15 JULY 2020 F E N G E T A L . 5919

Reexamination of the Climatology and Variability of the Northwest Pacific Monsoon Trough Using a Daily Index

TAO FENG College of Oceanography, Hohai University, Nanjing, and Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai), Zhuhai, China

XIU-QUN YANG AND XUGUANG SUN China Meteorological Administration–Nanjing University Joint Laboratory for Climate Prediction Studies, and Jiangsu Collaborative Innovation Center of Climate Change, School of Atmospheric Sciences, Nanjing University, Nanjing, China

DEJIAN YANG College of Oceanography, Hohai University, Nanjing, China

CUIJIAO CHU China Meteorological Administration–Nanjing University Joint Laboratory for Climate Prediction Studies, and Jiangsu Collaborative Innovation Center of Climate Change, School of Atmospheric Sciences, Nanjing University, Nanjing, China

(Manuscript received 24 June 2019, in final form 5 March 2020)

ABSTRACT

This study developed a daily index to represent the northwest Pacific monsoon trough using westerly re- lated cyclonic vorticity after removing tropical cyclones (TCs) from the reanalysis dataset. This index suffi- ciently captures the spatial and temporal variations in the monsoon trough. The use of this daily index revealed new features in the monsoon trough, including daily statistical characteristics, the active period over a year, and the main periodicity. A monsoon trough can be identified as active when the daily index is 2 2 greater than 2.0 3 10 4 s 1. Active monsoon troughs occur during half of the summertime, and these is no monsoon trough on one-third of days, with the remaining days categorized as inactive. The most active month is August, in which approximately 20 days exhibit an active monsoon trough. Using this index, an active monsoon trough period, which is related to vigorous TC activity, was determined by identifying the estab- lishment and decay dates for each year from 1979 to 2016. During most years, the active monsoon trough is established in mid-July and decays in late October, persisting for 3–5 months during the boreal summer. Moreover, spectral and wavelet analyses demonstrated the presence of intraseasonal, interannual, and in- terdecadal variabilities in the monsoon trough. The dominant periodicity for the interannual variability varied from 1.5 to 4 years in different decades. The relationship between the monsoon trough and TCs is also revealed using this index, showing that approximately 60% of TC formations were related to an active monsoon trough.

1. Introduction monsoon trough is sometimes called a monsoon de- pression as it displays a region with low sea level The lower-tropospheric monsoon trough in the west- pressure. The trough is the main region in which most ern North Pacific is a vital component of the East of the monsoon rainfall is induced over the northwest Asian summer monsoon system. The monsoon trough Pacific during boreal summer due to enhanced con- exhibits a planetary-scale cyclonic flow, which consists vective activity (e.g., Wang and LinHo 2002). The of southwesterlies from the Southern Hemisphere and trough region is also the main development region for trade easterlies from the tropical central Pacific. The tropical cyclones (TCs) due to its favorable dynamic and thermodynamic environments (e.g., Gray 1968; Corresponding author: Tao Feng, [email protected] Harr and Chan 2005).

DOI: 10.1175/JCLI-D-19-0459.1 Ó 2020 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses). Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 5920 JOURNAL OF CLIMATE VOLUME 33

The climatology and variability of the northwest In most of these studies, identification of a monsoon Pacific monsoon trough have been studied for decades. trough relied on subjective analysis of the lower- As revealed in previous studies (Ueda and Yasunari tropospheric circulation pattern on a synoptic map. 1996; Wu and Wang 2001; Wu 2002), the monsoon Some studies have attempted to produce an objective trough is first observed over the South China Sea during index to represent the intensity and location of a mon- mid-May. The eastward migration occurs in mid-June soon trough. Gao et al. (2011) defined intensity and lo- when the monsoon rainfall extends to the southwestern cation indices to investigate multi-TC formation events Philippine Sea. The northwest Pacific monsoon trough in association with a monsoon trough. Wu et al. (2012) usually forms near mid-July, and disappears during late used the June–November mean positive relative vor- October and is accompanied by the eastward advance ticity averaged from 58 to 208N to represent the strength and westward withdrawal of the monsoonal southwest- of a monsoon trough. This index was subsequently used erlies. Due to the complexity of the East Asian summer in several studies (Wu et al. 2015; Huangfu et al. 2017; monsoon system, the monsoon trough varies over mul- C. Li et al. 2017). Cao et al. (2014a) used a region- tiple time scales (Harr and Wu 2011). Over synoptic to averaged OLR instead of relative vorticity to reveal the intraseasonal scales, the monsoon trough interacts with interannual variation in the monsoon trough. In addi- equatorial waves (Takayabu and Nitta 1993; Dickinson tion, some East Asian summer monsoon indices, which and Molinari 2002; Molinari et al. 2007; Feng et al. 2016), are often expressed by a north–south gradient of zonal cross-equatorial surges (Love 1985; Feng et al. 2017), winds, actually measure the strength of the northwest the Madden–Julian oscillation (MJO) (Maloney and Pacific monsoon trough (e.g., Wang and Fan 1999; Wang Dickinson 2003; Gao and Li 2011; Cao et al. 2013), etc. et al. 2008). Most of these indices are applicable when Over a longer time scale, the interannual variation in the average period is longer than one month. If these the monsoon trough is related to the El Niño–Southern indices are applied to define the monsoon trough on a Oscillation (ENSO) system (e.g., Wang and Chan 2002; daily basis, several critical drawbacks are encountered. Lau and Nath 2006; Wu et al. 2012; Cao et al. 2014a)or Molinari and Vollaro (2013) summarized these draw- tropical tropospheric biennial oscillation of the mon- backs as follows: 1) too much influence from TCs in the soon system (Tomita and Yasunari 1996; Chang and Li vorticity fields, 2) noncontiguous vorticity regions re- 2000). An interdecadal shift in the monsoon trough lated to higher-frequency perturbations, and 3) a lack of affecting the interdecadal change in TC frequencies in separability from tropical low-frequency oscillations the late 1990s was also revealed by Huangfu et al. because the variability in the monsoon trough overlaps (2017). The multiple time scale variations in the mon- with various oscillation phenomena at similar frequen- soon trough strongly modulated TC formations (Lander cies in the tropics (e.g., the MJO). To avoid these 1994, 1996; Holland 1995; Wu et al. 2012; Cao et al. drawbacks, they suggested that the monsoon trough is a 2014a; Feng et al. 2014; Zong and Wu 2015a). climatology phenomenon and should be defined using However, the definition of a monsoon trough differs multimonth mean vorticity. between these studies, resulting in difficulties in under- Although the climatology and variability of the standing the climatology and variability of the northwest monsoon trough are described in the literature, some of Pacific monsoon trough. Some studies identified a their basic features remain vague without a reliable monsoon trough as a zonally elongated sheared region definition of the monsoon trough on a daily basis. For with northwesterlies in its southern part and easterlies in instance, daily weather maps show that the monsoon its northern part (e.g., Lander 1994, 1996; Holland 1995). trough does not always exist over the northwest Pacific Ritchie and Holland (1999) counted monsoon gyres, during boreal summer (not shown). What is the per- which are characterized by a near-circular cyclonic cir- centage of days when an active monsoon trough exists culation, as a part of a monsoon trough. In other studies over the western North Pacific? Sometimes we assess the (e.g., Chen and Chang 1980), the mei-yu front during the establishment of the monsoon trough by analyzing presummer and early summer periods from mid-May to the monsoon trough line from a weather map. What are mid-June is also counted as an early monsoon trough. the criteria for identifying the establishment of the mon- Molinari and Vollaro (2013) discussed several defini- soon trough? Moreover, what are the criteria for deter- tions and defined a monsoon trough as follows: regions mining the TC genesis capacity in a monsoon trough? with a positive relative vorticity averaged over at least Previous studies mentioned that monsoon trough pos- several months. However, Molinari and Vollaro (2013) sesses intraseasonal, interannual, or interdecadal vari- identified those positive relative vorticity regions on the abilities. Are these periods statistically significant? Can equator induced by trade northeasterlies as a monsoon the monsoon trough have other significant periods? At trough when the boreal winter monsoon prevails. the interannual time scale, the monsoon trough can be

