Journal of the Meteorological Society of Japan, Vol. 82, No. 2, pp. 695--709, 2004 695

Formation and Characteristics of a Summertime Hailstorm over

Northern Taiwan

George Tai-Jen CHEN and Iu-Man TANG

Department of Atmospheric Sciences, National Taiwan University, Taipei, Taiwan

(Manuscript received 26 August 2003, in final form 12 December 2003)

Abstract

Hailstorm is a rather unusual weather phenomenon in Taiwan particularly in the summer season.The hailstorm event occurring over northern Taiwan on 2 July 1998 was investigated using conven-tional data, Doppler radar observations, and satellite cloud winds. Results showed that an upper-levelcold vortex provided favorable conditions for the development and evolution of the storm. The storm ap-peared to be triggered by the low-level convergence associated with local circulations coupled with theupper-level divergence forced by a jet streak of cold vortex. Backing of winds and vertical shears withheight provided by cold vortex appears to be instrumental for the westward propagation and intensi-fication of the convective system. A couplet of mesoscale vortices was observed in the low- and mid-troposphere and was found to be generated by the tilting process on the vertical wind shear through astrong updraft in the convection.

1. Introduction

During the Meiyu season of May and June,the weather over Taiwan area is mainly influ-enced by the Meiyu front, which causes con-tinuous rains mixed with convective showers,thunderstorms, and heavy rain events. As theseason proceeds into summer in July and Au-gust, thunderstorms and/or showers associatedwith the afternoon convection become majorweather phenomena if there is no typhoon in-fluence. Over northern Taiwan, the afternoonconvection in summer is closely related to theformation and development of local circula-tions such as sea breezes and upslope winds.Based on the studies of local circulations overnorthern Taiwan in June–August (e.g., Liu andSu 1997; Chen et al. 2001), sea breezes ap-

peared to converge over northern Taiwan fromthe coastal areas over the west, the north, andthe east. Meanwhile, the upslope winds alsodeveloped over both sides of the Snow Moun-tain Range (Fig. 1).

Although afternoon thunderstorm may causeheavy rain, which often produces flooding lo-cally, it rarely causes severe weather such ashail. Take Taipei City as an example (CentralWeather Bureau 1991), there were 225 and 177thunderstorm days observed in 30 years (1961–1990) in July and August, respectively. In com-parison, however, there were only a total of 36hailstorm events reported officially by the 24surface stations of the Central Weather Bureauall over the Taiwan island in the last 38 years(1961–1999; 1980 missing) and there were only6 out of these 36 events occurring in summer.This is perhaps due to the summer environ-mental condition, which is potentially unstableand only exhibits a relatively weak verticalwind shear unfavorable for severe convectivestorm to develop. Take the mean July soundingin 1989–1993 as an example, it has a CAPE

Corresponding author: George Tai-Jen Chen, De-partment of Atmospheric Sciences, National Tai-wan University, 61, Ln 144, Sec 4 Keelung Rd,01772, Taipei, Taiwan.E-mail: george@george2.as.ntu.edu.tw( 2004, Meteorological Society of Japan

value of about 700 m2s�2 and a relatively smallvertical shear of about 1:0 � 10�3 s�1 in themiddle to upper troposphere between 500 and300 hPa. The hailstorm occurring over TaipeiCity and its immediate adjacent areas in theafternoon hour of 1430–1500 LST 2 July 1998is therefore a rather rare event climatologi-cally. This case was selected in the presentstudy not only because it was a rare event,it also lasted for a long period of time about20 minutes and produced locally unusual largehailstones with a radius reaching about 2 cm.Therefore, the formation process and charac-teristics of this hailstorm deserve further in-vestigation.

Chen et al. (1990) studied the severe con-vective storms developing over Taiwan areain summer and suggested that the westwardpropagating upper-level cold vortices mighthave played an important role in changing theenvironmental conditions and in providing theforcing for triggering convection. They observedthat the upper-level cold vortex not only in-creased the potential instability by loweringupper-level temperature but would also providelifting for convection initiation by the accom-panied upper-level jet streak. Strong convectionin all three cases studied occurred over the areawhere the upper-level divergence would be ex-pected on the left-hand side of exit region of thejet streak in cold vortex. A thermally indirectcirculation over the exit region associated withthe jet streak in cold vortex was derived byChen and Chou (1994) indirectly from the ther-mal structure, the divergence pattern, and thecloud distribution over the vortex region. As-cending motion associated with upper-level di-vergence on the left-hand side of exit regionwas suggested to be instrumental for the for-mation of cloud and convection over the area.These studies all suggested that the westwardpropagating upper-level cold vortex from thewestern North Pacific would play an impor-tant role in the formation of severe convectivestorm over northern Taiwan in summer underweak synoptic forcing in the low- and mid-troposphere. However, specific processes pro-vided by cold vortex could not be obtained inthese studies due to the data sparsity over theoceanic areas. With the help of abundant satel-lite cloud derived winds particularly in theupper levels, investigation on the specific role

of the cold vortex on the formation of strongafternoon convection over northern Taiwan insummer becomes possible for the present hail-storm case.

