The unusual evolution of the multiple eyewall cycles in super typhoon Hinnamnor (2022)

Forecasting the changes in tropical cyclone (TC) intensity remains a challenging issue, mostly due to the complex external and internal dynamics related to the inner-core TC structure (Wang and Wu, 2004; Elsberry et al., 2013; Montgomery and Smith, 2014). In recent years, many studies have shown that the process of secondary eyewall formation (SEF) and the associated eyewall replacement cycle (ERC) can cause large variability in TC structure and intensity changes (Sitkowski et al., 2011; Zhang et al., 2017).

Observationally, Fortner (1958) was the first to describe the concentric eyewall (CE) in Typhoon Sarah (1956). Later, several studies documented the existence of a CE structure in some hurricanes via reconnaissance aircraft and radar observations (Jordan and Schatzle, 1961; Jordan, 1966; Gentry, 1970; Hoose and Colón, 1970; Hawkins, 1971; Black et al., 1972; Holliday, 1977; Shapiro and Willoughby, 1982; Willoughby, 1990; Black and Willoughby, 1992; Houze et al., 2006). Furthermore, Willoughby et al. (1982) revealed that a CE structure can be generated several times during a TC’s life cycle and that this mostly appears in intense hurricanes or typhoons.

With the emergence and subsequent advancements in satellite meteorology, passive microwave satellite imagery has allowed observation of the structure of TCs more clearly than before. Hawkins et al. (Hawkins and Helveston, 2004; Hawkins et al., 2006) used 10 years (1993–2005) of Special Sensor Microwave/Imager digital data to examine the multiple eyewall characteristics of TCs in the Atlantic basin. They found that more than 50% of all TCs with maximum winds reaching 120 kts presented multiple eyewalls during their lifespan. They also found that TCs with a CE structure mostly occurred in the western Pacific. McNoldy (2004) identified three CEs in Hurricane Juliette (2001) by analyzing aircraft reconnaissance data and satellite microwave imagery. However, his dataset was very short, meaning it was difficult to draw robust conclusions regarding the evolution of CEs. Using flight-level aircraft data, Sitkowski et al. (2011) showed that the occurrence of SEF in some hurricanes took place more than once. More recently, Molinari et al. (2019) investigated three distinct ERCs that occurred in Hurricane Frances (2004) by utilizing flight data and argued that both internal forcing (vortex Rossby waves) and external forcing (moderate wind shear) worked together during the two overlapping ERCs. Aside from these observational findings, multiple SEFs and ERCs have also been simulated under idealized settings (Wang et al., 2013), indicating that the internal dynamics has the potential to govern the process of SEF.

Several studies have proposed mechanisms for the internal and external dynamical mechanisms responsible for SEF and the associated ERC, including vortex Rossby waves (Montgomery and Kallenbach, 1997), beta-skirt axisymmetrization via an upscale cascade of vorticity perturbations (Terwey and Montgomery, 2008), boundary-layer processes (Huang et al., 2012; Abarca and Montgomery, 2013, 2014; Kepert, 2013; Kepert and Nolan, 2014; Abarca et al., 2015; Ahern et al., 2022), the axisymmetrization process during binary vortex interaction (Kuo et al., 2004, 2008), and the axisymmetrization of the outer spiral rainbands (Fang and Zhang, 2012; Didlake and Houze, 2013; Qiu and Tan, 2013; Li et al., 2014; Zhu et al., 2015; Wang et al., 2016, 2019; Dai et al., 2017; Zhang et al., 2017; Chen, 2018; Chen et al., 2018; Didlake et al., 2018; Tyner et al., 2018; Wang and Tan, 2020; Yu et al., 2021, 2022 文后未体现; Zhu et al., 2022). Some other studies also found that the interaction between the TC and trough can trigger SEF in TCs and lead to change in TC intensity (Molinari and Vollaro, 1989; Leroux et al., 2013, 2016; Komaromi and Doyle, 2018; Zhao et al., 2021, 2022).

