Further Studies of a Prefrontal Convective Rainband During TAMEX IOP 13 : Part I : Reflectivity History and Cell Evolution

Radar measurements obtained from both conventional and Doppler radars were used to study the life cycle of a prefrontal convective rainband during TAMEX IOP 13 on 25 June 1987. The conventional radar data were available at 20-min intervals, while Doppler data were available at 7-min intervals. Additionally, surface observations at 30-min intervals taken from nine stations over the west coast and the Taiwan Strait were also used for analysis to determine the approximate positions of the Mei-Yu front and the leading edge of the rainband (gust front). Results show that a prefrontal convective rain band formed in the vicinity of the Mei-Yu front when the system was located far north of the island. As the rainband moved slowly down the central west coast, it began to move away from the front as gust fronts were formed by cells at the leading edge of the system. The convective cells generated low-level cold outflows in the warm sector to the southeast of the front. Part of these cold outflows moved toward the southeast, interacting with the strong moisture-rich southwest monsoon flow to form a gust front. At the same time, the southwestern portion of the front, located 50-60 km west of the coast, continued to lift moist air, generating a new cell along the front. This new cell then traveled to the east at a more rapid speed following the prevailing westerly flow at low levels. It eventually merged with the main (old) cell, thereby prolonging the lifetime of the rainband. The analysis in this study, as a whole, further supports the conceptual model of this rainband reported in the study by Lin et al. (1992) using two volumes of dual-Doppler data obtained from CP-4 and TOGA (Tropical Ocean and Global Atmosphere). (


INTRODUCTION
In the studies by Lin et al. (1992) and Lin et al. (1993), the structure of a subtrop ical prefrontal rainband in the Taiwan Area Mesoscale Experiment (TAMEX) IOP 13 \\'as investigated in detail using dual-Doppler data at 0653 and 0700 LST (local standard time), 25 June 1987.This convecti\1e rainband de\1eloped on the warm side of a Mei-Yu front over northwestern Tai'A1an as the front was approaching the northwest coast.The front retained its baroclinic character before reaching the central and southern portions of Taiwan .The cold pool behind the Mei-Yu front mechanically 1ifted the warm, moist strong southwest monsoon flow in the boundary layer.The lifted parcel became saturated near I km.In the absence of mixing with the environment, this parcel (after about 100 mb of lifting) became positively buoyant and asce.nded freely above 200 mb.Hence, organized convection was initiated and maintained by the fr ontal lifting of a 10\:\1-level jet (LLJ) in the boundary layer, and the main moisture supply came from the high-B e monsoon air at low levels.The rainband formed in the vicinity of the Mei-Yu front in the earlier stage of the rainband's life cycle when the system was located far north of the.island.
As the system moved down the central west coast of Taiwan, the rainband gradually intensified and became more organized.Composed of many cells, each one was accompanied by a moderate.convective updraft (6-8 m/s) and a weak convective downdraft (2-4 m/s).As the updraft air reached 5 km and higher, it reversed its direction and merged with the northwesterly flow in the middle and upper troposphere.As a result, the reflectivity core was elongated toward the southeast with the environmental shear ve.ctor.This elongated re.ftectivity core induced the convective downdraft on the warm side of the front due largely to precipitation loading.C:onsequently, a distinctive convective rainband fori-ned in a broad area ahead of the Mei-Yu front.During the mature stage. of the rainband 's life cycle, the rain band was 5-10 km wide and 100-1 20 km long.It moved slowly along the central west coast toward the south-southeast at 2.5 m/s.The maintenance of this long-lived rainband was caused by the factors of 1) frontal lifting; 2) a gust front arising from the convective downdraft ahead of the front; and 3) discrete developments in advance ot' the line.For more details, see Figure 25 and Section 5e in Lin et <Ll.( 1 992).