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 15 JULY 2020 F E N G E T A L . 5921 categorized as active or inactive using the annual aver- 2008; Li et al. 2014), a daily genesis rate (DGR) of TCs is aged index (Wu et al. 2012). How can we define activity or defined as the number of TC genesis divided by the inactivity on a daily basis? A daily index, which repre- number of total days. A statistical test is then employed sents daily variations in a monsoon trough, is of great help to examine whether the DGR significantly differs from in addressing these issues. the climatology following the definition by Li and Zhou In this study, a feasible index that represents the daily (2013). The daily accumulated cyclone energy (ACE) is variation in the northwest Pacific monsoon trough was also calculated as the sum of the squares of the 6-hourly developed. The climatology and variability of the maximum sustained wind for all TCs in the northwest monsoon trough are investigated using this daily index Pacific basin for each day (Bell et al. 2000). The DGR to address the above-mentioned remaining issues. The and ACE are applied to show how this index can rep- rest of this paper is organized as follows: In section 2, resent the possible interactions between the monsoon the dataset and methodology are described. In section 3, trough and TCs. the definition of a daily monsoon trough index is pro- vided. In section 4, the climatology and variability of the 3. Development of a daily monsoon trough index monsoon trough is examined using this index to address a. How to define a monsoon trough conceptually some statistical results. Finally, in section 5, the con- clusions and discussion of this study are presented. As defined in the Glossary of Meteorology (Glickman 2000; AMS 2019a), a monsoon trough is ‘‘the line in a weather map showing the locations of relatively mini- 2. Datasets and methodology mum sea level pressure in a monsoon region.’’ We agree The daily atmospheric fields on a global grid are obtained with Molinari and Vollaro (2013) that this definition is from the National Centers for Environmental Prediction– problematic. We think we can refer to the definition of Department of Energy (NCEP–DOE) AMIP-II Reanalysis the term trough to develop a more appropriate defini- (Kanamitsu et al. 2002; NCEP et al. 2000) from 1979 to tion of monsoon trough. The Glossary of Meteorology 2016. The data have a resolution of 2.5832.58 and 17 (Glickman 2000; AMS 2019b) defines a trough as ‘‘an pressure levels. Extended reconstructed sea surface elongated area of relatively low atmospheric pressure; temperature (ERSST) v5 data (Huang et al. 2017a,b) are the opposite of a ridge.’’ We noticed the explanation used to provide the surface thermal state of the Pacific under the term trough saying ‘‘the axis of a trough is the Ocean. The El Niño index is calculated from the SST trough line.’’ Apparently, the original term monsoon anomalies averaged during the preceding December– trough in the Glossary of Meteorology is the definition February (DJF) in the Niño-3.4 region (58N–58S, 1208– of a monsoon trough line. Amending the ‘‘line’’ issue, 1708W). The DJF mean SST is used because El Niño the definition of monsoon trough should be revised to exerts a delayed impact on the western North Pacific ‘‘an elongated area in a weather map showing relatively summer monsoon (Zhang et al. 1996; Wang et al. 2000; low atmospheric pressure in a monsoon region.’’ However, T. Li et al. 2017). The velocity potential MJO (VPM) this definition would recognize an equatorial trough, which index produced by Ventrice et al. (2013), which is is induced by southward cross-equatorial flows in boreal available at https://www.esrl.noaa.gov/psd/mjo/mjoindex/, winter, as a monsoon trough. Following previous studies, is used to track MJO activity. Thus, the relationship be- the northwest Pacific monsoon trough should consist of tween the establishment of the monsoon trough and the low-level easterly trade winds and the westerly monsoonal MJO can be discussed. The VPM index is derived from the wind (e.g., Chang and Chen 1995; Wang and Xu 1997; widely used real-time multivariate MJO (RMM) index Chan and Evans 2002; Wu 2002; Chen and Huang 2008; (Wheeler and Hendon 2004)byreplacingtheinputfrom Wu et al. 2012). Finally, the term monsoon trough in this the OLR with the 200-hPa velocity potential. Compared to paperisdefinedasan elongated area in a weather map the RMM, the VPM index better discriminates MJO- showing relatively low atmospheric pressure consisting of associated signals during boreal summer (Ventrice et al. easterly trade winds and westerly monsoonal winds. 2013; Kiladis et al. 2014). The horizontal winds at 850 hPa and sea level pressure The genesis and intensity information of TCs are during boreal summer averaged from 1979 to 2016 are adopted from the International Best Track Archive for shown in Fig. 1a. Climatologically, the monsoon trough, Climate Stewardship (IBTrACS) version 10 (Knapp depicted by the red trough line between monsoonal et al. 2010; NOAA/NCDC 2017). The first occurrence of southwesterlies to the south and easterlies to the north, the maximum near-surface wind speed reaching 35 kt overlaps well with regions with low sea level pressure. 2 (1 kt ’ 0.51 m s 1) is defined as the genesis time. The climatology pattern of the relative vorticity is similar Following previous studies (Hall et al. 2001; Kim et al. to that of sea level pressure, with a spatial correlation

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 5922 JOURNAL OF CLIMATE VOLUME 33

monsoon trough is a synoptic phenomenon that can be identified from a weather map illustrating the combined contribution of various wave-like disturbances (such as the MJO), the seasonal evolution of the atmospheric general circulation, and other variations with longer time scales (interannual, interdecadal, etc.). Consequently, the monsoon trough can be defined from a daily weather map without separation with the MJO and tropical waves. Therefore, we can unify the definition of the terms trough and monsoon trough such that the latter is the trough in the monsoon region. However, TC signals—different from tropical waves, which are identified from their dispersion relations—are not counted as a part of the monsoon trough, which will be discussed next. b. Extracting trough signals The monsoon trough creates a favorable environment for TC genesis. After genesis, TC circulation has a prominent impact on the wind fields (Hsu et al. 2008; Bi et al. 2015; Cao et al. 2018), which interferes with the identification of the monsoon trough. The evolution of the monsoon trough after TC formation has been less discussed because the signal of the monsoon trough overlaps the TC signal. In this subsection, we show how the TC circulation obscures the identification of a mon- soon trough and how to extract trough signals from a reanalysis dataset. FIG. 1. Long-term mean (a) sea level pressure (shading; hPa) and 2 2 Several algorithms can estimate the contribution of (b) relative vorticity (shading; 10 5 s 1) averaged from May to October for each year from 1979 to 2016. The wind vector is the TCs to environmental fields (e.g., Frank and Roundy horizontal wind field at 850 hPa. The thick red line denotes the axis 2006; Kurihara et al. 1993, 1995; Chen et al. 2010; of the monsoon trough. Schreck et al. 2011). However, most TC circulations are asymmetric and embedded within noncircular monsoon at 20.61 (Fig. 1b). Previous studies have also demon- circulation; therefore, the horizontal boundary of a TC strated the viability of using relative vorticity to define the should be carefully determined to avoid overestimating northwest Pacific monsoon trough (Wu et al. 2012; the impact of the TC. Under this consideration, the al- Molinari and Vollaro 2013). In this study, we show the gorithm developed by Kurihara et al. (1993, 1995)is development of a daily index based on relative vorticity employed to remove TC components from the original and zonal winds. fields. This method objectively identifies an asymmetric Another issue is the conceptual separation between TC without monsoonal flow following this procedure: the monsoon trough and other wave-like perturbations, first, the wind fields are split into a basic field and a such as the MJO. Molinari and Vollaro (2013) define the disturbance field using a filtering operator; then, the TC monsoon trough as a climatological feature to avoid the boundary is determined from the disturbance field by separation between the monsoon trough and other examining the profile of the tangential wind; finally, the wave-like perturbations. This is a feasible definition, but TC component is extracted using an empirically deter- it may be not the best definition in terminology. Why mined function in association with the TC radius. Thus, does monsoon trough only refer to a climatology back- smoother large-scale environmental fields can be extracted ground but a trough can be identified from a daily syn- after removing TC signals. Figure 2 shows a case in which optic map? Why do we need a particular definition of the this TC removal method is applied. Two strong TCs appear term trough in the monsoon region? Conceptually, the with strong rotational flows around the TC centers over the separation between the monsoon trough and the MJO tropical western North Pacific Ocean on 22 August 2004 (or tropical waves) is not necessary because the former is (Fig. 2a). Whether a monsoon trough exists is difficult to defined based on synoptic analysis and the latter is de- determine from the lower-tropospheric winds. By remov- fined by the dispersion relation. We suggest that a ing these two TC components (Fig. 2b)fromtheoriginal

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 15 JULY 2020 F E N G E T A L . 5923

25 21 FIG. 2. The horizontal wind field (vectors) and relative vorticity (shading; 10 s ) at 850 hPa on 22 Aug 2004: (a) original fields, (b) TC components, (c) fields after removing TC components, (d) perturbations by spatial smoothing, (e) fields after removing TC and perturbation components, and (f) fields by applying a 10-day low-pass filter to the original data. Typhoon symbols denote the centers of TCs ‘‘Chaba’’ and ‘‘Aere’’ in the best track data. Orange circles denote the boundaries of each TC, which are determined by the TC removal algorithm (Kurihara et al. 1993, 1995). wind fields (Fig. 2a), a northwest–southeast elongated trough. Thus, a smoothed circulation is obtained for monsoon trough can be easily distinguished from the further identification of a monsoon trough (Fig. 2e). contiguous vorticity regions (Fig. 2c). The reliability of the current trough-isolation method After removing the TC components, a nine-point lo- was tested for all TC days from 1979 to 2016. This cal smoothing is applied inside the region with positive method is useful to isolate a monsoon trough from vorticity. This smoothing method is used to eliminate complex lower-tropospheric circulations. Figure 2f shows high-frequency noise (Fig. 2d) in the trough region the result of simply applying a 10-day low-pass filter to the without changing the circulation patterns outside the original fields, which was used in some previous studies to

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 5924 JOURNAL OF CLIMATE VOLUME 33

higher latitudes other than the monsoon trough. After removing the TC signals, the structure of the mean horizontal winds (Fig. 3b) is similar to that of the orig- inal winds (Fig. 3a). However, the standard deviation of the relative vorticity is greatly reduced, and the peak variance is confined to the monsoon trough region. Since TC signals are removed from the original fields, the relative vorticity over the western North Pacific can represent the variance in the monsoon trough. c. Definition of a daily index The western North Pacific monsoon trough is char- acterized by its large-scale cyclonic shear with westerlies to the south and easterlies to the north (Fig. 1). Thus, a daily monsoon trough index is calculated using the rel- ative vorticity after applying the above two-step pre- processing approach:

N 5 å z IMT 850 . (1) n51

In Eq. (1), z850 is the relative vorticity at 850 hPa, and N is the grid point that satisfies the following two criteria: 1) westerlies exist (u . 0) and 2) the relative vorticity exceeds a certain threshold b inside the tropical north- FIG. 3. Long-term mean horizontal wind field at 850 hPa (vector) 25 21 western Pacific domain (58–258N, 1208E–1808), that and standard deviation of the relative vorticity (shading; 10 s ) z . averaged from May to October for each year from 1979 to 2016: is, 850 b. The selection of this extended area is due to (a) original data and (b) data after removing TC and perturbation the daily movement of the monsoon trough over the components. The blue dashed rectangle denotes the region se- northwest Pacific. For example, Lander (1996) shows a lected for the calculation of the monsoon trough index. reverse-oriented monsoon trough that extends north- ward to approximately 258N. Wu et al. (2012) show that extract environmental fields (Feng et al. 2014; Zong and the monsoon trough can extend to the date line during 2 2 Wu 2015a). Although a monsoon trough can be recog- some months. The threshold b is set to 0.1 3 10 5 s 1. nized from the spread of positive vorticity, two cyclonic The choice of the threshold b will be discussed later. The vortices with centers at 248N, 1248Eand188N, 1398E sum of the relative vorticity is used rather than the av- appear in the trough region, complicating identification of erage because the former provides information regard- the monsoon trough line (Fig. 2f). These two cyclonic ing the horizontal extension of the monsoon trough. vortices are attributed to the contribution of TC signals to To simplify, IMT is the sum of the relative vorticity at the 10–25-day variances (Hsu et al. 2008; Aiyyer et al. each grid point where westerlies exist with a relative 2 2 2012) and cannot be entirely removed using a 10-day low- vorticity at 850 hPa greater than 0.1 3 10 5 s 1. This pass filter. Instead, using a 20-day low-pass filter, the wind value measures the strength of the cyclonic sheared flow pattern is similar to Fig. 2e, in which the TC signal is induced by westerlies over the tropical northwest mostly removed (not shown). Pacific. In the following, we will discuss other con- On the other hand, the daily variance in the relative siderations in the choice of the criteria. vorticity is substantially impacted by tropical cyclone d. Choice of the criteria tracks. Figure 3a shows the horizontal wind, relative vorticity, and standard deviation at 850 hPa without re- The chosen criterion of u . 0 is derived from the moving TCs, which are averaged from May to October seasonal cycle. The most problematic issue in develop- from 1979 to 2016. The largest standard deviation of ing a daily monsoon trough index is that the seasonal relative vorticity occurs in the subtropical regions near cycle of the monsoon trough is misrepresented by the 258N(Fig. 3a). If a daily index is directly calculated by relative vorticity in the lower troposphere. Another using the original relative vorticity without removing simple daily index (marked as IVR), which is defined by TCs, the index primarily represents the variability in the average of the positive relative vorticity at each grid

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 15 JULY 2020 F E N G E T A L . 5925

25 21 FIG. 4. The long-term mean (1979–2016) Pacific relative vorticity (10 s ) and the mean horizontal winds (vectors) at 850 hPa in (a) boreal summer, (b) boreal winter, (c) May, (d) June, (e) July, (f) August, (g) September, and (h) October. The thick dashed line denotes the axis of the monsoon trough. Blue dots denote grid points with 25 21 westerlies and positive vorticity greater than 0.1 3 10 s . The averaged index IMT is shown in the upper right of 2 2 each panel (10 4 s 1). point over the northwest Pacific (58–258N, 1208E–1808), (Figs. 4c–h) has been revealed in various studies (e.g., will be used to illustrate the interference of easterly re- Lander 1996; Wu and Wang 2001; Wang and LinHo lated points in representing the seasonal evolution of the 2002; Wu 2002; Chen et al. 2004; Wang et al. 2004). monsoon trough. However, this seasonal evolution is misrepresented by

As depicted in Fig. 4, a monsoon trough, which is the index IVR (Fig. 5). Two false peaks appear in the represented by the artificially analyzed trough line, ap- seasonal cycle of IVR, which occur during November and pears over the tropical northwest Pacific during boreal early January when winter-monsoonal northeasterlies summer (Fig. 4a) and vanishes during boreal winter prevail. In boreal winter, trade easterlies induce strong (Fig. 4b). The seasonal cycle of the monsoon trough cyclonic vorticity near the equatorial region with

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 5926 JOURNAL OF CLIMATE VOLUME 33

24 21 25 21 FIG. 5. The daily indices IMT (red curve; 10 s ) and IVR (blue curve; 10 s ) from January to December averaged from 1979 to 2016. The histogram denotes the monthly mean 25 21 using different thresholds from 0 to 0.20 (10 s ). The IMT and IVR data were smoothed using a 9-point running averaging. The raw data are shown as gray dashed curves. southward cross-equatorial flows (Fig. 4b). These re- Pacific summer monsoon. The better capability of IMT gions with cyclonic vorticity are overlapped with to characterize the seasonal cycle of the monsoon easterlies and are not related to the western North Pacific trough is ascribed to the elimination of the grid points summer monsoon. If these grid points with easterlies are with easterlies. This leads to the first criterion (u . 0) in utilized to calculate the monsoon trough index, the sea- the selection of points. sonal cycle will be distorted, as depicted by IVR. Practically, The second criterion is the threshold b. In previous 2 2 the period of the western North Pacific summer monsoon, studies, b was set to 0.0 3 10 5 s 1 so that all grid points which varies from year to year, should be predefined be- with positive relative vorticity were employed to calcu- fore calculating IVR to avoid the misrepresentation of the late the index for the seasonal mean or climatological evolution of the monsoon trough when the monsoon mean (Wu et al. 2012; Molinari and Vollaro 2013). shifted to the winter monsoon. Considering that the spatial distribution of vorticity is

The index IMT is applicable in an entire year. As much noisier in a daily map, a nonzero value should be shown in Fig. 4, the grid points with positive vorticity assigned to b to reduce the number of discrete grid 2 2 and westerlies nearly perfectly reproduce the evolution points. Values of b from 0.0 to 0.2 3 10 5 s 1, with an 2 2 of the monsoon trough. As illustrated in Fig. 5, the index increment of 0.02 3 10 5 s 1, were tested by creating

IMT remains low during January and May. After June, monthly averaged IMT values. As shown in Fig. 5, the the value of IMT increases. The abrupt increase in the seasonal cycle of IMT is not sensitive to the selection of IMT value occurs around mid-July, which indicates a the threshold b. Finally, the threshold b is set to 2 2 change in the large-scale background circulation over 0.1 3 10 5 s 1. the western North Pacific. This date is consistent with the sudden jump in the monsoon trough from the South 4. Climatology and variability China Sea to the Pacific Ocean. The intensity of the a. Climatology monsoon trough reached its maximum in early August and gradually decreased during September and October Using the daily index IMT, the occurrence, strength, as revealed by the variation in IMT. In late October, the and spatial distribution of the monsoon trough can be value of IMT sharply decreased. This result is consistent examined statistically. Figure 6 shows the probability with the onset and withdrawal of the western North distribution of the daily IMT value during boreal sum- 2 2 Pacific summer monsoon (Wang and LinHo 2002). mer. The mean value is 2.50 3 10 4 s 1, with a standard 24 21 Therefore, the new index IMT, which only utilizes deviation of 2.34 3 10 s . The distribution density westerly-related grid points, sufficiently captures this gradually decreases with the increase in the value of IMT. seasonal migration of the monsoon trough as the The values of one-third of the samples are less than 1.0 3 2 2 westerlies are an essential indicator of the northwest 10 4 s 1, and the values of another one-third are greater

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 15 JULY 2020 F E N G E T A L . 5927

FIG. 6. (a) Probability distribution of the monsoon trough index FIG. 7. Monthly averaged values (bar) of (a) the monsoon trough 24 21 3 IMT (10 s ) during boreal summer. The bin interval is 0.5 index IMT and (b) the occurrence frequency of an active monsoon 2 2 10 4 s 1. (b) Probability distribution of occurrence frequency of trough. The red line denotes the range of 61.0 standard deviation. active monsoon trough in each month during boreal summer. The bin interval is 1. Black dashed lines denote the values of the 10th, featuring a shear line between easterly trade winds in 33rd, 50th, 67th, and 90th percentiles. the north and monsoonal southwest flow in the south. Although such analysis of the trough axis remains 2 2 2 2 than 3.0 3 10 4 s 1. The median value is 1.88 3 10 4 s 1, subjective, the results are used to evaluate the mon- which divides the entire boreal summer into more-active soon trough index IMT. Figures 8a and 8b show com- days and more-inactive days based on the state of the posite maps when the daily value of IMT is equal to 2 2 2 2 monsoon trough. Furthermore, we define the value of 0.0 3 10 4 s 1 and less than 1.0 3 10 4 s 1,respec- 24 21 24 21 2.0 3 10 s , approximately representing the 52nd per- tively. When IMT is less than 1.0 3 10 s , easterlies centile value, as the criterion for determining whether an prevail over the entire tropical western North Pacific. active monsoon trough exists over the northwest Pacific. No monsoon trough can be identified in the composite The occurrence frequency of the active monsoon map, which accounts for 32.57% of the total days. 24 trough in a month is shown in Fig. 6b. The occurrence When the daily IMT value is between 1.0 3 10 and 2 2 frequency of an active monsoon trough in a month can 2.0 3 10 4 s 1, the composition shows that a weak be divided into the following three categories based on monsoon trough exists over the Philippine islands the percentage density distribution: lower tertile, middle (Fig. 8c). When the daily IMT value is greater than 2 2 tertile, and upper tertile. The value of the upper tertile is 2.0 3 10 4 s 1, the monsoon trough extends eastward, 19 days in a month; thus, the month is considered active with its eastern end reaching approximately 1508E if the monsoon trough exists over the western North (Figs. 8d,e). Figure 8f shows the pattern of extreme Pacific for more than 19 days in a month. The seasonal cases, which account for 14.47% of the total samples. cycle and standard deviation of the occurrence fre- Here, the daily IMT value is greater than 5.0 3 2 2 quency in each month (Fig. 7b) are similar to those of 10 4 s 1, which exceeds the value of one standard the monthly averaged IMT value (Fig. 7a). deviation. These cases represent abnormally intense How well do these statistical results match the monsoon troughs. An abnormally strengthened monsoon in situ distribution and intensity of the monsoon trough may be associated with the westerly phase of the trough? Usually, a monsoon trough is recognized MJO (Maloney and Hartmann 2001; Hogsett and Zhang from a weather map by analyzing the trough axis 2010; Molinari and Vollaro 2017), favorable interannual