Most previous studies on repeated CE structures have been based on limited flight-level aircraft data and microwave satellite data, especially in the Atlantic basin. Few studies have focused on repeated CE structures in the western North Pacific, even though the proportion of TCs with CEs is highest in this basin (Hawkins and Helveston, 2008 文后未体现). Accordingly, the purpose of this study was to objectively examine the repeated CE structures and subsequent intensity changes of Super Typhoon Hinnamnor (2022). Given the lack of conclusive observations of the secondary eyewall and ERC, this was achieved by using a combination of microwave satellite imagery [from MIMIC, i.e., Morphed Integrated Microwave Imagery at CIMSS (Cooperative Institute for Meteorological Satellite Studies)], Doppler radar observations [from the Kadena Air Force Base in Japan, radar mosaic of the Central Weather Bureau (CWB) in Taiwan, and radar mosaic of the Korea Meteorological Administration (KMA)], and best-track data [from the Joint Typhoon Warning Center (JTWC) and China Meteorological Administration (CMA)]. FY-4A visible imagery was downloaded from http://fy4.nsmc.org.cn/portal/cn/theme/FY4A.html. The vertical wind shear was calculated between 200 and 850 hPa by using the US National Centers for Environmental Prediction (NCEP) Global Forecast System (GFS) Final Analysis (FNL) data.

On 28 August 2022, a broad area of low pressure developed approximately 2000 miles southeast of Japan (Fig. 1). The system rapidly organized on the next day and became a severe tropical storm. On the same day, i.e., 29 August, it quickly upgraded to a typhoon and then a severe typhoon (Fig. 2a). A double eyewall structure generated for only a short duration on this day. Subsequently, Hinnamnor moved westwards and further strengthened into a super typhoon on 30 August. From 30 August to 1 September, a tropical depression developed to the south of Hinnamnor (Figs. 2a, c), the northwestward movement of which caused binary interaction between the two systems that can be seen clearly from the visible satellite imagery (Fig. 2d). The subsequent merger and axisymmetrization of Hinnamnor and the tropical depression led to a continuous increase in both the inner-core size and the outer-core circulation of Hinnamnor (Fig. 2e).

Figure 1.  Track of Typhoon Hinnamnor (2022) with every 6-h position indicated by solid circles in different colors for different storm intensity categories based on the JTWC best-track data. The two eyewall cycles are indicated by SEF1 and ERC1, and SEF2 and CE2.

Figure 2.  (a) Tracks of Hinnamnor (2211) and the tropical depression (13W). (b–e) FY-4A visible imagery (0.65 μm) and 500 hPa geopotential height (blue contours; dagpm) at (b) 1200 UTC 5 September, (c) 0600 UTC 30 August, (d) 0600 UTC 31 August, and (e) 0600 UTC 1 September 2022. The letter “L” in (b) denotes the midlatitude low.

During this period, due to the SEF and ERC processes, Hinnamnor experienced a significant structural adjustment between 1100 UTC 30 and 1200 UTC 31 August, before finally reaching its lifetime maximum intensity at 0000 UTC 1 September (Fig. 3a). It then weakened significantly up until 1200 UTC 2 September, at which point a second SEF process commenced. The CE structure maintained for approximately 84 h until Hinnamnor made landfall, which caused dramatic and unconventional intensity changes. Specifically, it intensified as the SEF process began, and then weakened after 0000 UTC 5 September. This intensity fluctuation did not follow the traditional pattern of SEF, which will be discussed later. During this period of structural change, Hinnamnor moved southwestwards on 31 August and took a generally southwards track after 1 September. It then sharply changed its direction to the north from 2 September onwards. Hinnamnor gradually turned to the northeast after 5 September and made landfall in the southern part of the Korean Peninsula on 6 September, whereupon it transformed into an extratropical cyclone on the same day. Hinnamnor was the fourth-strongest typhoon to hit South Korea according to JTWC records. It brought record-breaking rainfall and the worst flooding event over Jeju Island (1,058 mm) and mainland South Korea (447.5 mm). Later, Hinnamnor moved into the Sea of Japan as a strong extratropical storm and finally dissipated near the coast of Russia.

