The Implication of Outflow Structure for the Rapid Intensification of Tropical Cyclones under Vertical Wind Shear (original) (raw)

1. Introduction

Accurately forecasting tropical cyclone (TC) intensity remain a challenging task (Cangialosi et al. 2020; Knaff et al. 2020), especially for TCs that undergo rapid intensification (RI; defined as an increase of at least 30 kt (1 kt ≈ 0.51 m s−1) in maximum wind speed over a 24-h period; Kaplan and DeMaria 2003). The difficulties in forecasting RI stem from the complex controlling factors involved in the occurrence of RI, such as the large-scale environment (Kaplan and DeMaria 2003; Kaplan et al. 2010; Chen et al. 2015), internal dynamics (Zhang and Chen 2012; Miyamoto and Takemi 2015; Chang and Wu 2017; Wang et al. 2021) and multiscale interactions (Fang and Zhang 2011; Bhalachandran et al. 2020). Previous studies (Kaplan and DeMaria 2003; Shu et al. 2012) showed that the RI TCs usually experience higher SST, higher lower-level humidity and weaker vertical wind shear (VWS) than the non-RI TCs, implying the importance of the large-scale environment on RI. Evidence indicates that intensity forecasts for TCs under a moderate VWS (defined as 4.5–11 m s−1; Rios-Berrios and Torn 2017) exhibit large errors (Bhatia and Nolan 2013). Based on a series of ensemble experiments, Zhang and Tao (2013) found that the likelihood and timing of RI become highly uncertain in the presence of moderate VWS. Therefore, understanding the mechanism leading to RI under moderate VWS is crucial for the improvement of RI forecasts.

It is well known that moderate-to-strong VWS has a detrimental effect on the TC intensification. From the thermodynamic aspect, environmental VWS can reduce the efficiency of TC heat engine by midlevel ventilation (Tang and Emanuel 2010) or transport of lower-entropy air into the boundary layer via strong downdrafts (Riemer et al. 2010; Ge et al. 2013; Alland et al. 2021a). Dynamically, VWS would tilt the TC vortex from its upright alignment (Jones 1995), with the midlevel vortex generally displaced downshear relative to the low-level TC center. The balanced vertical motion in response to the tilt can then lead to a wavenumber-1 convective asymmetry, with active convection located in the down-tilt direction and suppressed convection in the up-tilt direction (Wang and Holland 1996; Frank and Ritchie 2001). The vertically tilted TC usually has a poorly organized and shallow secondary circulation, which is unfavorable for the inward transport of absolute angular momentum and thus the spinup of the TC primary circulation (Rogers et al. 2015; Tao and Zhang 2019).

To intensify rapidly under VWS, TCs must overcome the aforementioned negative effects. Observational and modeling studies documented that TCs experiencing RI under moderate VWS share some similar structural changes prior to RI onset (Rogers et al. 2015; Rios-Berrios et al. 2018; Ryglicki et al. 2018b; Tao and Zhang 2019). Namely, the midlevel vortex displaced downshear by VWS would rotate cyclonically to the upshear flank, followed by a significant decrease of vortex tilt. This process is generally termed as “precession” and is accompanied by the axisymmetrization of inner-core precipitation. A TC completing the precession (i.e., becoming vertically aligned) is expected to establish a compact eyewall with well-organized convection and thus is more likely to undergo RI (Chen et al. 2018a; Tao and Zhang 2019; Shi et al. 2020). On the other hand, other studies (Molinari et al. 2006; Nguyen and Molinari 2015; Chen et al. 2018b) suggested that for some sheared TCs the active convection in the downshear flank can produce a mesovortex in the lower-to-mid troposphere. If this mesovortex is strong enough, it will axisymmetrize and replace the weak parent TC vortex, leading to vortex realignment and RI. This mechanism is usually called “downshear reformation.”

All abovementioned mechanisms regarding RI under vertical wind shear mainly focus on the interaction between convection and circulation at the lower to midlevels, while relatively less attention has been paid to the upper troposphere. The upper layers of TCs are generally more asymmetric than the mid-to-lower layers, with the outflow concentrated in one or two outflow channels (Black and Anthes 1971). Previous studies have shown evidence that the interactions between the TC and upper-tropospheric systems can affect TC intensification under VWS. For instance, when a TC approaches an upper-level trough, it may be subject to an increasing VWS and experience slower intensification or weakening (DeMaria et al. 1993; Peirano et al. 2016), whereas in some cases the TC–trough interaction may favor TC intensification by enhancing the eddy flux convergence of angular momentum and potential vorticity (Molinari and Vollaro 1989; Hanley et al. 2001; Leroux et al. 2016). This dual influence of TC–trough interaction is generally known as the “bad trough–good trough” issue. Recently, Fischer et al. (2017) suggested that the distribution of the trough-induced quasigeostrophic (QG) ascent relative to the TC plays an important role in modulating the TC intensity under VWS. Specifically, the QG ascent located preferentially in the upshear (downshear) quadrants of the TC likely promotes (impedes) the axisymmetrization of convection and results in a higher (lower) intensification rate.