2. Data and analyses

Figure 1 presents the topography over north-ern Taiwan and the geographical locations forthe Chiang Kai-Shek (CKS) International Air-port, Pan-Chiao rawinsonde station, and otherrelevant locations discussed in this paper. Tai-pei City is located to the north of the SnowMountain Range (SMR), which is oriented ina NE-SW direction over northern Taiwan. Asindicated by the solid boundary, Taipei City islocated in the Taipei Basin. Radar reflectivitydata of vertical maximum indicator (VMI),range height indicator (RHI), and constant al-titude plan position indicator (CAPPI) and ra-dial winds observed by Doppler radar at theCKS International Airport at 0.5-h intervalson 2 July 1998 were used to study the forma-tion and evolution of convective system. TheVMI presents the projection of maximum re-flectivity in a vertical reflectivity profile at eachgrid on the horizontal plane for a complete

Fig. 1. Topography (grey scale in m) overnorthern Taiwan and some locationsrelevant to the hailstorm event. TaipeiCity as delineated by solid boundary islocated to the north of the Snow Moun-tain Range. The Sung-Shan Airportis located near the center of the City.Doppler radar is located at the CKS In-ternational Airport by ‘‘þ’’ sign andrawinsonde station at Pan-Chiao by ‘‘n’’sign.

Journal of the Meteorological Society of Japan696 Vol. 82, No. 2

volume scan. The CKS Doppler radar is locatedat 25.08�N, 121.22�E. The Doppler mode scanwith a 5-cm wavelength covers the area of 120km radius with 1-km resolution in both hori-zontal and vertical spaces. Mesoscale analy-ses were carried out using surface observationsof wind direction, wind speed, and hourlyrainfall amount obtained from the CentralWeather Bureau (CWB) and other governmen-tal agencies in Taiwan to reveal the local circu-lations and the distribution of convective rain-fall. Rawinsonde data at Pan-Chiao (Taipei)station, which is located at 25.0�N, 121.4�E,were used to analyze the environmental con-ditions. These data were also used to com-pute the convective available potential energy(CAPE) and convective Richardson number(Ric) as defined by Weisman and Klemp (1982).The surface and 300 hPa charts analyzed bythe CWB were used to illustrate the synopticsituations.

In order to better reveal the detailed envi-ronmental conditions, Barnes’ objective analy-sis scheme (Barnes 1973) was employed toobtain the reanalysis data. Rawinsonde obser-vations and satellite cloud derived winds overEast Asia (100�–140�E, 10�–40�N) were used.The latter wind data were provided by Uni-versity of Wisconsin-Madison Cooperative In-stitute for Meteorological Satellite Studies(CIMSS) using both infrared and water vaporchannels of Japan Geostationary Meteorologi-cal Satellite (GMS). According to Velden et al.(1997), these cloud-derived winds are of equiv-alent quality to the rawinsonde observationsand have been widely used in the daily opera-tional forecast models. The abundant cloud de-rived winds particularly in the upper levelsfor the present case would greatly enhancethe specific details of the cold vortex in thereanalysis data. Gridded operational analysesfrom the European Centre for Medium RangeWeather Forecasts [ECMWF; Tropical OceanGlobal Atmosphere (TOGA) advanced; EC/TOGA] with a horizontal resolution of 1.125�

long. � 1.125� lat. were taken as first guesses inthe objective analysis.

