The Year of Tropical Convection (YOTC) high-resolution global reanalysis dataset was analyzed to reveal precursor synoptic-scale disturbances related to tropical cyclone (TC) genesis in the western North Pacific (WNP) during the 2008–09 typhoon seasons. A time filtering is applied to the data to isolate synoptic (3–10 day), quasi-biweekly (10–20 day), and intraseasonal (20–90 day) time-scale components. The results show that four types of precursor synoptic disturbances associated with TC genesis can be identified in the YOTC data. They are 1) Rossby wave trains associated with preexisting TC energy dispersion (TCED) (24%), 2) synoptic wave trains (SWTs) unrelated to TCED (32%), 3) easterly waves (EWs) (16%), and 4) a combination of either TCED-EW or SWT-EW (24%). The percentage of identifiable genesis events is higher than has been found in previous analyses.
Most of the genesis events occurred when atmospheric quasi-biweekly and intraseasonal oscillations are in an active phase, suggesting a large-scale control of low-frequency oscillations on TC formation in the WNP. For genesis events associated with SWT and EW, maximum vorticity was confined in the lower troposphere. During the formation of Jangmi (2008), maximum Rossby wave energy dispersion appeared in the middle troposphere. This differs from other TCED cases in which energy dispersion is strongest at low level. As a result, the midlevel vortex from Rossby wave energy dispersion grew faster during the initial development stage of Jangmi.
Tropical cyclone (TC) genesis is a process through which random convective systems are organized into a mesoscale vortex under favorable large-scale conditions (Gray 1968, 1979; Montgomery and Enagonio 1998; also see Li 2012 for a review). Whereas the favorable environmental conditions are necessary, the timing of TC genesis depends on the occurrence of synoptic-scale wave disturbances that trigger individual cyclogenesis events. Different from the tropical Atlantic, where the typical perturbation type is African easterly waves (Burpee 1972, Landsea 1993), three major types of low-level precursor disturbances associated with tropical cyclogenesis in the WNP are 1) Rossby wave trains induced by energy dispersion of a preexisting TC (TCED), 2) northwest–southeast-oriented synoptic wave trains (SWTs, sometimes referred to as tropical depression type (TD type) disturbances) unrelated to TC energy dispersion, and 3) Pacific easterly waves (EWs; Fu et al. 2007).
The first type of precursor synoptic-scale disturbances for cyclogenesis in the WNP is associated with TCED. A Rossby wave train with alternating anticyclonic and cyclonic vorticity perturbations formed in the wake of a preexisting TC due to its Rossby wave energy dispersion (Frank 1982; Flier 1984; Holland 1995; McDonald 1998; Li et al. 2003; Krouse et al. 2008; Krouse and Sobel 2010). Although TCED is essentially of a barotropic vorticity dynamics nature (Carr and Elsberry 1995), a Rossby wave train associated with a realistic 3D TC structure has a baroclinic vertical structure (Ge et al. 2008, 2010), and vorticity asymmetry between upper- and lower-tropospheric wave trains depends strongly on the sign of the vertical shear of the zonal mean flow (Ge et al. 2007). A new TC may form in the cyclonic vorticity region of the wave train under favorable environmental conditions (Li and Fu 2006; Li et al. 2006).
The second type of precursor disturbance associated with TC genesis in the WNP is a northwest–southeast-oriented wave train that does not involve a preexisting TC. Lau and Lau (1990) showed that this TD type of perturbation is a dominant synoptic-scale mode in the summertime WNP. Chang et al. (1996) examined the wave trains in the Navy Operational Global Atmospheric Prediction System (NOGAPS) analysis data and found that the wave trains are well presented regardless of whether a TC bogus was used in the analysis. Dickinson and Molinari (2002) attributed the generation of SWTs to the development of equatorial mixed Rossby–gravity (MRG) waves located initially near the equator. A theoretical study by Li (2006) suggested that the SWT resulted from the instability of the summer mean flow in the WNP regardless of the initial perturbation pattern. Recently, Wang et al. (2012) applied a so-called pouch theory (Dunkerton et al. 2009; Wang et al. 2010) to the northwestward-propagating SWT system.