The obser\1ational evidence presented in the current study further confirms that the fro ntal lifting did play an important role in the initiation and maintenance of the rainband in the earlier stage of the rain band's life cycle.However, the Mei-Yu front gradually lost its direct influence on tl1e rainband in the area near the coastline as the system moved down the central west c. oast of Taiwan.During that period, the rainband began to move away from the front as gust fronts were formed by cells at the leading edge of the system.The convective cells generated low-level cold outflows in the wa1m sector to the southeast of the front.Part of these cold outflows moved toward the southeast, interacting with the strong high-8 e southwest monsoon flow to form a gust fr ont in the warm sector ahead of the front.As a result, new cells developed along the gust front.At the same time, the southwestern portion of the front, approximately 50-60 km we.st of the coast, continued to lift moisture rich air on the warn1 side of the front, generating new cells along the front.These new cells then traveled to the east at a more rapid speed following the prevailing westerly flow at low levels.They then merged with the old cells, thereby prolonging the lifetime of the rainband.The density•-current 1nechanism resulting from the cold outflows of convective cells in the lower la)1er was likely to be responsible for the more rapid movement of the rainband than the slow movement of the Mei-Yu front.The slow movement of the cold front during the IOP 13 was, in part, attributed to the relatively weak northwesterly wind in the lower layer behind the front as evidenced in the 850 mb flow chart (see Figure 3a in Lin et al. 1992).Additionally, the effect of the Central Mountain Range (CMR) also played a role in slowing down the system movement; for example, see studies by Mannouji and Kurihara (1990), etc.
The purpose of this study in Part I is to provide additional observational evidence to support the conceptual model of a prefrontal convective rainband in IOP 13 presented in the study by Lin et al. (1992) described above.The kinematic structure of the rainband over a period of more than one hour as seen from the TOGA radar is presented in Part II of this study.In Part I, the surface traces obtained from nine stations over the west coast and the Taiwan Strait were employed to identify the positions of the cold front and the leading edge of a prefrontal convective rainband.Additionally, the reflectivity history of the rainband revealed by both the Kaohsiung and TOGA radars were used to demonstrate line motion, precipitation pattern, cell merger and other related features.Such structural features make it possible for researchers to further understand the formation and maintenance of this long-lasting, heavy-rain-producing convective rainband during the TAMEX IOP 13.

DATA AND METHODOLOGY
The data used in this study included the surface observations at 30-min intervals from 0200 to 2400 LST 25 June as well as radar observations obtained from the Kaohsiung (744) conventional radar, and the TOGA (770) Doppler radar.The conventional radar data were available at 20-n1in intervals, while the Doppler data were available at 7-min intervals.Temperature, dewpoint temperature, station pressure, significant weather, wind direction and speed along with the rainfall rate at each station were carefully analyzed to determine the approximate locations of the Mei-Yu front and the leading edge of the rainband.
As mentioned in Lin et al. (1992), the TOGA radar was unable to begin observation until 0653 LST 25 June.At the time of data collection, the radar site had already been surrounded by widespread convection.As a consequence, radar measurements were set at the VAD (velocity-azimuth display) mode, collecting data at 20 elevations.It took approximately 7-8 min to complete each volume scan.Eleven consecutive volume scans at 0653, 0701, 0708, 07 16, 0724, 0732, 0740, 074 7, 0755, 0802 and 0810 LST 25 June were employed in this stud)1• These authors used NCAR's Research Data Support System (RDSS) to process raw Doppler data.Ground clutter was suppressed, and range and velocity foldings were eliminated .Only those data with high signal-to-noise ratio values were retained in the analysis.Since the intense convective regions were near the TOGA radar site at the times of analysis, the combined effects of earth curvature and attenuation on reflectivity were found to be ,,.ery small.

Surface Observations
The enhanced surface network began collecting data at 30-min intervals right after the IOP 13 commenced.With this volume of data available for analysis, several features of the Mei-Yu front and the associated rainband could be detennined at a high confidence level in their accuracy.The time variations of surface observations at every 30 min obtained from nine stations over the northwestern and central west coast and the Taiwan Strait are presented TA O, Vol. 7, No .1, Marcl1 1996 in Figure 13 of Lin ( 1993).In the figure, the passage times of the convective rainband gust front (GF) and cold front (F) are indicated.The positions of GF and F were subjecti\rely determined based on the variations of surface parameters prior to, during, and after the passage of the system (see the following discussion for details).Typically, the passage of the prefrontal squall line was accompanied by signs of the GF with rising pressure, temporary cooling and temporar)' wind shifts.On the other hand, th e passage of the Mei-Yu front was characterized by signs of steady pressure rise, temporary cooling and distinctive wind shifts from southwe.st to northeast.
Anal)'Ses of these.surface observations reveal that the rainband was located very close to the front at the CAA (Civil Aeronautic Administration) radar site around 0300 to 0400 LST.As the frontal system moved slowly toward the south-southeast along the northwest coast of Tai\\i •an, the rainband quickly moved away from the cold front.The leading edge of the rainband (GF) reached Tao-Yuan around 0400 LST, approximately 15 min ahead of the frontal passage.There was a rapid wind shift from southwesterly to northeasterly winds, and convective activity did not commence until after the passage of the cold front.Both temperature (T) and dewpoint temperature (Tcz) decrease.cl by nearly 5 °C within three hours of the frontal passage.Pressure (P) rose steadily after frontal passage and increased nearly 4 mb.De. wpoint depression did not decrease until 6-8 h after the Mei-Yu front passed� thus, it took several hours for the low-B e air from northern China to arrive.
