Specific Tropical Cyclone Track Types and Unusual Tropical Cyclone

Home , Monsoon trough, Typhoon Wayne (1983), Typhoon Wayne (1986)

170 WEATHER AND FORECASTING VOLUME 11

Speci®c Tropical Cyclone Track Types and Unusual Tropical Cyclone Motions Associated with a Reverse-Oriented Monsoon Trough in the Western North Paci®c

MARK A. LANDER University of Guam, Mangilao, Guam (Manuscript received 27 January 1995, in ®nal form 7 November 1995)

ABSTRACT In its simplest description, the large-scale low-level circulation of summer over the western North Paci®c Ocean can be described in terms of low-latitude southwesterlies, a monsoon trough, and a subtropical ridge. When the axis of the monsoon trough is in its normal orientation (NW±SE), tropical cyclones tend to move northwestward on tracks close to those expected from climatology. As an episodic event, the axis of the monsoon trough extends farther north and east than normal and acquires a reverse (SW±NE) orientation. When the monsoon trough becomes reverse oriented, tropical cyclones within it tend to exhibit north-oriented motion and other speci®c unusual motions such as eastward motion at low latitude and binary interactions with other tropical cyclones along the trough axis. Approximately 80% of the tropical cyclones that are associated with a reverse- oriented monsoon trough move on north-oriented tracks. A tropical cyclone track type, de®ned herein as the ``S''-shaped track, is primarily associated with reverse orientation of the monsoon trough: 23 of 35 cases (66%) of S motion during the period 1978±94 occurred in association with a well-de®ned reverse-oriented monsoon trough.

1. Introduction the low-level monsoon ¯ow of the tropical WNP: the reverse-oriented monsoon trough (RMT). The RMT is It is universally recognized that the large-scale pe- shown to be associated with a speci®c tropical cyclone riodic reversal of wind currents over the Indian sub- (TC) track type and with certain speci®c unusual TC continent and in other regions commonly acknowl- motions. edged to possess a monsoonal climate (e.g., northern Very few attempts have been made to discriminate Australia and sub-Saharan Africa) is due to the sea- TC track type based upon the large-scale pattern of the sonal changes in the differential heating of continents WNP monsoonal ¯ow. Harr and Elsberry (1991, 1993, and oceans. Lacking the sharp land±sea contrast, a sim- and 1995a, hereafter HE91, HE93, and HE95a) at- ilar lower-tropospheric low-pressure trough is often tempted to discriminate TC track type based upon the found over the tropical western North Paci®c (WNP) anomalies of the large-scale circulation of the WNP. in the summer. This trough has several dynamic and Their results showed that, under certain conditions, a kinematic features that distinguish it from the inter- discrimination of TC track type can be made based tropical convergence zones (ITCZs) of the Atlantic and upon the pattern of the large-scale circulation. Carr and eastern North Paci®c (see Atkinson 1971; Sadler 1975; Elsberry (1994, hereafter CE94) developed a forecast- and Sadler et al. 1987). In order to emphasize these er's guide in which subjective recognition of the large- differencesÐin particular, the presence of deeply scale ¯ow pattern is crucial to the proper application of moist southwesterly wind ¯ow to the south of the their forecasting techniques for TC motion. The RMT trough axis in the WNPÐthe lower-tropospheric low- has been adopted by CE94 as one of the environmental pressure trough there will herein be called a monsoon ¯ow patterns that corresponds with speci®c TC mo- trough. tions. The monsoon trough of the WNP undergoes sub- Data sources are listed in section 2. A climatology stantial migrations and major changes to its shape and of the WNP monsoon trough and a summary some of orientation. This paper de®nes a recurring and easy-to- the commonly observed patterns of the monsoon cir- recognize major departure from the normal pattern of culation of the WNP are found in section 3. Section 4 presents a synthesis of the TC track types found in HE91 and HE95a with TC track types described in other sources [e.g., Sandgathe 1987; JMA (Japan Me- Corresponding author address: Dr. Mark A. Lander, Water and Energy Research Institute, University of Guam, UOG Station, Man- teorological Agency) 1976] that was performed in or- gilao, Guam 96923. der to form a small set of TC track types that included E-mail: [email protected] the majority of TCs and that left as few TC tracks as

᭧ 1996 American Meteorological Society

/ams 3q04 0202 Mp 170 Thursday May 23 04:06 PM AMS: Forecasting (June 96) 0202

Unauthenticated | Downloaded 10/06/21 09:32 PM UTC JUNE 1996 LANDER 171

1) the hand-plotted surface synoptic charts plotted operationally at the Joint Typhoon Warning Center (JTWC), Guam, which contain SLP and winds re- ported from ships, land stations, and drifting buoys; cloud drift winds; and gradient-level wind reports from available upper-air stations; 2) high-resolution visible and infrared satellite imagery accessible at the JTWC; 3) the U.S. Navy's operational numerical analyses of SLP and other data ®elds (e.g., 1000-hPa streamlines); and 4) the Annual Tropical Cyclone Reports (ATCRs) issued by the JTWC. The charts of SLP and sequences of satellite imagery were the primary products used to determine the structure and the evolution of the selected cases of reverse orientation of the WNP monsoon trough. A survey of past ATCRs and an archive of full-disk satellite imagery helped to identify some of the commonly observed patterns of the monsoon circulation of the WNP. The positions and intensities (at 6-h intervals) of TCs in the WNP are published by the JTWC in their ATCRs. Also contained in issues of the ATCR are brief narrative discussions of the highlights of each TC.

3. The monsoon trough of the western North Paci®c a. Climatology FIG. 1. The low-level circulation during the summer in the Tropics of the western North Paci®c: (a) the long-term average and (b) a In its simplest description, the large-scale low-level schematic example of the low-level circulation associated with a re- circulation of summer over the WNP can be described verse-oriented monsoon trough. Bold zig-zag lines indicate ridge in terms of low-latitude southwesterlies, a monsoon axes, and the bold dashed line indicates the axis of the monsoon trough, and a subtropical ridge (Fig. 1a). The de®ning trough. Arrows indicate wind direction. The locations of Guam (G) and Tokyo (T) are indicated. characteristic of the WNP monsoon is the presence of deep, moist, southwesterly wind ¯ow to the south of the trough axis. Though loosely anchored to the region of highest sea surface temperature, the over-water mon- possible with indeterminate classi®cation (e.g., the soon trough of the tropical WNP undergoes substantial ``odd'' track type of Sandgathe). Typical characteris- migrations and major changes to its shape and orien- tics of the RMT are provided in section 5. In section 6, tation (unlike the monsoon trough over south Asia, the TC track type and speci®c unusual motions of TCs, which is ®rmly anchored by topography, and unlike the which form in an RMT, are discussed. The possible ITCZs of the eastern Paci®c and the Atlantic, whose bene®ts to TC forecasting of categorizing the large- scale ¯ow pattern are discussed in section 7. Section 8 summarizes the ®ndings. The general association of reverse orientation of the monsoon trough of the WNP with a speci®c TC track type and with certain speci®c unusual TC motions, and the domination of reverse orientation of the WNP monsoon trough throughout the summer and fall of 1994, prompted the writing of this paper.

2. Data sources The evolution of the low-level wind, sea level pressure (SLP), deep convective clouds, and the TC mo- FIG. 2. Normal migration of the axis of the monsoon trough indi- tion associated with an RMT were documented using cated by its mean monthly positions during June±November (after the following data sources: Atkinson 1971).