The third type of precursor synoptic-scale disturbance is Pacific EW (Ritchie and Holland 1999). The nondivergent barotropic model experiments by Kuo et al. (2001) suggested that the scale contraction of EWs could lead to the accumulation of kinetic energy at a critical longitude where monsoon westerlies meet trade easterlies. This energy accumulation mechanism may lead to the successive development of TCs at the critical longitude. The origin of the Pacific EWs is from the southward propagation of Rossby wave energy from the upper-tropospheric jet in the North Pacific (Tam and Li 2006). Based on the pouch theory (Dunkerton et al. 2009; Wang et al. 2010), EWs provide a favorable environment for TC development. Frank (1988) reported that only a small percentage of EWs in the WNP (~10%) developed into TCs. Using 8 yr of data from an Australia Bureau of Meteorology tropical analysis field, Ritchie and Holland (1999) attributed 18% of WNP cyclogenesis events to EWs. Using Quick Scatterometer (QuikSCAT) wind data, Fu et al. (2007) found that 21% of cyclogenesis events in the WNP were associated with EWs during the summers of 2000 and 2001. This is in contrast to Chen et al. (2008), who suggested that 80% of TC formations in the WNP are due to the influence of EWs either directly or indirectly. Therefore, what percentage of TC formation in the WNP is induced by EWs remains unclear.
In addition to the aforementioned three types of precursor disturbances, other pathways to cyclogenesis in the WNP include the influence of a tropical upper-tropospheric trough (TUTT; Sadler 1976, 1978; Briegel and Frank 1997), monsoon gyres (Lander 1994; Holland 1995; Ritchie and Holland 1999; Lee et al. 2008), equatorial waves (Frank and Roundy 2006; Schreck et al. 2011, 2012), and cross-equatorial flows (Love 1985; Xu 2011; Beattie and Elsberry 2012). Some of the upper- and lower-level precursor signals may occur simultaneously. For example, Briegel and Frank (1997) suggested that 49% of TC genesis events have both an upper- and a lower-level feature.
Fu et al. (2007) noticed that during the 2000–01 WNP summer seasons 70% of cyclogenesis events are associated with the aforementioned three types of precursor disturbances, whereas the remaining were unclassified due to uncertainty in the relatively coarse (2.5° × 2.5°) resolution of the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis data. Recently, the Year of Tropical Convection (YOTC) project has provided unprecedented high-resolution global analysis data with a horizontal resolution of 0.225° and 26 vertical levels expanding from the surface to 1 hPa. The unique high-resolution YOTC data make it possible to identify the finer 3D structures and patterns of evolution of precursor synoptic-scale disturbances associated with TC genesis events.
The lower-frequency atmospheric oscillations may significantly modulate TC genesis in the WNP. It has been suggested that the atmospheric intraseasonal oscillation [ISO, 20–90-day mode; Yamazaki and Murakami (1989); Hartmann et al. (1992); Liebmann et al. (1994); Sobel and Maloney (2000); Maloney and Hartmann (2000, 2001, 2002); Maloney and Dickinson (2003); Hogsett and Zhang (2010)] may exert a large-scale control on TC formation through enhancing or suppressing the TD-type disturbances in the lower-level troposphere (Zhou and Li 2010). The enhancement of the TD-type disturbances is attributed to either Rossby wave accumulation caused by large-scale confluent flows (Kuo et al. 2001) or barotropic energy conversion that involves both rotational and divergent winds (Hsu et al. 2011). The quasi-biweekly oscillation (QBW) with a period of 10–20 days is one of the dominant modes over the WNP during boreal summer (Li and Wang 2005; Chen and Sui 2010). The perturbation associated with the QBW has a northeast-tilted structure and propagates northwestward in the WNP (Gao and Li 2011). The QBW has been suggested to be closely associated with convectively coupled equatorial Rossby waves and affect TC genesis in the WNP (Li and Wang 2005; Kikuchi and Wang 2009; Gao and Li 2011).
The rest of the paper is organized as follows. In section 2, the data and analysis methods are described. Precursor synoptic-scale disturbances during the YOTC (2008–09) are analyzed in section 3. In section 4, the impact of the atmospheric QBW and ISO on TC genesis during the 2-yr period is further examined. A summary is given in section 5.
