Numerical Study of the Influences of a Monsoon Gyre on Intensity Changes of Typhoon Chan-Hom (2015)

Despite the development of numerical models and intense research on tropical cyclone intensity change over the past two decades, there have still been no significant improvements in intensity forecasting, especially when tropical cyclones experience unusual intensity changes (e.g., Elsberry et al., 2007; DeMaria et al., 2014). Tropical cyclone rapid intensification (e.g., Kaplan and DeMaria, 2003; Kaplan et al., 2010; Shu et al., 2012; Rogers et al., 2013) and weakening over tropical open oceans (e.g., Brand and Blelloch, 1973; Wood and Ritchie, 2015; Liang et al., 2016) are regarded as two major causes of large intensity forecasting errors. Therefore, understanding unusual intensity changes of tropical cyclones is important to the improvement of tropical cyclone intensity forecasts.

Tropical cyclone rapid intensification has been discussed in previous observational and numerical studies. Some focused on the positive effects of environmental factors (e.g., Kaplan and DeMaria, 2003; Kaplan et al., 2010; Shu et al., 2012; Susca-Lopata et al., 2015), while others concentrated on the importance of internal processes of tropical cyclones (e.g., Willoughby et al., 1982; Montgomery and Kallenbach, 1997; Heymsfield et al., 2001). (Willoughby et al., 1982) noted remarkable intensification in tropical cyclone intensity when the outer eyewall contracted during secondary eyewall replacement, which has been confirmed by observational and numerical studies (e.g., Camp and Montgomery, 2001; Zhu et al., 2004; Houze et al., 2007; Sitkowski et al., 2011). (Montgomery and Kallenbach, 1997) suggested that the axisymmetrization of a convectively induced region of positive potential vorticity near the radius of maximum wind can lead to a rapid intensification process through increasing eyewall tangential winds. Recent studies suggest an important role played by vortical hot towers or convective bursts with cold cloud tops and intense vertical motion near the tropical cyclone center in rapid intensification (e.g., Heymsfield et al., 2001; Hendricks et al., 2004; Reasor et al., 2009; Guimond et al., 2010; Rogers et al., 2013; Wang and Wang, 2014).

In contrast, our current understanding of tropical cyclone weakening over tropical open oceans comes from just a few observational studies. Besides the well-known effects of changes in environmental factors such as rapid reductions in sea surface temperature, dry air intrusion, and increasing vertical wind shear (e.g., Zhang et al., 2007; DeMaria et al., 2012; Qian and Zhang, 2013; Wood and Ritchie, 2015), (Titley and Elsberry, 2000) found that a shallower secondary circulation caused by the downward extension of the upper-level eddy flux convergence was responsible for the rapid weakening of Typhoon Flo (1990) over the open ocean. In general, previous studies have mainly focused on the effects of environmental factors and internal processes on tropical cyclone intensity changes. Little attention has been paid to the interaction between tropical cyclones and large-scale monsoon gyres, although this interaction can affect tropical cyclone intensity through changing tropical cyclone structure (Liang et al., 2016).

As a specific pattern of the low-level summertime monsoonal circulation on the intraseasonal time scale (e.g., Lander, 1994; Carr and Elsberry, 1995), the monsoon gyre in the tropical western North Pacific (WNP) basin can provide favorable large-scale environments and initial disturbances for tropical cyclone formation (e.g., Ritchie and Holland, 1999; Chen et al., 2004; Wu et al., 2013b; Liang et al., 2014), and interacts with the embedded tropical cyclones to result in sudden track changes (e.g., Carr and Elsberry, 1995; Wu et al., 2011a, 2011b, 2013a; Liang et al., 2011; Bi et al., 2015; Liang and Wu, 2015). Recently, (Liang et al., 2016) found in an observational study that the interaction of a monsoon gyre with Typhoon Chan-Hom (2015) affected Chan-Hom's intensity through changing the typhoon structure. Chan-Hom (2015) initially moved westward along ∼10^°N, and its intensity reduced by 10.3 m s^-1 within 12 h during a sharp northward turn in the Philippine Sea——a feature that was not predicted in any of the operational forecasts. During the coalescence of Chan-Hom (2015) with the monsoon gyre on the 15-30-day timescale, the strong convection on the eastern side of the monsoon gyre developed, which prevented the inward transportation of mass and moisture, leading to the collapse of the eastern part of the eyewall. In follow-up work, they further performed a composite observational study to examine rapid weakening events occurring in monsoon gyres in the tropical WNP basin during May-October 2000-14 (Liang et al., 2018). It was found that in excess of 40% of rapid weakening events happening south of 25^°N were observed near the center of monsoon gyres and were accompanied by a sudden northward turn, with large-scale environmental factors not the primary contributor.