/ams 3q04 0202 Mp 171 Thursday May 23 04:06 PM AMS: Forecasting (June 96) 0202

Unauthenticated | Downloaded 10/06/21 09:32 PM UTC 172 WEATHER AND FORECASTING VOLUME 11

FIG. 3. Illustration of the extreme year-to-year differences that can occur in the monthly mean low-level wind ¯ow of the tropical western North Paci®c. (a) The resultant ¯ow of an August that featured an episode of an RMT monsoon trough: streamline analysis of the surface wind during August 1989. (b) An August with persistent easterly wind anomalies in the low latitudes and with the monsoonal southwesterlies displaced far to the north: streamline analysis of the surface wind during August 1988.

axes do not normally stray far from their mean monthly mean, there is considerable interannual and episodic positions). The axis of the summer monsoon trough of variation. the WNP usually emerges from east Asia at about 20Њ± 25ЊN and extends southeastward to a terminus south- b. Interannual and episodic variation east of Guam (13ЊN, 145ЊE). Long-term averages of low-level wind ¯ow and sea level pressure (see Sadler Eliminated by the averaging process is information et al. 1987) show that, during the boreal summer, the concerning the character and geographical bounds of mean eastward penetration of low-level winds with a episodic excursions and systematic migrations of the westerly component is to 145ЊE. Eastward of this lo- monsoon trough of the WNP. Large year-to-year dif- cation, the low-pressure trough is associated with the ferences in the individual summer monthly averages of con¯uent easterly wind ¯ow of the ITCZ. the monsoon circulation are observed (Fig. 3), which The mean location of the monsoon trough of the may be related, in part, to El NinÄo and changes in the WNP exhibits a seasonal response (Atkinson 1971) Southern Oscillation (e.g., Ramage and Hori 1981; (Fig. 2). The mean trough axis reaches its highest lat- Rasmusson and Carpenter 1982; Luther et al. 1983; itude during August; however, the easternmost pene- Lander 1994a). Large month-to-month differences are tration of monsoonal winds with a westerly component sometimes observed (Fig. 4), which may be a mani- occurs in November when low-level westerly winds, festation of the 40±50-day Madden±Julian oscillation occurring near the equator, extend to 165ЊE. While the (Madden and Julian 1971), or which may be due to pattern of the low-level wind, sea level pressure, and episodic evolution of the monsoon circulation from one of other meteorological parameters (e.g., large-scale pattern to another (Harr and Elsberry 1995b). deep convection) associated with the monsoon trough of the WNP may resemble the long-term monthly c. Commonly observed pattern types In HE93 and HE95a, cluster analysis was used to de®ne summer circulation patterns of the low-level wind ®eld of the WNP. Using the 700-hPa wind ®elds of June±October of each year from 1979 to 1987, they obtained ®ve distinct clusters. These ®ve clusters primarily indicate the strength of the monsoon trough in the WNP. Clusters 2 and 3 represent an ``active'' monsoon trough, with cluster 3 representing an ``extremely active'' monsoon trough. Cluster 5 represents interim periods of weak ¯ow anomalies, and clusters 1 and 4 FIG. 4. Illustration of extreme month-to-month changes in the cir- represent periods during which the monsoon trough is culation of the tropical western North Paci®c. This Hovmoeller-type very weak. HE95a found that these clusters were useful diagram shows the SLP pro®le along 150ЊE from 10ЊSto40ЊN for discriminators of TC activity (i.e., more tropical cy- the period 1 July 1991±11 September 1991. Solid contour Å 1012 hPa, blackened region shows SLP of 1008 hPa or lower, and dashed clones formed during times of active monsoon ¯ow line Å 1010 hPa. Note the replacement of July's subtropical ridge by than when the monsoon circulation was weak). A ma- the monsoon trough axis during August. jor goal of their research, however, was to utilize the

/ams 3q04 0202 Mp 172 Thursday May 23 04:06 PM AMS: Forecasting (June 96) 0202

Unauthenticated | Downloaded 10/06/21 09:32 PM UTC JUNE 1996 LANDER 173 clusters to discriminate between TC track type. Toward this goal, a further cluster analysis was performed on the original cluster 3 (i.e., extremely active monsoon trough), yielding three subclusters that were better at discriminating between TC track type. The three subclusters of cluster 3, all representative of an extremely active monsoon trough, differed in the strength and location of the subtropical ridge poleward of the monsoon trough. A subjective examination of the low-level monsoon ¯ow patterns in the WNP for the period 1978±94 yielded the following recurring con®gurations:

1) the long-term average as a synoptic pattern, 2) the twin-trough pattern, 3) the active ``mei-yu'' (plum rain) pattern, 4) the monsoon gyre, and 5) the reverse-oriented monsoon trough.

This list is not an exhaustive summary of all possible con®gurations of the monsoon circulation in the WNP. It is, however, a summary of some of the patterns that WNP TC forecasters will often encounter. Further- more, these pattern types are based upon an examination of data sources that the forecaster in a typical WNP FIG. 5. (a) Schematic illustration of the twin-trough pattern of low- TC warning center (e.g., the JTWC) would most likely level circulation of the western North Paci®c. Bold dashed lines in- have the time and inclination to examine: geostationary dicate the axes of the monsoon troughs. Arrows indicate wind direc- satellite imagery, and the analyses and forecasts of a tion. The locations of Guam (G) and Darwin (D) are indicated. Bold Cs show cyclonic circulations. (b) Schematic illustration of the dis- limited number of ®elds such as SLP and 500-hPa tribution of deep convection (black silhouettes) during the formation winds and heights. of TC twins. Dashed lines indicate the axes of the monsoon troughs. Four of the ®ve aforementioned patterns of the monsoon circulation are associated with certain speci®c characteristics of TC motion and/or TC structural characteristics. In this paper, con®guration 5Ðthe reverse- respect to the equator (Dean 1954; Keen 1982; Lander oriented monsoon trough (see Fig. 1b) Ðis described, 1990) (Fig. 5b). and its association with speci®c characteristics of TC In con®guration 3, the monsoon trough becomes ac- motion is discussed. As will be shown, the RMT is tive at high latitude and resembles the ¯ow pattern of associated with a consistent suite of pattern-speci®c TC late spring when the mei-yu front is active. The term, motions. Before moving on with the discussion of the mei-yu (the Chinese expression for plum rains) has RMT, con®gurations 1±4 are brie¯y described. been used to describe the annual recurrence of this per- In con®guration 1, the long-term monthly average sistent zone of disturbed weather during spring. During patterns of wind and sea level pressure appear as the the months of July and August, the mei-yu frontal zone synoptic condition of the tropical WNP. On such days, usually shifts farther northward to lie between 30Њ and the low-level circulation is near normal, and anomalies 40ЊN and extend across northern China eastward are weak. Such patterns would likely fall into the through Japan and on into the North Paci®c (Akiyama HE95a cluster 5, wherein no useful discrimination of 1973, 1990). It is the occasional transformation (usu- TC track type could be made. ally in late spring or early summer) of the mei-yu fron- Another commonly observed con®guration of the tal zone into a major large-scale convective cloud band, monsoon circulation, particularly during the spring and its association with an extension of monsoon south- (March±May) and late fall/early winter (November± westerly ¯ow into subtropical latitudes (Fig. 6), that early January), is the twin-trough patternÐcon®gu- characterizes con®guration 3. For a brief period of ration 2Ðwhich features low-level westerly wind ¯ow time, as an RMT moves poleward, it can resemble con- along the equator, bounded by near-equatorial troughs ®guration 3; however, con®guration 3 is more com- in the Northern and Southern Hemispheres (Fig. 5a). monly a steady, long-lasting ¯ow pattern wherein the This wind pattern is often associated with the devel- low latitudes of the WNP are dominated by higher- opment of equatorial westerly wind bursts (Luther et than-normal sea level pressure and region-wide low- al. 1983; Keen 1988) and also with the less common level easterly wind ¯ow (for periods often exceeding occurrence of tropical cyclone twins symmetrical with 1 month; e.g., see Fig. 3b). During such times, no

/ams 3q04 0202 Mp 173 Thursday May 23 04:06 PM AMS: Forecasting (June 96) 0202

Unauthenticated | Downloaded 10/06/21 09:32 PM UTC 174 WEATHER AND FORECASTING VOLUME 11

come TCs in the WNP are ®rst detected somewhere along the axis of the monsoon trough (this is a personal observation that is supported by the WNP TC summaries in the monthly issues of the Tropical Diagnostic Statement, published by the WMO-designated regional specialized meteorological center at Darwin, Australia, and also the WNP TC summaries in the ATCRs published by the JTWC, Guam). This assertion is also corroborated by the ®nding in HE93 and HE95a that the status (active or inactive) of the monsoon trough is a good discriminator of TC activity in the WNP. Occa- sionally, some TCs of the WNP, roughly ®ve per year, form outside of the monsoon trough, for example, in

