlies tilted more and the southwesterly wind ascended toward the PV anomaly.Precipitation was limited under the axis of cyclonic rotation.

In the cell-forming period,the value of low-level convergence was approximately 1×10?5to 3×10?5s?1(Fig.9e).The shading represents the degree of divergence(light colored)and convergence(dark colored).At the location where the convective cell formed(marked by the arrow),air converged from the surface to 800 hPa and weak divergence appeared above 800 hPa.The boundary between convergence and divergence was distinct near the isopleth of 1 PVU,particularly in the area of TPF.The region of convergence and divergence maintained a tilted structure similar to that of the relative vorticity and PV anomaly.The lower-level convergence during the quasi-stationary period became stronger than that of the cell-forming period(Fig.9f).From the surface to 850 hPa at 37°N,convergence existed with a maximum of 6×10?5s?1.From 850 to 300 hPa,a divergence region existed and the value was slightly larger than in the cellforming period.Similarly,there was a clear boundary between convergence and divergence near the 1 PVU isopleth.On the other hand,the divergence was weaker than that of the cell-forming period in the region immediately above the 1 PVU isopleth.

The cross section of vertical velocity ofω(=dp/dt)is also shown in Fig.9e for the cell-forming period.The dark and light shading represent the area of updraft and downdraft,respectively.The vertical velocity was up to?3×10?5Pa s?1aloft,where the convective cell formed.The maximum vertical velocity was?7×10?5Pa s?1along the PV anomaly.During the quasi-stationary period,a downdraft existed above the quasi-stationary front due to rainfall(Fig.9f).The updraft was again stronger along the PV anomaly.The maximum was?11×10?5Pa s?1at 850 hPa above 40°N.The gradient of the vertical velocity in this period was larger andmoresignif i cantthanthatinthecell-formingperiod.This result is consistent with the study reported by Byun and Kim(1993),who explained the high wind shear and large temperature gradient at the axis of cyclonic rotation.

In summary,a strong southwesterly wind converged and ascended with strong relative vorticity over the mid Korean peninsula when convective cells developed.Ascending northeasterly and southwesterly from the surface to 800 hPa caused strong temperature gradients and wind shear.Under this boundary between the southwesterly and northeasterly was the area of concentrated precipitation.The lower-level convergence and upper-level divergence for the quasi-stationary period were stronger than those for the cellforming period at 37°N,where the surface front was located.

Figure 10 presents a schematic diagram of the quasistationary period,which is a composition of the results obtained by analyzing the surface weather maps,water vapor advection,3-D winds,and isentropic charts.The yellow surface and tube-like volume represent 1 PVU.The green,blue and red arrows represent the dry intrusion,cool northeasterly wind,and warm moist southwesterly wind,respectively.Continental polar(cP)and maritime tropical(mT)air masses occurred over the northwest and southeast of the Korean peninsula,respectively.This area of deformation contributed to the development of a stationary front along the region,allowing warm and moist air advection in a narrow zone toward the middle of the Korean peninsula.The quasi-stationary front contributed to the heavy precipitation by slowly moving the convective cells and increasing the duration of precipitation over Seoul metropolitan area.Precipitation was limited to this region between warm and moist southwesterly with an equivalent potential temperature＞345 K,mixing ratio＞ 11 g kg?1,relative vorticity ＞ 12×10?5s?1,and a relatively cool and dry northeasterly f l ow.In contrast to the lower troposphere,dry air intruded the mid and upper troposphere from600hPaandabove.Thisdifferenceinhumiditybetween the lower and upper troposphere destabilized the atmosphere.TPF extended to approximately 900 hPa,which induced cyclonic vorticity and maintained the MCB.The lower-level convergence and upper-level divergence also supported the vertical development of the MCB.

Fig.10.Schematic diagram of the stationary period.The yellow surface and tube-like volume represent 1 PVU.The green,red,and blue arrows represent the dry air intrusion,warm moist southwesterly and northeasterly wind,respectively.

4.Summary and conclusions

Various synoptic,thermodynamic,and dynamic features for a case of MCB developmentalong a quasi-stationary front were investigated.The MCB dumped a total precipitation of 259.5mm ina singleday,with 71.0 mm fallingin 1 h at Seoul and Gyeongi-do on 21 September 2010.

Surface observational data revealed a warm and moist southwesterly wind f l ow toward Seoul at 0300 UTC,and a rapid increase in precipitation rate to 75 mm h?1at 0400 UTC.Radar CAPPI imagery showed that the shape,size,and movement of this heavy precipitation system met the criteria of an MCB.As shown in the enhanced infrared imagery from MTSAT-2,a tapering cloud that narrowed in the upwind direction brought heavy precipitation.The quasi-stationary front formed due to the deformation zone in Korea,which showed that cP lay to the northwest of Korea and mT lay to the southeast of Korea in the surface weather maps.At 850 hPa,the isotherms and isohypses were at large angles to each other,showing strong baroclinicity and frontogenesis forcing.The quasi-stationary front moved southward,and was pushed by the relatively cold and dry air of cP along the northwest trough at 500 hPa,which further destabilized the troposphere.