Therefore, recent understanding of the effects of monsoon gyres on tropical cyclone intensity changes merely comes from a small quantity of observational evidence (Liang et al., 2016, 2018). In the present study, to advance our understanding of tropical cyclone intensity changes over the tropical open ocean, the influence of the monsoon gyre on the intensity change of Typhoon Chan-Hom (2015) is examined by analyzing the results from a series of numerical experiments with the Weather Research and Forecasting (WRF) model.

The rest of the paper is organized as follows: The model and experimental design are described in section 2. Section 3 compares the results of the numerical experiments and investigates the influences of the low-frequency monsoon gyre on intensity changes of Chan-Hom (2015) over the tropical open ocean. Finally, a brief summary is provided in section 4.

The numerical experiments in this study were conducted with version 3.7.1 of the Advanced Research version of WRF (ARW) (Skamarock et al., 2008), including three two-way interactive domains with horizontal resolutions of 27, 9 and 3 km. These three domains covered the regions (23.18^°S-41.72^°N, 120.13^°-179.86^°E), (12.62^°S-31.71^°N, 131.26^°-168.18^°E) and (3.51^°-19.98^°N, 141.04^°-162.88^°E), respectively. The model contained 41 vertical levels from the surface to 50 hPa.

The WRF single-moment three-class simple ice microphysics scheme (Dudhia, 1989) and the Kain-Fritch convective scheme (Kain and Fritch, 1993) were used in the outermost domain, and the WRF Double-Moment 6-class scheme (Lim and Hong, 2010), suitable for high-resolution simulations, was used in the middle and innermost domains, with no convective parameterization. The Yonsei University PBL scheme (Noh et al., 2003), the Dudhia shortwave parameterization (Dudhia, 1989), and the Rapid Radiative Transfer Model longwave parameterization (Mlawer et al., 1997) were used in all three domains.

The model was initialized with National Centers for Environmental Prediction (NCEP) Final (FNL) Operational Global Analysis data, on 1.0^°× 1.0^° grids, every 6 h (NOAA/NCEP, 2000). To retain the large-scale information at the interior of the domain in simulations, grid nudging was applied to the horizontal winds in the outermost and middle domains. No nudging was conducted within the PBL. Based on the results of (Stauffer and Seaman, 1990) and (Stauffer et al., 1991), this configuration for grid nudging can induce the least bias. The numerical simulations with the outermost and middle domains covered a 126-h period from 1200 UTC 30 June to 1800 UTC 5 July 2015, and the innermost domain was integrated up to 48 h, starting from 0000 UTC 2 July 2015.

Figure 1.  (a) The unfiltered 850-hPa initial wind field (vectors and shaded; units: m s^-1) and its components on the (b) synoptic scale, (c) the 15-30-day band-pass filtered scale, and (d) the 30-day low-pass filtered scale, at 1200 UTC 30 June 2015. Red dashed lines in (c) and (b) roughly outline the large-scale cyclonic circulations, with red dots indicating their centers. The typhoon symbol indicates the center of Chan-Hom (2015) at 1200 UTC 30 June 2015.