2. Data and analysis methods
The primary dataset used in this study is the high-resolution global analysis from the Year of Tropical Convection (YOTC) program, a coordinated research program jointly conducted by the World Weather Research Program [WWRP/The Observing System Research and Predictability Experiment (THORPEX)] and the World Climate Research Program (WCRP) including observing, modeling, and forecasting of organized tropical convection. Based on decadal substantial investments in the earth science infrastructure, including satellite observing systems and operational buoy arrays in each of the tropical oceans, the European Centre for Medium-Range Weather Forecasts (ECMWF) generated a global analysis dataset with a horizontal resolution of 0.225° at 26 standard pressure levels for a period of 2 yr from 1 May 2008 to 30 April 2010 (Waliser et al. 2011). A more detailed description of the YOTC dataset can be found online (www.ucar.edu/YOTC). The other data used in this work include TC best-track data from the Joint Typhoon Warning Center (JTWC) and observed daily outgoing longwave radiation (OLR) data from the National Oceanic and Atmospheric Administration (NOAA). The OLR field has a horizontal resolution of 2.5° × 2.5° and will be used to identify the atmospheric low-frequency oscillations including the QBW and ISO.
Bandpass filtering (Murakami 1979) is applied to separate the synoptic-scale (3–10 day), quasi-biweekly (10–20 day), and intraseasonal (20–90 day) components from the original dataset. In the 3–10-day bandwidth, the strongest response is around 5 days, with a maximum response of 1.0. By isolating the synoptic-scale (3–10 day) motion, one may examine precursor synoptic perturbation signals prior to TC genesis. In fact, the most significant spectrum of SWTs revealed by Lau and Lau (1990) appears in this bandwidth. In the 10–20- and 20–90-day bandwidths, the strongest response appears around 14 and 43 days, respectively. By isolating these two time-scale motions, we examine how the quasi-biweekly and intraseasonal oscillations modulate cyclogenesis activity.
Different types of synoptic-scale perturbations as TC precursors are identified based on the following approach. By examining the daily synoptic-scale maps of low-level winds prior to TC genesis, we group cyclogenesis events into four categories. If a TC formed in the cyclonic circulation embedded within a wave train produced by a preexisting TC, this genesis event belongs to the TCED category. If a TC formed within a synoptic wave train that does not involve a preexisting TC, it belongs to the SWT group. If a TC formed in association with the pronounced westward propagation of perturbation kinetic energy, vorticity, and total column water vapor, it belongs to the EW group. If two or more of the above genesis scenarios were involved, it belongs to a combined cyclogenesis group.
To examine to what extent TCs may affect 3–10-day filtered wind fields, we compared the results with and without the removal of TCs. The TC removal technique follows Schreck et al. (2011), in which a Gaussian function with a radius of 500 km is applied. The results show that the influence of TCs on the filtered synoptic-scale wave train pattern is small. This implies that the TC impact on the wind is not as significant as the rainfall field shown in Schreck et al. (2011).
3. Characteristics of four cyclogenesis groups
Twenty-five WNP TC genesis events during the boreal summer seasons of 2008 and 2009 were investigated. The selection of these 25 events was based on the following criteria: 1) the genesis location appears south of 30°N and 2) the maximum surface wind speed during its life cycle exceeds 35 kt (kt = 0.51 m s−1). Table 1 lists the names of the 25 TC cases selected, along with the characteristics of their identified pregenesis disturbance types. Among the 25 TCs, 24 (96%) are associated with the aforementioned four genesis groups, with 6 cases being associated with TCED, 8 associated with SWT, 4 cases are associated with EW, and 6 cases are associated with the combined genesis scenario. Among the six combined genesis cases, two belong to TCED-EW and four belong to SWT-EW. If one counts a combined genesis event as one-half of an event for each of the involved groups, the percentages for cyclogenesis associated with TCED, SWT, and EW during the 2 yr are 28%, 44%, and 24%, respectively. The relative contribution of the three genesis scenarios is quite similar to that analyzed by Fu et al. (2007), although the absolute value of the genesis percentage for each group is greater.
In the following we will examine each of the four genesis scenarios.
a. Cyclogenesis associated with TCED
The analysis of 3–10-day filtered YOTC wind fields shows that 16 out of the 25 TCs induced a Rossby wave train in their wakes, and among them 8 resulted in a subsequent TC genesis event.
By examining the vertical structure of these Rossby wave trains, we noted that the maximum vorticity associated with the wave trains mostly formed in the lower troposphere. This is consistent with 3D TC energy dispersion characteristics (Ge et al. 2008). Therefore, a general feature for this category is that TCED-induced Rossby wave trains are confined at lower levels throughout the development. However, energy dispersion associated with Typhoon Hagupit (2008) appears to be a special case.
