2006 MONTHLY WEATHER REVIEW VOLUME 136

Mesoscale Features Associated with Formations in the Western North Pacific

CHENG-SHANG LEE Department of Atmospheric Sciences, National University, , Taiwan

KEVIN K. W. CHEUNG* National Science and Technology Center for Disaster Reduction, Taipei, Taiwan

JENNY S. N. HUI Department of Atmospheric Sciences, National Taiwan University, Taipei, Taiwan

RUSSELL L. ELSBERRY Department of Meteorology, Naval Postgraduate School, Monterey, California

(Manuscript received 24 May 2007, in final form 13 September 2007)

ABSTRACT

The mesoscale features of 124 tropical cyclone formations in the western North Pacific Ocean during 1999–2004 are investigated through large-scale analyses, satellite infrared brightness temperature (TB), and Quick Scatterometer (QuikSCAT) oceanic wind data. Based on low-level wind flow and surge direction, the formation cases are classified into six synoptic patterns: easterly wave (EW), northeasterly flow (NE), coexistence of northeasterly and southwesterly flow (NE–SW), southwesterly flow (SW), monsoon conflu- ence (MC), and monsoon shear (MS). Then the general convection characteristics and mesoscale convective system (MCS) activities associated with these formation cases are studied under this classification scheme. Convection processes in the EW cases are distinguished from the monsoon-related formations in that the convection is less deep and closer to the formation center. Five characteristic temporal evolutions of the deep convection are identified: (i) single convection event, (ii) two convection events, (iii) three convection events, (iv) gradual decrease in TB, and (v) fluctuating TB, or a slight increase in TB before formation. Although no dominant temporal evolution differentiates cases in the six synoptic patterns, evolutions ii and iii seem to be the common routes taken by the monsoon-related formations. The overall percentage of cases with MCS activity at multiple times is 63%, and in 35% of cases more than one MCS coexisted. Most of the MC and MS cases develop multiple MCSs that lead to several episodes of deep convection. These two patterns have the highest percentage of coexisting MCSs such that potential interaction between these systems may play a role in the formation process. The MCSs in the monsoon-related formations are distributed around the center, except in the NE–SW cases in which clustering of MCSs is found about 100–200 km east of the center during the 12 h before formation. On average only one MCS occurs during an EW formation, whereas the mean value is around two for the other monsoon-related patterns. Both the mean lifetime and time of first appearance of MCS in EW are much shorter than those developed in other synoptic patterns, which indicates that the overall formation evolution in the EW case is faster. Moreover, this MCS is most likely to be found within 100 km east of the center 12 h before formation. The implications of these results to internal mechanisms of tropical cyclone formation are discussed in light of other recent mesoscale studies.

* Current affiliation: Climate Risk Concentration of Research Excellence (CORE) and Department of Physical Geography, Mac- quarie University, Sydney, Australia.

Corresponding author address: Kevin K. W. Cheung, Department of Physical Geography, Macquarie University, Sydney, NSW 2109, Australia. E-mail: [email protected]