The results of two numerical experiments are analyzed in this paper. In the control experiment (CTL), the unfiltered FNL-derived initial fields were used in the model. According to the spectral analysis of (Liang et al., 2016), which showed a spectral peak band of 15-30 days during the activity of Chan-Hom (2015), the initial wind field associated with Chan-Hom (2015) consisted of three components (Fig. 1), and they were obtained using Lanczos filtering (Duchon, 1979) at each grid point. A band-pass filter and a low-pass filter were applied to obtain the 15-30-day band-pass and 30-day low-pass timescale components, respectively. The synoptic-scale component was derived from the difference between the unfiltered field and the 15-day low-pass field. An evident cyclonic circulation of Chan-Hom (2015) is apparent in Fig. 1b. On the 850-hPa 15-30-day timescale (Fig. 1c), a monsoon gyre was located around (10^°N, 148^°E), with a zonal elongated axis of about 2500 km and dominant westerly winds along the southern periphery at 1200 UTC 30 June 2015. Note that the monsoon gyre was also overlapped with a 30-day low-pass cyclonic gyre with dominant westerly and easterly winds along the southern and northern peripheries (Fig. 1d). Typhoon Chan-Hom (2015) was observed to the east of their centers at 1200 UTC 30 June 2015. In this study, the center of the low-frequency gyre was defined as the position of the maximum positive relative vorticity in the vicinity of the circulation center at 850 hPa, and the gyre size was visually determined as the diameter of the outermost closed wind vectors (Wu et al., 2013b; Liang et al., 2016). A Lanczos filter was also used for isolating other variables in the initial and lateral boundary conditions, including temperature, water vapor, geopotential height, and pressure. To examine the influence of the 15-30-day timescale monsoon gyre on the intensity changes of Chan-Hom (2015), a sensitivity experiment (NOMG) was conducted by removing the 15-30-day timescale components from the initial and lateral boundary conditions. In other words, the tropical cyclone was embedded in the large-scale 30-day low-pass cyclonic gyre in the NOMG experiment. The model configuration and physical parameterization schemes were the same as those in the CTL experiment.

Figure 2 shows the simulated track and intensity in CTL during the period from 1200 UTC 30 June to 1800 UTC 5 July 2015, compared to the observations from the China Meteorological Administration (CMA) (Ying et al., 2014), Japan Meteorological Agency (JMA), and Joint Typhoon Warning Center (JTWC) best-track datasets. In this study, the simulated typhoon center was defined as that which maximized the symmetric tangential wind by a variational approach through maximizing the azimuthal-averaged tangential wind speed (Wu et al., 2006). The CTL experiment reproduced the movement of Chan-Hom (2015) well during the period from 1200 UTC 30 June to 1800 UTC 5 July 2015 (Fig. 2a). The simulated typhoon initially moved westward along 10^°N, with centers that were almost consistent with the observations. At 1800 UTC 2 July, it took a sharp northward turn at around (10^°N, 149.5^°E), with a slightly smaller turning angle and a more eastward turning point than observed. After 1200 UTC 3 July, the simulated typhoon made a northwestward movement.

Figure 2.  Observed and simulated (a) tracks and (b) intensities of Typhoon Chan-Hom (2015) from 1200 UTC 30 June to 1800 UTC 5 July 2015. The marks are at 6-h intervals. Red dashed vertical lines in (b) outline the period of the weakening of Chan-Hom (2015), which is the focus of this study.

The simulated intensity evolution was relatively close to that in the CMA best-track dataset. The simulated typhoon reached its peak intensity of about 34 m s^-1 at 1800 UTC 2 July 2015, which was 1 m s^-1 stronger than that observed in the CMA dataset. Then, the simulated intensity decreased in the following 48 h, with a reduction of about 12 m s^-1, and reintensified from 0000 UTC 5 July 2015. In contrast, the observed intensity in the CMA dataset decreased 10 m s^-1 in 12 h, and maintained at tropical storm intensity till late 5 July 2015. Based on the definition of rapid weakening suggested by (DeMaria et al., 2012), i.e., that the tropical cyclone intensity decreases 20 kt (10.3 m s^-1) or more in 24 h, the weakening of Chan-Hom (2015) shown in the CMA dataset can be classified as a rapid weakening event. Despite a slower weakening, the CTL experiment represented the important features of intensity changes of Chan-Hom (2015), including the magnitude and the occurrence time of the peak intensity and the subsequent weakening over the tropical WNP basin.