The evolution of the Rossby wave train associated with Typhoon Hagupit (2008) and subsequent genesis of TC Jangmi (2008) is shown in Fig. 1. Jangmi's first warning was issued by JTWC at 1800 UTC 23 September 2008. The precursor synoptic signal prior to the Jangmi formation can be traced back to 21 September 2008, about 3 days prior to the JTWC warning (Fig. 1a). At that time a northwest–southeast-oriented Rossby wave train was clearly seen, in association with a previous typhoon (Hagupit, during 2008, whose center is denoted by a black typhoon mark in Fig. 1). The Rossby wave train had a typical wavelength of 2500 km and was composed of alternate anticyclonic and cyclonic circulation patterns. During the initial stage of Hagupit, cyclonic circulation in its wake was weak (Fig. 1a). As the storm intensified, the wave train strengthened due to the TC energy dispersion. The weak cyclonic circulation in the wake seen in Figs. 1a and 1b developed into a strong cyclone with a closed circulation at 0000 UTC 23 September 2008 (Fig. 1c, with its center labeled as C). The strengthening of the wave train eventually led to the formation of Jangmi at 0000 UTC 24 September 2008 (Fig. 1d). Jangmi then developed into a supertyphoon on 27 September 2008, when the observed maximum surface wind exceeded 140 kt.
The vertical cross section of the TCED-induced wave train and its evolution are shown in Fig. 2. Here, each vertical cross section is plotted along the wave train axis (i.e., black line in each panel of Fig. 1). In Fig. 2 we show both the 3–10-day filtered vorticity and the wind component normal to the wave train axis. The latter is defined to be positive (negative) if air flows from southwest (northeast) to northeast (southwest). It is interesting to note that during the early development stage, the maximum vorticity and circulation of the wave train are confined in the middle troposphere (around 700–400 hPa; Fig. 2a). This feature appears in both the anticyclonic and cyclonic circulations of the wave train. With the continuing strengthening of Hagupit, the wave train further developed and stretched in the vertical, extending from the surface to about 200 hPa (Figs. 2c and 2d). Both the anticyclonic and cyclonic circulations of the wave train during this stage show a quasi-barotropic structure below 200 hPa.
The stronger midlevel vorticity during the initial developing stage implies that Jangmi underwent a typical “top down” developing process in which a precursor perturbation signal first appeared in the middle level (Fu et al. 2007; Li 2012) and then developed from a weak and shallow system (Fig. 2a) into a strong and deep one (Fig. 2d) prior to Jangmi's formation.
An E-vector approach (Hoskins et al. 1983; Trenberth 1986; Sobel and Bretherton 1999; Li and Fu 2006) may be used to measure the Rossby wave energy dispersion. The E vectors are defined as , where the overbar denotes time averaging, and and denote zonal and meridional wind components associated with synoptic-scale disturbances, respectively. The E vectors indicate energy propagation during a specified period. Figure 3a shows the horizontal map of the calculated E vector at 500 hPa during the wave train development stage (from 0000 UTC 20 September to 0000 UTC 24 September 2008). The results show that as the previous TC (denoted by a black typhoon mark each day in Fig. 3) moved northwestward, it emitted energy southeastward, leading to the strengthening of systems behind it, including not only the wake cyclone (C in Figs. 1c and 2c), in which TC Jangmi (a red typhoon mark in Fig. 3) formed later, but also the anticyclone (A in Figs. 1 and 2) before the cyclone (C). By comparing the E vectors at each vertical level (from the surface to 300 hPa), we note that the greatest energy propagation occurs at 500 hPa (Fig. 3b). This seems to explain why the anomalous circulation in the wake developed first in the middle troposphere.
The result above suggests a new midlevel vortex genesis scenario, in which TC Rossby wave energy dispersion plays an essential role. This differs from the classical scenario proposed by Simpson et al. (1997), in which evaporative cooling of raindrops under stratiform clouds set up a midlevel vortex.