In addition to the upper-level trough, the TCs may also be sheared by the synoptic-scale anticyclones in the upper troposphere. Using the satellite-based observations, Ryglicki et al. (2018a) investigated the atypical RI of six TCs under moderate VWS and found that these TCs were all influenced by the upper-level anticyclones. They hypothesized that the TC sheared by an upper-level anticyclone is more likely to undergo RI than that sheared by an upper-level trough, due to the fact that the strong upper-level wind produced by the anticyclone is shallower in depth than that by the trough. Later on, Ryglicki et al. (2019) proposed an outflow–environment interaction mechanism that contributes to the resilience of a TC to the moderate VWS. Specifically, they showed that the divergent upper-level outflow emanated from down-tilt convection acts to block the background environmental flow around the TC, which leads to a decrease in the local VWS in the TC inner-core region and thus promotes the occurrence of RI. They also suggested the outflow blocking is more effective to lead to RI when TC is sheared by the upper-level anticyclone whose circulation is mainly confined in the same layer of the TC outflow. Hereafter in this paper, the term outflow blocking will refer to the process that the developed TC outflow gradually expels the upper-level environmental flow from the TC inner-core region. Ryglicki et al. (2020) further demonstrated the outflow blocking mechanism based on the analysis of a hierarchy of datasets: reanalysis, an analytical model, a divergent shallow water model, and a full-physics model. This mechanism was later employed by Ryglicki et al. (2021) in an observational study to explain the RI of Hurricane Dorian (2019) under moderate VWS.

The role of outflow in resisting and deflecting the upper-tropospheric environmental wind has also been documented by an early observational study of Elsberry and Jeffries (1996). More recently, Dai et al. (2021) carried out a series of idealized simulations in which a strong VWS of 14 m s−1 is imposed on a TC of major-hurricane intensity and investigated the outflow–environment interaction therein. Specifically, the modulation effect of the TC outflow on the environment is quantified by a TC-induced shear difference (TCSD), defined as the vector difference between the local shear around the TC and the environmental shear (by their definition, the 200–850-hPa shear averaged over the radii of 0–500 and 500–1000 km, respectively). They found that the development of the outflow in the upshear quadrants produces a TCSD acting against the environmental shear and thus helps the resilience and reintensification of the TC, which is analogous to the results of Ryglicki et al. (2019). Furthermore, by performing a statistical analysis using the TC best track and reanalysis data, Dai et al. (2021) demonstrated a significant correlation between the TCSD and TC intensity change under VWS greater than 10 m s−1. Overall, the results of Dai et al. (2021) indicate that the outflow-blocking phenomenon is not limited to the regime of moderate VWS and may serve as a precursor for the intensification of the real-world sheared TCs.

The current study aims to extend the work of Ryglicki et al. (2019) and Dai et al. (2021) by performing a climatological analysis on the outflow–environment interaction under VWS. In particular, we intend to address the following questions: 1) how the outflow evolves relative to RI onset under environmental VWSs of different strengths, 2) how the outflow structures differ between sheared TCs with different intensification rates, and 3) why some TCs can become resilient to VWS via the outflow blocking mechanism while the others cannot. These intriguing issues will be investigated for Northern Hemisphere TCs between 1980 and 2019 using various datasets including TC best track, reanalysis and infrared brightness temperature (T IR). Storm-centered shear-relative composites will be generated to compare the outflow structure, convection and VWS at different stages relative to RI onset and in cases of different intensification rates, which is unique from previous studies (Ryglicki et al. 2019, 2020; Dai et al. 2021; Ryglicki et al. 2021). By answering the questions presented, we hope to confirm that the outflow blocking is a typical mechanism contributing to intensification and an important phenomenon differentiating between the RI and non-RI TCs under moderate-to-strong VWS.

The rest of this paper is organized as follows. Section 2 introduces the datasets and methods used in this study. Section 3 explores the outflow evolution in rapidly intensifying TCs under VWSs of various strengths and its relationship with the convection evolution. A comparison of the outflow structures between TC cases with different intensification rates and its implications for RI forecasts are described in section 4. A summary of the paper is presented in section 5.

2. Data and methods

a. Data

This study analyzes the TCs forming in the western North Pacific (WNP), eastern North Pacific (ENP), and North Atlantic (NA) basins during 1980–2019. Although the TCs in the Southern Hemisphere are not examined in this study, the outflow-blocking mechanism can also be applied to them. An example of the outflow blocking in Southern Hemisphere RI TCs can be found in Ryglicki et al. (2020) (see their Fig. 3d). The 6-hourly storm intensity and position estimates for the WNP TCs and the ENP/NA TCs are recorded from the best track datasets of the Joint Typhoon Warning Center (JTWC) and the National Hurricane Center (NHC), respectively. The atmospheric fields surrounding TCs are obtained from the ERA5 reanalysis of European Centre for Medium-Range Weather Forecasts with a horizontal resolution of 0.25° × 0.25° (Hersbach et al. 2018). In addition, the T IR data on a 0.07° × 0.07° grid from Gridded Satellite B1 (GridSat-B1; Knapp et al. 2011) is used to characterize the convective activities.