DOI: 10.1175/2007MWR2267.1

© 2008 American Meteorological Society

MWR2267 JUNE 2008 LEEETAL. 2007

1. Introduction thermodynamic contribution) may depend on near- surface processes where observations are rare, most of The formation of tropical cyclones (TCs) has long the proposed mechanisms are based on numerical simu- been a major area of research, and has resulted in the- lations that must be subjected to validation by obser- ories such as the convective instability of the second vations. kind (CISK; Charney and Eliassen 1964) and wind- One of these theories, the so-called top-down theory, induced surface heat exchange (WISHE; Emanuel is based on the classic MCS structure that usually de- 1986). Whereas these theories focus on the intensifica- velops in an environment with substantial low-level ver- tion process of TCs after the basic kinematic structure tical wind shear and possesses a midlevel mesoscale and, in some cases, the warm-core structure is already convective vortex (MCV) in the stratiform rain region established, the physical processes responsible for the for long-lasting MCSs. This MCV is anticipated to be development from weak or unorganized disturbances the potential focal point for TC formation. Such a struc- to the tropical depression stage are not well under- ture has been observed (e.g., Chen and Houze 1997) stood. and simulated (e.g., Zhang and Fritsch 1986, 1987; The generally accepted picture of TC formation is Chen and Frank 1993), particularly for MCSs over land. multiscale in nature (Holland 1995). That is, synoptic, The characteristic low-level dry layer behind the major subsynoptic-scale, and mesoscale circulations may all deep convective area is also identified in observed and contribute to the formation process. However, the de- simulated MCSs in a TC formation region (Harr and termining system (so-called on–off switch in some lit- Elsberry 1996; Cheung and Elsberry 2006). If the low- erature) is still not clear and is the focus of ongoing level winds are to strengthen to become a tropical de- research. Starting from the large scale, the basic envi- pression, two theories have been proposed: a MCV is ronmental conditions favorable for formation have either advected downward by continuous rain (Bister been known for years (Gray 1968, 1998). These condi- and Emanuel 1997); or a merging of two or more MCVs tions include high sea surface temperature (SST), con- (Ritchie and Holland 1997) occurs. Whereas these pro- ditional instability and high relative humidity in the cesses have been simulated in numerical models, veri- middle troposphere, cyclonic absolute vorticity in the fication with observations is lacking. lower troposphere, anticyclonic relative vorticity in the The bottom-up theory is somewhat based on the ob- upper troposphere, and low vertical wind shear (e.g., servations of Zehr (1992) that low-level vortex intensi- Cheung 2004). The monsoon trough region in the west- fication sometimes follows bursts of intense deep con- ern North Pacific Ocean (WNP) during the summer vection. Montgomery et al. (2006) suggest this deep- usually satisfies most of the above-mentioned condi- convective, low-level vortex enhancement is taking tions and is the source of many TCs in the basin. place within MCSs well before the system-scale vortex Very often TC formations originate from distur- becomes self-sustainable, and is then further intensified bances with enhanced convection and low-level relative through mechanisms such as CISK and WISHE. Mont- vorticity, which may originate from low-frequency os- gomery et al. suggest that vortical hot towers (VHTs) cillations in the tropics such as the Madden–Julian os- on scales of 10–20 km play an important role during the cillation. These wave activities in the tropics modulate process (see also Hendricks et al. 2004; Tory et al. the global TC activity on different time scales and their 2006a,b). relationships with TC frequency, location, and intensity Two questions then arise as to the mechanisms of are the subjects of various studies (e.g., Dickinson and low-level vortex enhancement in MCSs. Is the structure Molinari 2002; Frank and Roundy 2006). of tropical MCSs, especially those that contribute to TC Mesoscale convective systems (MCSs) are often formation, different from the usual characteristics of a found to develop in association with TC formations in tilted deep convection, an extensive stratiform rain re- the monsoon trough or in tropical waves. Contributions gion, and low-level cold pool? To what extent can cur- of MCSs to the formation process have been debated in rently available remote sensing data reveal the contri- the past decade. Due to extensive coverage of all ocean bution of MCSs in TC formation? Zehr (1992) provided basins by satellite imagery, the presence of MCSs dur- one of the first detailed reports utilizing satellite data to ing TC formation has been confirmed in various studies monitor and analyze early-stage TC development, and (Simpson et al. 1997; Ritchie and Holland 1997; Cheung he proposed a two-stage formation process. The first and Elsberry 2004) and monitoring of the life cycle of stage of enhanced convection occurs one or more days these systems is feasible. Because the exact mecha- before the tropical depression forms, and produces a nism(s) by which MCSs contribute to the formation distinct midlevel circulation center. Stage 2 begins process (e.g., whether predominantly a dynamical or a when another burst of deep convection occurs with a 2008 MONTHLY WEATHER REVIEW VOLUME 136 curved pattern of organized clouds closer to the system (ECMWF) are used to analyze the synoptic patterns center than those in stage 1, which indicates that an associated with TC formations. increase in deep-column vorticity may be taking place. Tropical depression intensity or even a tropical storm is b. Definition of formation usually attained within stage 2. It is generally agreed that TC formation is a continu- The primary objective of this paper is to extend the ous atmospheric process, and the exact time of transi- work of Lee (1986), Zehr (1992), and Ritchie and Hol- tion from unorganized cloud clusters to a self- land (1999) by using the latest remote sensing data sustainable system with a certain intensity is not well available. In addition, connections between large-scale defined. In this study, the formation time is taken to be patterns of TC formation and mesoscale convection the first time in the poststorm analysis (best track) of systems are established. Section 2 describes the data each TC that the intensity reached 25 kt, which is usu- used in the study, the period of formation to be exam- ally after the release by the Joint Warning ined, and the definition of formation. In section 3, the Center (JTWC) of a tropical cyclone formation alert formation cases are categorized according to their re- (TCFA) on the developing disturbance. This time is spective synoptic patterns. Section 4 is an examination preferred relative to the TCFA because the TCFA is of the spatial and temporal evolution of the mesoscale quite subjective and sometimes multiple TCFAs are is- convective features that are associated with TC forma- sued. The preformation (as defined above) positions of tions in different synoptic patterns. Identification of tropical disturbances in the JTWC best tracks are used MCSs and diagnosis of their contribution to formation to produce system-centered composites of environmen- processes are performed in section 5. A discussion of tal data and the positions of MCSs with respect to the the results is given in section 6. system center. For a few cases in which these prefor- mation positions are not available, the QuikSCAT winds are used to identify the center of the low-level 2. Data sources and definitions circulation. a. Data sources The study period of this study is from September 1999 to December 2004. According to the JTWC, a Hourly imagery from the Geostationary Meteorologi- total of 190 TC formations occurred in the entire WNP cal Satellite-5 (GMS-5) and Geostationary Operational during the period. Due to the eastern boundary of the Environmental Satellite-9 (GOES-9) infrared channel-1 satellite imagery, cases with formation positions east of (IR1) with wavelength 10.3–11.3 ␮m and 5-km resolu- 160°E are not included in the analysis. In addition, tion is used to monitor deep convective clouds and cases with formation north of 26°N may be affected by MCSs during TC formations in the WNP. This imagery midlatitude systems and are ignored. With these limi- from a data archive in Kochi University of covers tations, the total number of formation cases in this the area 20°S–70°N, 70°–160°E. The cloud-top height is study is 124. inferred from the infrared brightness temperature The time interval for diagnosis is 48 h before forma- (TB). Areas with TB lower than thresholds of Ϫ32°, tion time of each case, which is consistent with the av- Ϫ60°, and Ϫ75°C are considered to be convection in erage time of an early convective maximum that is 50.2 cloud clusters, deep convection in MCSs, and extreme h before the designation of a tropical depression as convection in vortical hot towers, respectively. identified in Zehr (1992). Based on early stage Dvorak For diagnosing near-surface features associated with analysis technique, Bessho et al. (2006) also found that TC formations, the Quick Scatterometer (QuikSCAT) the average time period from the detection of orga- oceanic winds derived from microwave scatterometer nized cloud clusters to designation as a tropical storm is data are used. The 25-km resolution swath data from 50.5 h for the 2004 WNP formation cases. Therefore, the Remote Sensing System (RSS) and 50-km resolu- the convective bursts and MCSs that occurred earlier tion global data from the Jet Propulsion Laboratory than 2 days before formation are considered to have (JPL) are available twice daily. Wind speeds and direc- less effect on the formation process and are not exam- tions from QuikSCAT generally agree well with ocean ined in this study. buoy data (Ebuchi et al. 2002; Pickett et al. 2003), and root-mean-square differences of wind speed and direc- Ϫ1 tion for the swath data from RSS are 1.01 m s and 3. Synoptic patterns associated with formation 23°, respectively. In addition, gridded analyses with 1.125° latitude–longitude resolution from the Euro- One of the major objectives in this study is to relate pean Centre for Medium-Range Weather Forecasts the convection patterns and occurrence of MCSs before JUNE 2008 LEEETAL. 2009

TABLE 1. Criteria used for classifying the six synoptic flow patterns associated with the 124 TC formations, which are applied to identify strong (Ͼ5msϪ1) and weak (Ͻ2msϪ1) average zonal (U) and meridional (V) wind components and their directions (ϩ/Ϫ) at low levels (850 and 925 hPa) in the four 5° ϫ 5° latitude–longitude quadrants with respect to the system center.