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 5928 JOURNAL OF CLIMATE VOLUME 33

25 21 FIG. 8. Composited relative vorticity (shaded; 10 s ) and the mean horizontal winds (vectors) at 850 hPa for

cases (a) IMT 5 0, (b) 0 , IMT , 1, (c) 1 , IMT , 2, (d) 2 , IMT , 3, (e) 3 , IMT , 5, and (f) IMT . 5. The units for 24 21 25 21 IMT are 10 s . Blue dots denote grid points with westerlies and positive vorticity greater than 0.1 3 10 s . and interdecadal background (Chou et al. 2003; Wu et al. 2004), the active period of the monsoon trough can be 2012; Huangfu et al. 2017), and other high-frequency determined by identifying the establishment and decay fluctuating phenomena. dates for each year from 1979 to 2016. An active monsoon Linking the circulation pattern of a monsoon trough trough period should be characterized by firm estab- to the daily index can help investigate the climatology. lishment of a steady circulation pattern similar to the Statistically, there are no monsoon troughs in one- monsoon trough for at least several weeks. As the third of the days during boreal summer (Figs. 8a,b). index IMT corresponds well with the daily advance of Approximately one-fifth of the days exhibit a weak the monsoon trough, some objective criteria are pro- monsoon trough confined near the Philippine islands duced to define the establishment of the monsoon (Fig. 8c). These samples are categorized as inactive. trough. The establishment of the monsoon trough is Active monsoon troughs appear in only approximately defined by the first day after 1 May (as we focused on one-half of the summer days. However, if such statistical boreal summer) that satisfies the following three cri- 24 21 analysis is performed by analyzing daily weather maps teria: 1) the IMT value is greater than 2.0 3 10 s on using a dataset without the removal of TCs, the percent- the establishment day, 2) the IMT value is greater than 2 2 age with active monsoon troughs may be greatly reduced, or equal to 2.0 3 10 4 s 1 on the following four con- as the trough signal is overridden by TCs in daily maps. secutive days after establishment, and 3) the IMT value 2 2 is greater than 2.0 3 10 4 s 1 on at least 20 of the b. Establishment and decay of the active monsoon subsequent 30 days. trough The first criterion is used to identify to active monsoon Similar to the consideration of the onset date of the troughs, as shown in Figs. 6a and 8. The third criterion South China Sea summer monsoon (e.g., Wang et al. mainly refers to the upper tertile value of the monthly

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 15 JULY 2020 F E N G E T A L . 5929 occurrence of an active monsoon trough (Fig. 6b), which ensures that the northwestern Pacific monsoon trough is firmly established. The conditions are also tested for a duration range of 15–25 days out of 30 days for this criterion. Although the establishment dates in several years change with the change in the criterion (using 15 or 25 days instead of 20 days), the results do not change in more than 75% of the years. Therefore, the active monsoon trough has a firmly established circulation background that is not sensitive to the definition in most years. The second criterion is chosen to avoid the mis- identification of the establishment date. In some cases, such as an isolated active day followed by 9 inactive days and 20 additional active days, the first day is rejected as being included in the active period by this criterion. Similar to the establishment date of a monsoon trough, a decay date is also defined to represent the end of the active monsoon trough period. The decay date of a monsoon trough is defined by the following criteria: 1) the 24 21 IMT value is greater than 2.0 3 10 s but less than 2.0 3 24 21 10 s on the following day, 2) the IMT value is greater 2 2 than 2.0 3 10 4 s 1 on at least 20 of the previous 30 days, and 3) if one specific day is satisfied by the above two FIG. 9. Variation in the monsoon trough index in (a) 2002 and criteria for the last time in a year, then the following day is (b) 2016. TC symbols denote TC formation on that day. Dashed lines denote the establishment or decay date of the active monsoon defined as the decay date. These criteria were chosen trough. The IMT and IVR data were smoothed using a 15-point based on those for the establishment date. running averaging. The raw data are shown as gray dashed curves. Figures 9a and 9b show the establishment and decay dates of the active monsoon trough in 2002 and 2016. active monsoon trough persists for 80 to 160 days over the The active monsoon trough periods were determined western North Pacific, which is approximately three to from 23 June to 23 October in 2002 and from 2 August to five months (Fig. 10c). Over six special years (1983, 1988, 11 November in 2016. These two cases imply that an 1995, 1998, 2008, and 2010), no active period can be de- active monsoon trough episode represents a period termined, as the monsoon trough remains inactive rather dominated by a persistent active monsoon trough (Fig. 9a) than absent throughout the boreal summer (not shown). or by the occurrence of an alternative establish–break– As revealed by the change in the Niño-3.4 index establish trough (Fig. 9b). (Table 1), five of the six no-MT events occurred during Table 1 shows the establishment and decay dates the transition years from El NiñotoLaNiña(1983, during each year from 1979 to 2016. After division of the 1988, 1995, 1998, and 2010), implying that the ENSO months into three 10-day periods, the establishment and system induced an abnormally weak summer monsoon decay dates of the active monsoon trough are summa- trough in the northwest Pacific during the El NiñotoLa rized in Fig. 10. During most years, the monsoon trough Niña transition years (Chou et al. 2003). begins to be active during late June to mid-July (Fig. 10a). The activity of the monsoon trough is related to the Among these years, seven events were established in mid- activity of the MJO during boreal summer (Maloney July, which corresponds to the climatological establish- and Hartmann 2001; Maloney and Dickinson 2003; ment date. In a particular year (e.g., in 1980 or 2013) the Hogsett and Zhang 2010; Cao et al. 2014b). Figure 11 establishment of the monsoon trough occurs very late in presents an MJO phase diagram showing the MJO September. This does not mean that there is no monsoon phases on the establishment days from 1979 to 2016. A trough before these dates. It means the monsoon trough large cluster containing 14 cases (37%) in a total of 38 was not active until September in these years (not years appeared during MJO phase 5. Twelve cases shown). After establishment, the monsoon trough usually (32%) occurred during MJO phases 3 and 4. Only six decays during mid-October, late October, and early cases occurred during the other five MJO phases. When November (Fig. 10b). The trough may also decay very the MJO-related convection moves from the Indian early in September or August (e.g., 1999 and 2000) or Ocean to the Maritime Continent, anomalous westerlies persist until late December (1986). In most years, the advance from the South China Sea to the western North

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 5930 JOURNAL OF CLIMATE VOLUME 33

TABLE 1. Dates of monsoon trough establishment and decay and the Niño-3.4 index from 1979 to 2016. A dash (—) denotes that no active monsoon trough existed during the year. The last line shows the mean and standard deviation of the establishment and decay dates.

Year Establishment Decay Niño-3.4 index Year Establishment Decay Niño-3.4 index 1979 3 Aug 11 Dec 0.0 1998 — — 2.2 1980 2 Sep 3 Oct 0.6 1999 7 Jul 17 Aug 21.5 1981 26 Jul 28 Oct 20.3 2000 28 Aug 25 Sep 21.7 1982 15 Jun 26 Oct 0.0 2001 22 Jun 7 Nov 20.7 1983 — — 2.2 2002 23 Jun 23 Oct 20.1 1984 15 Jul 10 Nov 20.6 2003 15 May 3 Dec 0.9 1985 10 Jun 7 Sep 21.0 2004 13 Jun 11 Dec 0.4 1986 10 May 25 Dec 20.5 2005 19 Jul 21 Oct 0.6 1987 9 Jun 12 Oct 1.2 2006 5 Jul 12 Nov 20.8 1988 — — 0.8 2007 27 Aug 15 Oct 0.7 1989 25 Jul 12 Dec 21.7 2008 — — 21.6 1990 1 Jul 4 Dec 0.1 2009 29 May 30 Oct 20.8 1991 16 Jul 3 Dec 0.4 2010 — — 1.5 1992 10 Aug 29 Nov 1.7 2011 30 Jun 21 Sep 21.4 1993 19 Jul 31 Oct 0.1 2012 20 Jul 20 Oct 20.8 1994 4 Jul 26 Nov 0.1 2013 13 Sep 26 Oct 20.4 1995 — — 1.0 2014 28 Jun 24 Oct 20.4 1996 18 Jul 30 Sep 20.9 2015 20 Jun 25 Oct 0.6 1997 14 Jul 8 Nov 20.5 2016 2 Aug 11 Nov 2.5 Average 11 Jul 1 Nov Std dev (day) 29.9 29.9

Pacific, which is accompanied by the eastward retreat of (Chen and Sui 2010; Wang and Chen 2017). This result anomalous easterlies over the central Pacific. This event demonstrates that the monsoon trough exhibits signifi- favored the establishment of an active monsoon trough cant intraseasonal, interannual, and interdecadal vari- with enhanced convective activities over the western North abilities, which supports previous studies (Cao et al. Pacific. Another notable feature is that for most cases oc- 2014a,b; Wu et al. 2012; Huangfu et al. 2017). curring during MJO phases 3 and 4, such as those occurring A wavelet analysis shows a similar periodicity to that in 1986, 1987, 1979, 2002, and 2003, the MJO amplitude is in Fig. 12, where the monsoon trough exhibits significant slightly larger than that for those cases occurring dur- periods of approximately 2 and 10 years and a weaker ing phase 5, which may be ascribed to stronger westerly peak at approximately 50 days (Fig. 13). However, the 10- anomalies induced by a stronger MJO, thus inducing yr period is significant only after the early 1990s. This the early establishment of an active monsoon trough. finding is consistent with the interdecadal shift in the thermal conditions of the tropical Pacific Ocean, which c. Multiscale variability have been reported in a large number of papers (e.g., Liu