FIG. 6. Schematic illustration of the active mei-yu pattern of low- the mei-yu trough (JTWC 1982) or in the low-level level circulation of the western North Paci®c. Bold dashed line in- easterly ¯ow to the north of the monsoon trough in dicates the axis of the mei-yu/monsoon trough. Arrows indicate wind direct association with cyclonic disturbances in the direction. The locations of Guam (G) and Tokyo (T) are indicated. tropical upper-tropospheric trough (JTWC 1992a, A ridge of high pressure (zig-zag line) is associated with easterly wind ¯ow throughout the low latitudes. Southwesterly monsoon ¯ow 1994a). occurs in the South China Sea and extends northeastward to the south According to Lander (1994a), the mean genesis lo- of the mei-yu/monsoon trough. When this ¯ow pattern occurs during cation (where the genesis location of individual TCs is the summer, TCs rarely form at low latitude but may form at un- de®ned as the position on the JTWC best track where usually high latitude along the trough axis. the TC ®rst attained 25-kt intensity) of all WNP TCs during the period 1970±92 is near 14ЊN, 142ЊE. This location is only a few hundred kilometers to the north- winds with a westerly component penetrate into low west of the long-term mean eastward penetration, dur- latitudes anywhere east of the Philippine islands. This ing summer, of the monsoon trough of the WNP. Major con®guration would likely fall into the very weak mon- shifts of the mean annual genesis locations of TCs in soon clusters 1 or 4 in HE95a (although in some re- the WNP, these are strongly linked to the status of the spects, it may be argued that the monsoon trough has El NinÄo±Southern Oscillation, are closely associated become anchored extremely far to the north of its nor- with major changes in the location of the monsoon mal position, as in Fig. 3b). Tropical cyclone formation trough. in this ¯ow pattern may occur at unusually high latitudes along the axis of the mei-yu/monsoon trough. Low-latitude cloud clusters, which form during times 2) TROPICAL CYCLONE MOTION of region-wide low-level easterlies, rarely form TCs (a The character and evolution of the midtropospheric personal observation corroborated by the ®nding in subtropical ridge and the midlatitude disturbances pass- HE95a of a lack of TC activity found under these conditions). Con®guration 4, the monsoon gyre (Fig. 7), has been described by Lander (1994b), and its affects on the motion and structure of TCs in the WNP is therein explored. A monsoon gyre typically affects TC motion by capturing the TC in a binary interaction (e.g., Dong and Neumann 1983). In some cases, the TC may merge with the gyre and undergo a sudden track change (Carr and Elsberry 1995, see section 7). One of the more striking manifestations of a large monsoon gyre is the formation of a large ``®shhook'' cloud pattern that forms as a TC is advected ®rst northward and then northwestward around the eastern periphery of the gyre (Fig. 8). d. Relationship with tropical cyclones FIG. 7. A monsoon gyre. Contours show sea level pressure at 2- 1) TROPICAL CYCLONE ACTIVITY hPa intervals for 0000 UTC 14 August 1991. Tropical cyclones Ellie and Tropical Depression 13 (T.D. 13) are embedded in the outer Of the roughly 80 named TCs that form worldwide circulation of a monsoon gyre whose low pressure center is located each year, approximately 30 of them form in the WNP. near Guam (the star labeled G). [Adapted from Fig. 2 in Lander With few exceptions, the tropical disturbances that be- (1994b).]

/ams 3q04 0202 Mp 174 Thursday May 23 04:06 PM AMS: Forecasting (June 96) 0202

Unauthenticated | Downloaded 10/06/21 09:32 PM UTC JUNE 1996 LANDER 175

FIG. 8. The ®shhook cloud pattern that is a hallmark signature of some monsoon gyres. Wind ¯ow is indicated by arrows: the location of an unnamed tropical cyclone is shown (TC). [0031 UTC 27 July 1994 visible Geostationary Meteorological Satellite (GMS), Japan, imagery].

ing by poleward of this ridge have long been used to 4. Tropical cyclone track types develop the rationale for forecasts of TC motion. These are the cited causative agents of the behavior of storms The vast majority of TCs of the WNP move west- in most post analyses (e.g., Matsumoto 1984; Sand- northwestward at low latitude (i.e., 15ЊN or lower). gathe 1987; and the Annual Tropical Cyclone Reports About one-third continue on west-northwestward issued by the JTWC, Guam) and in forecast guides tracks and make landfall in east Asia south of about (e.g., JMA 1976; Australian Bureau of Meteorology 25ЊN. The greater portion of the remaining two-thirds 1978; WMO 1993). Other factors deemed to be of im- turn northward, gain latitude, and eventually turn east- portance to TC motion in the WNP include binary in- ward as they enter the midlatitude westerlies (e.g., teraction (Brand 1970; Dong and Neumann 1983; Shanghai Typhoon Institute 1990). Lander and Holland 1993) and the effects of the large- Sandgathe (1987) divided the tracks of WNP TCs scale monsoon circulation (HE91; Lander 1994b; Carr into three major groups: ``straight moving,'' ``recurv- and Elsberry 1995). Herein, the speci®c TC motions ing,'' and ``odd.'' Straight-moving TCs are those that (i.e., track type and certain unusual TC motions) that move on long and uncomplicated west-northwestward accompany reverse orientation of the monsoon trough paths from their genesis location to their eventual land- of the WNP are discussed. fall in southeast Asia or southern China. Recurving TCs

/ams 3q04 0202 Mp 175 Thursday May 23 04:06 PM AMS: Forecasting (June 96) 0202

Unauthenticated | Downloaded 10/06/21 09:32 PM UTC 176 WEATHER AND FORECASTING VOLUME 11

TABLE 1. The redistribution of the straight-moving (SM), recurve-south (RS), recurve-north (RN), and odd track types of 56 WNP TCs, as described in HE91, into the track types straight moving, recurving (R), north oriented (NO), and S, as described in this paper. For example, of the 14 recurve-north tracks found in HE91, 7 were herein classi®ed as north oriented, and 7 were classi®ed as S.

SM (20) RS (12) RN (14) Odd (10) HE91 type/(number)

19 Ð Ð 8 Ð Ð 1 1 key: SM R 1 Ð 4 Ð 7 7 5 3 NO S

are those that move initially west-northwestward, then tics2 in order to sieve the north-oriented TC tracks from slow and turn north, and ®nally are captured in the other track types. A TC track exhibiting any one, or westerly ¯ow of the higher latitudes, within which ¯ow more, of the following characteristics during the they accelerate toward the northeast. All TC motion months of July±October is considered to be north ori- that did not ®t within the straight-moving or recurving ented: category was thrown into the category labeled by Sand- gathe as odd. Additional categorizations of TC motion 1) the TC moved steadily northward (330Њ±30Њ) for are found in HE91, HE93, and HE95a. They de®ne one more than 3 days and gained at least 15Њ of latitude new track type: South China Sea, a category based on during its period of northward motion; geography rather than any speci®c motion. The South 2) the TC formed south of 20ЊN and meandered China Sea track type contains all TCs that spent all, or greatly but gained at least 20Њ of latitude and had most, of their life in the South China Sea. The logic of a westerly component to its motion anywhere north dispensing with all South China Sea TCs as a separate of 30ЊN; track category is that most of them live very short lives 3) the TC had an eastward component of motion for before making landfall, or else they move erratically its entire track; within the small con®nes of the region of the South 4) the TC, while north of 15ЊN and moving north China Sea. HE91, HE93, and HE95a add a latitudinal through eastward (360Њ±90Њ), turned leftward at least strati®cation to Sandgathe's recurving track type: re- 30Њ, intensi®ed, and reached peak intensity during or curve north (recurving tropical cyclones that formed after making the leftward turn; and north of 20ЊN) and recurve south (recurving tropical 5) the TC track was S-shaped. cyclones that formed south of 20ЊN). Herein, the recurve-north and recurve-south track types are not used. The latter track characteristic within the north-ori- The reason for this is that in HE91 and HE95a, where ented track type (the S-shaped track) will herein be TCs undergoing these track types are presented, several given heightened emphasis. This S motion is a north- of the tracks did not recurve but, rather, moved north- oriented motion of a TC that features eastward mo- northwestward into Japan or east China. In fact, all of tion at low latitude, a later bend to the north or north- the recurve-north, and one-third of the recurve-south, west, and then eventually northeastward motion as tracks presented in HE91 (see Table 1) and HE95a are the TC enters the midlatitude westerlies (Fig. 10). clear-cut examples of another major track type: the Twenty-three of 35 cases (66%) of observed S mo- north-oriented track. The north-oriented track is a spe- tion during the years 1978±94 occurred in associa- ci®c major track type recognized by JMA (JMA 1976). tion with a well-de®ned RMT. Since the motion of nearly 80% of TCs embedded in In summary, we now have four major track classi- an RMT is north oriented, such motion warrants further ®cations for TCs of the WNP: 1) straight moving, 2) discussion. recurving, 3) north oriented, and 4) South China Sea. TCs that move on north-oriented tracks move gen- These are a synthesis of track types found in JMA erally on long, northward paths from their genesis lo- (1976), Sandgathe (1987), and HE95a. Using this cation (Fig. 9) and may feature large meanders and scheme, less than 9% of TC tracks during the period abrupt turns to the left or right. North-oriented tracks 1979±94 are relegated to an indeterminate status, or occur predominantly during July±October. Herein, the some ad hoc minor subcategory (see Table 2). With tracks of TCs that formed in the WNP1 were examined sections 1±4 serving as a background, the RMT and with respect to a suite of speci®c motion characteris- its association with north-oriented TC motion can now be discussed.