3. Synoptic situations

Surface analysis on the hailstorm day at0800 LST (0000 UTC) 2 July is presented inFig. 2a. Note that local time (LST) is UTC

plus 8 hours. A frontal system was locatedover the Japan Sea, the Korea Peninsula, theYellow Sea, and the Yangtze River Valley.To the south of this system, the Pacific sub-tropical high pressure center was located near27�N, 135�E with a ridge extending westwardpassing through Taiwan into southern China.Weak pressure gradient and weak south-southeasterlies prevailed over Taiwan and itsvicinity in ridge area. Synoptically, this is afavorable condition for the local circulations todevelop in summer. Figure 2b presents the 300hPa analysis at the same time. A NE-SW ori-ented trough was located over northern Chinato the northwest of the surface frontal systemindicating the structure of a typical midlati-

30

20

110 130 120

30

20

110 130 120

0000 UTC

2 July 1998

(b)

9720

9720

9600

c

9660-35

9720

-35.5

1012

1016

1008

1004

0000 UTC

2 July 1998

1004

(a)

Bashi Channel

Fig. 2. (a) Surface and (b) 300 hPa anal-yses at 0800 LST (0000 UTC) 2 July1998 of the Central Weather Bureauin Taiwan. Isobars and geopotentialheight contours analyzed at 4 hPa and120 gpm intervals.

G.T.-J. CHEN and I.-M. TANG 697April 2004

tude baroclinic wave system. A cold vortex waslocated over the Bashi Channel so that thesoutheasterlies prevailed over Taiwan and theeast adjacent area and the northerlies pre-vailed to the west of the vortex center oversouthern China and the northern South ChinaSea. It originally formed in the western NorthPacific near 18�N, 148�E on 26 June as re-vealed by the cyclonic circulation indicated inGMS-5 water vapor channel cloud images. Itmoved westward then northwestward along theperiphery of the Pacific subtropical high pres-sure.

Vertical time cross section of winds and tem-perature anomalies observed at the Pan-Chiaorawinsonde station in 0800 LST 29 June–0800LST 7 July is illustrated in Fig. 3. The temper-ature anomalies were computed using the timemean over the period of 15 June–15 July 1998.The significant veering of the northeast to thesoutheast winds at 200–300 hPa in 1–2 Julyreflected the approach of cold vortex into theBashi Channel to the south of the station. Windmaximum occurred at 250 hPa and graduallyincreased from 0800 LST 30 June and reacheda peak southeast wind of 33 ms�1 at 0800 LST3 July. The vertical wind shear between 500–300 hPa gradually increased from 3.1 � 10�3

s�1 (29�) at 0800 LST 1 July to a relativelystrong value of 4.3 � 10�3 s�1 (106�) at 0800LST 2 July, then reached a peak value of5.7 � 10�3 s�1 (153�) at 0800 LST 3 July. Thesevalues were much greater than that in the Julymean of 1.0 � 10�3 s�1 as discussed previously.The cold vortex influence was also manifestedby the remarkable negative temperatureanomalies in the middle to upper troposphere(600–200 hPa) from 2000 LST 1 July. This wasalso the time when the lower troposphericwinds weakened and backed from the south-westerlies to the southerlies/southeasterliesunder the influence of the Pacific subtropicalhigh pressure ridge. Meanwhile, the wind veer-ing at each level was evident in 600–200 hPalayer, reflecting the circulation influence of coldvortex. The region of cold temperature anomalycoincided with the region of wind veering sug-gested the vertical extension of cold vortex inthis case. The convective instability was en-hanced by lowering temperature in the uppertroposphere. The effect of cold vortex on theenvironmental instability is clearly demon-

strated in the vertical profiles of mean equiva-lent potential temperature ðyeÞ before, during,and after the influence of cold vortex as illus-trated in Fig. 4. Convective instability wasenhanced during the influence of cold vortex in2–3 July mainly by the decrease of ye in the600–200 hPa level. This is consistent with thevertical extension of upper-level cold vortex asrevealed in the vertical distribution of temper-ature anomaly and wind veering presented inFig. 3.

Figure 5 presents the sounding at Pan-Chiaostation at 0800 LST 2 July. Troposphere wasvery moist with a nearly saturated conditionbelow 800 hPa. The level of free convection(LFC) at 900 hPa was much lower than theclimatological average of 823 hPa, and the un-stable layer (i.e., positive area) was so deepwith a high CAPE value of 2490 m2s�2. ThisCAPE value was greater than that of the ma-ture squall line environment (1330 m2s�2) ob-served in Taiwan Area Mesoscale Experiment(TAMEX) in May–June 1987 (Chen and Chou1993) and was much greater than the climato-logical mean in July of 700 m2s�2 discussedpreviously. The convective Richardson number(Ric) was large (81.8) due to the moderate ver-tical shear and the very high CAPE value. It

Fig. 3. Vertical time cross section ofwinds (conventional symbols) andtemperature anomalies (�C, negativedashed) from the time mean of 15June–15 July at Pan-Chiao station in0800 LST (0000 UTC) 29 June–0800LST (0000 UTC) 7 July 1998. Arrow inthe bottom indicates the time of hail-storm occurrence.