As the current index IMT is successively defined on et al. 2012; Gu and Philander 1997). In addition, the 2-yr daily time scale, spectral and wavelet analyses have been period, which represents the relationship between the applied to determine the dominant periodicity of the monsoon trough and ENSO, is significant during 1980–87, multiscale variabilities in the monsoon trough. Figure 12 1993–2000, and after 2002. However, from 1987 to 1993, shows the spectrum of IMT after removing the seasonal no significant 2-yr fluctuation can be found. Instead, a cycle from the original time series. There are three main significant 4-yr periodicity exists since 1979 and disap- periodicities in the variation of the monsoon trough. A pears after 2000. The 4-yr periodicity is identical to the significant periodicity is found at 1.5–4 years, with a peak period of the South China Sea summer monsoon during near 2.5 years, which is the most dominant period of the the same decades, which was revealed by Zhou and Chan monsoon trough. Another significant periodicity ap- (2007). The change in the dominant interannual period pears at approximately 10 years, indicating that an in- implies that an unstable relationship exists between the terdecadal shift exists in the monsoon trough. In the ENSO and the northwest Pacific monsoon trough. Such intraseasonal band, the spectrum shows several signifi- an unstable relationship may be ascribed to the inter- cant periodicities ranging from 10 to 60 days, indicating decadal shift of environmental backgrounds. that the monsoon trough is impacted by tropical intra- When investigating the variability on the interannual seasonal oscillations in various time scales (e.g., the time scale or longer, an annual index is more convenient MJO and the tropospheric quasi-biweekly oscillations) than a daily index. Following Wu et al. (2012), a simple

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 15 JULY 2020 F E N G E T A L . 5931

FIG. 11. An MJO phase diagram with the VPM index on the establishment day of the monsoon trough from 1979 to 2016. The numbers on the dots denote different years. The establishment date is shown in Table 1. Different colors denote the establish month from May to September.

missing. Therefore, an annual index should be produced by only considering those periods when the monsoon trough is active. Thus, the annual monsoon trough index

is calculated by averaging the daily index IMT during the active period of the monsoon trough (hereafter IMTA). In six years (1983, 1988, 1995, 1998, 2008, and 2010), no active monsoon trough can be identified. Therefore, the

index IMTA is zero in these six years. Figure 14 illustrates the interannual variation in the

standardized IMTA and IAVG. These two indices show similar variations with a correlation of 0.88 during most of the years from 1979 to 2016. Nevertheless, in some FIG. 10. Histogram of the (a) establishment date, (b) decay date, years, the indices differ considerably. For example, in and (c) duration in days. The ‘‘No MT’’ category denotes no active the year 2000, the standardized I was 20.19, monsoon trough period during a year. AVG whereas the standardized IMTA was 0.79. A negative IAVG implies that the monsoon trough is slightly weaker annual index (hereafter IAVG) can be calculated by av- than those during normal years. A positive IMTA indi- eraging the positive relative vorticity at 850 hPa in the cates that the monsoon trough is stronger than normal region 58–258N, 1208E–1808 from 1 June to 31 October after trough establishment. However, the active period of each year. This definition is based on a fixed period is quite short, lasting only 29 days during 2000 (Table 1). that approximately covers the boreal summer season. Remarkable differences between IMTA and IAVG can However, as is discussed in section 3, the establishment also be found in the years 1980 and 2013. The correlation date of the monsoon trough varies from year to year. If a coefficient reaches 20.32 between the IMTA and the fixed season is used for averaging, then the signal of the Niño 3.4 index, which exceeds the 95% significance active monsoon trough may be smoothed by those pe- level. In contrast, the correlation is only 20.17 between riods without an active monsoon trough. The intensity of the index IAVG and the Niño 3.4 index. The newly de- the monsoon trough may be weakened, and the infor- fined annual monsoon trough index IMTA correlates better mation regarding the establishment and decay dates is with the ENSO phases, implying that the ENSO-related

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 5932 JOURNAL OF CLIMATE VOLUME 33

late July, thus causing the abnormal TC formation in 2016 (Zhan et al. 2017; Chen and Wang 2018; Huangfu et al. 2018).

Figure 15a illustrates the statistics of the daily IMT value on each genesis day during boreal summer (June– November) from 1979 to 2016. A total of 493 (62.2%)

TCs formed when the daily index IMT was greater than 2 2 2 3 10 4 s 1. As discussed in section 4, a monsoon

trough is defined as active when the IMT value is greater 2 2 than 2 3 10 4 s 1, implying that approximately 60% of TC formations may be associated with an active mon- soon trough during boreal summer. This proportion is slightly less than that reported in previous studies, which suggested a percentage of 70% (Ritchie and Holland FIG. 12. The spectrum of the daily monsoon trough index IMT 1999; Molinari and Vollaro 2013; Feng et al. 2014; from 1979 to 2016. The seasonal cycle has been removed from the Yoshida and Ishikawa 2013; Zong and Wu 2015b). One original time series. The dotted lines are the spectrum of the 95% confidence level. of the possible reasons is the different methods for identifying an active monsoon trough. The present def- inition of an active monsoon trough is slightly stricter oceanic thermal state in winter impacts both the in- than those in previous studies because a constraint is tensity and active period of the monsoon trough. The placed on the intensity of the monsoon trough in which 24 21 index IMTA provides a proxy to investigate the effect the strength must be greater than 2 3 10 s . Another of ENSO on the establishment and intensity of the issue is that the TC formations outside the trough region monsoon trough. were counted as being in association with the monsoon trough if the monsoon trough is classified as active on d. Interaction with TCs that day. Nevertheless, the present study suggests that The establishment of the monsoon trough indicates over 60% of TC formation events are related to the the outbreak of the western North Pacific summer monsoon trough. There are 300 genesis events without monsoon. More importantly, the establishment also an active monsoon tough. These TC genesis events may marks the beginning of an active season of TCs over the be attributed to the tropical upper tropospheric trough western North Pacific, especially for multiple tropical (Briegel and Frank 1997) or tropical waves (Ritchie and cyclone events (Gao and Li 2011; Krouse and Sobel Holland 1999; Wang et al. 2012; Frank and Roundy 2010; Schenkel 2017). Figures 9a and 9b show that TC 2006; Chen and Chou 2014; Wu and Takahashi 2018). It formations were more active during the active periods in should be noted that the signals of tropical waves con- 2012 and 2016. The establishment of the active monsoon tribute to the identification of the monsoon trough, and trough in early August 2016 supports the findings that these waves may also affect TC genesis events with an the monsoon trough strengthened substantially after active monsoon trough.

FIG. 13. Wavelet power of the daily monsoon trough index IMT from 1979 to 2016. The seasonal cycle has been removed from the original time series. Stippling denotes anomalies significantly above the 95% confidence level.

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 15 JULY 2020 F E N G E T A L . 5933

FIG. 14. Variation in the annual monsoon trough indices IMTA (red) and IAVG (blue) from 1979 to 2016. The black curve is the Niño-3.4 index. The histogram denotes the duration of the active monsoon trough period during the year. The correlation coefficients between the monsoon trough indices and the Niño-3.4 index are shown in the upper-right corner, in which a star symbol denotes the value exceeding the 95% confidence level.

Figure 15a also shows the change in the DGR with the This result indicaes that the definition of the active mon- monsoon trough index IMT. The DGR is 8.3% when IMT soon trough period better captures the outburst of TC 2 2 is less than 2.0 3 10 4 s 1. Approximately eight TCs activities during an individual year. form every 100 days. This percentage is slightly less than The present index can also be used to investigate the the climatological DGR (11.4%). When IMT increased interaction between the monsoon trough and TCs due to 2 2 2 to between 2.0 3 10 4 and 4.0 3 10 4 s 1, the DGR increased to 13.0%, and the significance exceeded the 95% level. Further increases in the DGR were accom- panied by an increase in the IMT value. The peak DGR 24 (20.9%) appears when IMT is between 6.0 3 10 and 2 2 8.0 3 10 4 s 1, indicating that approximately one TC forms every five days. This finding indicates positive correlation between an active monsoon trough and TC genesis when the IMT value is greater than 2.0 3 2 2 10 4 s 1. This result also supports the selection of a 2 2 threshold of 2.0 3 10 4 s 1 to represent the active monsoon trough. Section 4 defined an active period of the monsoon trough by tracing the abrupt change in the index IMT. This active period can be used as a proxy to distinguish the active period of TC formations. Table 2 shows the DGR during the 15 days before and after the estab- lishment or decay date. Before the establishment of an active monsoon trough, the DGR is 10.2%. However, the DGR increased to 19.0% in the 15 days after the establishment date. Before decay, the DGR is 11.0%. Subsequently, the DGR decreased to 7.5% after the decay date of the trough. The changes in TC genesis FIG. 15. (a) The number of tropical cyclogenesis events (bar) and frequency before the establishment date and after the DGRs (solid curve) in association with the index IMT during June decay date are 18.8% and 23.5%, respectively. In and November from 1979 to 2016. The dashed line denotes the 95% confidence level. (b) The monsoon trough index IMT in association comparison, if the establishment and decay dates are set with the daily ACE on TC days during June and November from to the climatological dates (15 July and 31 October, re- 1979 to 2016. The solid red line is the linear regression of IMT as a spectively), the frequency changes are 14.9% and 21.5%. function of the daily ACE.