In the JMA and JTWC best-track datasets, Chan-Hom (2015) only intensified to tropical storm intensity, with a peak intensity of about 28 m s^-1 at 1200 UTC 2 July and 25 m s^-1 at 0600 UTC 2 July 2015, respectively, which were much weaker and earlier than observed in the CMA dataset and in the CTL simulation. However, the weakening after reaching peak intensity was observed in both best-track datasets. The intensity decreased 5.3 m s^-1 in the JMA dataset and 2.5 m s^-1 in the JTWC dataset in 12 h, showing slower weakening than in the CMA dataset. Then, according to the JMA data, Chan-Hom (2015) maintained an intensity of about 24 m s^-1 from 1200 UTC 3 July 2015, whereas JTWC showed reintensification. Despite their different weakening rates, the weakening processes after the peak intensity in the JMA and JTWC best-track datasets also confirm the reproduction of the weakening of Chan-Hom (2015) over the tropical open ocean in CTL. For comparison with observations, we focus mainly on the 24-h weakening process from 1800 UTC 2 July to 1800 UTC 3 July 2015 in this study.

A large intensity deviation during the interaction of Chan-Hom (2015) and the monsoon gyre not only occurred in the three best-track datasets, but also between the JTWC best-track dataset and the real-time intensity information used in (Liang et al., 2016). The intensity in the best-track datasets was mainly estimated using the Dvorak method (Dvorak, 1975, 1984, 1995), whereas the real-time intensity information used in (Liang et al., 2016) was derived from the operational warnings issued by JTWC. The implication, therefore, is that a challenge exists in the current technique used for estimating tropical cyclone intensity during the complex interaction between tropical cyclones and large-scale monsoonal systems.

During the weakening of the typhoon, the enlargement of the eye size observed in (Liang et al., 2016) was simulated in CTL. Figure 3 compares the simulated 700-hPa radar reflectivity in the innermost domain at the onset and end of the 24-h weakening in CTL. At 1800 UTC 2 July (Fig. 3a), it shows an open eye with a radius of less than 50 km, and the eyewall is mainly located to the southeast of the simulated typhoon center. The deep convection in the eyewall has a maximum radar reflectivity exceeding 50 dBZ. After the 24-h weakening (Fig. 3b), no evident deep convection remains within a radius of 100 km from the simulated typhoon center.

Previous studies found that the interaction between a monsoon gyre and a tropical cyclone can enhance the southwesterly flows in the southeastern periphery of the monsoon gyre, which can cause the development of convection on the eastern side of the monsoon gyre (e.g., Carr and Elsberry, 1995; Liang et al., 2011, 2014; Wu et al., 2011a, 2011b, 2013). (Liang et al., 2016) proposed the importance of this eastern deep convection associated with the monsoon gyre in tropical cyclone intensity changes.

In the CTL experiment, the eastern deep convection associated with the monsoon gyre developed, which agreed with that in the observation (Fig. 4). Figures 4a and c show the cloud top temperature from the FY2F satellite. It shows a relatively well-organized convective area with a cold cloud top of -70^°C (green) 700 km east of the typhoon center at 1230 UTC 2 July, which rapidly decays in the following 18 h. This convection area is also apparent in the results from CTL, as shown in Figs. 4b and c, which display the simulated radar reflectivity with the wind field without the typhoon circulation at 700 hPa in the 9-km domain in CTL. The typhoon circulation was removed from the wind field based on (Kurihara1993,Kurihara1995). The convection with maximum radar reflectivity exceeding 45 dBZ was in the southeastern periphery of the large-scale monsoon circulation at 1200 UTC 2 July, and no evident convection remained in-situ by 0600 UTC 3 July.

Figure 3.  Simulated radar reflectivity (shaded; units: dBZ) at 700 hPa in the 3-km domain in CTL at (a) 1800 UTC 2 July and (b) 1800 UTC 3 July 2015. Red circles are 50 and 100 km away from the typhoon center, indicated by the typhoon symbol, respectively.

Figure 4.  (a, c) Infrared brightness temperature (shaded; units: ^°C) from the FY2F satellite at (a) 1230 UTC 2 July and (c) 0630 UTC 3 July 2015. The blue area denotes temperature lower than -60^°C. (b, d) Simulated radar reflectivity (shaded; units: dBZ) and wind fields (vectors; units: m s^-1) without typhoon circulations at 700 hPa in the 9-km domain in CTL at (b) 1200 UTC 2 July and (d) 0600 UTC 3 July 2015. The typhoon center is indicated by the typhoon symbol.