Journal of the Meteorological Society of Japan698 Vol. 82, No. 2

was much larger than the values ðRic ¼ 15–35Þfavorable for the occurrence of supercell thun-derstorm as suggested by Weisman and Klemp(1982). Thus, the environmental conditions

for the formation of the hailstorm in the pres-ent case were quite different from those forthe hailstorm associated with supercell overthe Great Plains in the United States (e.g.,Bunkers 2002). The 0�C level at 4.7 km wasclose to the climatological average of 4.8 km.This suggested that the bigger hailstone size inthe present case was not due to the loweringof freezing level, which is a favorable conditionto keep the hailstone from melting. Winds wereweak in the lower troposphere under the influ-ence of the Pacific subtropical high pressureand were backing from middle to upper tropo-sphere reflecting the influence of cold vortex.

4. Characteristics of convection

Figure 6 presents the spatial distributionof radar reflectivity of VMI in 1330–1530 LST2 July at 0.5-h intervals. Convection occurredmainly over northwestern slope area of theSMR with a maximum intensity of 30–40 dBZat 1330 LST. This was consistent with the sur-face rainfall observation at 1400 LST as will bediscussed in the later section. In the following0.5 h, convective system A occurred over Kung-Kuan and Yong-Ho areas to the southwest ofSung-Shan Airport (cf., Fig. 1) with a maximumintensity of 40 dBZ.

In 1430–1500 LST, system A slowly movedwestward to Yong-Ho/Pan-Chiao area and in-tensified to a value over 50 dBZ (Figs. 6c, d).During the same period, hail was observed inmany locations of Taipei City and Yong-Ho/Pan-Chiao areas where the heavy rainfall wasalso observed. Meanwhile, a new convectivesystem B with an intensity of 40 dBZ alsoformed to the east of the system A. System Bappeared to be generated over the convergentarea within the cyclonic vortex which formed tothe east of system A. The eastward propagationof system B was perhaps associated with thisvortex as well. The generation of this cyclonicvortex will be discussed in Chapter 5. Basedon the locations of surface hail reports, systemB did not cause hail in this case. From 1500to 1530 LST, system A moved westward andweakened, while system B continued to moveeastward and intensified to 50 dBZ.

Radar reflectivity of CAPPI at 3 km and ra-dial winds at 0.5 and 2 km at 1430 LST 2 Julyprior to the occurrence of hail and rainfall arepresented in Fig. 7. Convective system A devel-

6/29 ~ 7/1

7/2 ~ 7/3

7/4 ~ 7/5

qe

hPa

Fig. 4. Profiles of the mean equivalentpotential temperature (ye; K) before (29June–1 July), during (2–3 July), andafter (4–5 July) the influence of coldvortex.

T

Td

Fig. 5. Sounding in skew T-ln P diagramat Pan-Chiao station at 0800 LST (0000UTC) 2 July 1998. Solid arrow indicatesthe adiabatic ascending air parcel.

G.T.-J. CHEN and I.-M. TANG 699April 2004

(d)

A

B

1500(c)

A

1430

(b)

A

1400(a) 1330

(e)

B

1530

Fig. 6. Radar reflectivity (dBZ) of verticalmaximum indicator (VMI) observed atthe CKS International Airport (‘‘þ’’sign) at (a) 1330, (b) 1400, (c) 1430, (d)1500, and (e) 1530 LST 2 July 1998.Sung-Shan Airport is indicated by ‘‘l’’sign. The radius is 25, 75, and 120 kmfor inner, middle, and outer circle, re-spectively.

Journal of the Meteorological Society of Japan700 Vol. 82, No. 2

oped to an intensity of 50 dBZ over the areawhere strong horizontal convergence would beexpected by the large gradient of reversed ra-dial winds at 0.5-km level. A signature of anti-cyclonic vortex was suggested by radial windsat 2-km level to the immediate southwest ofsystem A. Figure 8 presents the similar analy-ses at 1445 LST when rainfall and hail startedto occur. System A moved westward and in-tensified in the last 15 minutes over the low-level convergent area with a maximum inten-sity of 58 dBZ. An echo-free area appeared toextend in an NW-SE direction to the southeastside of system A. It was collocated with thestrong northwestward inflows into the convec-tive system at 0.5 km height. In other words,the NW-SE oriented echo-free area coincidedwith the area of strong northwestward inflowsinto the convective system (Fig. 8b). This sug-gested that the strong convective updraft oc-curring over the area of large reflectivity gradi-ent located over the northwestern boundary ofstrong inflows as indicted by a star sign in thefigure. No hook echo in the vicinity of strongupdraft core was observed in this case. This issimilar to the hailstorm case studied by Nelson(1987) and is different from a classical supercellthunderstorm case studied by Browning (1965).Existence of a couplet of vortices in the im-mediate vicinity of strong updraft core wassuggested by the radial wind pattern. An anti-cyclonic vortex to the immediate southwest ofthe strong updraft core was indicated not onlyat 2 km level as at the previous time but also at5-km level at this time. A weak cyclonic vortexwas also discernible to the northeast of strongupdraft core at both 2-km and 5-km levels. Fig-