NE quadrant NW quadrant SW quadrant SE quadrant Synoptic pattern U V U V UVUV EW (Ϫ) strong (Ϫ) weak (Ϫ) strong (Ϫ) weak NE (Ϫ) strong (Ϫ) strong (Ϫ) strong NE–SW (Ϫ) strong (ϩ) strong SW (Ϫ) strong (Ϫ) weak (ϩ) strong (ϩ) strong MC (ϩ) strong (ϩ) strong MS (Ϫ) strong (ϩ) strong

formation to the associated synoptic pattern for diag- winter monsoon. The other monsoon-related patterns nosing possible mechanisms of formation. For this pur- (SW, MC, and MS) have distributions that peak in ei- pose, the synoptic patterns of the 124 formation cases ther July or August, which is similar to the overall sea- are classified based on QuikSCAT oceanic winds and sonal distribution for the entire basin. Cases in these 925- and 850-hPa ECMWF analyses. To identify strong three patterns account for over 60% of all cases. winds in the vicinity of the formation, a set of criteria is The geographic distributions of the formation posi- established (Table 1) on the magnitude of the zonal and tions of these 124 cases (Fig. 1) are also consistent with meridional wind components in the four 5° ϫ 5° lati- their relationship with the monsoon environment in the tude–longitude quadrants. For example, strong easter- WNP. No EW case (Fig. 1a) is found in the South lies to the north are found prior to formation associated Sea (SCS) because easterly waves seldom reach a re- with easterly waves or a northeasterly flow, and the gion as far west as the SCS. Instead, formation cases in criteria in Table 1 distinguish these two patterns by the SCS are often associated with the SW (Fig. 1d) and imposing conditions on the meridional wind compo- NE–SW (Fig. 1c) patterns since strong cross-equatorial nent. Another example is the wind shear environment southwesterly flows frequently occur in that region that is common in the summer monsoon trough with (Love 1985). Some of the SW and NE–SW cases are strong westerlies to the south and easterlies to the located at latitudes as high as 20°N when strong south- north. The formation is classified as a southwesterly westerlies persist in the entire monsoon trough region flow case if the meridional wind to the south of the such that the criteria for these two categories are satis- center is stronger than in a simple monsoon shear situ- fied. The formation positions in the MS pattern (Fig. 1f) ation. Although these proposed criteria may not be ex- are distributed quite evenly along the monsoon trough. clusive, these classifications will hopefully provide a Note that the meridional spread of the formation posi- better understanding of mesoscale activities relative to tions of these MS cases is quite large, which is related to identifiable synoptic patterns. the monthly variation of the monsoon trough location As a result of this classification scheme, six charac- (McBride 1995). On the other hand, a few low-latitude teristic low-level flow patterns associated with TC for- mations are identified: easterly wave (EW; 10 cases), northeasterly flow (NE; 19 cases), coexistence of north- TABLE 2. Monthly distribution of the number of formation cases easterly and southwesterly flow (NE–SW; 19 cases), associated with the six synoptic flow patterns. southwesterly flow (SW; 23 cases), monsoon confluence EW NE NE–SW SW MC MS (MC; 23 cases), and monsoon shear (MS; 30 cases). Jan 1 Based on the seasons with TC formations in the WNP, Feb 2 these flow patterns are mostly monsoon related except Mar 2 for the EW. For example, the monthly distribution of Apr 2 the six flow patterns suggests that the NE cases may be May 1 5 1 3 4 related to the surges during Asian winter monsoon as Jun 1 3 3 5 Jul 3 5 5 5 they all occurred from October to March (Table 2). Aug 1 1 9 5 7 Five of the NE–SW cases occurred in May during the Sep 1 3 4 3 4 early stage of the summer monsoon and the other nine Oct 2 4 6 1 2 2 occurred in October and November, which suggests Nov 2 5 3 1 1 that formations in this category are also affected by the Dec 5 1 2010 MONTHLY WEATHER REVIEW VOLUME 136

FIG. 1. Positions of formation (dots) and best tracks associated with the six synoptic flow patterns: (a) EW, (b) NE, (c) NE–SW, (d) SW, (e) MC, and (f) MS. formation cases [e.g., tropical depression 32W (1999), formations are classified as the NE synoptic pattern tropical depression 29W (2001), and Typhoon Vamei (Fig. 1b). (2001)] between 100° and 120°E were associated with Some low-level flow patterns in this study resemble northeasterly cold surges during the boreal winter. The the large-scale patterns identified in Ritchie and Hol- formation positions of these cases may also be associ- land (1999). For example, the composite 850-hPa flows ated with the climatological monsoon trough position. for the EW, MC, and MS (Figs. 2a,e,f) patterns are However, criteria in Table 1 require strong wind similar to those in Ritchie and Holland. However, the magnitudes in the northern quadrants. Therefore, these classification scheme used here emphasizes more on the JUNE 2008 LEEETAL. 2011

Ϫ FIG. 2. Composite 850-hPa streamlines and wind speed (m s 1, shaded) based on ECMWF analyses centered at the formation position 48 h before formation for the (a) EW, (b) NE, (c) NE–SW, (d) SW, (e) MC, and (f) MS synoptic patterns. synoptic flow within about 1000–1200 km of the forma- (Fig. 2d), which is defined solely according to the wind tion location. Therefore, some formation cases that were field. For the same reason, the new NE and NE–SW in a large-scale monsoon gyre as classified by Ritchie patterns (Figs. 2b,c) highlight cases that also form in the and Holland may be classified here as a SW pattern monsoon trough, but mostly during the cold season. 2012 MONTHLY WEATHER REVIEW VOLUME 136

Ϫ FIG. 3. Time series of average QuikSCAT wind speed (m s 1) in the four quadrants within 5° latitude–longitude radius for the (a) EW, (b) NE, (c) NE–SW, (d) SW, (e) MC, and (f) MS synoptic flow pattern. Four QuikSCAT passes are defined: pass 1 is the earliest time (about 36 h before formation) and pass 4 is the closest time to the formation time.