Figure 3.  (a) The maximum sustained wind speed (m s⁻¹) from 0600 UTC 28 August to 0600 UTC 6 September 2022 based on JTWC (red) and CMA (blue) best-track data. (b) Time series of the deep-layer shear between 200 and 850 hPa (green; m s⁻¹). The two eyewall cycles are indicated by ERC1 and CE2.

Figure 4 shows snapshots of the brightness temperature from MIMIC and the reflectivity field captured by the radar at Kadena Air Force Base in Japan, which together depict a typical evolution of CE features in this case. Specifically, isolated convective cells generated and spiraled outside the inner eyewall at 1100 UTC 30 August (Fig. 4a). Then, the outer convective perturbations elongated in the azimuthal direction and were much stronger in the down-shear half of the storm (Fig. 4b). These signals mark the beginning of SEF. Eventually, an individual annular eyewall structure with almost no convective activity in the moat region appeared at 0057 UTC 31 August (Fig. 4c). Convection in the outer eyewall enhanced progressively with the weakening of the inner eyewall convection, which marks the beginning of ERC. The outer eyewall intensified and contracted, with its strength of maximum radar reflectivity catching up with the primary eyewall (Fig. 4d), eventually revealing a convective ring structure. At 1302 UTC (Fig. 4e), both the inner and outer eyewall had an asymmetric structure, with the strongest eyewall reflectivity occurring in the eastern part of the eyewall (in the left-of-shear quadrants) and the weakest in the western half. After the ERC, the newly organized eyewall was larger in size than the original primary eyewall (Fig. 4f). Therefore, the double eyewall structure lasted for about 28 h in the first period of structural change. This eyewall behavior constitutes a typical SEF and ERC process. According to the CMA track data, the super typhoon maintained its intensity during the SEF and ERC period, which weakened in the JTWC track data gradually after the SEF process. The completion of SEF1 and ERC1 was followed by a short re-intensification period (from 1800 UTC 31 to 0100 UTC 01 September) (Fig. 3a).

Figure 4.  The MIMIC-IR brightness temperature (shading; K) at (a) 1100 UTC 30 August, (b) 1700 UTC 30 August, and (f) 1900 UTC 31 August 2022; and the observed reflectivity of the radar at Kadena Air Force Base in Japan at (c) 0057 UTC 31 August, (d) 0657 UTC 31 August, and (e) 1302 UTC 31 August, related to Typhoon Hinnamnor (2022). The upper-left corner of each panel shows the JTWC-reported maximum sustained winds interpolated in time; the upper-right corner of each panel shows the time to the nearest actual image either before or after (whichever is closer); and the bottom-left corner of each panel shows the vertical wind shear. For each image, the red arrow is the 200–850-hPa environmental wind shear vector. The × symbol represents the location of the radar at Kadena airport (26°21.34’N, 127°46.06’E).

From the axisymmetric mean perspective, the azimuthally averaged brightness temperature from MIMIC during SEF1 and ERC1 demonstrated well the continuous inward-contracting outer eyewall, the diminishing of the inner eyewall, and the strengthening of the new primary eyewall (Fig. 5). Hinnamnor was located far away from the monsoon and the westerly trough during this time period (not shown), so all its cloud systems were self-generated and all its energy was packed around its eye. In ERC1, asymmetric rainbands tended to form in the down-shear half quadrant under the influence of the strengthened vertical wind shear (Fig. 3b) (Hence and Houze, 2012; Yu et al., 2021). These dynamics of the asymmetric rainband that led into the SEF process have been demonstrated in detail previously in an idealized simulation (Wang et al., 2019) and then later verified in a case study (Yu et al., 2021). They found that when the top-down development and inward propagation of a strong outer rainband approached the outer edge of the rapid filamentation zone outside the primary eyewall, the downwind sector of the rainband in the boundary layer then rapidly axisymmetrized, leading finally to the SEF. However, not all the spiral rainbands that formed in the downshear region were effective in forming the secondary eyewall structure. Thus, despite the external forcing (vertical wind shear) being effective in triggering spiral rainbands, the SEF mechanism may still have depended on the internal vortex dynamics in Wang et al. (2019).