(b)

CKS Sung-Shan

(c)

(a)

A

Fig. 7. Radar reflectivity (dBZ) of con-stant altitude plan position indicator(CAPPI) at 3 km (a), and radial wind(ms�1) at 0.5 km (b) and 2 km (c) at1430 LST 2 July 1998. Radial windsanalyzed at 2 ms�1 intervals with neg-ative values (dashed) towards radarand positive values (solid) away fromradar. X (EW) and Y (NS) axes are inkm. The CKS International AirportDoppler radar is located at ð0; 0Þ with a‘‘þ’’ sign and the ‘‘l’’ sign 40 km to theeast is Sung-Shan Airport.

G.T.-J. CHEN and I.-M. TANG 701April 2004

ure 9 presents the similar analyses at 1500LST. Slightly weakening of system A was in-dicated in CAPPI reflectivity at 3 km (Fig. 9a)and also in the vertical extension of high re-flectivity in the range height indicator (RHI,not shown). The anticyclonic vortex intensifiedat 2-km and 5-km levels and extended down-ward to 0.5-km level. As illustrated in the

figure, the radial wind pattern in the immedi-ate vicinity of strong updraft region appearedto shift northeastward with height. This sug-gested that the vortex center, which was lo-cated at the central position between the areasof maximum reversed radial winds, tended totilt slightly towards northeast with height aswell. To the northeast of strong updraft core

(a)

A

CKS Sung-Sang

B

(b)

Inflow

(c) (d)

Fig. 8. Radar reflectivity (dBZ) of CAPPI at 3 km (a) and radial wind (ms�1) at 0.5 km (b), 2 km (c),and 5 km (d) at 1445 LST 2 July 1998. Strong northwestward inflow is indicated by an arrow in (b)and strong updraft region is indicated by a star sign (?) in (c) and (d).

Journal of the Meteorological Society of Japan702 Vol. 82, No. 2

the weak cyclonic vortex again was discernibleat 5 km.

A theoretical study by Klemp (1987) sug-gested that a cyclonic-anticyclonic couplet canbe generated by tilting process on the verticalwind shear with a cyclonic vortex on the right-hand side and an anticyclonic vortex on theleft-hand side of the vertical shear vector. Itwas also found in that study that the pertur-bation pressure is related to the vertical shearwith the high pressure on the upstream andlow pressure on the downstream of the vertical

shear vector. Based on Klemp’s theoreticalstudy, the relationship among vortices, pertur-bation pressure, and vertical shear vectors inthis case as would be expected from the morn-ing sounding at Pan-Chiao rawinsonde stationis presented in Fig. 10. The anticyclonic (cy-clonic) vortex formed to the southwest (north-east) of strong updraft and was located on theleft (right) hand side of the vertical shear vector(facing towards the shear direction) in the layer850–600 hPa and 600–400 hPa. To reveal theformation process of mesovortices, the tilting

(a)

A

B

CKS Sung-Shan

(b)

(c) (d)

Fig. 9. Same as in Fig. 8, except for 1500 LST 2 July 1998.

G.T.-J. CHEN and I.-M. TANG 703April 2004

term in the vorticity equation was estimatedby using the environmental vertical wind shearand Doppler radial winds. The vertical windshear at Pan-Chiao rawinsonde station in 600–400 hPa was 2.27 � 10�3 s�1. The convergenceover the echo-free area was estimated to be1.2 � 10�3 s�1 at 0.5 km and 1 � 10�3 s�1 at5 km at 1445 LST. As the strong convectionindicated by the strong radar reflectivity ex-tended up to 10 km height, the strong updraftof 10 ms�1 would be expected by continuityequation. Therefore, the vorticity generation inthe vorticity equation through tilting processwas estimated to be 9.1 � 10�6 s�2 in this case.Thus, it only needed 2–3 minutes to generate

a mid-level cyclonic vortex of 1.25 � 10�3 s�1 asobserved in Fig. 8d. The vortices in this casewere different from those vortices frequentlyobserved in mesoscale convective systems atthe horizontal scale of 100–200 km, which canbe explained by balanced dynamics (Davis andWeisman 1994). The westward propagationand intensification of convection system A werealso supported by the upward perturbationpressure gradient over the area of downshearside of the updraft region in the middle andupper troposphere. On the other hand, theconvection system B on the upshear side ofthe updraft region would be suppressed bythe downward perturbation pressure gradient.Therefore, the backing of the winds and verticalshears under the influence of an upper-levelcold vortex appears to be instrumental for thewestward propagation and intensification ofthe convective system A in this case.