Based on the available QuikSCAT winds, all of the in the NE pattern (Fig. 3b), and between the southeast formation cases are accompanied by a gradual increase and northwest quadrants in the SW pattern (Fig. 3d). in near-surface wind (Fig. 3). Due to the classification The maximum winds increase to about 9–11 m sϪ1 36 h scheme, the wind magnitudes in different quadrants de- before formation in these two categories, which is al- viate substantially. For example, the average wind dif- ready close to the intensity of a tropical depression. ferences between the opposite quadrants are 4–5msϪ1 Similar magnitudes of quadrant-average maximum JUNE 2008 LEEETAL. 2013 winds are found in the NE–SW, MC, and MS patterns, TABLE 3. Average IR temperatures (°C) in the 24-h period be- but without the large asymmetry between quadrants. fore formation computed within circular domains with radius from 1° to 5° latitude.

4. General convection characteristics Radius EW NE NE–SW SW MC MS 0°–1° Ϫ15.5 Ϫ24.7 Ϫ29.9 Ϫ21.0 Ϫ27.5 Ϫ26.3 a. Mean cloud-top temperature 0°–2° Ϫ12.3 Ϫ23.6 Ϫ27.0 Ϫ19.5 Ϫ26.5 Ϫ23.5 0°–3° Ϫ8.8 Ϫ21.9 Ϫ23.8 Ϫ17.8 Ϫ24.4 Ϫ20.5 The development from tropical disturbance to TC 0°–4° Ϫ5.6 Ϫ20.0 Ϫ21.0 Ϫ15.7 Ϫ22.1 Ϫ18.0 may occur over several hours or more than 2 days, 0°–5° Ϫ3.0 Ϫ18.0 Ϫ18.2 Ϫ13.5 Ϫ19.8 Ϫ15.6 during which the convective behavior is modulated by the diurnal cycle. Previous studies have shown that the diurnal variation of tropical convection over the ocean the convection closer to the center, only those TBs typically has a convective maximum between 0600 and within 3° latitude–radius are examined. (The convec- 0800 LT and a convective minimum between 1800 and tion profiles at the outer radii are also examined but 2000 LT (Gray and Jacobson 1977; Miller and Frank they usually fluctuate with low correlation with the in- 1993; Chen and Houze 1997). To evaluate the effect of ner-core processes. Therefore, TBs within only 3° lati- the diurnal cycle on the formation cases, a composite tude–radius are discussed.) For example, some cases time series of TB ϽϪ32°C averaged in a circular area have the classic two-stage process of formation (Zehr centered on the system with a radius of 600 km is com- 1992) in which the first deep convection event is then puted according to local times. The difference in TB followed by a second event when formation time is ap- between the convective maximum at around 0500 LT proached (Fig. 4a). However, other cases have three and the minimum at 1800 LT is about 5.3°C (not convective periods (Fig. 4b), or a gradual development shown). The composite percentage area of deep con- of convection with a single TB minimum (Fig. 4c). In vection (TB ϽϪ60°C) also has a similar daily variation some cases, the TB near formation is actually higher (not shown), with a maximum of about 18% at 0400 LT than that in the early stage (Fig. 4d). Thus, a high vari- and minimum of about 8% at 1800 LT. ability in the convective events before formation oc- The average TBs for the EW cases have the highest curs. temperatures (least convection) among the synoptic Detailed comparison of the temperature time series patterns (Table 3, column labeled EW). The average of the 124 formation cases suggests five characteristic TB 24 h before formation within a radius of 1° latitude convective evolutions: (A) single deep convection, (B) is Ϫ15.5°C and this value increases rapidly within a two deep convection events, (C) three deep convection larger circular domain, which indicates that convection events, (D) gradual decrease in TB, and (E) fluctuating is mainly concentrated near the center in the EW cases. TB or a slight increase in TB before formation (Fig. 5). The average TBs of the other synoptic patterns are all Except for a few formation cases in which analysis of much lower than for the EW pattern and thus more convection is incomplete due to missing satellite data, deep convection is found in these monsoon-related situ- the case distributions for the five convective evolutions ations (Table 3, columns labeled NE, NE–SW, SW, are summarized in Table 4. Although at the beginning MC, and MS). These average TBs are nearly Ϫ30°C the easterly wave cases have higher TBs (less deep con- near the center and increase to just above Ϫ20°C within vection), a gradual decrease in TB (profile D) occurs a larger domain. Note that the temperatures for the SW before formation. Two or three periods of deep con- cases are all higher than the corresponding tempera- vection (profiles B and C) are more common for the tures in other monsoon-related patterns, which is prob- monsoon-related formations. No clear difference in the ably due to weaker forcing from the southwesterlies. convective evolution exists between the monsoon- However, the number of cases in each pattern is quite related synoptic patterns with different dominant wind limited and therefore further statistical significance directions (e.g., NE versus SW). In an appreciable num- tests are not performed for these differences in TB. ber of cases, the preformation period is not character- ized by a strong signature of deep convection, and this b. Temporal evolution may make prediction of the formation time using only The temporal variation of convection is examined satellite imagery difficult. through the 48-h time series of area-averaged TB be- The average times of the deep convection events in fore a few selected TC formation cases (Fig. 4). Since profiles A, B, and C are computed using the cases as- convection beyond 4° or 5° latitude from the center sociated with each of the profiles (Fig. 5, top panels). possibly has less effect on the formation process than On average, the deep convection in profile A occurs at 2014 MONTHLY WEATHER REVIEW VOLUME 136

FIG. 4. Time series of average TB (°C) within 1° (dash), 2° (solid), and 3° (dot) latitude radius for (a) TD 10W during 2000 in a SW pattern, (b) during 2000 in a MC pattern, (c) TD 29W during 2000 in a EW pattern, and (d) TS 27W during 2003 inanNE pattern.