A calculation of E vectors for other TCED cases shows that maximum energy dispersion indeed occurred in the lower troposphere. The energy dispersion feature is consistent with the vertical structure of corresponding Rossby wave trains. A composite E vector calculated based on a 4-day period prior to the genesis of each TC in the TCED group included in Table 1 is presented in Fig. 3c. Figure 3c indicates that an averaged new TC appeared to the east-southeast of the previous existing composite TC. Krouse et al. (2008) suggested that nonlinearity could alter the structure of the wave trains, making them more zonally oriented, as opposed to northwest–southeast oriented.
b. TC genesis associated with SWT
During the 2008 and 2009 typhoon seasons, 8 out of the 25 TC genesis cases were related to SWTs. The formation of TC Molave (2009) is an example (Fig. 4). The first warning by JTWC for TC Molave was issued at 0600 UTC 15 July 2009. At 1800 UTC 16 July 2009, 36 h later, it became a tropical storm with maximum surface winds of 35 kt. It later developed into a typhoon (TY) and made landfall in Hong Kong. The 3–10-day filtered 850-hPa wind field can clearly identify the synoptic wave train pattern as early as 1800 UTC 11 June 2009, 5 days prior to the TC genesis. Figure 4 shows the time sequence maps (every 36 h) of the SWTs from 0600 UTC 12 July to 1800 UTC 16 July. During the early stage, cyclonic (labeled as C) and anticyclonic (labeled as A) centers were not aligned along a straight line and the systems appeared weak and shallow (Figs. 4a and 5a). The cyclonic and anticyclonic centers gradually aligned into a northwest–southeast-oriented line, as they intensify and deepen (Figs. 4b and 5b). Note that there is no preexisting tropical cyclone associated with these waves. On 15 July, the SWT became a typical northwest–southeast-oriented wave train and was composed of well-defined cyclonic and anticyclonic circulation patterns with a wavelength of about 2500 km, covering a region between 0°–25°N and 120°–150°E (Fig. 4c). As the SWT continues intensifying and moving northwestward, TC Molave formed at 1800 UTC 16 July in the cyclonic vorticity region of the wave train (Fig. 4d).
Figure 5 illustrates the vertical cross section of the synoptic-scale vorticity along the wave train axis. Different from the TCED case shown in Fig. 2, the strongest perturbation vorticity was initially confined in the lowest level (Fig. 5a). It gradually extended upward, and by 1800 UTC 13 July, the positive vorticity has penetrated into the upper troposphere. The maximum vorticity, however, is still kept in the lower troposphere (Figs. 5c and 5d). It appears that this cyclogenesis event resembles a typical “bottom up” development process. The time evolution of area-averaged vorticity, divergence, vertical motion, and relative humidity fields averaged over a 1.8° × 1.8° domain centered at the maximum 850-hPa vorticity are shown in Fig. 6. At day −5, positive vorticity is confined in the lower troposphere (below 700 hPa). It gradually expands upward as the TC develops (Fig. 6a). The increase in the synoptic-scale vorticity is closely linked to low-level convergence (Fig. 6b). A marked increase in relative humidity throughout the troposphere happened prior to genesis time, indicating the important role of deep-layer moistening in preconditioning TC genesis (Nolan 2007; Ge et al. 2013). These features support the bottom-up development hypothesis.
The genesis event of TC Molave shown above is representative of all of the genesis events in the SWT group. In these cases, maximum vorticity was confined at low levels, suggesting that the bottom-up mechanism may play a dominant role in the SWT group as shown by the developing process for Molave.
c. TC genesis induced by EW
A subjective but strict method was applied to identify the EW-induced cyclogenesis events. To avoid overestimating or underestimating the influence of the EW, in addition to examining the horizontal maps of synoptic-scale disturbances prior to TC formation, we also examined the time–longitude cross section of the 3–10-day filtered kinetic energy, vertical vorticity, and total-column water vapor fields along the latitudinal band (5° width) where a TC formed. The results show that 4 out of 25 TC genesis cases were associated with the EW during the 2008–09 typhoon seasons.