5. Surface mesoanalyses

Surface streamline and rainfall analyses arepresented in Fig. 11 for the period of 0800–1600 LST 2 July. Offshore flows associated withland breezes and/or downslope winds were ob-served along the northern and northwesterncoastal areas at 0800 LST. This indicated thatlocal circulations had developed well under theweak synoptic forcing with weak pressure gra-dient in the subtropical Pacific high pressureridge (cf., Fig. 2a). Sea breezes and upslopewinds developed in the following hours. By1400 LST, sea breezes prevailed over northernTaiwan and tended to converge in the TaipeiBasin and its immediate vicinities from thenortheast, the northwest, and the southwest.Meanwhile, convective activities as revealed byradar reflectivity (cf., Figs. 6a, b) started to occurover the Taipei Basin and the sloping areas ofthe Snow Mountain Range (SMR). Apparently,these convective activities were triggered bythe convergent flows of sea breezes and the up-slope flows over the mountain slope. Convectiverainfall occurring over the southern portion ofthe SMR was not related to the hailstorm andthus was not discussed in this paper.

Over the convergent area to the southwestof the Taipei Basin, convective rainfall occurredat 1500 LST with a maximum value of 20 mmh�1 as observed at Kung-Kuan station. Out-flows from the convective downdrafts prevailed

400~200 hPa

Shear Direction 60°

H

L

+

-

850~600 hPa

Shear Direction ~ 133° H

L

+

-

600~400 hPa

Shear Direction ~ 102° H

L

-

+

1000~850 hPa

Shear Direction ~ 203°

H

L

-

+

Fig. 10. Perturbation pressure and vor-ticity perturbation generated by verti-cal wind shear and convective updraftcore in each layer at 0800 LST (0000UTC) 2 July 1998. Arrows indicate ver-tical shear directions. H and L indicatethe high and low perturbation pressure,respectively. Circles illustrate vorticityperturbation with plus sign for cyclonicand minus sign for anticyclonic circula-tion. Star sign indicates the location ofthe convective updraft core.

Journal of the Meteorological Society of Japan704 Vol. 82, No. 2

(a)

0800 LST

2 JULY

(b)

1400 LST

2 JULY

5

(c)

1500 LST

2 JULY

10

10

5

5

(d)

1600 LST

2 JULY 10

105 5

Fig. 11. Surface streamline analyses at (a) 0800, (b) 1400, (c) 1500, and (d) 1600 LST 2 July. A fullbarb indicates 1 ms�1, half wind barb 0.5 ms�1 and pennant 5 ms�1. Dashed lines are isohyets inmm h�1 analyzed for 1, 5, 10 mm h�1. Hourly rainfall (mm h�1) is plotted at each station.

G.T.-J. CHEN and I.-M. TANG 705April 2004

over the area to the east of the convectiverainfall center. This mesoscale feature was in-dicated by the wind shift from the previousnorthwesterly winds to the westerly winds.Meanwhile, a cyclonic vortex formed to the eastof the convective rainfall center. It appears tobe generated by the downdraft outflows fromthe west and the sea breezes from northerncoastal area. Convective system B as shown inFig. 6d formed over the convergent area ofthis cyclonic vortex. The eastward propagationof system B appeared to be associated with thisvortex. At 1600 LST, sea breezes weakened andthe outflows prevailed to the west and north-west of the convective rainfall area. As a re-sult, onshore flows over northwestern coastalareas at earlier time periods changed to off-shore flows. Two rainfall centers were analyzedwith the primary one of 18 mm h�1 occurringat Shin-Zhuang and secondary one to the eastof Kung-Kuan. They were consistent with theconvective systems A and B discussed in theprevious section.