As the system continued to mO\le down the coast, the rainband moved more rapidly away from the cold front.The GF passed Hsin-Chu at about 04 15 LST, but the front did not pass until around 0830 LST, a 4 h difference from Tao-Yuan despite the close proximity of the two stations.The southwest flow was able to feed moisture-rich air along the GF, thus enabling convection to sustain itself for nearly three hours.Upon frontal passage, the \\1ind shifted in a similar fashion to the Tao-Yuan station \Vith there only being a gradual decrease of 3 °C in T and T d from the GF passage until I h after the frontal passage.Ho\\i•ever, it took 2 h after the frontal passage before the de\vpoint depression dropped significantly ( 4-5 ° C).
Wu-Chi is located right along the coast and displays some interesting features.GF passage and the Mei-Yu front passage \Vere nearly 15 h apart with con\1ection lasting only a couple of hours.However, several reports of heavy rain after the convective acti\'ity ceased indicated substantial moisture advection into the system, thus enhancing heavy precipitation.Temperature, dewpoint temperature and pressure all maintained a quasi-steady observation until the t• rontal passage.The Ching-Chung-Kang (CCK) station, where the TOGA radar was located, is slightly farther inland then Wu-Chi.The.GF passed at about 0745 LST with convective activity being reported for nearly six hours.Just as in Wu-Chi, the frontal passage \Vas about 15 h after the GF passage.Persistent strong 'Ai'inds (about 10 mis) obviously trans ported moist air into the convective area, enhancing the probability of ne\v cells developing ahead of the front.At Tai-Chung, convection preceded the GF passage (at 0845 LST), but heav•y rains w•ere consistently• reported for five hours, thus leading to the understanding that it was a slow movement with persistent rains and not the intensity of the convection itself (Lin et.<1,l.1992) that led to the.flash floods.
Looking at two coastal island stations over the strait, Peng-Hu and Tung-Chi-Tao, it is noticed that the GF passage and tl1e Mei-Yu front passage were much closer together than at the central stations (only 4-5 h apart).Most likely the speed of the front was not influenced as much by the CMR in these locations.The lack of convective activity and the overall short duration of the preGipitation should also be noted.Temperature, dewpoint temperature and pressure all remained steady throt1ghout the passage of the system, indicating that the southern portion of the fr ont underwent more air mass modification than the other regions (Trier et al. 1990).
The aforementioned surface traces further reveal that high-8 e air dominated before the frontal passage at most northern stations.The dewpoint depression reduced to nearly zero 1-4 h after the frontal passage, showing cool, saturated air near the surface.On the other hand, the dewpoint depression became larger several hours after the frontal passage, indicative of the dominance of low-8 e air from northern China.The surface traces displayed all show that the passage of the prefrontal rainband was accompanied by signs of a gust front with temp orary cooling and wind shifts.
A synopsis of 16 coastal stations is depicted in Table 1.The stations are displayed such that as you read down the table, you are heading south in a direction pertaining to the location of each station.It is interesting to note the time differential between the gust front and the Mei-Yu front passage between the northern and the central stations.The gust front progressed steadily southward along the west coast, while the cold front moved much more slowly.Table 1 also indicates that southwesterly winds dominated, thus enabling moist air to feed the system.Temperature changes were almost null after the GF passage over most of the central west coast stations.However, there was generally a 2-5 °C drop in temperature after the fr ont passed over the northwest coast.These surface traces clearly show that the frontal system modified its properties from more baroclinic to less baroclinic in behavior as it traveled slowly down the west coast.

Rainfall Distribution
The rainfall distributions (in 30-min intervals) in relation to the GF and F are presented in Figure 14 of Lin (1993).Notice that the frontal position (F) trailed the leading edge of the heavy precipitation by 3 ha t Hsin-Chu by 14 ha t Wu-Chi and by 15 ha t Tai-Chung.In co ntrast, the rainband gust front (GF) passed through each station before the leading edge of the precipitation maximum.The heavy precipitation occurred 1-2 h after the passage of the GF, indicative of the prefrontal convective rainband.Among the five stations shown, CAA (686) and Tao-Yuan (697) received their heaviest amount of rainfall (30-50 mm) within one an d-a-half hours after the Mei-Yu front passed.However, these totals were small compared to those of the other stations.