Figure 5.  (a) Time-longitude cross section of the simulated radar reflectivity (shaded; units: dBZ) along the latitude of the typhoon center, and (b) time-radius cross section of the simulated azimuthal averaged radial wind speed at 950 hPa (shaded and contours; units: m s^-1), in the 9-km domain in CTL. The black and blue dots in (a) indicate the simulated typhoon and monsoon gyre positions, respectively. The two solid lines in (a) indicate 500 km away from the typhoon center, respectively. The horizontal red dashed lines outline the period of the weakening of Chan-Hom (2015), which is the focus of this study. Red circles outline the convection associated with the monsoon gyre and the corresponding inflows.

Figure 5a shows a time-longitude cross section of the simulated radar reflectivity along the latitude of the typhoon center. We can see a cluster of deep convection in a band-shaped area between 500 and 1000 km east of the typhoon center between 0000 UTC 2 July and 1200 UTC 3 July 2015. Before 0000 UTC 2 July, weak and loose eastern convection mainly occurs about 750 km east of the typhoon center. Subsequently, the eastern convection extends westward to a radius of about 500 km from the typhoon center by 1800 UTC 2 July and is enhanced to a strength with maximum radar reflectivity exceeding 55 dBZ. After the onset of the weakening process, the eastern deep convection gradually weakens and withdraws eastward in 18 h. Simultaneously, the deep convection that was initially about 40 km west of the typhoon center shifts rapidly westward and can be seen about 250 km away from the typhoon center after the 24-h weakening, which agrees with the enlargement of the eye size.

Corresponding to the enhancement of the eastern convection associated with the monsoon gyre, simulated strong inflow with a peak value exceeding 5 m s^-1 occurs within the band region between 600 and 1000 km between 0000 UTC 2 July and 0600 UTC 3 July 2015, as shown in Fig. 5b, which is a time-radius cross section of the low-level azimuthal-averaged radial wind speed. The strong inflow in the inner core of the simulated typhoon is enhanced with the typhoon intensification and attains its peak value exceeding 10 m s^-1 at 1800 UTC 2 July. Subsequently, the strong inner inflow rapidly weakens and shifts outward during the 24-h weakening. Figure 5 suggests that the eastern deep convection of the monsoon gyre played an important role in preventing the inward transportation of mass and energy.

The influence of the 15-30-day timescale monsoon gyre on the intensity changes of Chan-Hom (2015) in the tropical WNP basin is examined by comparing the results of the NOMG experiment with those of the CTL experiment. Figure 6 shows the simulated tracks and intensities in both experiments during the period from 1200 UTC 30 June to 1800 UTC 5 July 2015. The simulated track in NOMG does not deviate significantly from that in CTL (Fig. 6a). During the first 48 h, the simulated typhoon in NOMG takes a westward track consistent with that in CTL, but has a slightly slower translation speed. At around 1800 UTC 2 July 2015, a slow northwestward turn is apparent in NOMG, with the turning point less than about 100 km to the east of that in CTL, as shown in Fig. 7a. We can also examine, based on the results presented in Fig. 7b, the track direction change during the 6-h period, in which positive (negative) values indicate a northward (southward) turn. The simulated typhoon in NOMG takes a less northward shift of about 25^° within 6 h around 1800 UTC 2 July, as compared to that reaching about 45^° in CTL. According to the definition of sudden northward turning track changes in Wu et al., (2013), i.e., that a track direction change exceeds 37^° during a 6-h period, the sudden northward track change happening in CTL does not appear in NOMG. The simulated slow northwestward turn in NOMG may result from the interaction between the typhoon and the cyclonic gyre on a timescale lower than 30 days, as shown in Fig. 1d. We can see the simulated typhoon in NOMG gradually moves into the interior of the cyclonic gyre on a timescale lower than 30 days, and coalesces with it at 1800 UTC 2 July 2015 (Fig. 7c). The enhanced southwesterly wind in the southern periphery steers the simulated typhoon northwestward. Then, the simulated typhoon keeps a northwestward track that is very close to that in CTL. It is noted that the monsoon gyre plays a major role in the sudden northward track change of Chan-Hom (2015), as suggested by previous studies (e.g., Liang et al., 2011; Wu et al., 2011a, 2011b; Liang and Wu, 2015).