6. Upper-level cold vortex

Figure 12 presents the reanalysis wind fieldsof 250 hPa at 0800 and 1400 LST 2 July usingBarnes’ scheme as discussed in Chapter 2. Thereanalysis data are in a better agreement withthe observations as compared to the EC/TOGAgridded data. For example, 22.5 ms�1 wind wasreanalyzed at the grid over northern Taiwan(24.75�N, 121.50�E) at 1400 LST. It is closer to25 ms�1 as reported at the nearby Pan-Chiaostation (25.0�N, 121.4�E) as compared to 15ms�1 of EC/TOGA gridded wind. At 0800 LST,a NNW-SSE oriented jet streak over the north-eastern quadrant of cold vortex was located tothe east of Taiwan. Northern Taiwan was lo-cated on the cyclonic side of the exit regionwhere horizontal divergence prevailed. As thecold vortex moved northwestward in the fol-lowing 6 h, the accompanying jet streak alsomoved towards northern Taiwan. Meanwhile,divergence intensified on the cyclonic side of jetstreak over exit region with the maximum di-vergence center located over northern Taiwan.

Cross-contour ageostrophic winds computedtogether with areas of divergence at 0800 and1400 LST 2 July are presented in Fig. 13. At0800 LST, cross-contour ageostorophic flowsprevailed over the exit region of jet streak from

cyclonic to anticyclonic side. Divergence wasobserved on the cyclonic side of the exit region.Thus, it is clear that the jet streak was accom-panied by a rather strong thermally indirectsecondary circulation over the exit region witha forced upward motion of cold air. As the jetstreak moved northwestward in the following

Fig. 12. Reanalyses of wind fields at250 hPa at (a) 0800 LST (0000 UTC)and (b) 1400 LST (0600 UTC) 2 July1998. Wind speeds > 20 ms�1 (40 kt)are shaded. Grey scales unit in kt(0.5 ms�1). Isolines illustrate diver-gence field at 1 � 10�5 s�1 intervals.Solid lines indicate divergence anddashed lines convergence.

Journal of the Meteorological Society of Japan706 Vol. 82, No. 2

6 h, the area of divergence and accompaniedupward motion also moved over northern Tai-wan. During this period convection developedover the Taipei Basin and its vicinity as re-vealed by radar reflectivity (Figs. 6a, b). Diver-gence over northern Taiwan reached a maxi-mum at 1400 LST, and this was the time priorto the occurrence of convective rainfall andhail over the Taipei Basin and its vicinity. Theexistence of jet streak and the accompaniedsecondary circulation in the cold vortex was

suggested to be important for cloud formationby Chen and Chou (1994).

To further reveal the influence of jet streakon the occurrence of hailstorm over northernTaiwan, vertical P-velocity at the grid overnorthern Taiwan (24.75�N, 121.5�E) was com-puted kinematically using O’Brien’s adjust-ment scheme (O’Brien 1970). Figure 14 illus-trates the vertical profiles of divergence andvertical velocity at 1400 LST July 2. Diver-gence prevailed in the upper troposphere witha maximum value of about 2.8 � 10�5 s�1 at250 hPa, while convergence prevailed in themiddle- and lower-troposphere with a relativelysmall value as compared to the large diver-gence in the upper troposphere. Although up-ward motion existed throughout the whole tro-posphere, it was much stronger in the uppertroposphere and reached a maximum valueof about �12 mb s�1 at 300 hPa. Apparently,the strong horizontal divergence as well as thestrong accompanying upward motion in theupper troposphere was associated with the sec-ondary circulation of the jet streak over exitregion. It is important to note that the develop-ment of local circulations produced large con-

Fig. 13. Cross-contour ageostrophicwinds at 250 hPa at (a) 0800 LST (0000UTC) and (b) 1400 LST (0600 UTC) 2July. Solid lines are isotaches of totalwinds > 40 kt (20 ms�1). Areas with di-vergence greater than 1 � 10�5 s�1 areshaded with the grey scale in the bot-tom.

D

hPa

w

Fig. 14. Vertical profiles of P-velocity(o; solid; Pa s�1 ¼ 10 mb s�1) and diver-gence (D; dashed; 10�5 s�1) at 24.75�N,121.5�E at 1400 LST (0600 UTC) 2July.

G.T.-J. CHEN and I.-M. TANG 707April 2004

vergence value near the surface. The relativelyweak upward motion together with the de-crease of convergence with height in the lowertroposphere was consistent with the surfacemesoanalyses (cf., Fig. 11b), which illustratedthe sea breeze convergence over northern Tai-wan under the weak synoptic forcing of thesubtropical Pacific high pressure ridge. There-fore, the convective storm in this case appearedto be triggered by the low-level convergence as-sociated with sea breeze circulations coupledwith the upper-level divergence forced by a jetstreak of cold vortex.