9 h before formation with a TB of Ϫ48.6°C. The two c. Spatial patterns of strong convection convection events in profile B are at 34 and 6 h before formation, and the TBs are Ϫ39.5° and Ϫ47.9°C, re- Since the six synoptic patterns have been classified spectively. The corresponding figures for the three con- according to the low-level wind distribution, a strong vection events in profile C are 37, 20, and 3 h, and association with the dominant wind direction is found Ϫ33.7°, Ϫ42.5°, and Ϫ47.8°C. These average values in- when the spatial distribution of convection is examined. dicate that for formation cases with distinguishable con- Two TB thresholds are used to evaluate convection ar- vection events, the deepness of convection is often in- eal coverage within 2° latitude radius of the center for creasing when approaching the formation time. The last each flow pattern (Table 5). A Ϫ60°C TB threshold is deep convection may be quite crucial in intensifying the chosen to isolate pixels in satellite imagery that contain low-level circulation because formation occurs after deep convective clouds, and a threshold of Ϫ32°Cis only a few hours. used to include moderate cumulus convection. Whereas JUNE 2008 LEEETAL. 2015

FIG. 5. Schematics of the five characteristic convection (brightness temperature) temporal evolutions identified for the formation cases in this study: single deep convection (profile A), two deep convection events (profile B), three deep convection events (profile C), gradual deepening in TB (profile D), and fluctuating TB or a slight increase in TB before formation (profiles E1 and E2). The average times (hours before formation) and TBs (°C) of the deep convection events in profiles A, B, and C are shown. convection with TB ϽϪ32°C and with TB ϽϪ60°C emphasize the likely significant effects of deep convec- are concentrated at the southern side of the center in tion on the formation, a MCS is only counted if it per- the NE–SW, SW, and MC patterns, convection devel- sists more than 3 h. ops mainly at the northern side in the EW and NE More MCSs are found in the monsoon-related pat- patterns, although the differences are small in the EW terns with an average number of about two, whereas in pattern. Since most convection in the EW pattern is the EW pattern on average only one MCS is identified close to the center, the area percentage of deep con- in each formation (Table 6, second column). Recall that vection is on average lower than for the other synoptic two or three deep convection events were found with patterns. In the MS pattern, convection is concentrated most of the monsoon-related scenarios (Table 4), and on the western side of the center. These monsoon- these deep convection regions served as candidates for related patterns have areas of moderate convection MCS development. On average, the MCS in the EW ranging from about 32% to 55%, and deep convection pattern appears at a shorter time (ϳ13 h) to formation ranging from about 10% to 25%. This overall correla- time, and the MCS mean lifetime (5.7 h) in the EW tion between low-level synoptic patterns and convec- pattern is much shorter than those in the monsoon- tion distribution does not change when the domain is related patterns with their mean lifetimes of about 14– extended to as large as the 5° latitude radius. 16 h (Table 6, third and fourth columns). In addition, the first MCS in the monsoon-related patterns develops 5. Contributions from mesoscale convective

systems TABLE 4. Number of synoptic pattern cases classified in the five IR temperature temporal profiles as defined in Fig. 5. The total a. Statistics of MCSs associated with formation number of cases in each pattern is less than in Table 1 because of Following Ritchie and Holland (1999), a MCS is some missing satellite data. identified in the satellite imagery by deep convection Profile type EW NE NE–SW SW MC MS with TB Ͻ 214K(Ϫ56.15°C), an area larger than 4 ϫ 4 2 A132434 10 km , and eccentricity larger than 0.5. Without add- B1368106 ing a duration threshold in this definition, one to sev- C2665613 eral MCSs are identified in most of the 124 formations, D410311 except for two cases in the NE and NE–SW patterns. E045201 However, some of these MCSs only lasted 1–2h.To Tot 8 17 19 22 20 25 2016 MONTHLY WEATHER REVIEW VOLUME 136

TABLE 5. Percentage of convective area coverage in the four TABLE 7. Percentage of formation cases with MCS at multiple quadrants within 2° latitude radius for two TB thresholds (Ϫ32° times and that of multiple MCSs identified at a single time for and Ϫ60°C) and for the six synoptic patterns. The percentages are each of the six synoptic patterns. arranged according to the northeast, northwest, southwest, and southeast quadrants, respectively. The two quadrants with the Synoptic Percentage of MCS Percentage of multiple highest area percentage of convection in each synoptic pattern are flow pattern at multiple times MCSs at a single time boldface. EW 40 10 NE 67 26 TB ϽϪ32°CTBϽϪ60°C NE–SW 56 33 NW NE NW NE SW 61 30 quadrant quadrant quadrant quadrant MC 68 54 MS 87 57 Synoptic SW SE SW SE All cases 63 35 flow pattern quadrant quadrant quadrant quadrant EW 38 37.3 17.4 14.6 32.8 34.8 14.2 11.1 NE 61.2 52.2 30.7 25 between the MCVs, or even their merging to become a 43.3 36.3 16.7 13 stronger vorticity cell, is most likely to be found in the NE–SW 50.5 46.2 22.5 17.5 MS pattern. The monsoon confluence pattern is also 21.4 58.7 51.2 25.7 quite favorable for generating multiple MCSs. As only SW 38.8 32.1 16.5 10.6 54.2 46.1 24.8 17.4 one MCS is identified in nearly all EW cases in the MC 48.2 50.6 18.3 18.7 pattern, the likelihood of interaction or merger is 57.5 52.7 23.8 22.5 smaller. MS 51.3 45.5 22.9 18.8 The MCS position in each synoptic pattern is simply 44.5 15.5 53.8 22.7 defined as the grid point with the lowest brightness temperature. Most of the MCSs in the EW cases first appear as far as 300–400 km east of the system center at an earlier time with values up to about 1 day before and then they migrate toward the center. From 6–12 to formation. 0–6 h before formation, the MCS positions in the EW In some of the formations, successive MCS develop- cases tend to cluster within 100 km east of the center ments occur. The overall percentage of cases with (Figs. 6a,b). By comparison, the first MCSs in the mon- MCSs at multiple times is 63% (Table 7). On the other soon-related formations are more scattered around the hand, 35% of the formation periods have more than center. When the formation time is approaching in the one MCS coexisting at a time. Note that these two per- NE–SW pattern, the convergence between the south- centages are similar to those (70% and 44%, respec- westerlies and northeasterlies is predominantly in the tively) in the study of Ritchie and Holland (1999). The region about 100–200 km east of the center, and the percentages for individual synoptic patterns indicate MCSs tend to develop in the region (Figs. 6c,d). Be- that the MS pattern is the most favorable for generating cause of the larger longitudinal spreading of the con- more than one MCS and the highest tendency for mul- vergence associated with the generally east–west- tiple MCSs existing at a time during the formation pe- oriented shear line in the MS pattern, the correspond- riod. It is speculated that if MCVs are generated in ing longitudinal spreading of the MCS positions is these MCSs, and they are sufficiently close, interaction larger (Figs. 6e,f).