1 Tropical Cyclones that spent most of their lives in or near the South China Sea (i.e., south of 25ЊN and west of 122ЊE) were ex- 2 In JMA (1976), characteristics of north-oriented motion are not cluded. As in HE91, such TCs were considered to be a separate track rigidly codi®ed: an illustration of the tracks of several north-oriented type. TCs is shown therein.

/ams 3q04 0202 Mp 176 Thursday May 23 04:06 PM AMS: Forecasting (June 96) 0202

Unauthenticated | Downloaded 10/06/21 09:32 PM UTC JUNE 1996 LANDER 177

the trough axis acquires a reverse orientation, the TCs begin to move on north-oriented tracks (JTWC 1994b). As a point of interest, reverse orientation of the monsoon trough in the WNP is analogous to the normal orientation of the monsoon trough of the tropical western South Paci®c during the austral summer. In the mean, the axis of the Australian summer monsoon trough extends from northern Australia eastward over water toward the international date line, at which point it bends southeastward into the subtropics of the central South Paci®c. This southeastward extension is called the South Paci®c convergence zone (SPCZ). By turn- ing Fig. 1b upside down and left-for-right, one obtains a reasonable facsimile of the SPCZ during much of the austral summer. A typical reverse-oriented monsoon trough in the WNP has the following characteristics: 1) the orientation of its axis (and associated cloud band) is the reverse of the orientation of the axis of the long-term mean monthly trough axis; 2) southwesterly low-level monsoon winds, which occur within a 1000-km swath to the south of the trough axis, are brought far to the north and east of normal FIG. 9. A sample of some north-oriented tropical cyclone tracks. into regions that normally feature low-level easterly Dots are at 24-h intervals. Tropical cyclone name and year of occur- ¯ow; rence are indicated. 3) two or more TCs form simultaneously along the trough axis; and 5. The reverse-oriented monsoon trough The normal summer monsoon trough of the WNP extends southeastward from near Taiwan (23ЊN, 120ЊE) to a region southeast of Guam (13ЊN, 145ЊE), whereas a reverse-oriented trough extends from its western end (say, Luzon, 15ЊN, 125ЊE) northeastward into subtropical latitudes. An RMT in the WNP is an episodic event that occurs on average about once each typhoon season sometime between mid-July and mid- October. During some years it is not seen, but during other years (such as 1989 and 1994) the monsoon trough repeatedly organizes in a reverse orientation. The distinguishing characteristics of an RMT are the reverse orientation of its axis with respect to the long- term monthly mean, its excursion into subtropical areas normally the province of easterly ¯ow (Figs. 1a and 1b), and its tendency to migrate northward to higher latitudes of the WNP. When the monsoon trough acquires a reverse orientation, TCs along it tend to move on north-oriented tracks and may undergo binary interactions with other TCs along the trough axis. The cloud band of the monsoon trough may become reverse oriented before tropical cyclogenesis occurs (Fig. 11), or the trough axis can become reverse oriented as two (or more) individual TCs, which are located along a zonally oriented monsoonal cloud band, move relative to one another so as to bring the trough FIG. 10. A sample of some S tracks. Dots are at 24-h intervals. axis into reverse orientation. In such cases, as soon as Tropical cyclone name and year of occurrence are indicated.

/ams 3q04 0202 Mp 177 Thursday May 23 04:06 PM AMS: Forecasting (June 96) 0202

Unauthenticated | Downloaded 10/06/21 09:32 PM UTC 178 WEATHER AND FORECASTING VOLUME 11

TABLE 2. Individual 1978±94 tropical cyclone (TC) track types. The observed track classes are de®ned as straight moving (SM), recurving (R), north oriented (NO), S track (S), and other. Further subdivisions of the other category are indicated by icons: ✌ Å TC remained in or near South China Sea for its whole life, ☞ Å TC formed over open Paci®c and died over water after a short track, ❡ Å TC made many loops and meanders but made little overall forward progress, ➦ Å TC formed in mei-yu cloud band and tracked rapidly to the northeast, and ✈ Å TC formed in the lee of Taiwan during conditions of monsoonal southwesterly ¯ow and tracked northward then westward around the top of Taiwan to make landfall on the China coast. Note that the subdivisions within the other category are not mutually exclusive; for example, a South China Sea TC (✌) might also have looped and meandered (❡).

Year SM R NO S Other ✌☞❡➦✈

1978 5 10 9 1 7 4 3 Ð Ð Ð 1979 10 12 2 Ð 4 2 1 1 1 Ð 1980 10 7 5 1 6 5 1 1 Ð Ð 1981 11 7 4 1 6 2 2 1 Ð 1 1982 9 5 6 1 7 3 1 2 2 Ð 1983 11 5 4 Ð 4 4 Ð 1 Ð Ð 1984 6 5 7 4 8 5 3 1 Ð Ð 1985 7 3 3 6 8 5 2 1 Ð Ð 1986 9 12 2 Ð 3 1 1 1 1 Ð 1987 8 2 9 Ð 5 2 3 Ð Ð Ð 1988 9 4 7 1 5 2 Ð 1 2 1 1989 15 5 5 4 6 3 3 Ð Ð Ð 1990 8 11 1 4 7 5 1 Ð Ð 1 1991 11 14 2 Ð 4 3 1 1 Ð Ð 1992 8 11 5 4 4 3 1 2 Ð Ð 1993 15 10 4 Ð 8 2 6 Ð Ð Ð 1994 15 5 6 8 5 4 1 Ð Ð Ð TOTALS: 167 128 81 35 97 53 30 13 6 3