7. Summary

A rare weather event of summertime hail-storm occurring over northern Taiwan on 2July 1998 was investigated using conventionalobservations, EC/TOGA gridded data, GMScloud derived winds, and Doppler radar ob-servations. Environmental conditions and localcirculations were analyzed to reveal the role ofupper-level cold vortex and the sea breezes inthe development of the storm. Doppler radarobservations at the CKS International Airportwere used to reveal the structure and evolutionof the convective system. Results can be sum-marized as follows.

1) The approach of an upper-level cold vortexincreased vertical shear and enhanced con-vective instability favorable for developing asevere convective storm which produced hailover northern Taiwan.

2) The convective storm appeared to be trig-gered by the low-level convergence forcedby local circulations coupled with the upper-level divergence forced by a jet streak of thecold vortex.

3) Backing of the winds and the vertical windshears provided by the cold vortex appearedto be instrumental for the westward propa-gation and intensification of the convectivestorm.

4) A mesoscale vortex couplet formed in themid-level with the anticyclonic vortex tothe southwest and the cyclonic vortex to thenortheast of the strong updraft core. Thiscouplet of vortices was generated efficientlyby tilting process through strong updraftover the moderate vertical shear environ-ment provided by a cold vortex.

Acknowledgments

Thanks are to Dr. C.T. Terng of the CWB forproviding satellite cloud derived winds, to Mr.H.C. Chou of the CAA for providing Dopplerradar data, and to Mr. A.S. Wang and Mr. C.S.Chang for preparing the manuscript. This workwas supported by the National Science Councilunder the Grant NSC 92-2111-M-002-006.

References

Barnes, S.L., 1973: Mesoscale objective map anal-ysis using weighted time series observations.NOAA. Tech. Memo. ERL NSSL-62, Norman,Oklahoma, 60 pp.

Browning, K.A., 1965: Some inferences about theupdraft within a severe local storm. J. Atmos.Sci., 22, 669–677.

Bunkers, M.J., 2002: Vertical wind shear associatedwith left-moving supercells. Wea. Forecasting,17, 845–855.

Chen, G.T.J. and H.C. Chou, 1993: General charac-teristics of squall lines observed in TAMEX.Mon. Wea. Rev., 121, 726–733.

———, H.C. Chou, T.C. Chang, and C.S. Liu, 2001:Frontal and non-frontal convection over north-ern Taiwan in Meiyu season. Atmos. Sci., 29,37–52. (in Chinese with English abstract)

——— and L.F. Chou, 1994: An investigation of vor-tices in the upper troposphere over the West-ern North Pacific during the warm season.Mon. Wea. Rev., 122, 1436–1448.

———, L.F. Chen, and L.F. Chou, 1990: A study onthe upper-tropospheric cold vortices: Casesaccompanying thunderstorms. Proc. Conf. onWeather. Radar and Flight Safety, Civil Aero-nautics Administration, Taipei, Taiwan, 329–349. (in Chinese with English abstract)

Central Weather Bureau, 1991: Summary report ofTaiwan meteorological data. Vol. 3, 4, 5, 1961–1990.

Davis, C.A. and M.L. Weisman, 1994: Balanceddynamics of mesoscale vortices produced in si-mulated convective systems. J. Atmos. Sci., 51,2005–2030.

Klemp, J.B., 1987: Dynamics of tornadic thunder-storms. Ann. Rev. Fluid Mech., 19, 369–402.

Liu, C.M. and W.C. Su, 1997: Analysis of local cir-culation and high ozone generation. Atmos.Sci., 25, 27–50. (in Chinese with English ab-stract)

Nelson, S.P., 1987: The hybrid multicellular—supercellular storm—an efficient hail pro-ducer. Part II. General characteristics and im-plications for hail growth. J. Atmos, Sci., 44,2060–2073.

Journal of the Meteorological Society of Japan708 Vol. 82, No. 2

O’Brien, J.J., 1970: Alternative solutions to the clas-sical vertical velocity problem. J. Appl. Meteor.,9, 197–203.

Velden, C.S., C.M. Hayden, S.J. Nieman, W.P. Men-zel, Wanzong, and J.S. Goerss, 1997: Upper-tropospheric winds derived from geostationary

satellite water vapor observations. Bull. Amer.Meteor. Soc., 78, 173–195.

Weisman, M.L. and J.B. Klemp, 1982: The depen-dence of numerically simulated convectivestorms on vertical wind shear and buoyancy.Mon. Wea. Rev., 110, 504–520.

G.T.-J. CHEN and I.-M. TANG 709April 2004