b. Cases classification and RI event

The 6-hourly records in the best track data are classified into four groups according to the changes in maximum-sustained 10-m wind in the subsequent 24 h (Δ_V_ max), including RI (Δ_V_ max ≥ 30 kt), slow intensification (SI; 10 ≤ Δ_V_ max < 30 kt), neutral (NT, −10 < Δ_V_ max < 10 kt) and weakening (WK; Δ_V_ max ≤ −10 kt). Note that RI criterion is defined as the 95th percentile of 24-h intensity changes of the TCs in the Atlantic basin (Kaplan and DeMaria 2003). Each 24-h period following a synoptic time is treated as a separate case. In this study, we focus on the cases that are weaker than major hurricanes (i.e., TCs ≤ 95 kt), because the intensity fluctuation of major hurricanes is commonly associated with eyewall replacement cycles (Willoughby et al. 1982). In addition, the cases to the north of 30°N are not considered in this study, since they are likely subject to extratropical influences and relatively cold SST such that VWS may be a less important factor controlling their intensity changes. The cases making landfall within 24 h are also excluded.

Following Tao et al. (2017), a RI event is defined as a continuous period during which the TC intensification rate always satisfies the RI criterion. In general, a RI event can last as long as 48–60 h. The onset of the RI event is defined as the first time point that a TC intensifies at least 30 kt in the subsequent 24 h. By definition, the RI event can be composed of one RI case (when the RI event only lasts for 24 h) or a sequence of RI cases. A composite analysis of TC outflow evolution will be performed with respect to the onset of each RI event in section 3. The main advantage of the RI-event-based composite is that the non-RI stage and RI stage can be strictly separated by RI onset. Thus, the RI-event-based composite is very suitable for studying the storm evolution leading up to RI onset (Zagrodnik and Jiang 2014; Tao and Jiang 2015). In contrast, the RI case does not really differentiate between the non-RI and RI stages, since TCs may have already started RI before a RI case. This is why we adopt the RI-event-based composite to explore the outflow blocking leading up to RI in section 3.

Although the case-based analysis is not suitable for studying the evolution the RI storm, it would cause no problem when only comparing RI cases with non-RI cases (Tao et al. 2017). Moreover, the case-based approach can increase the sample size for composite analysis and thus enhance the robustness of the results. Numerous studies (e.g., Kaplan and DeMaria 2003; Hendricks et al. 2010) have used the case-based analysis to investigate the differences in environmental conditions and storm characteristics between the RI and non-RI TCs. In section 4, a case-based composite analysis will be performed to illustrate how the outflow structures differ among different intensity change categories. In this study, the statistical significance of differences will be evaluated using a bootstrap test similar to Chen and Chou (2014).

c. Shear definitions

To better explore the interaction between the TC outflow and environmental flow, two types of VWS are examined in this study. Specifically, the environmental VWS is defined as the 200–850-hPa vector difference between the area-averaged winds within the 400–800-km radii, and the local VWS is calculated using the area-averaged winds within the 0–200-km radii. The definition of the local VWS is consistent with Ryglicki et al. (2019), whereas our radial extent used to calculate the environmental VWS is located farther from the TC center than that in their idealized numerical study (200–500 km) in order to better characterize the large-scale wind field. Since TC position in the reanalysis may deviate from the best track data, the TC centers in ERA5 need to be reassessed before calculating the VWS. Here, the TC center is identified as the 850-hPa geopotential centroid within a 500 km × 500 km domain centered in the estimated center, following the method of Nguyen et al. (2014). Results indicate that, among most of the cases, the TC center in the reanalysis is displaced 30–100 km from that in the best track data (not shown). It is worth noting that the results shown in the rest of this study are not sensitive to reasonable choices of radial extent used to calculate the environmental VWS. We have repeated our analysis based on the definitions of environmental VWS from other studies, such as the 500–750-km mean shear (Tao et al. 2017), the 300–800-km mean shear (Chen et al. 2018a) and the 200–800-km mean shear used in the Statistical Hurricane Prediction Intensity System (SHIPS; DeMaria et al. 2005), and the obtained results are qualitatively consistent (not shown).