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 5934 JOURNAL OF CLIMATE VOLUME 33

TABLE 2. Genesis frequency (%) before establishment or decay of year, most of the establishments occur when the MJO the monsoon trough. The dates 15 Jul and 31 Oct refer to the cli- phase is between 3 and 5. After trough establishment, the matology dates for the establishment and decay of the monsoon DGR increased one time more than before establish- trough, respectively. Establishment and decay dates during individ- ual years are obtained from Table 1. The symbols * and ** denote ment. Therefore, the establishment of a monsoon trough significance at the 95% and 99% confidence levels, respectively. is a valid indicator for the beginning of enhanced TC activities. A stronger monsoon trough results in a greater 15 days 15 days Frequency capacity for tropical cyclogenesis. before after change The variability in the monsoon trough is also revealed 1 Varied 10.2 19.0 8.8** by applying spectral and wavelet analyses to the daily establishment dates Varied decay dates 11.0 7.5 23.5* monsoon trough index. There are three main periodic- 15 Jul 9.3 14.2 14.9* ities in the variation of monsoon trough, including 31 Oct 9.4 7.9 21.5 intraseasonal, interannual, and interdecadal variabil- ities. The peak period of the monsoon trough is ap- proximately 2.5 years, implying a close relationship the removal of TC signals. Figure 15b shows the index between the monsoon trough and ENSO. However, the

IMT varies with the daily ACE during TC days and dominant interannual variability varied from 1.5 to provides some clues regarding the influence of TCs on 4 years during different decades, which implies an un- the monsoon trough. The linear regression shows a sig- stable relationship between ENSO and the interdecadal nificant positive correlation between the daily IMT and variation of the monsoon trough. This issue could be ACE (which exceeds the 95% confidence level). It is further investigated in the future. The present results speculated that TCs support the maintenance of the support previous studies on the intraseasonal, interan- monsoon trough through upscale dynamic processes. nual, and interdecadal variabilities in the monsoon This result may differ from that of Ko et al. (2012), who trough (Cao et al. 2014a,b; Wu et al. 2012; Huangfu speculated that TCs neither feed energy to nor gain et al. 2017). energy from large-scale circulation. Some detailed cal- Additional analyses were also performed to show the culations should be carried out to validate this upscale advantage of this index. By defining the annual index effect in the future. based on the establishment and decay dates, the mon- soon trough correlates better with the ENSO than the index averaged during a fixed period from June to 5. Conclusions and discussion October. This implies that the ENSO influences not only In the present study, a daily index was developed to the intensity but also the active period of the monsoon illustrate the state of the northwest Pacific monsoon trough. The feedback between the monsoon trough and trough. A two-step preprocessing approach is employed TCs can also be represented using this daily index, al- on the original wind field by removing the TC circulation though this result needs further investigation. Previous and high-frequency noises. After preprocessing, the index studies have revealed that breaks in the Pacific summer is defined as the sum of the relative vorticity at each of monsoon (e.g., Xu and Lu 2015) may lead to an inactive the grid points where westerlies and remarkable cyclonic period monsoon trough. It is convenient to identify such circulation exist over the tropical western North Pacific. breaks based on this daily monsoon trough index to in- This daily index can sufficiently illustrate the seasonal vestigate the cause of an inactive trough. Furthermore, migration of the northwest Pacific monsoon trough. As occurrences of an abnormally intense monsoon trough this index is defined on a daily basis, the statistical char- (Fig. 8f), cases with a late-established monsoon trough acteristics of the monsoon trough are examined. These is or cases without an active monsoon trough (Table 1) are no monsoon trough on one-third of the days of the boreal interesting findings. Examination of these cases may summer. Active monsoon troughs occur during half of the improve our understanding of the western North Pacific days. The most active month is August, in which ap- summer monsoon system. proximately 20 days exhibit an active monsoon trough. Because the TC signals impact the large-scale circu- Using this index, active monsoon trough periods were lation on the daily synoptic maps, the present study defined by determining the establishment and decay suggests that TCs should be removed before calculating dates for every single year from 1979 to 2016. During the monsoon trough index. Another advantage of this most years, the active monsoon trough period begins in definition is to better illustrate the evolution of the mid-July and ends in late October, persisting for three to monsoon trough after TC formation. Therefore, the five months during boreal summer. Although the estab- definition can be employed to investigate the interaction lishment date of the monsoon trough varies from year to between TCs and the monsoon trough. However, the

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 15 JULY 2020 F E N G E T A L . 5935 requirement of TC removal in the definition may limit ——, T. Li, M. Peng, W. Chen, and G. Chen, 2014a: Effects of the usage of this daily index. The new index is defined monsoon trough interannual variation on tropical cyclogene- based on the situation in the southern half of the mon- sis over the western North Pacific. Geophys. Res. Lett., 41, 4332–4339, https://doi.org/10.1002/2014GL060307. soon trough, while the situation in the northern half is ——, ——, ——, ——, and ——, 2014b: Effects of monsoon trough neglected. Although preprocesses were employed prior intraseasonal oscillation on tropical cyclogenesis over the to calculating the index, identifying the shape of the western North Pacific. J. Atmos. Sci., 71, 4639–4660, https:// monsoon trough objectively remains difficult. The doi.org/10.1175/JAS-D-13-0407.1. pattern-identifying method, which was developed by ——, R. Wu, and M. Bi, 2018: Contributions of different time-scale variations to tropical cyclogenesis over the western North Yoshida and Ishikawa (2013), may be a valid strategy to Pacific. J. Climate, 31, 3137–3153, https://doi.org/10.1175/ objectively determine whether an organized monsoon JCLI-D-17-0519.1. trough exists. Moreover, a combined index, including Chan, S. C., and J. L. Evans, 2002: Comparison of the structure relative vorticity, sea level pressure, and precipitation, of the ITCZ in the west Pacific during the boreal summers may be more reliable in determining the establishment of 1989–93 using AMIP simulations and ECMWF re- analysis. J. Climate, 15, 3549–3568, https://doi.org/10.1175/ and decay of the monsoon trough. Notably, the value 1520-0442(2002)015,3549:COTSOT.2.0.CO;2. of this index is sensitive to the horizontal resolution of Chang, C.-P., and G. T.-J. Chen, 1995: Tropical circulations the original dataset because this index is a sum of associated with southwest monsoon onset and westerly certain grid points. If datasets with different hori- surges over the South China Sea. Mon. Wea. Rev., 123, , zontal resolutions are used for comparison, the cal- 3254–3267, https://doi.org/10.1175/1520-0493(1995)123 3254: TCAWSM.2.0.CO;2. culation of the index should include a scale parameter ——, and T. Li, 2000: A theory for the tropical tropospheric bi- that equals the square of the horizontal resolution of ennial oscillation. J. Atmos. Sci., 57, 2209–2224, https://doi.org/ the dataset. 10.1175/1520-0469(2000)057,2209:ATFTTT.2.0.CO;2. Chen, G., and R. Huang, 2008: Influence of monsoon over the Acknowledgments. The authors appreciate Prof. John warm pool on interannual variation on tropical cyclone ac- Chiang, Prof. Tim Li, and three anonymous reviewers tivity over the western North Pacific. Adv. Atmos. Sci., 25, 319–328, https://doi.org/10.1007/s00376-008-0319-7. for their helpful comments. The authors also thank Prof. ——, and C.-H. Sui, 2010: Characteristics and origin of quasi- Xuejuan Ren for the helpful discussion. This study was biweekly oscillation over the western North Pacific during supported by the National Natural Science Foundation boreal summer. J. Geophys. Res., 115, D14113, https://doi.org/ of China (41621005, 41775056) and the Fundamental 10.1029/2009JD013389. Research Funds for the Central Universities (2019B03314). ——, and C. Chou, 2014: Joint contribution of multiple equatorial waves to tropical cyclogenesis over the western North Pacific. Mon. Wea. Rev., 142, 79–93, https://doi.org/10.1175/ MWR-D-13-00207.1. REFERENCES ——, and K. Wang, 2018: Why is the tropical cyclone activity over Aiyyer, A., A. Mekonnen, and C. J. Schreck, 2012: Projection of the western North Pacific so distinct in 2016 and 1998 following tropical cyclones on wavenumber–frequency-filtered equato- super El Niño events? J. Meteor. Soc. Japan, 96, 97–110, https:// rial waves. J. Climate, 25, 3653–3658, https://doi.org/10.1175/ doi.org/10.2151/jmsj.2018-013. JCLI-D-11-00451.1. Chen, J.-M., T. Li, and C.-F. Shih, 2010: Tropical cyclone– and AMS, 2019a: Monsoon trough. Glossary of Meteorology, American monsoon-induced rainfall variability in Taiwan. J. Climate, 23, Meteorological Society, http://glossary.ametsoc.org/wiki/ 4107–4120, https://doi.org/10.1175/2010JCLI3355.1. Monsoon_trough. Chen, T.-C., S.-Y. Wang, W.-R. Huang, and M.-C. Yen, 2004: ——, 2019b: Trough. Glossary of Meteorology,American Variation of the East Asian summer monsoon rainfall. Meteorological Society, http://glossary.ametsoc.org/wiki/ J. Climate, 17, 744–762, https://doi.org/10.1175/1520-0442(2004) Trough. 017,0744:VOTEAS.2.0.CO;2. Bell, G. D., and Coauthors, 2000: Climate assessment for 1999. Chen,T.-J.G.,andC.-P.Chang,1980:Thestructureandvor- Bull. Amer. Meteor. Soc., 81 (6), S1–S50, https://doi.org/ ticity budget of an early summer monsoon trough (mei-yu) 10.1175/1520-0477(2000)81[s1:CAF]2.0.CO;2. over southeastern China and Japan. Mon. Wea. Rev., 108, Bi, M., T. Li, M. Peng, and X. Shen, 2015: Interactions 942–953, https://doi.org/10.1175/1520-0493(1980)108,0942: between Typhoon Megi (2010) and a low-frequency mon- TSAVBO.2.0.CO;2. soon gyre. J. Atmos. Sci., 72, 2682–2702, https://doi.org/ Chou, C., J.-Y. Tu, and J.-Y. Yu, 2003: Interannual variability of 10.1175/JAS-D-14-0269.1. the western North Pacific summer monsoon: Differences be- Briegel, L. M., and W. M. Frank, 1997: Large-scale influences on tween ENSO and non-ENSO years. J. Climate, 16, 2275–2287, tropical cyclogenesis in the western North Pacific. Mon. Wea. https://doi.org/10.1175/2761.1. Rev., 125, 1397–1413, https://doi.org/10.1175/1520-0493(1997) Dickinson, M., and J. Molinari, 2002: Mixed Rossby–gravity 125,1397:LSIOTC.2.0.CO;2. waves and western Pacific tropical cyclogenesis. Part I: Cao, X., G. Chen, and W. Chen, 2013: Tropical cyclogenesis in- Synoptic evolution. J. Atmos. Sci., 59, 2183–2196, https://doi.org/ duced by ITCZ breakdown in association with synoptic wave 10.1175/1520-0469(2002)059,2183:MRGWAW.2.0.CO;2. train over the western North Pacific. Atmos. Sci. Lett., 14, Feng, T., G. Chen, R. Huang, and X. Shen, 2014: Large-scale cir- 294–300, https://doi.org/10.1002/asl2.452. culation patterns favourable to tropical cyclogenesis over the