Hsin-Chu (756) received 10-20 mm of rain every half hour from the time the gust front pas sed until one hour prior to the frontal passage.This totaled nearly 100 mm of rainfall within three and a half hours.Wu-Chi (777) reported approximately 80 mm .. of rainfall between the period of 0730-0900 LST.Moderate to heavy rain continued for several hours adding nearly 65 mm more to that total.The largest amount of rainfall was received by Tai-Chung (749) where convection was followed by five hours of heavy rain (about 170 mm).Between 0900 and 1100 LST alone, 80 mm was reported.Coupled with the drainage from the mountainous regions nearby, it is easy to understand why extensive flash flooding occurred in the central west coast region of Taiwan.
The 24-h accumulated rainfall distribution for 25 June over Taiwan is depicted in Figure 9 of Lin et al. (1992).The contour interval was 40 mm per 24 h.Two areas of up to 200 mm occurred near the TOGA radar with the northern and central west coast of Taiwan receiving the most rain.
TA O, vTol. 7, No.1, l'vfarch 1996 Table 1.Surface observations obtained from several west coast stations and the stations over the Taiwan Strait during the IOP 13, showing, winds, pres sure, temperature and significant weather.The approximate positions of the Mei-Yu front and the leading edge of the rainband (gust front) were subjectively deterrnined from the surface traces.28-27 The 1 O-c1n Kaohsiung radar reftectivit)' distributions during the IOP 13, in relation to the gust front (indicated by the heavy dashed line) and the Mei-Yu front (denoted by the heavy solid line), at the constant altitude (approximately 3 km above sea level) are shown in Figure 8 of Lin ( 1993).The CAPPI (constant altitude plan-position indicator) displays of reflectivity covered the period from 0300 to l 000 LST 25 June.As stated earlier, the gust front and Mei Yu front locations were subjectively determined base.ct on the surface observations discussed previously in Section 3.1.A similar plot without 'the gust front superimposed for the period from 0400 to 1100 LST can also be found in Figure 5 of Lin et al. ( 1992).
In summary, the precipitation signatures were located along and behind both the gust front and Mei-Yu front at 0300 LST.By 0400 LST, the signatures started to elongate in an east-northeast to west-southwest orientation ahead of the front with a cell located near the CAA radar.By 0500 LST, a line of cells (constituting the first semblance of a prefrontal rainband) developed in the are.a between the gust front and Mei-Yu front.At 0600 LST, the gust front moved along the coast of west central Taiwan.Several stronger cells developed within the dual-Doppler coverage area and were again located ahead of the Mei-Yu front.By 0700 LST, a distinctive prefrontal rainband had developed along this coast.At this time, the rainband was located in the southeast comer of the dual-Doppler coverage area.By 0800 LST, the convective cells began to dissipate; however, a new line of cells appeared to be developing along the gust front from 0900 to 1000 LST.These new cells continued to grow over the next several hours, and perhaps a new gust front for1ned to the south of these cells due to their outflows.Unfortunately, these cells were outside the dual-Doppler coverage area.By 1800 LST, almost all conve .cti\'e activity had dissipated as the front pushed onto land, thus cutting off its source of moisture.
Figure la is reproduced from Figure 9 of Lin (1993) and shows the successive hourly positions of the squall-line leading edge (GF) as determined from the surface traces described above .The position of the Mei-Yu front at 2000 LST 24 June is also shown.For comparison, the corresponding positions of the cold front at 3-h intervals, adapted from the study by Jou and Deng (1990), are a1so shown (Figure I b) .Notice that the frontal position trailed the gust front by many' hours as the system moved slowly down the central west coast.--+-0:--� - . . -..: Fig. 1.The successi\1e positions of (a) the squall-line leading edge in 1-h inter \'als, and (b) the surface cold front in 3-h intervals during the IOP 13 as determined from the surface data.Adapted from Lin (1993) and Jou and Deng (1990).

Doppler Radar Signatures
The reflectivity history of the rain band, as the system approached the central west coast of Ta iwan, can be studied from the PPI displays of TOGA at low elevation angles.The 3 ° PPI scans for eleven successive times/occ.asions of analysis from 0653 to 08 10 LST are rl�4.0,\lo].7, 11\/0.l, J_ \1.arcb 1996 presented in Figures 2 and 3.The cold front has been s11perimposed on the first two ''olume scans from the dual-Doppler analysis results of Lin et al. ( 1992) and Lin et al. (1993).Notice a mesoscale ""'ave-like pattern with a typical wavelength of about 15 km.Strong low-level winds across the front can produce small-scale vortices nearly I 3 km along the gust front (Carbone 1982).