Following Rios-Berrios and Torn (2017), the moderate VWS is defined as 4.5–11 m s−1, which corresponds to the 25th and 75th percentiles of the VWS experienced by TCs globally. The VWSs less than 4.5 m s−1 and greater than 11 m s−1 are defined as the weak VWS and strong VWS, respectively. According to the strength of environmental VWS at the initial time of cases, the cases in each intensity change category are further divided into three groups: the weak-shear cases, moderate-shear cases and strong-shear cases. The numbers of different types of intensity change cases are summarized in Table 1. Also, the RI events are divided into the weak-shear RI events, moderate-shear RI events and strong-shear RI events. However, the three types of RI events are classified according to the mean environmental VWS during 0–24 h before RI onset. It will be shown that with this approach the magnitudes of environmental VWS in each group do not vary substantially before RI onset. Note that the temporal mean VWS is calculated by averaging the environmental VWS at the synoptic times with a 6-h interval. There are total 715 RI events during 1980–2019, among which 252, 431 and 32 RI events are identified as the weak-shear, moderate-shear and strong-shear events, respectively (Table 1). To validate the reasonability of the classification, the box-and-whisker plots of the environmental VWS for the three categories of RI events from t = −24 to 0 h relative to RI onset are shown in Fig. 1. It is clear the distribution of the environmental VWS of each category differs considerably from those of the other categories. The difference in the environmental VWS between any two categories is statistically significant at 99% confidence level at each time point. Moreover, the evolution of mean VWS from t = −24 to 0 h is generally steady for all three categories of RI events. These results indicate that the RI events influenced by sustained weak, moderate or strong shears preceding RI onset are reasonably separated.

Table 1.

The numbers of the four categories of cases and the RI events classified under the different strengths of environmental VWS.

Table 1.

Table 1.

Fig. 1.

Fig. 1.

Fig. 1.

Box-and-whisker plots of environmental VWS from −24 to 0 h relative to RI onset for the RI events under (a) weak VWS, (b) moderate VWS, and (c) strong VWS. Whiskers extend from the 5th to 95th percentile, boxes extend from the 25th to 75th percentile, horizontal lines within the boxes represent medians, and black curves represent means of the distributions.

Citation: Monthly Weather Review 149, 12; 10.1175/MWR-D-21-0141.1

3. Composite results for RI events

a. Evolutions of outflow structure and VWS

Figure 2 depicts the evolution of composite 200-hPa wind for the RI events in each VWS group from t = −24 to 24 h centered on RI onset. Note that before being composited the wind field of each event has been rotated such that the environmental VWS points to the west. Here, the 200-hPa wind field is used to characterize the TC outflow, such that the evolutions of outflow can be readily related to the changes in 200–850-hPa shear. Note that in specific cases the strongest signal of outflow blocking may occur at other vertical levels since the maximum outflow can be above or below 200 hPa (Ditchek et al. 2017). Thus, it is necessary to look at multiple levels when studying or observing the outflow blocking in a certain TC. Nevertheless, in the context of composite study, we will show that the wind evolution at 200 hPa is sufficient to depict the outflow blocking process.

Fig. 2.

Fig. 2.

Fig. 2.

Composite 200-hPa streamline and total wind velocity (shading, m s−1) from −24 to 24 h relative to RI onset with 12-h intervals for the RI events under (top) weak, (middle) moderate, and (bottom) strong environmental VWS. The red typhoon symbol denotes the 850-hPa TC center. The magenta X represents the location of the stagnation point. The gray circles mark the 200-, 400-, and 600-km radii, respectively. The environmental shear direction is denoted by the vector in the bottom right of the figure.

Citation: Monthly Weather Review 149, 12; 10.1175/MWR-D-21-0141.1

For the weak-shear RI events, the outflow has been established in all shear-relative quadrants and extended outside the 600-km radius since t = −24 h (Fig. 2a). With time, the outflow gradually strengthens and becomes more symmetric (Figs. 2b–e). Unlike the weak-shear RI events, there is evident inflow penetrating into the inner-core region in the upshear flank at t = −24 h for the moderate-shear and strong-shear RI events (Figs. 2f,k), indicating a stronger impact of the upper-level environmental wind on TC. As found in previous studies, the environmental flow tends to split as it approaches the TC center (Ryglicki et al. 2019, 2021), indicating the environmental flow is rerouted by the TC outflow.

For the convenience of describing the outflow–environmental interaction, a stagnation point (magenta X in the figure) is defined as the location of minimum wind velocity in the upshear side of the storm center. As suggested by Ryglicki et al. (2019), the stagnation point represents the location where the TC outflow and environmental flow cancel each other out and it would be gradually pushed away from the TC center as the outflow develops and blocks the upper-level environmental flow. Note that the stagnation point is not plotted in Figs. 2a–e, since there is no discernable interface between the outflow and environmental flow in the composite wind field of the weak-shear RI events.

For the moderate-shear RI, the stagnation point is originally in the left-of-shear side and within the radius of 100 km at t = −24 h (Fig. 2f), and then moves outward continuously to 275 km upshear (Figs. 2g,h) at RI onset. The environmental easterly flow is well blocked from the inner-core region at RI onset (Fig. 2h). During the 24-h period after RI onset, the stagnation point is pushed outward further to 350 km and strong cyclonic winds are observed in the region within the 200-km radius (Figs. 2i,j). As a result, the storm is protected from the intrusion of upper-level environmental flow during RI. For the strong-shear RI events, the stagnation point remains very close to the TC center until t = −12 h (Figs. 2k,l) and then starts to be pushed upshear continuously, reaching 125 km upshear at RI onset (Fig. 2m) and 300 km upshear at t = 24 h (Fig. 2o). Overall, the evolution of composite wind fields at the outflow layer for the moderate-shear and strong-shear RI events bears resemblance to those found in previous numerical and observational studies (Ryglicki et al. 2019, 2020; Dai et al. 2021; Ryglicki et al. 2021), indicating the outflow blocking is a common feature of the RI events under moderate-to-strong VWS.