4) the TCs that form in it tend to move on north- Of the 35 cases of S motion during the period 1978± oriented tracks and may exhibit one or more of a suite 94, 23 cases (66%) occurred in association with a well- of pattern-speci®c unusual TC motions. de®ned RMT. A look at the tracks of TCs that formed in the WNP Of particular interest herein is the association of a spe- during 1994 (JTWC 1994a) indicates an unusually ci®c TC track type and speci®c unusual TC motions high number of north-oriented tracks (see Table 2). Of with reverse orientation of the monsoon trough of the 39 TCs that formed in the WNP during 1994, 15 the WNP. (38%) were straight moving, only 5 (13%) were re- curvers, 14 (36%) moved on north-oriented tracks, and 6. Tropical cyclone motion associated with a 4 (10%) were South China Sea TCs. Only one had a reverse-oriented monsoon trough track too short to be classi®ed. Of the 14 TCs that During the period 1978±94, there were 508 TCs that moved on north-oriented tracks during 1994, 8 under- were numbered or named by the JTWC in the WNP went S motion. The majority (10 of 14 or 71%) of the (JTWC 1994a). The tracks of these TCs were broken TCs during 1994 that moved on north-oriented tracks down into our four general categories: straight moving, occurred in association with three episodesÐone in recurving, north oriented, and South China Sea (see July, one in September, and one in OctoberÐof re- Table 2). Characteristics of the straight-moving and verse orientation of the monsoon trough. All S motion recurving track types are those developed by Sandgathe during 1994 was associated with an RMT. (1987). Characteristics of the north-oriented track type Using the JTWC's ATCRs and geostationary satel- are those described by the JMA (1976) and which are lite imagery for the period 1978±94, some well-de®ned further ampli®ed in section 5. Tropical cyclones whose periods of reverse orientation of the monsoon trough tracks did not conform to these four track types tended were established. The WNP monsoon circulation of the to be TCs that had very short lifetimes or moved on years 1981, 1984, 1985, 1986, 1987, 1989, 1990, 1992, very short tracks. Of the aforementioned 508 TCs, 167 and 1994 was punctuated by one or more clear-cut ep- (33%) were straight movers, 128 (25%) were recurv- isodes of reverse orientation of the axis of the monsoon ers, 116 (23%) moved on north-oriented tracks, and 53 trough. During these episodes of reverse orientation of (10%) were South China Sea TCs. Forty-four (9%) of the monsoon trough, most of the TC motion (30 of 38 the TC tracks did not conform to these four main track TCs or 79%) was north oriented. The 30 TCs that types. Within the north-oriented track type, a subcate- moved on north-oriented tracks during episodes of re- gory of track type, S motion, has herein been added. verse orientation of the monsoon trough during these Of the 116 north-oriented tracks during the period years accounted for 26% of the 116 cases of north- 1978±94, 35 (30%) could be categorized as S motion. oriented TC motion for the period 1978±94. Although

/ams 3q04 0202 Mp 178 Thursday May 23 04:06 PM AMS: Forecasting (June 96) 0202

Unauthenticated | Downloaded 10/06/21 09:32 PM UTC JUNE 1996 LANDER 179

FIG. 11. High-contrast GMS infrared imagery showing the development of Tropical Cyclones Owen (O) and Nancy (N) within the preexisting cloud band associated with a reverse-oriented monsoon trough: (a) 0000 UTC 12 July 1989, (b) 0000 UTC 13 July 1989, (c) 0000 UTC 14 July 1989, and (d) 0000 UTC 15 July 1989.

/ams 3q04 0202 Mp 179 Thursday May 23 04:06 PM AMS: Forecasting (June 96) 0202

Unauthenticated | Downloaded 10/06/21 09:32 PM UTC 180 WEATHER AND FORECASTING VOLUME 11 most north-oriented motion does not occur within a clear-cut RMT ¯ow pattern, the RMT ¯ow pattern is nonetheless a useful discriminator of TC track type (i.e., the existence of an RMT implies north-oriented motion, but not vice versa).

Selected examples Twelve well-de®ned cases of reverse orientation of the axis of the monsoon trough of the WNP are now brie¯y described. The relationship of the TC tracks to the RMT is discussed. A summary of these tracks during 1978±92 is shown in Fig. 12. The north-oriented

TC tracks associated with the RMTs of 1994 are shown FIG. 12. Twenty-seven tropical cyclone tracks that occurred in as- in Figs. 13a±c. sociation with selected episodes of reverse orientation of the monsoon trough during 1981, 1984±87, 1989, 1990, and 1992.

1) AUGUST 1981 The WNP monsoon trough was particularly active 4) AUGUST 1986 during early and mid-August 1981 (JTWC 1981). By By 12 August 1986, an intense monsoon trough ex- 16 August 1981, an active reverse-oriented monsoon tended east-northeastward from the central Philippines trough had become established in the WNP. At this to near Wake Island (JTWC 1986). Typhoon Tip de- time, two TCsÐThad and VanessaÐwere forming in veloped at the easternmost reaches of this reverse-ori- the monsoonal cloud band. Both of these TCs moved ented monsoon trough and moved on a complex north- on north-oriented tracks, and each also exhibited an- oriented track. This very active reverse-oriented mon- other unusual motion common to TCs that form in a soon trough also encompassed Tropical Storm Vera reverse-oriented monsoon trough: an eastward com- and Typhoon Wayne. Vera remained quasi-stationary ponent of motion at the time of genesis. for several days before it moved westward and then recurved. Typhoon Wayne was one of the longest-lived 2) OCTOBER 1984 TCs in the WNP. It meandered back and forth between the northern South China Sea and the western Philip- By the last week of September 1984, the monsoon pine Sea throughout its entire 22-day life. Slow, con- trough of the WNP became reverse oriented (JTWC voluted, meandering tracks, con®ned to a small region 1984). On 28 September, the ®rst two (of ®ve) TCsÐ in or near the northern South China Sea, are a recurring Maury and NinaÐformed at the eastern reaches of the behavior of TCs that form in the South China Sea at reverse-oriented monsoon trough. Several days later, the western end of reverse-oriented monsoon troughs. while the monsoon trough remained in a reverse orientation, TCs Ogden, Phyllis, and Roy formed. All ®ve 5) SEPTEMBER 1987 of these TCs moved on north-oriented tracks. The tracks of TCs Nina, Ogden, Phyllis, and Roy all could In early September 1987, an active monsoon trough be categorized as S tracks. was associated with a three-TC outbreakÐFreda, Ger- ald, and Holly (JTWC 1987). As these three systems matured, the monsoon trough became displaced well to 3) AUGUST 1985 the north of its normal location and acquired a reverse orientation. During the ®rst few days in the lifetimes By the third week of August 1985, the eastern end of these three TCs, the trough was oriented in a normal of the monsoon trough extended across the Philippine NW±SE orientation, and the three TCs were moving Sea to the vicinity of Guam. A multiple outbreak of on typical northwesterly tracks. After Gerald made TCsÐOdessa, Pat, and RubyÐoccurred in this trough landfall in China, Holly, the easternmost of the TCs, (JTWC 1985). By the time all three of these TCs had gained latitude faster than Freda. Eventually, Holly sur- become named, the monsoon trough was located north passed Freda in latitude causing the monsoon trough to of normal and it had acquired a reverse orientation. become reverse oriented, and both Freda and Holly Each of these TCs moved on S tracks. The tracks of turned northward to follow slow north-oriented tracks. Pat and Ruby appear in Fig. 10 as de®nitive examples of the S track. Odessa was involved in a binary inter- 6) AUGUST 1989 action with Ruby and later, escaping from the interaction with Ruby, became involved in a binary interaction During the ®rst week of August 1989, an active mon- with Pat (JTWC 1985; Lander and Holland 1993). soonal cloud band acquired a reverse orientation. Three

/ams 3q04 0202 Mp 180 Thursday May 23 04:06 PM AMS: Forecasting (June 96) 0202

Unauthenticated | Downloaded 10/06/21 09:32 PM UTC JUNE 1996 LANDER 181

TCsÐNancy, Owen, and PeggyÐeventually developed in this reverse-oriented monsoon trough (Fig. 11) and moved on north-oriented tracks (JTWC 1989a). Nancy and Owen underwent a binary interaction.

7) SEPTEMBER 1989 On the ®rst day of September 1989, the monsoon trough stretched across the WNP in a southwest±northeast (i.e., reverse) orientation between 10Њ and 20ЊN, and supported several discrete convective cloud masses (JTWC 1989b). Three TCs eventually formed in this troughÐSarah, Tip, and Vera. Of these three TCs, only Tip moved on a genuine north-oriented track. Sarah and Vera moved west-northwestward on some- what erratic straight-moving tracks.

8) JULY 1990 Typhoon Steve, with Tropical Storm Tasha and Ty- phoon Vernon, made up the only three-TC outbreak to occur in the WNP during 1990 (JTWC 1990). The axis of a reverse-oriented monsoon trough extended eastward from Tasha through Vernon and Steve to a low east of Japan. Steve and Vernon moved on north-oriented tracks, while the westernmost TC, Tasha, meandered along a complex track near the Philippines. The tracks of Steve and Vernon are good examples of the S-track subcategory of north-oriented motion.