Inspection of Figures 2 and 3 reveals that the rainband had a more circular pattern than that of a typical squall line.This is attributed to the fact that the rainband observed in the IOP 13 was dominated by a strong northwesterly flow in the middle and upper layers, causing the convecti\1e cells along the cold front to be e]ongated toward the southeast.As a result, convective downdrafts fanned in a broad area ahead of the front in the \\ ' 'arm sector, due largel)' to precipitation loading (Lin et al., 1992 and Lin et al., 1993).It is seen that the width of the rainband (with reflectivities > 30 dBZ) was about 15-20 km.The rainband was oriented in a northeast-southwest direction.There were many cells embedded within the rainband.The maximum reflectivity was 45 dBZ.
A large area of high re.flectivities (Z > 30 dBZ) \Vas located to the northwest of TOGA.This convecti\1e region was mainl)1 responsible for the heavy precipitation reported at Wu-Chi and CCK in the period from 0700 to 0800 LST.It is seen that the frontal system moved very slowly toward the south-southe.ast,traveling at a speed of only 2 mis over a period of about 80 min.It is shown later that the front became almost stationary after 0_ 7 16 LST as it was approac.hing the coastline.Also evident was a weak stratifom1 precipitation behind the f'ront.We ak echoe.s (Z < 15 dBZ) were observed beyond the 60 km range, but they advanced to within 40 km b)' the final scan.

Cell Merger
The conceptual model of this rainband, based on the dual-Doppler data obtained from the CP-4 and TOGA radars at 0653 and 0700 LST 25 June, is presented in Lin et al. (I 992) (their Figure 25).The rainband developed on the wann side of the front over northwestern Taiwan as the Mei-Yu front was approaching the northwest coast.The front provided lifting \vhich initiated the convection along the cold front .It traveled slowly at 2.5 mis from the north-northwest due to the relatively weak environmental wind (only 4-5 mis from the northwest) in the lower layer.The rainband was about 100 km long and was composed of many cells each of which was accompanied by a moderate convective updraft and a weak convective downdraft.The descending air of this weak convecti\'e downdraft produced a horizontally diverging flow in the boundary layer.Part of this c.ool diverging flow moved southeastward, interacting with the incoming high-8 e environmental air to form a gust front in the 'Ai'arm sector.As a result, new cells de\teloped ahead of the GF.These new cells moved tO\\'ard the east and northeast at a speed faster than the system speed following the prevailing flow at low lev•els.These cells eve.ntually merged with the main (old) cells near the west coast, thereby prolonging the life span of' the rainband.
Figure 4 is the PPI display of reflectiv•ity for the 10V;1est elevation (0. 3 degree).This display presents an excellent example of cell merger illustrating how a new cell, located to the west of the main cell, merged with it during a time span of 46 min.This new c.ell, identified as cell A in the figure, apparently formed before 065 3 LST over the southwestern edge of the rainband.It was observed about 35 km west of TOGA at 0653 LST (Figure 4a) and 1noved toward the east at approximately 7 mis fo llowing the westerl)' Vt1ind in the lower • layer.In contrast, the main cell, located to the north\\1est of TOGA, traveled very slowly at  2-3 mis toward the south-southeast (Figures 4b-4d).Cell A reached the south�'estern edge of the mai11 cell around 0724 LST (Figure 4e) and eventually ine .rged \Vith the main cell after 0732 LST (Figure 4f).A sitnilar feature of� cell 1nerger� but for the earlier ti1ne periods in the IOP 13: is also reported in the recent study b)1 Li el (Li. ( 1995).This e .. xample� together with those surf' ace t' eatures presented earlier, appear to prov•ide ft1rther evidence to support the conceptual model presented 1n f__,in el.ri,l.( 1992.).It strl)ngly suggests that the Je.ading edge of the convec.tive rainband play•s an important role in initiating a new cell 'ts the lo�•-level c. old outflow, associated with the con\1ecti\1e downdraft ahead of the f\.1ei-Yu front, interacts with the 11igh-B e southwest 1nonsoon flow in the war1n sector.This 1nechanis111 continues to prevail until the syste111 reaches the southwest coast of Tai \van.

RHI Display
The RH! (range-height indicator) slices ot' the rainband were not pa11 of the scanning strategy.The RHis were deri \'ed from the entire volume of verticall)' stacked PPI scans� \Vi th the objective analysis scheme developed at Saint Louis Universit)1 (Pasken and Lin, 1991) being used.The azimuth angle chosen for investigation was ne,1rly nc)rrnal to the advancing rainband.Reftectivities and radial velocitie .s were taken t' rom the RHI slices, while the horizontal velocities were determined by geometry using range and height.