Ryglicki et al. (2019) suggested that the outflow blocking usually happens when the TC is sheared by the upper-tropospheric anticyclone. To test this hypothesis, we visually inspected the nondivergent wind fields at t = −12 h of all the moderate-shear and strong-shear RI events. It is found that about 76% (353 out of 463) of RI events under moderate-to-strong environmental VWS are sheared by the upper-level anticyclones, which strongly supports the hypothesis of Ryglicki et al. (2019).

To illustrate the evolution of the wind field in more detail, the differences between t = −24 h and the subsequent times are shown in Fig. 3. For the moderate-shear RI events, the most prominent change in the composite wind field appears to be the continuous enhancement of outflow in the upshear flank of the storm center. The region of statistically significant positive radial flow changes extends from 300-km radius at t = −12 h to more than 600-km radius at RI onset (Figs. 3c,d), indicating the development of TC upshear outflow is responsible for the outward migration of the stagnation point. In addition, there are strengthened cyclonic winds within the 400-km radius at RI onset compared to t = −24 h, reflecting the upward development of the TC primary circulation. The mean intensity change over the 24-h period leading up to RI onset for the weak-shear, moderate-shear, and strong-shear RI events are 14, 15, and 8 kt, respectively, This is consistent with Chang and Wu (2017), who suggested that the strengthening of primary circulation in the mid-to-upper troposphere is a necessary condition of RI.

Fig. 3.

Fig. 3.

Fig. 3.

Deviations of the composite 200-hPa winds (vectors; m s−1) at (left) t = −12 h and (right) RI onset from that at t = −24 h for the RI events under (top) weak, (middle) moderate, and (bottom) strong VWS. The magnitude of the total wind speed is proportional to the vector key in the bottom-right inset of each panel. Shading represents the radial component of the wind deviation (m s−1). Green hatching denotes the wind difference is significant at the 90% confidence level. The typhoon symbol denotes the 850-hPa TC center. The gray circles mark the 200-, 400-, and 600-km radii, respectively. The environmental shear direction is denoted by the vector in the bottom right of the figure.

Citation: Monthly Weather Review 149, 12; 10.1175/MWR-D-21-0141.1

Similar to the moderate-shear RI events, the strong-shear RI events also exhibit significant positive radial flow changes in the upshear flank (Figs. 3e,f), especially within the 200-km radius. The strongest positive changes are located at the upshear-right quadrant, extending outward to 600 km. In contrast, for the weak-shear RI events, the changes in the radial flow are quite weak within the 200-km radius (Figs. 3a,b). The most significant changes in radial flow, although still located in the upshear quadrants, are mainly outside of the 400-km radius (Fig. 3b). Therefore, the strengthening of the upshear outflow in the inner-core region generally precedes the RI events under moderate-to-strong shear, but is not a precursor for the weak-shear RI events.

Previous studies have shown that the outflow development in the upshear flank can modulate the VWS in the inner-core region (Ryglicki et al. 2019; Dai et al. 2021; Ryglicki et al. 2021). To see how the local VWS responds to changes in outflow, Fig. 4 depicts the box-and-whisker plots of the local VWS from t = −24 to 24 h relative to RI onset in the three categories of RI events. For the moderate-shear RI events, the local VWS shows a continuous decrease prior to RI onset (Fig. 4b), which is consistent with the modeling results (Ryglicki et al. 2019). The mean local VWS at RI onset is about 1.2 m s−1 smaller than that at t = −24 h, statistically significant at 95% confidence level, which is attributed to the significantly enhanced outflow in the upshear flank within a 200-km radius (Fig. 3d). Note that the significant decrease in the local VWS mainly happens from t = −24 to −6 h, during which the stagnation point moves from near the TC center to more than 200 km upshear (Figs. 2f–h). After that, the local VWS no longer shows significant change since the upper-level environmental flow has already been largely blocked from the inner-core region.

Fig. 4.

Fig. 4.

Fig. 4.

As in Fig. 1, but for the local VWS from −24 to 24 h relative to RI onset. The asterisk (circle) indicates the difference in the mean local VWS between a certain time and RI onset is significant at the 95% (90%) confidence level.