The evolution of the MCB can be divided into f i ve periods:(1)the cell-forming period,when the convective cell formed over Shandong;(2)the frontogenetic period,when the stationary front was over the middle of the Korean peninsula;(3)the quasi-stationary period,when the MCB remained over Seoul for about 3 h;(4)the mature period,when the cloud cover was largest and the precipitation rate was more than 90 mm h?1;and(5)the dissipating period,when the MCB diminished and disappeared.During the formation and development of the MCB,an area of equivalent potential temperature greater than 345 K advected toward the Korean peninsula from the southwest of Korea.At Baengyeongdo,dry air intruded above 600 hPa,and the atmosphere became unstable with a K-index and SSI of 37.3 and 0.1,respectively,during the frontogenetic period.

On the isentropic surface,the area greater than 1 PVU and the area of MCB did not necessarily coincide.Thorough analyses of the zonal,meridional,and vertical winds,relative vorticity and divergence f i elds showed that the MCB was formed by “tilted”updraft that headed toward the PV anomaly.Precipitation was concentrated under this area,where a tilted ascending southwesterly converged with a tilted ascending northeasterly,at the axis of cyclonic rotation.On the isentropic surface of 310 K,moist and warm southwesterly f l ow with a mixing ratio greater than 12 g kg?1ascended and advected toward the Korean peninsula.

In conclusion,the formation of a convective cell was due in part to TPF,which enhanced the cyclonic vorticity at the surface,and by the low-level convergence and upper-level divergence of warm moist air.Southwesterly f l ow ascended in aregionwith highmoisturecontentandstrongrelativevorticity maintained the development of an MCB along the quasistationary front.The effect of the quasi-stationary front was to make the convective cells move slowly and increase the duration of precipitation over a localized area.Thus,the extreme precipitation and f l ash f l ooding was further enhanced.Based on the case study,our research suggests how upperlevel and lower-level dynamics interact to form heavy precipitating MCBs over Korea during fall,outside of the summer monsoon period.

It should be noted that there are some limitations to this study.The initiation of convective cells and its vertical structure were not discussed in detail because high-resolution observational data were not available.In particular,how the scattered convection organized into a line along the quasistationary front needs to be studied further.To generalize the concept presented here and to better understand the characteristics of MCB formation,we hope to present more case studies in the future.

Acknowledgements.This study was funded by the Korea Meteorological Administration Research and Development Program under grant CATER 2012-2072.The authors appreciate the constructive comments of the two anonymous reviewers and the editor.

Bithell,M.,L.J.Gray,and B.D.Cox,1999:A three-dimensional view of the evolution of midlatitude stratospheric intrusions J.Atmos.Sci.,56,673–688.

Bluestein,H.B.,1993:Synoptic-Dynamic Meteorology in Mid-Latitudes.Vol.II,Oxford University Press,594 pp.

Bluestein,H.B.,and M.H.Jain,1985:Formation of mesoscale lines of precipitation:Severe squall lines in Oklahoma during the spring.J.Atmos.Sci.,42,1711–1732.

Bolton,D.,1980:The computation of equivalent potential temperature.Mon.Wea.Rev.,108,1046–1053.

Browning,K.A.,1997:The dry intrusion perspective of extratropical cyclone development.Meteorological Applications,4,317–324.

Byun,H.-R.,and Y.-S.Kim,1993:The role of the con fl uence of the upper and low level jet.Journal of the Korean Meteorological Society,29,263–279.

Davis,C.A.,and K.A.Emanuel,1991:Potential vorticity diagnostics of cyclogenesis.Mon.Wea.Rev.,119,1929–1953.

Doswell,C.A.,H.E.Brooks,and R.A.Maddox,1996:Flash fl ood forecasting:An ingredients-based methodology.Wea.Forecasting,11,560–581.

Ertel,H.,1942:Ein neuer hydrodynamischer Wirbelsatz.Meteor.Z.,59,271–281.

Gold,D.A.,and J.W.Nielsen-Gammon,2008:Potential vorticity diagnosis of the severe convective regime.Part II:The impact of idealized PV anomalies.Mon.Wea.Rev.,136,1582–1592.

Hassan,E.H.,and A.H.Meshkatee,2011:Low-level jet(LLJ)and western Iran heavy rain:case study Dec.1991.Proceedings of International Conference on Environmental Science and Technology(ICEST),V1163–V1167.

Hoskins,B.J.,M.E.McIntyre,and A.W.Robertson,1985:On the use and signif i cance of isentropic potential vorticity maps.Quart.J.Roy Meteor.Soc.,111,877–946.

James,R.P.,and J.H.E.Clark,2003:The diagnosis of vertical motion within dry intrusions.Wea.Forecasting,18,825–835.

Keyser,D.,M.J.Reeder,and R.J.Reed,1988:A generalization of Petterssen’s frontogenesis function and its relation to the forcing of vertical motion.Mon.Wea.Rev.116,762–780.