TABLE 6. Mean number, mean lifetime, and mean time of iden- b. Case studies tifying the first MCS in each of the six synoptic formation patterns. Two formation cases are examined in detail to illus- trate the application of satellite data to diagnose for- Mean time of mation mechanisms in a way that is consistent with the identifying the above statistical results. The first case is Tropical Storm Synoptic flow Mean No. Mean lifetime first MCS before pattern of MCS of MCS (h) formation (h) 01W (2004) that formed at 1800 UTC 11 February (first warning time) from a westward-traveling disturbance at EW 0.8 5.7 13 NE 1.7 13.7 20 about 8°N. The formation was accompanied by a stron- NE-SW 1.6 15.6 17 ger northeasterly wind and so is classified as a NE case. SW 2.0 14.0 20 A MCS developed in the northwestern quadrant 33 h MC 2.2 14.7 24 before formation and this MCS subsequently enlarged MS 2.2 14.7 25 at about 0000 UTC 11 February (Fig. 7a). A new MCS JUNE 2008 LEEETAL. 2017

FIG. 6. Locations of the first MCS identified in a TC formation with respect to the system center for all cases associated with the (a) EW pattern at 6–12 h, (b) EW pattern at 0–6 h, (c) NE–SW pattern at 6–12 h, (d) NE–SW pattern at 0–6 h, (e) MS pattern at 6–12 h, and (f) MS pattern at 0–6 h before formation. The distances are in degrees of latitude–longitude. started to develop at 0000 UTC 11 February east of the strong convection (e.g., Fig. 7b). The northwest quad- system center (also shown in Fig. 7a). This new MCS rant-averaged relative vorticity increases rapidly during was not as long lived as the first one and was slowly the hours of the first and second MCS development, advected into the northwestern quadrant by the low- and reaches a value as high as 5.65 ϫ 10Ϫ5 sϪ1, which is level cyclonic circulation. much higher than the average value of around 3 ϫ 10Ϫ5 These MCS developments can be monitored by the sϪ1 in the other quadrants (Fig. 8b). percentage area of convection (with several TB tem- Another case is (30W) in 2000 perature thresholds) in the same quadrant. A sharp in- that developed in a monsoon confluence (MC) environ- crease in area started about 33 h before formation (Fig. ment at about 10°N with a first warning time of 1200 8a). When the second MCS ceased to develop, a third UTC 25 October. Because of low-level convergence in MCS developed at about the same location, which led such an MC pattern, deep convection started to de- to another rise in the percentage area of convection just velop in the southeast quadrant from about 2.5 days before formation. During these convective develop- before formation. Several centers of convection are ments associated with the MCSs, near-surface conver- identified in this quadrant, including a major one that gence as derived from QuikSCAT winds was large and lasted for about 18 h and moved near the center during concentrated northwest of the center, and large relative formation (Fig. 7c). As indicated by the percentage ar- vorticity values were also collocated with the areas of eas of convection with various TB thresholds (Fig. 8c), 2018 MONTHLY WEATHER REVIEW VOLUME 136

FIG. 7. (a) IR1 imagery of tropical storm 01W (2004) at 0000 UTC 11 Feb, which is 18 h before formation. The white dot marks the center of the disturbance and the two MCSs are indicated. (b) QuikSCAT-derived relative vorticity (10Ϫ5 sϪ1, color shaded) and convergence (10Ϫ5 sϪ1, contours) associated with 01W at the same time. (c), (d) As in (a), (b), respectively, but for Typhoon Xangsane (30W) at 0000 UTC 25 Oct 2000, which is 12 h before formation. two major episodes of convection are identified in the MCS shown in Fig. 7c. By comparison, the increases in southeast quadrant. Although the formation time was near-surface relative vorticity in the other quadrants not during a deep convection episode, the area percent- are minor. ages are beginning to increase by that time. The low- This case then appears to be a gradual strengthening level intensification may be related to the simultaneous of the circulation with superposed weaker convective strengthening of convection west and southwest of the episodes than in the case of 01W in 2004 for which a center. In addition to a major convergence maximum rapid strengthening in surface vorticity is found during southeast of the center, convergence maxima seem to 20–40 h before formation. The relative vorticity in the surround the center (Fig. 7d). A similar ring of near- southeast quadrant of Typhoon Xangsane fluctuated surface relative vorticity (Fig. 7d) exists around the cen- during 12–50 h before formation and then began a ter with only a slightly larger relative vorticity in the slight increase to 5.32 ϫ 10Ϫ5 sϪ1 after the MCS devel- southeast quadrant that is evidently associated with the opment in that region (Fig. 8d). Despite this small in-