Citation: Monthly Weather Review 149, 12; 10.1175/MWR-D-21-0141.1

Likewise, in the strong-shear RI events the local VWS shows significant decrease from t = −6 to 12 h (Fig. 4c) when the stagnation point moves from the inside to the outside of the 200-km radius (Figs. 2l–n). The decrease in the local VWS prior to RI onset in the strong-shear RI events is stronger than that in the moderate-shear RI events (2.0 m s−1 versus 1.2 m s−1), which is consistent with the larger positive radial flow changes within the 200-km radius (Fig. 3f). It is worth mentioning that insignificant decreases of 0.2 and 0.8 m s−1 in the environmental VWS are observed from t = −24 to 0 h for the moderate-shear and strong-shear RI events, respectively (Figs. 1b,c). The fact that the local VWS decreases at a greater rate than the environmental VWS suggests that the decrease of local VWS preceding RI onset is mainly due to the blocking of the environmental flow, rather than the changes in the large-scale environment.

Unlike the RI events under moderate-to-strong VWS, the local VWS in weak-shear RI events only exhibits a decrease of 0.4 m s−1 during the 24 h prior to RI onset (Fig. 4a), consistent with the weak change of the radial wind within the 200-km radius (Figs. 3a,b). Although the local VWS decrease of 0.4 m s−1 is statistically significant and may play a positive role in RI, it is less attributed to the outflow blocking mechanism because the upshear outflow has already extended outward to the 600-km radius at t = −24 h (Fig. 2a). The decrease in local VWS in the weak-shear RI events is likely associated with the axisymmetrization of the inner-core wind field, as shown in Figs. 2a–c.

The above results indicate that the blocking effect due to the TC upper-level outflow on the environmental flow and the resultant decrease of local VWS can be viewed as a precursor for the RI events under moderate-to-strong shear, in good agreement with the mechanisms proposed by recent numerical studies (Ryglicki et al. 2019, 2020; Dai et al. 2021). Overall, the composite from the reanalysis proves that the outflow blocking may be one of the key mechanisms responsible for the occurrence of RI of TCs under moderate-to-strong VWS. By comparison, the process in which the developed upshear outflow gradually expels the upper-level environmental flow from the TC inner-core region is not observed in the weak-shear RI events, indicating that the dynamics controlling the RI under weak environmental shear may be different from that under moderate-to-strong environmental shear.

b. Convective evolution

Previous studies suggested that the convection related to rainbands of TC serves as an important source for outflow (Dai et al. 2019, 2021). To investigate the relationship between outflow and convection, the composite T IR for the RI events in each VWS group is examined (Fig. 5). As suggested in previous studies, sheared TCs exhibit a wavenumber-1 convective asymmetry with active convection occurring preferentially downshear (Corbosiero and Molinari 2003; Chen et al. 2006; Zhang and Tao 2013). Specifically, for all three categories of RI events, the areal coverage of T IR < −30°C is larger in the downshear flank than in the upshear flank at t = −24 h (Figs. 5a,d,g). Note that the convective asymmetry acts to strengthen with increasing VWS magnitude, consistent with Chen et al. (2006). During the 24 h prior to RI onset, all three types of RI events experience an intensification of the inner-core convection (Fig. 5). Moreover, the moderate-shear and strong-shear RI events undergo evident convective axisymmetrization before RI onset, as reflected by the increase in the coverage of T IR < −30°C within the 200-km radius (Figs. 5f,i).

Fig. 5.

Fig. 5.

Fig. 5.

Composite T IR (°C) from −24 to 0 h relative to RI onset with 12-h intervals for the RI events under (top) weak, (middle) moderate, and (bottom) strong VWS. The dashed black circles mark the 200- and 400-km radii. The environmental shear direction is denoted by the vector in the bottom right of the figure.

Citation: Monthly Weather Review 149, 12; 10.1175/MWR-D-21-0141.1

To better illustrate the convective axisymmetrization process, Fig. 6 shows the deviations of the composite T IR at t = −12 h and RI onset from that at t = −24 h. For the moderate-shear RI events, a significant decrease of T IR is observed in the upshear-left quadrant at t = −12 h (Fig. 6c), hinting at a tendency of convection symmetrization. At the RI onset, the significant decrease of T IR extends further into all shear-relative quadrants within the 200-km radius (Fig. 6d), indicating an overall strengthening of the inner-core convection. Nevertheless, the changes in T IR in the downshear flank is much weaker than those in the upshear flank. As a result, the distribution of T IR within a radius of 200 km becomes more symmetric relative to the storm center at RI onset (Fig. 5f). Note that the enhanced convection is collocated with the positive radial flow changes in the upshear flank, suggesting that the outflow blocking is closely associated with the convection axisymmetrization. Similarly, there is also significant decrease in T IR in the upshear quadrants for the strong-shear RI events (Figs. 6e,f), and the magnitude of T IR decrease is larger than that for the moderate-shear RI events, consistent with the stronger changes in radial flow (Fig. 3f). In contrast, the T IR decrease for the weak-shear RI events (Fig. 6b) is weaker than those for the strong-shear and moderate-shear RI events (Figs. 6d,f). The significant T IR decrease is observed in all shear-relative quadrants in the inner core, with the largest value upshear left. The inner-core convection also experiences axisymmetrization before RI onset in the weak-shear events. However, since the convective structure of the weak-shear RI events is already relatively symmetric at t = −24 h (Fig. 5g), the change in the convective structure prior to RI onset is less evident than those observed in the moderate-shear and strong-shear RI events.