The simulated intensity in the NOMG experiment is significantly different from that in CTL (Fig. 6b). Prior to 1800 UTC 1 July, both simulations show a similar intensity evolution. With the westward movement, the simulated typhoon in NOMG intensifies to a peak intensity of about 27 m s^-1 at 1800 UTC 2 July, which is much weaker than that in CTL. It then maintains this peak intensity in the following 30 h, while the simulated typhoon in CTL undergoes weakening during this period. During 4-5 July, the simulated intensity in NOMG shows a weakening and subsequent reintensification process similar to that in CTL. The NOMG experiment suggests a significant influence of the 15-30-day timescale monsoon gyre on the weakening of Chan-Hom (2015) during the period from 1800 UTC 2 July to 1800 UTC 3 July 2015, and the preceding intensification from 1800 UTC 1 July 2015.

Corresponding to the maintained intensity between 1800 UTC 2 July and 1800 UTC 3 July 2015, the eye size of the simulated typhoon in NOMG shows no significant change, as shown in Fig. 8. The simulated typhoon keeps an open eye with a radius of less than 50 km and the eyewall to the southeast of the typhoon center during 24 h.

As expected, the cluster of deep convection in the band-shaped area between 500 and 1000 km east of the typhoon center between 0000 UTC 2 July and 1200 UTC 3 July 2015 that occurs in CTL does not appear in NOMG, as shown in Fig. 9a, which is a time-longitude cross section of the simulated radar reflectivity along the latitude of the typhoon center in NOMG. Prior to 1200 UTC 2 July, the eastern convection outside a radius of 500 km is associated with the cyclonic gyre on the timescale lower than 30 days. It then decays rapidly and no evident convection remains in-situ. Correspondingly, there are no strong inflows appearing in the band-shaped area between 600 and 1000 km from the typhoon center before and during the weakening process observed in CTL. The strong inflow occurs within a radius of 100 km from the typhoon center and maintains the inner core of the simulated typhoon. As a result, due to the absence of the development of the eastern convection associated with the 15-30-day monsoon gyre, the weakening happening from 1800 UTC 2 July to 1800 UTC 3 July in CTL is not observed in NOMG.

In addition, (Liang et al., 2018) suggested an effect of monsoon gyres on the outer size of tropical cyclones, insofar as monsoon gyres can enlarge the outer size of tropical cyclones embedded in them. Therefore, Fig. 10 compares the evolution of the outer sizes of the simulated typhoons in CTL and NOMG from 1200 UTC 1 July to 0000 UTC 4 July 2015. In this study, the radius of the azimuthal-averaged wind intensities of 17 m s^-1 was used to quantify the outer typhoon size. Prior to 2 July, both simulated typhoons have similar outer sizes. Then, with the typhoon gradually moving westward into the monsoon gyre in CTL, its outer size increases to around 150 km by 1800 UTC 2 July. In NOMG, meanwhile, the simulated typhoon shrinks slightly in 3 h and maintains a smaller outer size of about 70 km on 2 July. During the following 24 h, the outer sizes of the simulated typhoons in both experiments increase rapidly, with almost identical rates of increase. This agrees with the negligible effects of the monsoon gyre on the rate of increase of the outer size suggested in (Liang et al., 2018), the dynamic implication of which needs further investigation. It is clear that the typhoon has a larger outer size when it interacts with the monsoon gyre, confirming the composite result of (Liang et al., 2018).

Figure 6.  Simulated (a) tracks and (b) intensities of Typhoon Chan-Hom (2015) from 1200 UTC 30 June to 1800 UTC 5 July 2015, with dots indicating 6-h intervals. Red dashed vertical lines in (b) outline the period of the weakening of Chan-Hom (2015), which is the focus of this study.