Figure 5.  Radius–time series of MIMIC-IR brightness temperature (shading; K). The two eyewall cycles are indicated by SEF1 and ERC1, and SEF2 and CE2.

Both the binary interaction between the tropical depression (13W) and Hinnamnor and the ERC1 process led to an increasing of the inner-core size of Hinnamnor and enhancement of its active spiral rainbands (Fig. 5). The main features of the transition from the spiral rainbands to the secondary eyewall were demonstrated well by the brightness temperature from MIMIC; plus, the maintenance of the CE structure was occasionally captured by the radars in Taiwan and Korea (Fig. 6).

Figure 6.  The MIMIC-IR brightness temperature (shading; K) at (a) 0000 UTC 1 September, (b) 1200 UTC 1 September, (c) 0200 UTC 2 September, (d) 1200 UTC 2 September, (g) 0000 UTC 4 September, (h) 1200 UTC 5 September, (i) 0000 UTC 5 September, (k) 0000 UTC 6 September; the observed reflectivity of the radar at CWB in Taiwan at (e) 0100 UTC 3 September and (f) 1200 UTC 3 September; and the observed rainfall of the radar at KMA in Korea at (j) 1200 UTC 5 September, related to Typhoon Hinnamnor (2022). The upper-left corner of each panel shows the JTWC-reported maximum sustained winds interpolated in time; the upper-right corner of each panel shows the time to the nearest actual image either before or after (whichever is closer); and the bottom-left corner of each panel shows the vertical wind shear. For each image, the red arrow is the 200–850-hPa environmental wind shear vector.

Isolated convective cells developed progressively and rotated cyclonically around the vortex center in the form of outer spiral rainbands (Fig. 6a, b). Both the eyewall and the outer spiral rainbands were most prominent in the downshear half of the storm where nascent convective cells weakened in the right-of-shear quadrants (Fig. 6c). There were two banded rainbands outside the inner eyewall that developed in the downshear quadrants (Fig. 6d), exhibiting strengthening maxima of brightness temperature. The CWB radar reflectivity occasionally captured the existence of the CE structure from 0100 to 1600 UTC 2 September (Fig. 6e, f). The outer spiral rainbands evolved into the outer eyewall with strong surrounding asymmetric rainbands. Both the inner and outer eyewall exhibited a circular shape with strong asymmetric rainbands spiraling outside (Fig. 6g–i). The inner and outer eyewalls began to exhibit asymmetric structures near the coast of South Korea (Fig. 6j) and record-breaking rainfall over mainland South Korea was brought by the asymmetric outer rainbands. The CE structure maintained for a very long period of about 84 h from 1200 UTC 2 to 0000 UTC 6 September (Fig. 6k).

The azimuthal mean brightness temperature of the outer rainband not only displays an increase in areal coverage, but also a clear inward contraction from a radius of about 300 to 150 km (Fig. 5). This inward contraction demonstrates well the merger of the tropical depression with Hinnamnor. At around 1200 UTC 2 September, the rainband started to axisymmetrize (Fig. 6d) when the outer spiral rainband continuously wrapped around the center. These signals mark the beginning of SEF and indicate that the secondary eyewall was indeed sourced from this intensifying outer rainband. The above definition of SEF allows us to distinguish the SEF signal from the azimuthal mean projection of the asymmetric rainband. Both the inner eyewall and outer eyewall convection were strengthened from the beginning of the SEF process to 0000 UTC 5 September, corresponding to the general intensification of Hinnamnor during this period. Later, Hinnamnor weakened owing to the dissipation of both the inner and outer eyewall. This dissipation may have been influenced by external forcing such as the eastward movement of the westerly trough (Fig. 2b) and the low sea surface temperature because Hinnamnor moved towards the Korean Peninsula.