Radial velocities for six successive times of analysis along the 310 degree radial in a direction nearly perpendicular to the rainband are shown in Figu1•e 5.The gust fro11t , as reported in Lin et rLl.(1993), was located about 9-10 km northwest of TOGA and \Vas characterized by we.ak converge.nee and slight wind shift.This gust front \vas related to the leading edge of the rainband as it moved slowly to\\1ard the TOGA radar.
An area of -20 mis in the upper levels \vas due to the high-le\1el north\\1esterly flow, which \Vas nearly parallel to the radar beam allo\\1ing for a nearly true indication of the ve.locity.
At 0701 LST (Figure Sb), both the GF and front had ad vc1nced 1-2 km toward the radar.There was also a region of strong mid-le\lel flow bet\vecn 6 and 12. km similar to that at 0653 LST.The f' ront was Iocate.d about 18.5 km from TOGA at 0708 I_.ST (Figure 5c), \\lhile the OF was seen at 8 km, approxi1nc1tely• 11 km southeast ()f the Mei-Yu t•ront.
Tl1e cold front re1nai11ed nearly st<ltionary in the period between 0716 and 073 2 l .
) move slo�•ly tc>ward the radar site.
as indicated.I11 the 1niddle a11d upper layers, the northVv.'esterly\Vind dominated in a bro<1ci area between the GF and the t' ront.
Figure 6 displays reflectivity along tl1e same radial (310 degree) vlith contot1rs every 5 ciBZ.Areas with Z > 40 dBZ c:1re hatched.Notice thclt <:t bro<1d a1•e!1 C)f-35 d BZ or greater occurred in the region ahead of the f' ront and extended t<.) the r•adar• site, indicative of tl1e convective region.Conversel)1, reflectivities were \:Ve£1k in a shallo\v regio11 behind the f1•c)nt corresponding to the stratif0r1n region.These .results shO\\/ t\vo types of precipitation i11 tt1e  regions separated by the fr ont, that is, the convective type in front and the stratiform type behind.Several cells were obser\1ed within the rainband with a maximum reflectivity near 45 dBZ.These reflectivity cores were confined to heights below 5.5 km.As noted in Lin et al. (1992), the warm-rain coalesc.ence processes dominated in the IOP 13 since the freezing level was at 5 .5 km.
As mentioned previously, a careful analysis of the radial \telocity in the lowest layers can pr0\7ide insight as to the mechanisms sustaining the long-liv•ed system.The elongated downdraft in a widespread area ahead of the front produced a gust fr ont characterized by negatively buoyant air which interacted with the high-0 e air transported into the prefrontal region by the LLJ (Lin et al., 1992).This would require a region of weak convergence co located with the GF in the lowest layer.Figure 7 shows the radial velocity at 0.25 km along the 310 degree radial for the same time periods as in Figure 5 \\7ith the arrow denoting the approximate position of the GF as re\'ealed by the single-Doppler analysis.Convergence may hav•e be.en produced by speed convergence and not only by a change in sign within the two dimensional fr ame of reference.This point is discussed in greater detail later.Examination of Figure 7 reveals that movement of the fr ont (F) and the gust front (arrow) was relati\'ely slow as the system gradually approached the coastline, which was in contrast to that observed T�4-0, Vol. 7, No.l, Afarch 1996 at earlier times when the system was located over northwestern Taiwan (Lin, 1993).These authors believe that the effect of the warm waters over the strait in late June continued to modif)' the cold air associated with the frontal system.This, in turn, affected the system movement as it tra\1eled slowly down the central west coast (Lin et al., 1993).
The values of the estimated horizontal convergences (divergences) along the 310 de gree radial at 0.25 km, based on the radial velocity shear term in the radar viewing direction, are presented in Figure 8.The justification for employing this estimated convergence field to calculate vertical velocity in a direction nearly nonnal to the rainband can be found in Part II of this study.Units are in 10-3 /s with positive values (divergences) hatched.For comparison, the values of reflectivity in dBZ along the same radial are superimposed.In general , relatively weak convergence and strong reflectivity (30-40 dBZ) were observed at the GF (see the vertical arrow in Figure 8).This result is consistent \vith the rainband's structure since the GF was formed by cells at the leading edge. of the rainband in the con vective region ahead of the Mei-Yu front (Lin et al., 1993).In contrast, the Mei-Yu front (F) was accompanied by relatively strong horizontal convergence (up to 4x 10-3 /s) at the leading edge of the front (see the vertical dashed line in Figure 8) and weak reflectivity (Z < 20 dBZ) in the stratiform region behind the fr ont.Between the two convergence zones at the F and GF, there was a broad area of divergence (hatched line) in the high reflectivity region with Z > 3 5 dBZ.This divergence zone near the surface was mainly caused by the horizontal spreading of descending air associated with the convecti\1e downdraft in the warm . , .