9) AUGUST 1992 During mid-August 1992, the monsoon trough acquired a reverse orientation. Three TCsÐMark in the South China Sea, Lois northeast of the Philippines, and Nina east of JapanÐwere aligned SW±NE in this trough. Lois was one of only two TCs in 1992 that had a persistent eastward component of motion during its period of warning (JTWC 1992b). This is an exclusive characteristic of some north-oriented TC tracks. Mark moved slowly on a meandering track in the northern South China Sea. Nina moved north, then recurved, and moved rapidly eastward along 40ЊN.

10) JULY ±OCTOBER 1994 The period July±October 1994 was dominated by north-oriented TC motion in the WNP and also by episodes of reverse orientation of the monsoon trough. Diagnosis of the reverse orientation of the monsoon trough in real time allowed for forecasters at the JTWC to anticipate north-oriented TC motion. The ®rst occurrence during 1994 of reverse orientation of the mon- FIG. 13. The north-oriented TC tracks of 1994 in the western North soon trough in the WNP was during July. By mid-July, Paci®c that occurred in association with three episodes of reverse an active reverse-oriented monsoon trough stretched orientation of the monsoon trough: (a) July (W Å Walt, Y Å Yunya, across the Tropics of the WNP from the South China Z Å Zeke); (b) September (M Å Melissa, N Å Nat, O Å Orchid, P Å Pat, R Å Ruth), note that Ruth and Pat merged at the location Sea northeastward into subtropical latitudes near the shown by the bold arrow; and (c) October (V Å Verne, W Å Wilda). international date line (Figs. 14a and 14b). Three TCs Dots indicate TC locations at 12-h intervals. formed in this troughÐWalt, Yunya, and Zeke. Each

/ams 3q04 0202 Mp 181 Thursday May 23 04:06 PM AMS: Forecasting (June 96) 0202

Unauthenticated | Downloaded 10/06/21 09:32 PM UTC 182 WEATHER AND FORECASTING VOLUME 11

FIG. 14. A reverse-oriented monsoon trough. (a) The cloud band associated with a reverse-oriented monsoon trough that stretched across the western North Paci®c during latter half of August 1994 (GMS infrared imagery at 0100 UTC 19 August 1994). (b) The sea level pressure analysis at 1200 UTC 19 August 1994. The locations of tropical cyclones Walt (W), Yunya (Y), Zeke (Z), and a subtropical disturbance (ST) are indicated. Pressure contours are at 4-hPa intervals. The region in Walt of SLP below 992 hPa is black. of these TCs moved on north-oriented tracks. Each of September. For a full 2 weeks in the middle of Septem- these TCs exhibited unusual eastward motion at low ber, an active reverse-oriented monsoon trough domi- latitude (Fig. 13a). nated the large-scale circulation of the WNP. Five TCs The second occurrence during 1994 of reverse ori- formed in this troughÐMelissa, Nat, Orchid, Pat, and entation of the monsoon trough in the WNP was during Ruth. All of them moved on S tracks. Nat, Orchid, and

/ams 3q04 0202 Mp 182 Thursday May 23 04:06 PM AMS: Forecasting (June 96) 0202

Unauthenticated | Downloaded 10/06/21 09:32 PM UTC JUNE 1996 LANDER 183

Ruth exhibited unusual extended eastward motion at low latitude (Fig. 13b). The third occurrence during 1994 of reverse orientation of the monsoon trough in the WNP occurred in October. By mid-October, the monsoon trough was oriented zonally at low latitude and in the eastern portion of the basin. Three TCsÐTeresa, Verne, and WildaÐ formed in succession in this monsoon trough and initially moved westward. By the end of October, Teresa, Verne, and Wilda had moved relative to each other so as to bring the axis of the monsoon trough into a reverse orientation. As soon as this occurred, the westward motion of Verne ceased, and Wilda began to move on a north-oriented S track (Fig. 13c).

7. Implications for TC motion forecasting a. Physical mechanisms The hypothesized physical mechanism responsible for the preponderance of north-oriented TC motion in an RMT is a strengthening of a ridge to its southeast FIG. 16. Typical track changes associated with the interaction of a by Rossby wave dispersion. Carr and Elsberry (1995) tropical cyclone with a monsoon gyre. Dots show positions at 6-h show, with a barotropic model, that a very large axi- intervals for 96 h of the simulated tracks of three tropical cyclones symmetric vortex in the Tropics (e.g., a monsoon gyre) placed at 600, 700, and 800 km, respectively, east of the center of a monsoon gyre. [Figure adapted from Fig. 12b in Carr and Elsberry alters the large-scale environmental ¯ow through (1995).] Rossby wave dispersion. The initial large axisymmetric vortex emanates a wave train into the environment whose prominent feature is a ridge axis located east- clockwise-through-south of the original vortex (Figs. tion of a southwesterly jet between the monsoon gyre 15a±c). Between the vortex and this ridge, a south- (or the RMT) and the induced ridge account for the westerly jet becomes established. When another steering ¯ow that drives the TC on a northward track. smaller vortex (representing a TC) is added on the east- TCs that form in association with a monsoon gyre or ern side of the monsoon gyre, the smaller vortex moves an RMT are observed to form and move along the cy- on a north-oriented track (Fig. 16). The north-oriented clonic shear side of the low-level jet (R. Elsberry 1994, motion of the TC is caused by its advection by the wind personal communication). ¯ow of the large-scale vortex and by its advection by the southwesterly ¯ow that becomes established be- b. A conditional climatology for TC motion in a tween the large-scale vortex and the ridge to the south- reverse-oriented monsoon pattern east. While the mechanisms leading to the formation of the large-scale vortex (and its rough analog, the RMT) Most TCs (79% of the cases) associated with an remain a mystery, the induced ridging and the forma- RMT move on north-oriented tracks. Subjective and objective diagnoses of reverse orientation of the monsoon trough should prove to be relatively easy given its extreme departure from the long-term monthly average circulation pattern. An RMT is easily detected in satellite imagery (Fig. 14a) and in analyses of synoptic data (e.g., SLP) (Fig. 14b). Using examples of the TC motion associated with reverse orientation of the WNP monsoon trough (Figs. 12 and 13), a generalized template emerges for the character of the motion with respect to the position of the TC along the axis of the trough. The ``Systematic FIG. 15. (a)±(c) Simulated streamfunction for the case of a TC and Integrated Approach to Tropical Cyclone Track [black dot in (a)] embedded within the circulation of a monsoon gyre Forecasting'' devised by CE94, which was presented at the indicated times. A realistic value for the contour interval is 1 1 106 m2 s01, negative (positive) contours are dashed (solid), and to JTWC forecasters during 1994, contains templates the zero line has been suppressed. [Figure adapted from Fig. 17 in for TC motion organized by four ``synoptic patterns'' Carr and Elsberry (1995).] and six ``synoptic regions'' that appear within the syn-