Figure 7 shows the formation of TC Hagupit (2008) as an example. A warning for Hagupit was first issued by JTWC at 1200 UTC 17 September 2008. After 42 h, at 0600 UTC 19 September 2008, it developed into a tropical storm with maximum surface winds of 35 kt. It later became a category 4 typhoon with a maximum surface wind speed of 125 kt. Hagupit made its landfall in Guangdong Province, China, and was the first known category 4 typhoon to hit the province. The estimated damage was around $1 billion (U.S. dollars) and at least 67 people were killed (Bell and Montgomery 2010). The easterly wave signal associated with the formation of Hagupit can be traced back 7 days prior to its genesis. The 3–10-day filtered 850-hPa wind field shows a closed cyclonic circulation pattern at low levels 7 days prior to Hagupit's genesis (Fig. 7a). During the early 4-day period (Figs. 7a–c), the disturbance moved westward but its intensity was rather weak. A closed cyclonic circulation can be seen in Fig. 7a, and it becomes open in Figs. 7b and 7c. The disturbance started to intensify rapidly at 0600 UTC 17 September, when a closed cyclonic circulation pattern could be clearly seen (Figs. 7d and 7e) and the TC formed on 19 September (Fig. 7f). No clear northwest–southeast-oriented wave train was observed in the genesis region. The time-sequence maps of the synoptic perturbations clearly indicate that Hagupit's formation arose from a precursor EW perturbation.
To confirm that Hagupit's genesis originates from an EW perturbation, we examined the longitude–time section of the perturbation kinetic energy, vorticity, and total-column water vapor (Fig. 8). The EW signals from these fields can be traced back to 6–7 days prior to the genesis time. In the early stage, the wave intensity appears weak, as it propagates slowly westward. From day −3 to day 0, the wave perturbation energy and vorticity intensify rapidly, while the wave phase speed increases slightly compared to the earlier period. It is worth noting that there are zonal phase differences among the wave perturbation energy, vorticity, and total-column water vapor fields. It appears that the perturbation energy leads the vorticity, while the latter leads the total-column water vapor. The exact cause of such a phase relationship is unknown at the moment. It is speculated that the vorticity leading may be attributable to the Rossby wave response to the convective heating (Hsu et al. 2011), whereas the perturbation energy leading may be attributable to zonal and meridional wave structures.
The time evolution of the vertical profile of the area-averaged synoptic-scale vorticity shows a bottom-up developing process for this EW-induced cyclogenesis case (figure not shown). The maximum cyclonic vorticity was mainly confined in the lower troposphere; positive vorticity penetrated into the upper troposphere as the disturbance developed. As in the SWT case (Fig. 6), there was oscillatory development in the area-averaged vorticity, divergence, vertical velocity, and relative humidity fields.
Various factors may contribute to the EW intensification. First, as the wave moves westward to warmer SST regions, environmental moisture increases; this favors greater latent heat release. Second, as the wave moves toward a critical longitude where the monsoon westerly meets the trade easterly, the wave energy may accumulate (Kuo et al. 2001). The decrease in zonal wavelength associated with the deceleration of the mean easterly may also lead to wave activity flux convergence near the critical longitude (Tam and Li 2006).
d. Combined genesis type
Among the 25 TC genesis events, 6 were associated with the combination of different genesis scenarios, either combined SWT-EW cases or combined TCED-EW cases.
The genesis of TC Sinlaku is an example of the combined SWT-EW scenario. It is worth mentioning that Sinlaku eventually developed into a supertyphoon and became one of the most impacting TCs in the WNP during 2008. The time sequence of the 3–10-day filtered 850-hPa wind field shows clearly the evolution of an EW, from 0600 UTC 5 September to 1800 UTC 6 September 2008 (Figs. 9a–c). The EW was composed of an open anticyclonic circulation (ridge, denoted by an A at 20°N, 145°E) and a cyclonic circulation (denoted by a C at 17°N, 160°E) at 0600 UTC 5 September 2008 (Fig. 9a). Eighteen hours later, a trough (denoted by a red dashed line) associated with the EW developed (Fig. 9b), as it slowly moves westward. Note that during the same period, a northwest–southeast-oriented SWT with a combined cyclonic and anticyclonic circulation pattern appeared to the west of 130°E (purple line in Fig. 9a). As the SWT intensified, it overlapped with the EW, leading to the formation of a closed cyclonic circulation at 14°N, 130°E (denoted by a C at 14°N, 131°E in Fig. 9c). After the formation of the cyclonic circulation, the EW seemed to dissipate slightly, while the SWT continued intensifying as it moved northwestward. At 1200 UTC 8 September 2008, a new TC, identified as Sinlaku, formed in the cyclonic vorticity region of the SWT. The rapid development of Sinlaku during its genesis stage may be attributable to the reenforced cyclonic flow from the EW and SWT.