;
.;  , l., 1993).This mesol1igh, in turn, spread the cooler air horizontally in all directions.Portions of these cold horizontal outflows move.d toward the southeast and northwest along the.310 degree radial, (see the negativ•e and positive radial v•elocities,) respectively, in the area between the F and GF in Figure 5.These diverging outtlo\\'S \\1ere la1�gely responsible for maintaining the GF at the leading edge ot• the rainband and enhancing the low-le\iel convergence at the front, respective1y (Lin et al., 1993).These findings c. lear1y demonstrate that both the front and GF associated with the rainband can be properl)' identified \Vith the single-•Doppler anal)1sis procedures described in this study.This point is t• urther elaborated below.As in Figure 7 except for convergence (solid line) and reflectivity (dashed line).Units for convergence and reflectivity are in 10-3 /s and dBZ, respectively.

Horizontal View of Doppler Measurements
Figure 9 displays the field of estimated horizontal convergence (divergence) at 0.25 km in the sector northwest of TOGA.This field was constructed using the information obtained at every kilometer along each of the nine radials between 290 and 330 degrees with the interval between any two 5 degrees.The sector is in the direction approximately normal to the front.Convergence is denoted by lined and hatched areas.The approximate positions of the front (heavy dashed line) and the GF (dotted line) at each an alysis time deter1nined from both the dual-Doppler analysis at 0653 and 0701 LST (Lin et al., 1992 and Lin et  al., 1993) and single-Doppler analysis at 0708, 07 16, 0724 and 0732 LST are superimposed.This "kink'' is commonly observed in a typical cold front.Reflecti\1it)1 was relative!)'strong at tt1e fro11t due to the frontal lifting of the warm, inoist monsoon air in the warm sector.However, it beca1ne inuch weake.r in a widespread area behind the front due to stratifonn precipitation.It is to be 11oted that the.divergence zone between the front and the GF was due to the downward motion associated with the convective downdraft over that region mentioned previously.These findings, as a whole, are in good agreement with those reported in the dual-Dopple1• study by Lin et (L l. (1992).

Discussion
The structural features of a cold front dt1ring the TAMEX IOP 8 were reported in the study by Trier et al. ( 1989) using single-Doppler radar data� uppe.r air and surf.aceobservations.Results showed that the structure and propagation properties of a frontal system bo1•e rese.mblance to those of• a squall-line gust front in conj unction with the gravity/density current mechanism.The system was cha1�acterized b)1 the 1noderate to high baroclinicity• and abrupt wind shit't associated with the front.Acc.ording tc> Trier el r1, l. (1 989), this t'rontal system originated in a synoptic-scale defor1nation zone at 1nid-latitudes and then propagated toward the southeast at a moderate.spee.d (8-9 1n/s).Ev•en though the amount of con\1ective avai lable potential energy (CAPE) was comparably s1nall, strong frontal 1ifti11g was adequate to sustain a quasi-steady con,rective S)'Stem at the. le.ading edge of the front.
Unlike the case in the IOP 8, the fro11tal system in the.IOP 13 traveled at a much slower speed in comparison \\1ith that estimated from the gravity/density mechani sm.As explained in Lin el r1, l. ( 1992), the presence of the CMR ofte11 divides a Mei� Yu front into two parts.The eastern part over the Pacific Ocean continues to tra\1el at a (normal) rapid speed toward the southeast, while the western part (west and south\\1est ot• Ta iwan) remains quasi-stationary, especially in the area close to the west coast of' Tai\van.Based on the obser\1ations evidenc presented in this study, it was found that the system was tra\1eling only at 2-3 mis when it -vv •as located O\'er the central west coast.The slow system speed was largely attributed to the presence of the CMR and the weak en\1ironmental north\\i'esterly flow in the low layer behind the front.