Fig 7 live 4/C JUNE 2008 LEEETAL. 2019

FIG. 8. (a) Time series from 48 h before formation of the percentage area of convection for TB ϽϪ32°C (dash), ϽϪ60°C (dot), and ϽϪ75°C (solid), respectively, in the northwest quadrant within 5° latitude radius of tropical storm 01W (2004). (b) Time series of the average value of vorticity (only values larger than 10Ϫ5 sϪ1 are included) and convergence (10Ϫ5 sϪ1) in the same quadrant of 01W. (c), (d) As in (a), (b), respectively, but for Typhoon Xangsane (30W) in 2000 and values are in the southeast quadrant. crease in relative vorticity, one might also question the flow (NE), coexistence of northeasterly and southwest- formation time derived from the JTWC best track. erly flow (NE–SW), southwesterly flow (SW), monsoon confluence (MC), and monsoon shear (MS). Except for 6. Discussion the EW pattern, the other five patterns are closely re- lated to the monsoonal flow, which is revealed in their a. Summary respective seasonal distributions. In summary, the mesoscale features associated with The general convection characteristics and MCS 124 TC formations in the western North Pacific during characteristics associated with these 124 formation 1999–2004 are investigated with large-scale analyses, cases in these six synoptic patterns are studied. Con- geostationary satellite infrared brightness tempera- vection characteristics in the EW formations vary con- tures, and QuikSCAT oceanic winds. Based on low- siderably from the monsoon-related formations in that level winds, the formation cases are classified into six the convection depth is less and convection occurs synoptic patterns: easterly wave (EW), northeasterly closer to the center. Rather than the two-stage forma- 2020 MONTHLY WEATHER REVIEW VOLUME 136 tion process described in Zehr (1992), five temporal for monitoring the upper-level and near-surface pro- evolutions of convective events are identified: (A) cesses in the tropical disturbances. single deep convection, (B) two deep convection One of the major findings here with an implication as events, (C) three deep convection events, (D) gradual to TC formation mechanisms is that near-surface rela- deepening in TB, and (E) fluctuating TB or a slight tive vorticity enhancements revealed by QuikSCAT increase in TB before formation. Although no domi- surface winds appear to be associated with almost all nant profile is associated with each synoptic pattern, 124 formations in association with the deep convection profiles B and C seem to be the common temporal events indicated by the cloud-top brightness tempera- evolutions in the monsoon-related formation patterns. tures. Two examples are illustrated in the case studies In addition, the spatial distribution of convection is cor- in section 5. Since the low-level vorticity centers are related with the low-level wind in the sense that the usually identified within hours of the time of the low percentages of moderate and deep convection in the cloud-top temperatures, the low-level vortex enhance- four quadrants agree with the direction and magnitudes ment process may be quite rapid. of the stronger winds in each of the six synoptic pat- Tory et al. (2006a) discussed the distinguishing terns. It is noteworthy that the average area percentage mechanisms of stratiform vortex enhancement (SVE) of convection in the EW cases is lower than all of the and convective vortex enhancement (CVE). The SVE monsoon-related formations. is based on the midlevel MCV downward development Whereas on average only one MCS is identified in within MCSs that was proposed to be the primary me- the EW formation cases, the mean number of MCSs is soscale mechanism for TC formation in Simpson et al. around two for the monsoon-related formations. Both (1997) and Ritchie and Holland (1997). The downward the mean lifetime and time of first appearance before penetration of vorticity field is hypothesized to be formation of an MCS in the EW cases is much shorter through merging of two MCVs as simulated in Ritchie than those developed in the monsoon-related patterns. and Holland, or via advection by long-lasting precipi- In these EW cases, this MCS is most likely to be found tation as simulated in Bister and Emanuel (1997). within 100 km east of the center only 12 h before for- Based on this 5-yr period of satellite imagery, the pos- mation. The overall percentage of cases with MCS ac- sibility of MCS merging seems to be small for most of tivity at multiple times during the 48 h prior to forma- these synoptic patterns. Although cases in the SW, MC, tion is 63%, and in 35% of formation periods more than and MS patterns have 2–2.2 MCSs on average (Table 6) one MCS coexisted. These percentages are similar to during their formation processes, these MCSs were the Ritchie and Holland (1999) study of cases during an mostly quite far apart to have merged. earlier period. Most of the MC and MS formations in- The size of meso-␤ to meso-␥-scale VHTs (Hen- cluded multiple MCSs that are related to episodes of dricks et al. 2004; Montgomery et al. 2006) and the deep convection identified in the temporal evolution. response time necessary for the low- to midlevel CVE The MC and MS synoptic patterns also had the highest process are more consistent with the satellite-based ob- percentage of coexisting MCSs so that interaction be- servations in this study. For example, VHT-like meso- tween these systems could potentially have a role in the vortices that have been observed in aircraft radar re- formation process, although this aspect was not exam- flectivity data and rapid-scan satellite imagery (Reasor ined in this study. Unlike the EW pattern, the MCSs in et al. 2005; Sippel et al. 2006; Hendricks and Montgom- the monsoon-related patterns are generally in different ery 2006) might be related to the deep convective locations around the center. One exception is the NE– events and the low-level convergence patterns from sat- SW formations in which clustering of MCSs is found ellite-based study. Whereas strengthening of low-level about 100–200 km east of the center in the 12 h before vorticity has been observed in the reconnaissance flight formation. data by Zehr (1992), given the coarse resolution of the flight data it is difficult to determine if the vorticity b. Implications to internal mechanisms of enhancement contributes to the SVE or CVE pro- formation cesses. The observations in this study only allow moni- Mesoscale processes are traditionally divided into toring the lower- and upper-level processes and thus meso-␣ (200–2000 km), meso-␤ (20–200 km), and cannot provide an answer to this issue. meso-␥ (2–20 km) scales (Orlanski 1975). Given the In addition to dynamical considerations, a thermody- resolution of the satellite brightness temperature data namical problem during tropical cyclone formation is and QuikSCAT oceanic winds used in this study, the the origin of the warm-core structure. However, few focus here is on the meso-␤ processes during TC for- observations have been available to determine the ther- mations. These two datasets also provide “snapshots” modynamics processes in TC formations. Bessho et al. JUNE 2008 LEEETAL. 2021