From Fig. 6b, we note a weaker simulated intensity during the intensification stage from 1800 UTC 1 July to 1800 UTC 2 July 2015 in NOMG, while Chan-Hom (2015) was located in the eastern part of the monsoon gyre during this period in CTL. Thus, the possible effects of changes in the environmental factors caused by the absence of the monsoon gyre are examined. As important parameters to tropical cyclone intensity changes (e.g., DeMaria, 1996; Shu and Wu, 2009; Shu et al., 2013; Wang et al., 2015), the environmental humidity and vertical wind shear from 1200 UTC 30 June to 1800 UTC 2 July 2015 are shown in Figs. 11a and b. Typhoon circulations were first removed from the fields based on (Kurihara1993,Kurihara1995), and then the environmental parameters obtained were all averaged in the area with a radius of 500 km from the typhoon center. Figure 10a shows the evolution of mid-level relative humidity in CTL and NOMG, which were averaged over the layers between 700 hPa and 500 hPa. Apart from a slightly higher relative humidity prior to 1 July in CTL, similar mid-level relative humidity evolutions were observed during the intensification stage in these two experiments, suggesting negative effects of the mid-level humidity associated with the monsoon gyre on the intensity change. The evolution of the 200-850-hPa and low-level vertical wind shears can be examined based on the results presented in Fig. 11b. These vertical wind shears were measured by the differences in the horizontal winds between 200 hPa and 850 hPa (200-850) and between 850 hPa and 1000 hPa (low-level), respectively. In CTL, the 200-850 vertical wind shear increased to a peak of about 16 m s^-1 at 1800 UTC 2 July when the typhoon reached its peak intensity, in agreement with the observational results in (Liang et al., 2016). The typhoon in NOMG was affected by lower 200-850 vertical wind shears, with the value between 6 and 9 m s^-1 during its intensification stage, which is usually favorable for tropical cyclone intensification (e.g., DeMaria, 1996; Frank and Ritchie, 2001; Wong and Wong, 2004). The low-level vertical wind shears in CTL and NOMG both remained lower than 2.5 m s^-1 before 1800 UTC 2 July. According to (Wang et al., 2015), tropical cyclones prefer to intensify under the effect of low-level vertical wind shear lower than 2.5 m s^-1. Figure 10b indicates that vertical wind shear was not a crucial factor as well.

Figure 7.  (a) Simulated tracks from 0600 UTC 2 July to 1800 UTC 3 July 2015, with dots indicating 3-h intervals. (b) Track direction change during a 6-h period, with positive (negative) values indicating a northward (southward) turn. (c) Simulated wind field (vectors and shaded; units: m s^-1) at 850 hPa in the 9-km domain in the NOMG experiment at 1800 UTC 2 July 2015. Blue dots in (a) highlight northward turning times. Horizontal and vertical lines in (b) mark the direction change of 37^° and the northward turning time. The red dashed line in (c) roughly outlines the large-scale cyclonic circulation, with the typhoon symbol indicating the simulated typhoon center.

Figure 8.  As in Fig. 3 but for the NOMG experiment.

Figure 9.  As in Fig. 5 but for the NOMG experiment.

Moreover, the inertial stability is also an important factor in the development of tropical cyclones, which is mainly related to the relative vorticity (e.g., Schubert and Hack, 1982). In the NOMG experiment, the removal of the monsoon gyre on the 15-30 day timescale in the initial field may have changed the environmental vorticity, further affecting the inertial stability. Thus, in Fig. 11c we present the evolution of the inertial stability at 950 hPa. Prior to 0000 UTC 2 July, the inertial stability in NOMG, with a value between 1 and 2× 1.0<disp-formula>×10 ^-8 s^-1 is almost consistent with that in CTL. Subsequently, the inertial stability in CTL enhances rapidly and reaches its peak value of about 5.5× 1.0</disp-formula>×10^-8 s^-1 in 12 h, which is much higher than that in NOMG. The strong environmental inertial stability provides more resistance to the radial displacement for the typhoon (e.g. Schubert and Hack, 1982). Further investigation shows that the strong inertial stability in CTL was associated with the monsoon gyre on the 15-30-day timescale. As shown in Fig. 10d, rapid increases in the positive relative vorticity and the inertial stability on the 15-30-day timescale were observed from 0000 UTC 2 July. The suggestion, therefore, is that, as a cyclonic circulation, the monsoon gyre may have been favorable for the development of the tropical cyclone moving westward in its eastern part by providing enhanced inertial stability.