It is likely that both the EW and SWT may affect the cyclogenesis through wave energy dispersion and accumulation. To demonstrate this, we calculated E vectors during the period centered at 1800 UTC 6 September. Figure 10 shows that during the period there was clear southeastward energy dispersion associated with the SWT, as well as westward energy dispersion associated with the EW. The convergence of these wave activity fluxes led to energy accumulation in the genesis region.
The formation of TC Higos is a genesis example that was influenced by both the EW and TCED. Figure 11 illustrates the daily sequence of 3–10-day filtered 850-hPa wind fields from 22 to 29 September. On 22 September, there was a clear Rossby wave train associated with the energy dispersion of the preexisting TC Hagupit (Fig. 11a). Two days later, TC Jangmi formed at 13°N, 135°E (Fig. 11c). Meanwhile, a weak EW may be identified east of 150°E (with the A and C denoting anticyclonic and cyclonic vorticity, respectively). As Jangmi rapidly intensified and moved northwestward, it emitted Rossby wave energy southeastward, leading to the development of a closed anticyclone circulation (labeled as A at 8°N, 140°E in Fig. 11d). Meanwhile, the EW also strengthened as it moved westward. As a result of the combined forcing of the TCED and EW, a closed cyclonic circulation (labeled as C at 8°N, 140°E in Fig. 11f) developed over the previous red triangle region (Figs. 11d and 11e). Two days later, a new TC named Higos (labeled as a red typhoon mark in Fig. 11h) formed in the closed cyclonic circulation region. A calculation of E vectors shows that indeed there was wave activity flux convergence in the cyclone-developing region due to energy dispersion from Jangmi (figure not shown).
While the TCED-induced wave train effect is quite obvious, as seen from in Fig. 11, the EW effect appears weaker. To demonstrate the EW effect, we plotted the time–longitude section of perturbation kinetic energy, vorticity, and total-column water vapor in Fig. 12. There was clear westward propagation of wave kinetic energy, vorticity, and water vapor. The EW signals from these fields can be clearly seen 4 days prior to the formation of Higos. Compared to the pure EW case in Fig. 8, the amplitude of the EW in this combined genesis case is weaker, implying that the TCED process may play a more important role in this particular TC genesis event.
4. Modulation of cyclogenesis by low-frequency waves
In this section we examine the impacts of the 10–20-day (QBW) and 20–90-day (ISO) modes on WNP TC genesis during the 2008–09 summer seasons. The combined influence of the two modes is also examined. Figure 13 shows the area-averaged low-frequency OLR and zonal wind anomalies for the 25 TC genesis cases. Here, the OLR anomaly was calculated based on a 5° × 5° box centered at the TC genesis location, whereas the zonal wind anomaly was calculated based on a same-sized box centered at 2.5° south of the TC genesis location. As one can see from Fig. 13, 92% and 80% of TC genesis cases during the 2 yr occurred in the region where the QBW and ISO, respectively, have a negative OLR anomaly. This suggests that TC activities are, to a large extent, coupled to both the QBW and ISO modes, consistent with previous studies (e.g., Fu et al. 2007; Gao and Li 2011). As seen from Figs. 13e and 13f, there is only one exception in which the combined 10–20- and 20–90-day OLR anomaly is positive. It was related to the formation of TC Higos (last TC shown in Fig. 13) in 2008. Higos formed due to the triggering of synoptic perturbations associated with both an EW and a Rossby wave train, as shown in Fig. 11.
To more strictly define an active phase of QBW or ISO, we use half of the OLR standard deviation averaged over 5°–20°N, 130°–160°E during the 2008 and 2009 typhoon seasons as a threshold for an active phase. The results show that 68% and 64%, respectively, of TC genesis cases during the 2 yr occurred in the region where the QBW and ISO are in an active phase, and 80% of the cases occurred in the region where the combined QBW and ISO modes are in an active phase. The same methodology is also applied to the wind field. We note that 60% of genesis cases occurred in the region where the combined QBW and ISO modes have a westerly anomaly of greater than 2 m s−1.
The composite patterns of the filtered OLR and 850-hPa wind fields associated with the QBW and ISO modes during the cyclogenesis time are shown in Fig. 14. The TC genesis is associated with the active phase of QBW and ISO, with favorable low-level cyclonic circulation, ascending vertical motion, and higher low-level and midtropospheric relative humidity. Note that the horizontal scale of the QBW mode is less that its ISO counterpart, while their intensities are comparable.