Based on multi-source observational and analysis datasets, two SEF cycles occurred during the lifetime of Super Typhoon Hinnamnor (2022), making it a highly unique and prominent typhoon in 2022. SEF1 occurred close to the time of Hinnamnor's maximum intensity and the internal dynamics [as discussed in Wang et al. (2019) and Yu et al. (2021)] seemed to dominate the process of SEF after the formation of shear-induced asymmetric spiral rainbands. The binary interaction between a tropical depression (13W) and Hinnamnor along with the completion of ERC1 led to growth of the inner-core size and active development of the outer spiral rainbands. The effect of the outer spiral rainbands on the storm intensity was to weaken it. Once the spiral rainbands had wrapped around the inner eyewall, SEF2 began and the weakening trend of the intensity ceased. During what was a long period with a CE structure (approximately 84 h), the intensity of Hinnamnor fluctuated, first increasing and then decreasing. The external dynamics may have played important roles in triggering the SEF2 process.

Based on the structural evolution of Hinnamnor, SEF1 and ERC1 belongs to the “ERC type” as categorized by Yang et al. (2013), while SEF2 and the CE2 belongs to the CEM type (a CE structure that is maintained for an extended period). In their study, the ERC was defined as having finished in less than 20 h (vs 28 h in Hinnamnor) after the SEF, and the mean duration of the CE structure was 31h (vs 84 h in Hinnamnor). This long duration of the CE structure in Hinnamnor is very unusual and different from the majority of TCs in the western North Pacific that underwent SEF. In the present case, SEF1 did indeed occur near the time of Hinnamnor's maximum intensity, whereas SEF2 happened in the period of weakening. This indicates that there was still potential for a secondary eyewall structure to be generated and maintained as the storm weakened, although this new eyewall was established owing to the development of the spiral rainbands following the binary interaction and completion of ERC1.

It is an interesting finding that the intensity of Hinnamnor weakened continuously with time during the process of binary interaction, while similar binary interaction led to the rapid intensification of Typhoon Megi (2010) (Wang et al., 2013; Wang and Wang, 2014). This intensity difference between these two cases can be explained by the different evolution of the active spiral rainbands. The binary interaction in these two cases was highly effective in triggering active spiral rainbands. The diabatic heating in the spiral rainbands of Megi played a critical role in driving the low-level inflow, further enhancing eyewall updrafts and convection and ultimately leading to its rapid intensification. In the case of Hinnamnor, the diabatic heating in the spiral rainbands may have brought absolute angular momentum inwards, thereby increasing the tangential wind speed outside the eyewall and causing the generation of the secondary maximum wind that helped form the secondary eyewall.

The increasing vertical wind shear during SEF1 contributed to generation of the asymmetric spiral rainbands in the downshear half of the storm, and the steady evolution of the moderate environmental wind shear may have helped maintain the CE structure for a relatively long time. Based on an idealized simulation, Wang et al. (2019) found that an inward-propagating outer rainband generated in a quiescent environment and was able to trigger SEF through the asymmetric dynamics. With the vertical wind shear added in the idealized simulation, Wang and Tan (2022) showed that the evolution of the outer rainbands can be influenced, further impacting upon the SEF process. It is hard to distinguish the relative importance of the vertical wind shear and the internal dynamics in SEF1. However, previous studies have verified that the internal dynamics are key to the SEF process because not all the asymmetric rainbands (no matter how they developed) can evolve into a secondary eyewall.

Our results highlight the TC features and evolving structures during SEF and ERC/CE processes that are likely to be associated with storm intensity change. Additional work must be performed to analyze these internal and external dynamics to determine their relative contributions to SEF in the proposed theories of this mechanism. On a final note, it is important to acknowledge that the analysis in this paper is based on limited observational datasets. Therefore, achieving a better understanding of the detailed dynamics of the two SEF processes will remain as a topic to be tackled in future work.