Figure 10.  Time series of simulated typhoon sizes in CTL (black) and NOMG (red). Red dashed vertical lines outline the period of the weakening of the simulated tropical cyclone, which is the focus of this study.

Figure 11.  Time series of the simulated (a) mid-level relative humidity (units: %), (b) 200-850 hPa (solid line with dots) and low-level (dashed line) vertical wind shear (units: m s^-1), (c) 950-hPa inertial stability (units: 1.0×10^-8 s^-1) averaged in the region with a radius of 500 km from the typhoon center in the 9-km domain in CTL (black) and NOMG (red), and (d) the FNL-derived 15-30-day timescale inertial stability (solid; units: 1.0×10^-9 s^-1) and relative vorticity (dashed; units: 1.0×10^-5 s^-1) at 950 hPa averaged in the region with a radius of 500 km from the typhoon center. The vertical line marks 1800 UTC 1 July 2015.

The unusual intensity changes of tropical cyclones, including rapid intensification and the weakening in the tropical open ocean, can result in large forecasting errors (e.g., Brand and Blelloch, 1973; Kaplan and DeMaria, 2003; Kaplan et al., 2010; Shu et al., 2012; Rogers et al., 2013; Wood and Ritchie, 2015; Liang et al., 2016). Based on a recent observational analysis by (Liang et al., 2016) that the weakening of Typhoon Chan-Hom (2015) in the Philippine Sea was primarily caused by its interaction with a monsoon gyre on the 15-30 day timescale, in this study, the influence of the monsoon gyre on the intensity change of Chan-Hom (2015) was investigated by analyzing the results of two numerical experiments performed using the ARW-WRF model.

The control experiment reproduced the track changes and intensity evolution of Typhoon Chan-Hom (2015) well——particularly the weakening process during the sudden northward turn in the Philippine Sea. In agreement with the observation in (Liang et al., 2016), the eye size of the simulated typhoon enlarged from less than 50 km to about 100 km during the weakening process. The deep convection on the eastern side of the monsoon gyre was also captured, which occurred prior to the weakening, and enhanced and shifted westward during the weakening process. Corresponding to the development of the eastern deep convection associated with the monsoon gyre, strong inflow occurred within a band region between 600 and 1000 km from the simulated typhoon center and, simultaneously, the strong inner inflow rapidly weakened and shifted outward. In the sensitivity experiment without the monsoon gyre on 15-30-day timescale, in the initial and lateral boundary conditions, the simulated typhoon showed a weaker intensity during the proceeding intensification stage and maintained a peak intensity and small eye size during the northward turn with a smaller turning angle, indicating the important role played by the monsoon gyre on the 15-30-day timescale in the intensity changes of Chan-Hom (2015) in the Philippine Sea. The eastern strong convection associated with the large-scale circulation decayed rapidly, and no strong outer inflow occurred during this period.

It was found that during Chan-Hom's westward movement in the eastern part of the monsoon gyre, the monsoon gyre with a cyclonic environment was favorable for the intensification of Chan-Hom (2015). The enhancing inertial stability associated with the monsoon gyre provided strong resistance to the radial displacement for Chan-Hom (2015). With Chan-Hom (2015) coalescing with the monsoon gyre and the convection on the eastern side of the monsoon gyre developing to its peak strength, the corresponding outer inflow enhanced, which prevented the inward transportation of moisture and mass. Thus, the eyewall collapsed and the weakening happened in the tropical WNP basin.

In agreement with the observational analysis, the numerical simulations confirm the important effects of the interaction between tropical cyclones and monsoon gyres on tropical cyclone intensity change in the tropical open ocean through changing the tropical cyclone structure. Moreover, we also note that the influence of monsoon gyres depends upon the tropical cyclone location relative to the monsoon gyre center, which needs further investigation.