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 5936 JOURNAL OF CLIMATE VOLUME 33

western North Pacific and associated barotropic energy con- Huangfu, J., R. Huang, W. Chen, T. Feng, and L. Wu, 2017: versions. Int. J. Climatol., 34, 216–227, https://doi.org/10.1002/ Interdecadal variation of tropical cyclone genesis and its rela- joc.3680. tionship to the monsoon trough over the western North Pacific. ——, X.-Q. Yang, W. Zhou, R. Huang, L. Wu, and D. Yang, 2016: Int. J. Climatol., 37, 3587–3596, https://doi.org/10.1002/joc.4939. Synoptic-scale waves in sheared background flow over the ——, ——, ——, and ——, 2018: Causes of the active typhoon western North Pacific. J. Atmos. Sci., 73, 4583–4603, https:// season in 2016 following a strong El Niño with a comparison doi.org/10.1175/JAS-D-16-0064.1. to 1998. Int. J. Climatol., 38, e1107–e1118, https://doi.org/ ——, X. Shen, R. Huang, and G. Chen, 2017: Influence of the in- 10.1002/joc.5437. terannual variation of cross-equatorial flow on tropical cy- Kanamitsu, M., W. Ebisuzaki, J. Woollen, S.-K. Yang, J. J. Hnilo, clogenesis over the western North Pacific. J. Trop. Meteor., 23, M. Fiorino, and G. L. Potter, 2002: NCEP–DOE AMIP-II 68–80, https://doi.org/10.16555/J.1006-8775.2017.01.007. Reanalysis (R-2). Bull. Amer. Meteor. Soc., 83, 1631–1644, Frank, W. M., and P. E. Roundy, 2006: The role of tropical waves in https://doi.org/10.1175/BAMS-83-11-1631. tropical cyclogenesis. Mon. Wea. Rev., 134, 2397–2417, https:// Kiladis, G. N., J. Dias, K. H. Straub, M. C. Wheeler, S. N. Tulich, doi.org/10.1175/MWR3204.1. K. Kikuchi, K. M. Weickmann, and M. J. Ventrice, 2014: A Gao, J., and T. Li, 2011: Factors controlling multiple tropical cy- comparison of OLR and circulation-based indices for tracking clone events in the western North Pacific. Mon. Wea. Rev., 139, the MJO. Mon. Wea. Rev., 142, 1697–1715, https://doi.org/ 885–894, https://doi.org/10.1175/2010MWR3340.1. 10.1175/MWR-D-13-00301.1. ——, J. Yu, X. Zhang, and R. Zhang, 2011: Variation of the Kim, J.-H., C.-H. Ho, H.-S. Kim, C.-H. Sui, and S. K. Park, 2008: monsoon trough intensity in the South China Sea and west- Systematic variation of summertime tropical cyclone activity ern North Pacific Ocean and its relationship to tropical cy- in the western North Pacific in relation to the Madden–Julian clone. J. Trop. Meteor., 17, 209–220, https://doi.org/10.3969/ oscillation. J. Climate, 21, 1171–1191, https://doi.org/10.1175/ J.ISSN.1006-8775.2011.03.003. 2007JCLI1493.1. Glickman, T., Ed., 2000: Glossary of Meteorology. 2nd ed. Amer. Knapp, K. R., M. C. Kruk, D. H. Levinson, H. J. Diamond, and C. J. Meteor. Soc., 855 pp., http://glossary.ametsoc.org/. Neumann, 2010: The International Best Track Archive for Gray, W. M., 1968: Global view of the origin of tropical distur- Climate Stewardship (IBTrACS). Bull. Amer. Meteor. Soc., bances and storms. Mon. Wea. Rev., 96, 669–700, https:// 91, 363–376, https://doi.org/10.1175/2009BAMS2755.1. doi.org/10.1175/1520-0493(1968)096,0669:GVOTOO.2.0. Ko, K.-C., H.-H. Hsu, and C. Chou, 2012: Propagation and mainte- CO;2. nance mechanism of the TC/submonthly wave pattern and TC Gu, D., and S. G. H. Philander, 1997: Interdecadal climate fluc- feedback in the western North Pacific. J. Climate, 25, 8591–8610, tuations that depend on exchanges between the tropics and https://doi.org/10.1175/JCLI-D-11-00643.1. extratropics. Science, 275, 805–807, https://doi.org/10.1126/ Krouse, K. D., and A. H. Sobel, 2010: An observational study of science.275.5301.805. multiple tropical cyclone events in the western North Pacific. Hall,J.D.,A.J.Matthews,andD.J.Karoly,2001:Themodu- Tellus, 62A, 256–265, https://doi.org/10.1111/j.1600-0870. lation of tropical cyclone activity in the Australian region 2010.00435.x. by the Madden–Julian oscillation. Mon. Wea. Rev., 129, Kurihara, Y., M. A. Bender, and R. J. Ross, 1993: An initialization 2970–2982, https://doi.org/10.1175/1520-0493(2001)129,2970: scheme of hurricane models by vortex specification. Mon. Wea. TMOTCA.2.0.CO;2. Rev., 121, 2030–2045, https://doi.org/10.1175/1520-0493(1993) Harr, P. A., and J. C. L. Chan, 2005: Monsoon impacts on tropical 121,2030:AISOHM.2.0.CO;2. cyclone variability. WMO Tech. Doc. 1266, TMPR Rep. 70, ——, ——, R. E. Tuleya, and R. J. Ross, 1995: Improvements in the WMO, 512–542. GFDL hurricane prediction system. Mon. Wea. Rev., 123, ——, and C.-C. Wu, 2011: Tropical cyclone characteristics and 2791–2801, https://doi.org/10.1175/1520-0493(1995)123,2791: monsoon circulations. The Global Monsoon System: Research IITGHP.2.0.CO;2. and Forecast, C.-P. Chang et al., Eds., World Scientific Lander, M. A., 1994: Description of a monsoon gyre and its effects Publishing, 357–372. on the tropical cyclones in the western North Pacific during Hogsett, W., and D.-L. Zhang, 2010: Genesis of Typhoon Chanchu August 1991. Wea. Forecasting, 9, 640–654, https://doi.org/ (2006) from a westerly wind burst associated with the MJO. Part 10.1175/1520-0434(1994)009,0640:DOAMGA.2.0.CO;2. I: Evolution of a vertically tilted precursor vortex. J. Atmos. Sci., ——, 1996: Specific tropical cyclone track types and unusual tropical 67, 3774–3792, https://doi.org/10.1175/2010JAS3446.1. cyclone motions associated with a reverse-oriented monsoon Holland, G. J., 1995: Scale interaction in the western Pacific trough in the western North Pacific. Wea. Forecasting, 11, monsoon. Meteor. Atmos. Phys., 56, 57–79, https://doi.org/ 170–186, https://doi.org/10.1175/1520-0434(1996)011,0170: 10.1007/BF01022521. STCTTA.2.0.CO;2. Hsu, H.-H., C.-H. Hung, A.-K. Lo, C.-C. Wu, and C.-W. Hung, 2008: Lau, N.-C., and M. J. Nath, 2006: ENSO modulation of the interannual Influence of tropical cyclones on the estimation of climate and intraseasonal variability of the East Asian monsoon—A variability in the tropical western North Pacific. J. Climate, 21, model study. J. Climate, 19, 4508–4530, https://doi.org/10.1175/ 2960–2975, https://doi.org/10.1175/2007JCLI1847.1. JCLI3878.1. Huang, B., and Coauthors, 2017a: NOAA extended reconstructed Li, C., R. Lu, and G. Chen, 2017a: Promising prediction of the sea surface temperature (ERSST), version 5. NOAA National monsoon trough and its implication for tropical cyclone ac- Centers for Environmental Information, accessed 3 September tivity over the western North Pacific. Environ. Res. Lett., 12, 2017, doi:10.7289/V5T72FNM. 074027, https://doi.org/10.1088/1748-9326/aa71bd. ——, and Coauthors, 2017b: Extended Reconstructed Sea Surface Li, R. C. Y., and W. Zhou, 2013: Modulation of western North Temperature, version 5 (ERSSTv5): Upgrades, validations, Pacific tropical cyclone activity by the ISO. Part I: Genesis and and intercomparisons. J. Climate, 30, 8179–8205, https:// intensity. J. Climate, 26, 2904–2918, https://doi.org/10.1175/ doi.org/10.1175/JCLI-D-16-0836.1. JCLI-D-12-00210.1.