Fig. 6.

Fig. 6.

Fig. 6.

Deviations of the composite T IR (°C) at t = −12 h and RI onset from that at t = −24 h for the RI events under (top) weak, (middle) moderate, and (bottom) strong VWS. Green dots denote the deviation is significant at 95% confidence level. The dashed black circles mark the radii of 200 and 400 km. The environmental shear direction is denoted by the vector in the bottom right of the figure.

Citation: Monthly Weather Review 149, 12; 10.1175/MWR-D-21-0141.1

As mentioned above, the extension of inner-core precipitation from downshear to upshear prior to the RI of sheared TCs has been well documented by observational (Zagrodnik and Jiang 2014; Rogers et al. 2015) and numerical (Riemer et al. 2010; Zhang and Tao 2013; Rios-Berrios et al. 2018; Ryglicki et al. 2018b, 2020) studies. A symmetric convective structure is suggested to favor the establishment of a deep secondary circulation and thus trigger TC RI (Miyamoto and Takemi 2013). In a sequence of recent works (Ryglicki et al. 2018b, 2019, 2020, 2021), the authors have proposed that the upshear migration of the inner-core tilt-modulated convective asymmetries is coupled with the outflow blocking. Specifically, as the tilt-modulated convective asymmetries precess from the downshear to the upshear side, the associated upper-level divergent wind also skews upshear, which enhances the outflow upshear to block the environmental wind. This provides an extra explanation for why in many sheared TCs the RI onset is preceded by the axisymmetrization of inner-core convection. Note that there are also RI events accompanied by the asymmetric convective structure (Molinari et al. 2006; Nguyen and Molinari 2012, 2015). For example, Nguyen and Molinari (2012) showed that a TC is capable of intensifying rapidly when the asymmetric convection is strong enough to dramatically increase the azimuthally mean diabatic heating inside the radius of maximum wind. The outflow evolution in the RI TCs with asymmetric convective structure is beyond the scope of the current study, which will be left for future work. Overall, from the context of climatology, the large-sample composite validates the relationship among the inner-core convection, upper-level outflow and local VWS proposed by previous studies, implying the importance of the outflow–environmental interaction on the resilience of a sheared TC to moderate-to-strong VWS and the subsequent RI occurrence.

4. Comparison among different intensity change cases

a. Outflow and convective characteristics

Discussions above focus on the RI events under different VWS strengths and indicate that the onset of the RI events under moderate-to-strong VWS follows a significant decrease of local VWS, which results from the blocking of upper-level environmental flow by the enhanced outflow in the upshear flank of the storm. To fully examine the importance of the outflow blocking mechanism on the occurrence of RI, comparisons between the RI and non-RI (including WK, NT and SI) cases are performed in this section. Since the RI under weak VWS has been proved to be less attributed to the outflow blocking mechanism, the following analyses only focus on cases under moderate and strong VWS. Note that the non-RI cases occurring after a RI event in a TC’s lifespan are excluded from our analysis. This is because some of these cases can have similar outflow features to the RI cases, but their intensification may be suppressed by other unfavorable conditions such as the dry air intrusion, low SST, and the proximity to the maximum potential intensity. Therefore, including these non-RI cases may obscure the differences in the composite results between the RI and non-RI cases.

Figure 7 shows the composite 200-hPa winds at the initial time of the four intensity change categories under moderate and strong environmental VWS. For both the moderate-shear and strong-shear RI cases, the composite wind structure (Figs. 7d,h) is similar to that at the onset of RI events (Figs. 2h,m), exhibiting the blocking of upper-level environmental flow by the outflow in the upshear flank of the storm. The stagnation points of the moderate-shear and strong-shear RI cases are located outside the 200-km radius, suggesting the environmental wind is well blocked from the storm inner-core region. By comparison, the stagnation points of the non-RI cases are all within the 200-km radius (Figs. 7a–c,e–g), hinting at a greater influence of environmental flow on these TCs. For the moderate-shear regime, it is interesting to note that the distance between the stagnation point and storm center increases with intensification rate, implying that the faster-intensifying storms are less susceptible to the influence of upper-level environmental flow. By comparison, for the strong-shear regime, the upper-level environmental wind intrudes in to the TC center in all three categories of non-RI cases, with no clear interface between the TC and the environmental flow. In short, these results suggest that the evident outflow blocking with a stagnation point outside the TC inner-core region is a hallmark of the RI TCs under moderate-to-strong VWS.

Fig. 7.

Fig. 7.

Fig. 7.

Composite 200-hPa streamline and total wind velocity (shading; m s−1) for (a),(e) WK; (b),(f) NT; (c),(g) SI; and (d),(h) RI cases under (top) moderate and (bottom) strong environmental VWS. The magenta X symbols represent the stagnation points. The gray circles mark the 200-, 400-, and 600-km radii, respectively. The environmental shear direction is denoted by the vector in the bottom right of the figure.