(2006) used Advanced Microwave Sounding Unit ob- of the formation of Typhoon Robyn (1993). Terr. Atmos. servations to detect the existence of a mid- to upper- Oceanic Sci., 17, 53–89. level warm core in organized cloud clusters. Those Dickinson, M., and J. Molinari, 2002: Mixed Rossby–gravity waves and western Pacific . Part I: Syn- tropical lows that did not acquire a warm core did not optic evolution. J. Atmos. Sci., 59, 2183–2196. further develop. Without observations of the thermo- Ebuchi, N., H. C. Graber, and M. J. Caruso, 2002: Evaluation of dynamical processes on the meso-␤ scale to supplement wind vectors observed by QuikSCAT/SeaWinds using ocean these satellite-based influences, much remains to be un- buoy data. J. Atmos. Oceanic Technol., 19, 2049–2062. derstood about tropical cyclone formation. Therefore, Emanuel, K. A., 1986: An air–sea interaction theory for tropical cyclones. Part I: Steady-state maintenance. J. Atmos. Sci., 43, observational studies of TC formation are required to 585–605. diagnose thermodynamic processes during formation. Frank, W. M., and P. E. Roundy, 2006: The role of tropical waves In particular, do the VHT-scale warm core and large in tropical cyclogenesis. Mon. Wea. Rev., 134, 2397–2417. vertical wind speed in the model simulations actually Gray, W. M., 1968: Global view of the origin of tropical distur- occur during the convective bursts observed in this bances and storms. Mon. Wea. Rev., 96, 669–700. study? Further study is also required to detect specific ——, 1998: The formation of tropical cyclones. Meteor. Atmos. Phys., 67, 37–69. favorable configurations of MCSs and VHTs leading to ——, and R. W. Jacobson Jr., 1977: Diurnal variation of deep tropical cyclone formations in the synoptic patterns de- cumulus convection. Mon. Wea. Rev., 105, 1171–1188. fined. Harr, P. A., and R. L. Elsberry, 1996: Structure of a mesoscale convective system embedded in Typhoon Robyn during Acknowledgments. Comments and suggestions from TCM-93. Mon. Wea. Rev., 124, 634–652. the two anonymous reviewers are appreciated. This re- Hendricks, E. A., and M. T. Montgomery, 2006: Rapid scan views of convectively generated mesovortices in sheared Tropical search is supported by the National Science Council of Cyclone Gustav (2002). Wea. Forecasting, 21, 1041–1050. the Republic of China (Taiwan) under Grants NSC 95- ——, ——, and C. A. Davis, 2004: The role of “vortical” hot tow- 2111-M-002-002-AP1 and NSC 96-2745-M-002-005. ers in the formation of Tropical Cyclone Diana (1984). J. The participation of R. L. Elsberry was supported by Atmos. Sci., 61, 1209–1232. the Office of Naval Research Marine Meteorology sec- Holland, G. J., 1995: Scale interaction in the western Pacific mon- tion. Penny Jones assisted in the preparation of the soon. Meteor. Atmos. Phys., 56, 57–79. Lee, C.-S., 1986: An observational study of tropical cloud cluster manuscript. evolution and cyclogenesis in the western North Pacific. Col- orado State University, Department of Atmospheric Science Paper 403, 250 pp. REFERENCES Love, G., 1985: Cross-equatorial interactions during tropical cy- clogenesis. Mon. Wea. Rev., 113, 1499–1509. Bessho, K., N. Tetsuo, S. Nishimura, K. Kato, and S. Hoshino, 2006: Statistical analysis of organized cloud clusters on west- McBride, J. L., 1995: Tropical cyclone formation. Global Perspec- ern North Pacific and their warm core structure. Preprints, tives on Tropical Cyclones, WMO/TD 693, R. L. Elsberry, 27th Conf. on Hurricanes and Tropical Meteorology, Ed., World Meteorological Organization, 63–105. Monterey, CA, Amer. Meteor. Soc., 9B.6. [Available online Miller, R. A., and W. M. Frank, 1993: Radiative forcing of simu- at http://ams.confex.com/ams/pdfpapers/108684.pdf.] lated tropical cloud clusters. Mon. Wea. Rev., 121, 482–498. Montgomery, M. T., M. E. Nicholls, T. A. Cram, and A. B. Saun- Bister, M., and K. A. Emanuel, 1997: The genesis of Hurricane ders, 2006: A vortical hot tower route to tropical cyclogen- Guillermo: TEXMEX analyses and a modeling study. Mon. esis. J. Atmos. Sci., 63, 355–386. Wea. Rev., 125, 2662–2682. Orlanski, I., 1975: A rational subdivision of scales for atmospheric Charney, J. G., and A. Eliassen, 1964: On the growth of the hur- processes. Bull. Amer. Meteor. Soc., 56, 527–530. ricane depression. J. Atmos. Sci., 21, 68–75. Pickett, M. H., W. Tang, L. K. Rosenfeld, and C. H. Wash, 2003: Chen, S. S., and W. M. Frank, 1993: A numerical study of the QuikSCAT satellite comparisons with nearshore buoy wind genesis of extratropical convective mesovortices. Part I: Evo- data off the U.S. west coast. J. Atmos. Oceanic Technol., 20, J. Atmos. Sci., lution and dynamics. 50, 2401–2426. 1869–1879. ——, and R. A. Houze Jr., 1997: Diurnal variation and life-cycle Reasor, P. D., M. T. Montgomery, and L. F. Bosart, 2005: Meso- of deep convective systems over the tropical Pacific warm scale observations of the genesis of Hurricane Dolly (1996). J. pool. Quart. J. Roy. Meteor. Soc., 123, 357–388. Atmos. Sci., 62, 3151–3171. Cheung, K. K. W., 2004: Large-scale environmental parameters Ritchie, E. A., and G. J. Holland, 1997: Scale interactions during the associated with tropical cyclone formations in the western formation of Typhoon Irving. Mon. Wea. Rev., 125, 1377–1396. North Pacific. J. Climate, 17, 466–484. ——, and ——, 1999: Large-scale patterns associated with tropical ——, and R. L. Elsberry, 2004: Characteristic fields associated cyclogenesis in the western Pacific. Mon. Wea. Rev., 127, with tropical cyclone formations. Preprints, 26th Conf. on 2027–2043. Hurricanes and Tropical Meteorology, Miami, FL, Amer. Me- Simpson, J., E. Ritchie, G. J. Holland, J. Halverson, and S. Stew- teor. Soc., 2C.4. [Available online at http://ams.confex.com/ art, 1997: Mesoscale interactions in tropical cyclone genesis. ams/pdfpapers/76014.pdf.] Mon. Wea. Rev., 125, 2643–2661. ——, and ——, 2006: Model sensitivities in numerical simulations Sippel, J. A., J. W. Nielsen-Gammon, and S. E. Allen, 2006: The 2022 MONTHLY WEATHER REVIEW VOLUME 136

multiple-vortex nature of tropical cyclogenesis. Mon. Wea. Zehr, R. M., 1992: Tropical cyclogenesis in the western North Rev., 134, 1796–1814. Pacific. NOAA Tech. Rep. NESDIS 61, 181 pp. Tory, K. J., M. T. Montgomery, and N. E. Davidson, 2006a: Pre- Zhang, D.-L., and J. M. Fritsch, 1986: Numerical simulation of the diction and diagnosis of tropical cyclone formation in an meso-␤ scale structure and evolution of the 1977 Johnstown NWP system. Part I: The critical role of vortex enhancement flood. Part I: Model description and verification. J. Atmos. in deep convection. J. Atmos. Sci., 63, 3077–3090. Sci., 43, 1913–1943. ——, ——, ——, and J. D. Kepert, 2006b: Prediction and diagno- ——, and ——, 1987: Numerical simulation of the meso-␤ scale sis of tropical cyclone formation in an NWP system. Part II: structure and evolution of the 1977 Johnstown flood. Part II: A diagnosis of Tropical Cyclone Chris formation. J. Atmos. Inertially stable warm-core vortex and the mesoscale convec- Sci., 63, 3091–3113. tive complex. J. Atmos. Sci., 44, 2593–2612.