/ams 3q04 0202 Mp 183 Thursday May 23 04:06 PM AMS: Forecasting (June 96) 0202

Unauthenticated | Downloaded 10/06/21 09:32 PM UTC 184 WEATHER AND FORECASTING VOLUME 11 optic patterns. Of interest herein is one of their north- oriented synoptic patterns (Fig. 17), which clearly shows the large-scale ¯ow pattern of an RMT (they identify it as such). The motion of the TCs in this pattern is north oriented, and in this particular illustration the TCs are shown to be undergoing S motion. Another application of the general association of north-oriented motion with reverse orientation of the monsoon trough is to produce a climatology of TC motion strati®ed by this synoptic pattern. The production of a separate climatology for TC motion during episodes of reverse orientation of the monsoon trough (e.g., Fig. 18) may help improve simple statistical pre- dictions (e.g., pure climatology, CLIPER, half-persist- FIG. 18. A conditional climatology for TC motion associated with ence and climatology, etc.). reverse orientation of the monsoon trough of the western North Pa- ci®c. The arrows show mean motion in 5Њ 1 5Њ squares composited from the 27 tropical cyclone tracks shown in Fig. 12. c. Improved objective guidance Statistical-dynamic models, such as those developed by Matsumoto (1984) and Neumann (1988, 1992), spect to the location of the axis of the subtropical ridge. predict TC motion with regression equations that use Position of the TC relative to the subtropical ridge was as predictors the values of a set of preselected environ- determined based upon the current motion of the TC: mental parameters (e.g., the 24-h fall of SLP 500 km if the TC was moving westward of 330Њ, it was consid- northeast of the TC, the u component of the 200-hPa ered to be south of the ridge axis; if the TC was moving wind 800 km northwest of the TC, etc.) Certainly the between 330Њ to 30Њ, it was considered to be on the output of these models is affected by differences in the ridge axis; if the motion of the TC was eastward of 30Њ, large-scale circulation within which the TC is embed- it was considered to be north of the ridge axis. Mat- ded; however, the large-scale circulation changes in the sumoto's model works well on average, but under cer- WNP are large, and a blanket set of regression equa- tain extreme large-scale circulation anomalies (e.g., the tions utilizing far-®eld variables would be expected to RMT) its forecasts can be poor (e.g., Fig. 19). A con- perform poorly under conditions of circulation patterns ceptual example of how this model can fail is illustrated that deviate far from the average. by its application to a TC that is moving eastward at Recognizing the need for some form of strati®cation low latitude in an RMT. Because the TC is moving based upon the structure of the large-scale environ- eastward of 30Њ, the model assumes that it is north of ment, Matsumoto developed three sets of regression the axis of the subtropical ridge and uses the north-of- equations based upon the location of the TC with re- the-ridge regression equations to make the track forecasts. Most of the data used to derive the north-of-the- ridge regression equations were compiled for cases of recurving TCs at high latitude. For eastward motion of a TC at low latitude under the steering in¯uence of the southwesterly monsoon ¯ow of an RMT, the north-of- the-ridge regression equations may be a misapplication. The argument to stratify statistical and statistical-dynamic models by the characteristics of the large-scale ¯ow patterns is compelling. Even the simplest of statistical forecast aids such as climatology could proba- bly be improved with a strati®cation based upon the character of the large-scale environment.

8. Summary and conclusions In its simplest description, the large-scale low-level FIG. 17. Template for determining tropical cyclone motion during circulation of summer over the WNP can be described episodes of reverse orientation of the monsoon trough of the western in terms of low-latitude southwesterlies, a monsoon North Paci®c. [Adapted from Carr and Elsberry (1994).] Streamlines trough, and a subtropical ridge. The axis of the summer depict large-scale midtropospheric wind ¯ow. Dashed line shows the monsoon trough axis; zig-zag lines show ridge axes. When the tem- monsoon trough of the WNP usually emerges from east plate is matched with the real-time synoptic pattern, the future motion Asia at about 20Њ±25ЊN and extends southeastward to of a tropical cyclone at any of locations 1±5 can be anticipated. a terminus southeast of Guam (13ЊN, 145ЊE). Most of

/ams 3q04 0202 Mp 184 Thursday May 23 04:06 PM AMS: Forecasting (June 96) 0202

Unauthenticated | Downloaded 10/06/21 09:32 PM UTC JUNE 1996 LANDER 185

the ¯ow pattern of the WNP monsoon to be an RMT (and its association with speci®c unusual TC motions) has been used by forecasters at the JTWC, Guam, to develop the prognostic reasoning for their TC track forecasts. CE94 use the RMT ¯ow pattern as one of only six basic synoptic ¯ow patterns of the WNP useful for predicting TC motion (Fig. 17). Stratifying subjective TC track forecast techniques (e.g., those presented in CE94), and objective forecast techniques (e.g., the CLIPER forecast aid) by the large-scale ¯ow pattern types is an idea that should be pursued. The demonstration herein (and in HE95a) that some patterns of the large-scale monsoon ¯ow are associated with certain TC track types and other speci®c unusual TC motions (and speci®c pattern-related biases in the TC track aids derived from dynamic models) shows that the concept of stratifying forecast techniques by the large-scale ¯ow pattern type may succeed in improving those techniques. The ®rst step in this process must be to identify the ¯ow pattern type. Herein, and in CE94, the pattern types are identi®ed subjectively. The TC forecaster identi®es the pattern type and uses this knowledge to re®ne the forecast of TC motion (and perhaps also of the TC structure and intensity). Whether the pattern types should be identi®ed subjectively (as is done herein and in CE94) or objectively (as in HE95a) will need to be decided by the relative success of the applied techniques that are FIG. 19. The output of Matsumoto's (1984) statistical-dynamic developed. Regardless of the method for determining model (known as CSUM) at various positions along the north-oriented track of Typhoon Wilda (October 1994). Large dots connected the ¯ow pattern types, it is important that the TC fore- by the bold line show the TC position at 24-h intervals. Open circles caster be able to recognize the pattern type and to un- connected by thin lines show the CSUM forecasts. The three open derstand its implications for the forecast. This need for circles on each of the CSUM forecasts show the 24-, 48-, and 72-h forecaster recognition of the pattern type is the reason forecast positions. The CSUM forecast a was made using the south- for the subjective nature of this paper and for giving of-the-ridge set of regression equations. Forecast b was made using the on-the-ridge set of equations. As Wilda moved eastward of 30Њ the pattern types speci®c names (e.g., the monsoon on the ®rst leg of the S portion of its track, the CSUM forecasts c±f gyre, the twin-trough pattern, and the reverse-oriented were made using the north-of-the-ridge equations. On a later bend to monsoon trough). the northwest, the CSUM forecasts g and h were made using the south-of-the-ridge equations. On Wilda's ®nal turn toward the north- Acknowledgments. Full support for this research was east, the CSUM forecast i was made using the north-of-the-ridge from the Of®ce of Naval Research through Grant equations. N00014-91-J-1721. The support of the personnel at the Joint Typhoon Warning Center in allowing me access to their satellite imagery and other meteorological data the TCs that develop in the WNP form in the monsoon is greatly appreciated. Conversations with Mr. Frank trough. When the axis of the monsoon trough is in its Wells (JTWC technical assistant), Mr. Charles Guard normal orientation (NW±SE), TCs tend to move (director of the JTWC, 1989±93), and Dr. Les Carr northwestward on tracks close to those expected from (Naval Postgraduate School instructor and deputy di- climatology. As an episodic event, the axis of the mon- rector of the JTWC, 1991±92) were invaluable. soon trough acquires a reverse (SW±NE) orientation. When this happens, TCs within it tend to move on REFERENCES north-oriented tracks and may undergo binary interac- Akiyama, T., 1973: The large-scale aspects of the characteristic fea- tions with other TCs along the trough axis. Approxi- tures of the Baiu Front. Pap. Meteor. Geophys., 24, 157±188. mately 66% of cases of a TC track type, de®ned herein , 1990: Large, synoptic and mesoscale variations of the Baiu as the S track, are associated with an RMT. Front during July 1982, Part II: Frontal structure and distur- The RMT is an easily recognized large-scale ¯ow bances. J. Meteor. Soc. Japan, 68, 557±574. Atkinson, G. D., 1971: Forecasters guide to tropical meteorology. pattern on analyses of the low-level wind and the SLP, Air Weather Service (MAC) U.S. Air Force Tech. Rep. 240, and its associated cloud pattern on satellite imagery 360 pp. [Available from NTIS, U.S. Dept. of Commerce, Sills also renders its diagnosis an easy task. Recognition of Bldg., 5285 Port Royal Rd., Spring®eld, VA 22161.]