Ko,H.-Y.,2008:A case study on formation mechanism of mesoscale convective complex along the Changma front.MS thesis,Kyungpook National University,80 pp.

Korea Meteorological Administration(KMA),2010:Monthly weather report 2010(10).Korea Meteorological Administration,116 pp.(in Korean)

Lee,H.-R.,2007:Development mechanism of a wintertime explosive cyclone:A case study.Ph.D.dissertation,Kyungpook National University,52 pp.

Lee,H.-Y.,H.-Y.Ko,K.-E.Kim,and I.-H.Yoon,2010:An analysis of characteristics of heavy rainfall events over Yeongdong region associated with tropopause folding.Journal of the Korean Earth Science Society,31,354–369.

Lee,T.-Y.,and Y.-H.Kim,2007:Heavy precipitation systems over the Korean peninsula and their classif i cation.Journal of the Korean Meteorological Society,43,367–396.

Maddox,R.A.,1983:Large-scale meteorological conditions associated with midlatitude mesoscale convective complexes.Mon.Wea.Rev.,111,1475–1493.

Matsumoto,S.,K.Ninomiya,and S.Yoshizumi,1971:Characteristic features of Baiu front associated with heavy rainfall.J.Meteor.Soc.Japan,49,267–281.

Meteorological Satellite Center of Japan Meteorological Agency,1991:Analysis and Use of Meteorological Satellite Images.Meteorological Satellite Center of Japan Meteorological Agency,45–47.

Moore,J.T.,1993a:Isentropic Analysis and Interpretation:Operational Applications to Synoptic and Mesoscale Forecast Problems.National Weather Service Training Center,6 pp.

Moore,J.T.,1993b:Isentropic Analysis and Interpretation:Operational Applications to Synoptic and Mesoscale Forecast Problems.National Weather Service Training Center,11 pp.

Morgan,M.C.,and J W.Nielsen-Gammon,1998:Using tropopause maps to diagnose midlatitude weather systems.Mon.Wea.Rev.,126,2555–2579.

Namias,J.,1983:The history of polar front and air mass concepts in the United Statesan eyewitness account.Bull.Amer.Met.Soc.,64,734–755.

Namias,J.,and H G.Stone 1940:An Introduction to the Study of Air Mass and Isentropic Analysis.American Meteorological Society,232 pp.

NationalEmergencyManagementAgency(NEMA),2010:Chronology of Disasters.National Emergency Management Agency,31–35.(in Korean)

Park,J.-H.,2006:A study on two heavy snowfall events in the springtime by the meso-αscale snow clouds developed over the East Sea.M.S.thesis,Kyungpook National University,86 pp.

Park,S.-U.,and S.-S.Kim,1983:Low level jets associated with severe storms over Korea.J.Korean Meteor.Soc.,19,20–36.Parker,M.,D,and R.H.Johnson,2000:Organizational modes of midlatitude mesoscale convective systems.Mon.Wea.Rev.,128,3413–3436.

Petty,G.W.,2008:A First Course in Atmospheric Thermodynamics.Sundog Publishing,336 pp.

Rossby,C.G.,and Collaborators,1937:Isentropic Analysis.Bull.Amer.Meteor.Soc.,18,201–209.

Santurette,P.,and C.Georgiev,2005:Weather Analysis and Forecasting:Applying Satellite Water Vapor Imagery and Potential Vorticity Analysis.Academic Press,179 pp.

Sawyer,J.S.,1956:The vertical circulation at meteorological fronts and its relation to frontogenesis.Proc.Roy.Soc.London A,234,346–362.

Schumacher,R.S.,and R.H.Johnson,2005:Organization and environmental properties of extreme-rain-producing mesoscale convective systems.Mon.Wea.Rev.,133,961–976.

Sun,J.H.,and T.-Y.Lee,2002:A numerical study of an intense quasi-stationary convection band over the Korean Peninsula J.Meteor.Soc.Japan,80,1221–1245.

Sun,W.-Y.,K.-H.Min,and J.-D.Chern,2011:Numerical study of 1998 late summer f l ood in East Asia.Asia-Pacif i c Journal of Atmospheric Sciences,47,123–135.

Uccellini,L.W.,and P.J.Kocin,1987:The interaction of jet streak circulations during heavy snow events along the East Coast of the United States.Wea.Forecasting,2,289–308.

Uccellini,L.W.,D.Keyser,K.F.Brill,andC.H.Walsh,1985:The president’s day cyclone of 18–19 February 1979:Inf l uence of upstream trough amplif i cation and associated tropopause folding on rapid cyclogenesis.Mon.Wea.Rev.,113,941–961.

Wilson,L.J.,S.Siok,and B.Marois,1980:Operational application of isentropic analysis to the diagnosis of severe convective weather threat areas.Preprints,Conference on Weather Forecasting and Analysis,Denver,CO.Amer.Meteor.Soc.,166–73.