As noted earlier, the leading edge ot• the rai nband in the IOP 13 traveled at a more rapid speed \\1he11 the syste1n Vv•as locate.d over the northwestern coast.At that time, the system \\1as characterized by 1noderate baroc linicity.As the system continually traveled down the cen tral west c. oast, its barocl inicity '"' • eakened considerably due to the effect of the warm "''aters over the strait (Trier l: i al. 1990).As a result, the movement of the gust t'ront associated with   the leading edge of the rainband slowed down drastically when compared to that over the northwestern coast (Lin, 1993 ).It is of interest to estimate the propagating speed based on the gra\1ity/density current mec.hanism mentioned earlier.
A calculation of the density-current propagation speed similar to that of a gravity current driven squall-line.gust front (Charba, 1974)  (1) where g is gravitational acceleration, 8z is layer depth, Tv (K) is the� virtual temperature and subscripts 1 and 2 refer to the lighter and denser masses, respectively.The value of k is chosen to be 0.84 (Lemaitre et al., 1989), while the layer depth is 1 km (Lin et al., 1992 . ).With the aid of (1) and the surface observed temperatures prior to and after line passage, it was calculated that c = 4.80 b T (mis).Table 2 shows the relationship between the temperature difference 8 T and the propagation speed c obtained from (1).When the temperature difference before and after the gust-front passage was 1 °C or greater, similar to that observed at several northern stations, the propagation speed was approximately 5 mis or larger.On the other hand, V\ 1he.n the temperature difference was less than 0.5 °C, as reported at several west coast stations, the speed of the GF was reduced to only 2-3 mis.These estimated speeds for gust-front propagation appear to agree well with those determined from the surface traces reported earlier in this study.
Table 2.The gravity-density propagation speed, similar to that of a gravity current driven squall-line gust front (Charba 1974), in relation to the temperature difference between the lighter (warmer) and denser (colder) masses.

SUMMARY AND CONCLUSIONS
The surface traces at 30-min intervals, obtained from the nine c. oastal stations, and conventional and Doppler 1�adar measurements were used to further investigate the life cycle and cell e. volution of the convectiv•e rainband during the IOP 13.Careful analyses have revealed several additional observations which support the conceptual model of this rainband, reported in the study by Lin et al. (1992) using two-volume scans of dual-Doppler data obtained from CP-4 and TOGA.
Results show that the rainband formed in the vicinity of the cold front when the system '�as located 0\1er the northwestern coast.As the rainband traveled slowly toward the south, it began to move awa)' from the front as gust fronts were formed by cells at the system's leading edge.The density-current mechanism resulting from the cold outflows of convective cells in the lower layer 'A'as largely responsible for the more rapid movement of the rainband than the slow movement of the.Mei-Yu front.The convective cells generated low-level cold outflows to the southeast of the front.Part of these cold outflows mo\'ed toward the southeast, interacting w•ith the strong moisture-rich southwest monsoon flow to fonn a gust front.At the same time, the southwestern portion of the front, located 50-60 km west of the coast, continued to lift moist air, generating new cells along the fr ont.These new cells then traveled to the east at a speed of 6-7 mis, following the low-level westerly flow over that region.The new cells eventuall)' me14ged with the old (main) cells near the west coast, thereby prolonging the life span of the rainband.The slow movement of the fr ont was mainly caused by the combined effects of the weak low-level northwesterly wind behind the front, the presence of the CMR and the warm waters over the Taiwan Strait in late June.
Negati\1e values indicate the fto\:v to"'•ard TOGA, while positiv•e velocities (hatched line) indicate the flow aw•ay t' rom the radar.Contours are e\'ery 2 mis \\1ith the Mei-Yu front rep1•esented b]1 the heavy dashed line . .Tl1e range is northwest of' the radar abo,1e sea Ie .vel.At 0653 LST (Figure Sa), the front \\,as located about 24 kn1 northwest of TOGA.This \Vas dete.r .mined by the s�1ift from negati\1e to positive velocities, resulting in a maximum low-le\1el convergence at the leading edge of• the front.Note positi 'v'e velocities ahead of the (ront were associated with the southvv•esterly winds in the wann sector, while negative values behind the .front \Vere associated with the w•esterly and 11orthweste1-ly winds at low Ie.vels.
Fig•. 5. Radial velocities along the 310 degree radial for (a) 0653, (b) 0701, ( c) 0708, (d) 0716, (e) 0724, and (f) 0732 LST 25 June.The heavy dashed line and arrow indicate the position of the Mei-Yu fr ont and gust fr ont, respectively.The contour interval is 2 mis with a positive (negati\'e) value away from (toward) the radar.Distances are in kilometers from TOGA.

.
As in Figure.5 except for reflectivity (Z).The contour interval is 5 dBZ.Areas with Z > 40 dBZ are hatched.