/ams 3q04 0202 Mp 185 Thursday May 23 04:06 PM AMS: Forecasting (June 96) 0202

Unauthenticated | Downloaded 10/06/21 09:32 PM UTC 186 WEATHER AND FORECASTING VOLUME 11

Australian Bureau of Meteorology, 1978: Australian Tropical Cy- , 1992a: Synoptic discussion of Tropical Storm Helen. 1992 clone Forecasting Manual. Bureau of Meteorology Research Annual Tropical Cyclone Rep., JTWC, 269 pp. [NTIS AD Centre, 274 pp. A274464.] Brand, S., 1970: Interaction of binary tropical cyclones in the western , 1992b: Synoptic discussions of tropical cyclones Lois, Mark North Paci®c Ocean. J. Appl. Meteor., 9, 433±441. and Nina. 1992 Annual Tropical Cyclone Rep., JTWC, 269 pp. Carr, L. E., and R. L. Elsberry, 1994: Systematic and integrated ap- [NTIS AD A274464.] proach to tropical cyclone track forecasting, Part I: Description , 1994a: Summary of western North Paci®c and north Indian of basic approach. Naval Postgraduate School Publ. NPS-MR- Ocean tropical cyclones. 1994 Annual Tropical Cyclone Rep., 002. Naval Postgraduate School, Monterey, CA, 65 pp. plus ®gs. JTWC, 337 pp. and appendix. , 1994b: Synoptic discussions of Typhoons Verne, Teresa, and , and , 1995: Monsoonal interactions leading to sudden Wilda. 1994 Annual Tropical Cyclone Rep., JTWC, 337 pp. tropical cyclone track changes. Mon. Wea. Rev. 123, 265±289. Keen, R. A., 1982: The role of cross-equatorial tropical cyclone pairs Dean, G. A., 1954: An example of tropical cyclogenesis symmetrical in the Southern Oscillation. Mon. Wea. Rev. 110, 1405±1416. with respect to the equator. Oahu Research Center (Institute of , 1988: Proceedings of the U.S. TOGA Workshop on Western Geophysics), University of California Rep. 3, 6 pp. plus ®gs. Paci®c Air±Sea Interaction. UCAR, 207 pp. and appendix. [Available from Dept. of Meteorology, Univer- Lander, M. A., 1990: Evolution of the cloud pattern during the for- sity of Hawaii, 2525 Correa Rd., Honolulu, HI 96822.] mation of tropical cyclone twins symmetrical with respect to the Dong, K., and C. J. Neumann, 1983: On the relative motion of binary equator. Mon. Wea. Rev., 118, 1194±1202. tropical cyclones. Mon. Wea. Rev., 111, 945±953. , 1994a: An exploratory analysis of the relationship between Harr, P. A., and R. L. Elsberry, 1991: Tropical cyclone track char- tropical storm formation in the western North Paci®c and ENSO. acteristics as a function of large-scale circulation anomalies. Mon. Wea. Rev., 122, 636±651. Mon. Wea. Rev., 119, 1448±1468. , 1994b: Description of a monsoon gyre and its effects on the , and , 1993: Variability of tropical cyclone track charac- tropical cyclones in the western North Paci®c during August teristics and large-scale circulation regimes over the western Pa- 1991. Wea. Forecasting, 9, 640±654. ci®c Ocean. Preprints, 20th Conf. on Hurricanes and Tropical , and G. J. Holland, 1993: On the interaction of tropical-cyclone Meteorology, San Antonio, TX, Amer. Meteor. Soc., 97±100. scale vortices. I: Observations. Quart. J. Roy. Meteor. Soc., 119, , and , 1995a: Large-scale circulation variability over the 1347±1361. tropical western North Paci®c. Part I: Spatial patterns and trop- Luther, D. S., D. E. Harrison, and R. A. Knox, 1983: Zonal winds in ical cyclone characteristics. Mon. Wea. Rev. 123, 1225±1246. the central equatorial Paci®c and El NinÄo. Science, 222, 327± , and , 1995b: Large-scale circulation variability over the 330. tropical western North Paci®c. Part II: Persistence and transition Madden, R. A., and P. R. Julian, 1971: Detection of a 40±50 day characteristics. Mon. Wea. Rev., 123, 1247±1268. oscillation in the zonal wind in the tropical Paci®c. J. Atmos. JMA, 1976: The north-oriented track type. Forecasting Manual for Sci., 28, 702±708. Typhoons (in English), K. Takemoto, Ed., Japan Meteorologi- Matsumoto, C. R., 1984: A statistical method for one- to three-day cal Agency, 194±197. tropical cyclone track prediction. Colorado State University At- JTWC, 1981: Synoptic discussions of tropical cyclones Thad, Va- mospheric Science Paper 379. Department of Atmospheric Sci- nessa and Warren. 1981 Annual Tropical Cyclone Rep., JTWC, ence, Colorado State University, Fort Collins, CO, 201 pp. 196 pp. [NTIS AD A112002.] Neumann, C. J., 1988: The National Hurricane Center NHC83 model. , 1982: Synoptic discussions of tropical cyclones Skip and Val. NOAA Tech. Memo. NWS NHC-41, 44 pp. 1982 Annual Tropical Cyclone Rep., JTWC, 236 pp. [NTIS AD , 1992: Joint Typhoon Warning Center (JTWC92) model. Sci- A124860.] ence Applications International Corporation Final Rep., SAIC, , 1984: Synoptic discussions of tropical cyclones Maury, Nina, McLean, VA, 65 pp. Ogden, Phyllis and Roy. 1984 Annual Tropical Cyclone Rep., Ramage, C. S., and A. M. Hori, 1981: Meteorological aspects of El JTWC, 222 pp. [NTIS AD A153395.] NinÄo. Mon. Wea. Rev., 109, 1827±1835. , 1985: Synoptic discussions of tropical cyclones Odessa, Pat Rasmusson, E. M., and T. H. Carpenter, 1982: Variations in tropical and Ruby. 1985 Annual Tropical Cyclone Rep., JTWC, 276 pp. sea surface temperature and surface wind ®elds associated with [NTIS AD A168284.] the Southern Oscillation/El NinÄo. Mon. Wea. Rev., 110, 354± , 1986: Synoptic discussions of tropical cyclones Tip, Vera and 384. Wayne. 1986 Annual Tropical Cyclone Rep., JTWC, 191 pp. Sadler, J. C., 1975: The upper circulation over the global tropics. [NTIS AD A184082.] UHMET Publ. 75-05, Department of Meteorology, University , 1987: Synoptic discussions of tropical cyclones Freda, Gerald of Hawaii, Honolulu, HI, 35 pp. and Holly. 1987 Annual Tropical Cyclone Rep., JTWC, 213 pp. , M. A. Lander, A. M. Hori, and L. K. Oda, 1987: Tropical [NTIS AD A191883.] marine climate atlas. Vol. II: Paci®c Ocean. UHMET Publ. 87- , 1989a: Synoptic discussions of tropical cyclones Nancy, Owen 02, Department of Meteorology, University of Hawaii, Hono- and Peggy. 1989 Annual Tropical Cyclone Rep., JTWC, 253 pp. lulu, HI, 27 pp. [NTIS AD A232469.] Sandgathe, S. E., 1987: Opportunities for tropical cyclone motion , 1989b: Synoptic discussions of tropical cyclones Sarah, Tip research in the northwest Paci®c region. Naval Postgraduate and Vera. 1989 Annual Tropical Cyclone Rep., JTWC, 253 pp. School Paper NPS-63-87-006, Department of Meteorology, Na- [NTIS AD A232469.] val Postgraduate School, Monterey, CA, 36 pp. , 1990: Synoptic discussions of tropical cyclones Steve, Tasha Shanghai Typhoon Institute, 1990: Climatological Atlas for North- and Vernon. 1990 Annual Tropical Cyclone Rep., JTWC, 278 west Paci®c Tropical Cyclones. China Meteorological Press, pp. [NTIS AD A239910.] 316 pp. , 1991: Direction of approach and movement of tropical cy- WMO, 1993: Global guide to tropical cyclone forecasting. WMO clones affecting Guam. Tropical cyclones affecting Guam. Tech. Document 560, Rep. TCP-31, Secretariat of the World NOCC/JTWC Tech. Note 91±2, JTWC, 45 pp. Meteorological Organization, Geneva, Switzerland, 354 pp.

/ams 3q04 0202 Mp 186 Thursday May 23 04:06 PM AMS: Forecasting (June 96) 0202

Unauthenticated | Downloaded 10/06/21 09:32 PM UTC