5. Summary and discussion
The high-resolution (0.225°) YOTC analysis data were used to analyze precursor synoptic-scale wave features associated with TC genesis in the WNP during the 2008–09 typhoon seasons. Four genesis scenarios were identified based on low-level synoptic-scale precursor signals. They are associated with energy dispersion of a preexisting TC (TCED), synoptic-scale wave train (SWT), Pacific easterly wave (EW), and the combination of TCED-EW or SWT-EW.
Among the 25 cyclogenesis cases investigated, 6 (24%) are associated with TCED, 8 (32%) are associated with SWT, 4 (16%) are associated with EW, and 6 (24%) are associated with the combined genesis scenario. There is one unclassified case in which we could not identify a significant precursor signal in either the lower or upper levels.
One interesting cyclogenesis case, Typhoon Jangmi, associated with TCED indicates that the strongest Rossby wave energy dispersion from the previous Typhoon Hagupit does not happen in the lower level; instead, its maximum energy dispersion appeared in the midtroposphere. As a result, a midlevel vortex developed initially in the wake of Hagupit, and subsequent cyclogenesis underwent a “top down” development process. Genesis events associated with the SWT and EW scenarios, on the other hand, showed a typical “bottom up” development process, with maximum vorticity always appearing in the lower troposphere.
A common feature associated with all of the aforementioned genesis scenarios is that the precursor circulation was usually weak and shallow in the early stages, but developed into a deep-layer cyclonic system with closed circulation, strong vorticity, and a near-saturated column in the later stages prior to genesis. Intensification appeared not only in the cyclonic circulation (where a TC finally formed), but also in the anticyclonic circulation. For example, in the cases of TCED and SWT, not only did the cyclonic vorticity region of the wave train develop, but the anticyclonic region of the wave train developed as well. In the case of EW, both the trough and ridge intensify with time.
Another interesting finding based on the YOTC data analysis is that sometimes two or more genesis precursors can be identified and they worked together to trigger cyclogenesis. For instance, Sinlaku was a result of combined SWT and EW forcing; Higos formed in the both TCED and EW scenarios. While identifying the coexistence of two or more precursor disturbances, we do not know the relative importance of these disturbances in causing TC formation. Further sensitivity numerical experiments are needed in order to isolate different triggering processes.
The percentage of TC genesis events associated with TCED during the 2-yr (2008–09) period is 32%, which is consistent with the result (approximately 30%) obtained for a much longer period [1948–2005; Krouse and Sobel (2010)]. However, caution is needed when comparing the two results as the methods used to identify the TCED events differ. While a detailed wave train structure was tracked in the current study, in Krouse and Sobel (2010) any TCs that formed within a distance of 5000 km to the east of an existing TC were counted.
The YOTC data analysis reveals a close relationship between the atmospheric low-frequency oscillation (including QBW and ISO) and TC genesis. Twenty-four out of 25 genesis cases occurred when the combined QBW and ISO modes have a negative OLR anomaly. It was found that 80% of the genesis events occurred in the region where the combined QBW and ISO modes are in an active phase.
Caution is also needed in interpreting the temporally filtered wave train patterns, because the strength of these wave train patterns may be overestimated due to the presence of tropical cyclones in the original reanalysis data. Schreck et al. (2011) proposed a way to remove TC signals with the use of a Gaussian function for a 500-km radius. By applying this TC removal technique to the YOTC data, we noted that the wave train intensity indeed became weaker, even though the wave train pattern remained. An open issue related to this is how to make a clean removal of TC signals without impacting other scale motions. In this study a barotropic E-vector formula was used to illustrate energy dispersion characteristics. Caution is needed here as well, as there might be some nontrivial issues in cases where the time-mean flow is complex or the averaging period is too short. It might be improper to assess the vertical variation of Rossby wave energy dispersion using this barotropic E-vector formula.
Comments from three anonymous reviewers are greatly appreciated. This study was supported by ONR Grant N00014-0810256 and by the International Pacific Research Center that is sponsored by the Japan Agency for Marine–Earth Science and Technology (JAMSTEC), NASA (NNX07AG53G), and NOAA (NA17RJ1230). YX was also supported by the National Natural Science Foundation of China under Grant 40675026 and the National Basic Research Program of China under Grant 2009CB421504.
School of Ocean and Earth Science and Technology Contribution Number 8926 and International Pacific Research Center Contribution Number 979.