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 15 JULY 2020 F E N G E T A L . 5937

——, ——, and T. Li, 2014: Influences of the Pacific–Japan tele- atmosphere system. J. Meteor. Soc. Japan, 74, 493–508, https:// connection pattern on synoptic-scale variability in the western doi.org/10.2151/jmsj1965.74.4_493. North Pacific. J. Climate, 27, 140–154, https://doi.org/10.1175/ Ventrice, M. J., M. C. Wheeler, H. H. Hendon, C. J. Schreck, C. D. JCLI-D-13-00183.1. Thorncroft, and G. N. Kiladis, 2013: A modified multivariate Li, T., B. Wang, B. Wu, T. Zhou, C.-P. Chang, and R. Zhang, 2017: Madden–Julian oscillation index using velocity potential. Theories on formation of an anomalous anticyclone in western Mon. Wea. Rev., 141, 4197–4210, https://doi.org/10.1175/ North Pacific during El Niño: A review. J. Meteor. Res., 31, MWR-D-12-00327.1. 987–1006, https://doi.org/10.1007/s13351-017-7147-6. Wang, B., and X. Xu, 1997: Northern Hemisphere summer mon- Liu, H., T. Zhou, Y. Zhu, and Y. Lin, 2012: The strengthening East soon singularities and climatological intraseasonal oscillation. Asia summer monsoon since the early 1990s. Chin. Sci. Bull., J. Climate, 10, 1071–1085, https://doi.org/10.1175/1520-0442(1997) 57, 1553–1558, https://doi.org/10.1007/s11434-012-4991-8. 010,1071:NHSMSA.2.0.CO;2. Love, G., 1985: Cross-equatorial interactions during tropical cy- ——, and Z. Fan, 1999: Choice of South Asian summer monsoon clogenesis. Mon. Wea. Rev., 113, 1499–1509, https://doi.org/ indices. Bull. Amer. Meteor. Soc., 80, 629–638, https://doi.org/ 10.1175/1520-0493(1985)113,1499:CEIDTC.2.0.CO;2. 10.1175/1520-0477(1999)080,0629:COSASM.2.0.CO;2. Maloney, E. D., and D. L. Hartmann, 2001: The Madden–Julian ——, and J. C. L. Chan, 2002: How strong ENSO events affect oscillation, barotropic dynamics, and North Pacific tropical tropical storm activity over the western North Pacific. cyclone formation. Part I: Observations. J. Atmos. Sci., 58, J. Climate, 15, 1643–1658, https://doi.org/10.1175/1520-0442(2002) 2545–2558, https://doi.org/10.1175/1520-0469(2001)058,2545: 015,1643:HSEEAT.2.0.CO;2. TMJOBD.2.0.CO;2. ——, and LinHo, 2002: Rainy season of the Asian–Pacific summer ——, and M. J. Dickinson, 2003: The intraseasonal oscillation monsoon. J. Climate, 15, 386–398, https://doi.org/10.1175/ and the energetics of summertime tropical western North 1520-0442(2002)015,0386:RSOTAP.2.0.CO;2. Pacific synoptic-scale disturbances. J. Atmos. Sci., 60, ——, R. Wu, and X. Fu, 2000: Pacific–East Asian teleconnection: 2153–2168, https://doi.org/10.1175/1520-0469(2003)060,2153: How does ENSO affect East Asian climate? J. Climate, 13, TIOATE.2.0.CO;2. 1517–1536, https://doi.org/10.1175/1520-0442(2000)013,1517: Molinari, J., and D. Vollaro, 2013: What percentage of western PEATHD.2.0.CO;2. North Pacific tropical cyclones form within the monsoon ——, LinHo, Y. Zhang, and M.-M. Lu, 2004: Definition of South trough? Mon. Wea. Rev., 141, 499–505, https://doi.org/10.1175/ China Sea monsoon onset and commencement of the East MWR-D-12-00165.1. Asia summer monsoon. J. Climate, 17, 699–710, https:// ——, and ——, 2017: Monsoon gyres of the northwest Pacific: doi.org/10.1175/2932.1. Influences of ENSO, the MJO, and the Pacific–Japan pat- ——, Z. Wu, J. Li, J. Liu, C.-P. Chang, Y. Ding, and G. Wu, 2008: How tern. J. Climate, 30, 1765–1777, https://doi.org/10.1175/ to measure the strength of the East Asian summer monsoon. JCLI-D-16-0393.1. J. Climate, 21, 4449–4463, https://doi.org/10.1175/2008JCLI2183.1. ——, K. Lombardo, and D. Vollaro, 2007: Tropical cyclogenesis Wang, X., and G. Chen, 2017: Quasi-biweekly oscillation over the within an equatorial Rossby wave packet. J. Atmos. Sci., 64, South China Sea in late summer: Propagation dynamics and 1301–1317, https://doi.org/10.1175/JAS3902.1. energetics. J. Climate, 30, 4103–4112, https://doi.org/10.1175/ NCEP, National Weather Service, NOAA, and Department of JCLI-D-16-0533.1. Commerce, 2000: NCEP/DOE Reanalysis 2 (R2). National Wang, Z., T. J. Dunkerton, and M. T. Montgomery, 2012: Center for Atmospheric Research, Computational and Application of the marsupial paradigm to tropical cy- Information Systems Laboratory, accessed 9 September clone formation from northwestward-propagating dis- 2017, https://rda.ucar.edu/datasets/ds091.0/. turbances. Mon. Wea. Rev., 140, 66–76, https://doi.org/ NOAA/NCDC, 2017: IBTrACS-All data, version v03r10. 10.1175/2011MWR3604.1. NOAA/National Climatic Data Center, accessed 14 July 2017, Wheeler, M. C., and H. H. Hendon, 2004: An all-season real- https://www.ncdc.noaa.gov/ibtracs/index.php?name=ibtracs- time multivariate MJO index: Development of an index data. for monitoring and prediction. Mon. Wea. Rev., 132, 1917–1932, Ritchie, E. A., and G. J. Holland, 1999: Large-scale patterns as- https://doi.org/10.1175/1520-0493(2004)132,1917:AARMMI. sociated with tropical cyclogenesis in the western Pacific. 2.0.CO;2. Mon. Wea. Rev., 127, 2027–2043, https://doi.org/10.1175/ Wu, L., and M. Takahashi, 2018: Contributions of tropical waves to 1520-0493(1999)127,2027:LSPAWT.2.0.CO;2. tropical cyclone genesis over the western North Pacific. Climate Schenkel, B. A., 2017: Are multiple tropical cyclone events similar Dyn., 50, 4635–4649, https://doi.org/10.1007/S00382-017-3895-3. among basins? J. Climate, 30, 5805–5813, https://doi.org/ ——, Z. Wen, R. Huang, and R. Wu, 2012: Possible linkage be- 10.1175/JCLI-D-17-0088.1. tween the monsoon trough variability and the tropical cyclone Schreck, C. J., J. Molinari, and K. I. Mohr, 2011: Attributing tropical activity over the western North Pacific. Mon. Wea. Rev., 140, cyclogenesis to equatorial waves in the western North Pacific. 140–150, https://doi.org/10.1175/MWR-D-11-00078.1. J. Atmos. Sci., 68, 195–209, https://doi.org/10.1175/2010JAS3396.1. ——, ——, and R. Wu, 2015: Influence of the monsoon trough on Takayabu, Y. N., and T. Nitta, 1993: 3–5 day period disturbances cou- westward-propagating tropical waves over the western North pled with convection over the tropical Pacific Ocean. J. Meteor. Soc. Pacific. Part I: Observations. J. Climate, 28, 7108–7127, https:// Japan, 71, 221–246, https://doi.org/10.2151/jmsj1965.71.2_221. doi.org/10.1175/JCLI-D-14-00806.1. Tomita, T., and T. Yasunari, 1996: Role of the northeast winter Wu, R., 2002: Processes for the northeastward advance of the monsoon on the biennial oscillation of the ENSO/monsoon summer monsoon over the western North Pacific. J. Meteor. system. J. Meteor. Soc. Japan, 74, 399–413, https://doi.org/ Soc. Japan, 80, 67–83, https://doi.org/10.2151/jmsj.80.67. 10.2151/JMSJ1965.74.4_399. ——, and B. Wang, 2001: Multi-stage onset of the summer monsoon Ueda, H., and T. Yasunari, 1996: Maturing process of the summer over the western North Pacific. Climate Dyn., 17, 277–289, monsoon over the western North Pacific: A coupled ocean/ https://doi.org/10.1007/s003820000118.

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC 5938 JOURNAL OF CLIMATE VOLUME 33

Xu, K., and R. Lu, 2015: Break of the western North Pacific sum- ’91/92 events. J. Meteor. Soc. Japan, 74, 49–62, https://doi.org/ mer monsoon in early August. J. Climate, 28, 3420–3434, 10.2151/JMSJ1965.74.1_49. https://doi.org/10.1175/JCLI-D-14-00588.1. Zhou, W., and J. C. L. Chan, 2007: ENSO and the South China Sea Yoshida, R., and H. Ishikawa, 2013: Environmental factors con- summer monsoon onset. Int. J. Climatol., 27, 157–167, https:// tributing to tropical cyclone genesis over the western North doi.org/10.1002/joc.1380. Pacific. Mon. Wea. Rev., 141, 451–467, https://doi.org/10.1175/ Zong, H., and L. Wu, 2015a: Re-examination of tropical cyclone MWR-D-11-00309.1. formation in monsoon troughs over the western North Zhan, R., Y. Wang, and Q. Liu, 2017: Salient differences in tropical Pacific. Adv. Atmos. Sci., 32, 924–934, https://doi.org/10.1007/ cyclone activity over the western North Pacific between 1998 s00376-014-4115-2. and 2016. J. Climate, 30, 9979–9997, https://doi.org/10.1175/ ——, and ——, 2015b: Synoptic-scale influences on tropical cy- JCLI-D-17-0263.1. clone formation within the western North Pacific monsoon Zhang, R., A. Sumi, and M. Kimoto, 1996: Impact of El Niño on the trough. Mon. Wea. Rev., 143, 3421–3433, https://doi.org/ East Asian monsoon: A diagnostic study of the ’86/87 and 10.1175/MWR-D-14-00321.1.

Unauthenticated | Downloaded 09/30/21 12:11 AM UTC