Citation: Monthly Weather Review 149, 12; 10.1175/MWR-D-21-0141.1

Figure 8 shows the differences of the 200-hPa wind and radial flow fields between the RI and non-RI cases. As can be expected, all three categories of non-RI cases are characterized by the significantly weaker outflow (indicated by the anomalous inflow) than the RI cases in the upshear flank of the storm center. To quantitatively compare the upshear radial flow among different cases, the radial flow difference averaged within the 200-km radius in the upshear quadrants is shown in the top-right of each panel. For the moderate-shear regime, the difference in the upshear radial wind increases from SI through NT to WK categories (Figs. 8a–c), well accounting for the decrease in the distance between the stagnation point and storm center with decreasing intensification rate. A similar increasing trend in the upshear radial flow difference with decreasing intensification rate can also be observed in the strong-shear cases (Figs. 8d–f), indicating the importance of the upshear outflow in TC intensification under moderate-to strong VWS.

Fig. 8.

Fig. 8.

Fig. 8.

Deviations of the composite 200-hPa winds (vectors; m s−1) and radial flow (shading; m s−1) of the (a),(d) WK; (b),(e) NT; and (c),(f) SI cases from those of the RI cases under (top) moderate and (bottom) strong environmental VWS. The magnitude of the total wind speed is proportional to the vector key in the bottom-right inset of each panel. The radial flow difference averaged within the 200-km radius in the upshear quadrants is shown in the top right of each panel. The green hatching represents that the deviation of the wind vector is significant at the 95% confidence level. The gray circles mark the 200-, 400-, and 600-km radii, respectively. The environmental shear direction is denoted by the vector in the bottom right of the figure.

Citation: Monthly Weather Review 149, 12; 10.1175/MWR-D-21-0141.1

To further explain the relationship between the outflow structure and TC intensification, the box-and-whisker plots of the local and environmental VWS in the four intensity change categories under the moderate and strong VWS are depicted in Fig. 9. As can be expected, the RI cases has the smallest local VWS among the four groups (Figs. 9b,d), consistent with the most prominent outflow blocking in the upper troposphere (Figs. 7d,h). In the moderate-shear regime, the mean local VWSs of the WK, NT, and SI cases are 1.8, 1.7, and 1.0 m s−1 stronger than that of the RI cases (Fig. 9b), respectively, which are all significant at 95% confidence level. The results for the strong shear regime are largely similar, except for that the differences in the local VWS between the RI and three non-RI categories become larger (4.6, 4.2 and 2.5 m s−1). On the other hand, the mean environmental VWS of the RI cases is also significantly smaller than those of the non-RI cases (Figs. 9a,c). Therefore, it is possible that the strength of environmental VWS can impact the change of local VWS. Namely, under relatively lower environmental VWS the upshear outflow may be more readily established and thus reduces the local VWS. Nevertheless, the differences in environmental VWS are much smaller than the counterparts in local VWS, indicating that the internal process of TC (i.e., stronger upshear outflow) plays a major role in lessening the local VWS of the RI cases.

Fig. 9.

Fig. 9.

Fig. 9.

Box-and-whisker plots of the (a),(c) environmental VWS and (b),(d) local VWS at the initial time for the four categories of the cases under (top) moderate VWS and (bottom) strong VWS. Whiskers extend from 5th to 95th percentile, boxes extend from the 25th to 75th percentile, horizontal lines within the boxes represent medians, and black dots represent means. The letters W, N, S, and R in the parentheses represent that the parameter is significantly different from that of the WK, NT, SI, and RI category at the 95% confidence level, respectively.

Citation: Monthly Weather Review 149, 12; 10.1175/MWR-D-21-0141.1

To verify the importance of the upshear outflow on the local VWS, Figs. 10a and 10d show the scatterplots of the radial wind averaged within a 200-km radius in the upshear flank versus the local VWS for the moderate-shear and strong-shear cases, respectively. Clearly, the local VWS shows a decreasing trend with increasing upshear outflow, regardless of the strength of environmental shear. The correlation between the two parameters reaches −0.55 and −0.75 for the moderate-shear and strong-shear cases, respectively, which are both significant at 99% confidence level. This result strongly supports that the lower local VWS in the RI cases is associated with the enhanced upper-level outflow in the upshear flank.

Fig. 10.

Fig. 10.

Fig. 10.

Scatterplots of (a) local shear vs upshear radial wind averaged within a 200-km radius, (b) 24-h TC intensity change vs the environmental shear at t = 0 h, and (c) 24-h TC intensity change vs the local shear at t = 0 h. The regression line is shown in red, and the correlation coefficient is shown at the top left of each subplot. All correlation coefficients are significant at the 99% confidence level.

Citation: Monthly Weather Review 149, 12; 10.1175/MWR-D-21-0141.1

Another important finding from Fig. 9 is that the differences in local VWS between the SI and the other non-RI groups are also statistically significant at 95% confidence level (Figs. 9b,d), whereas the differences in environmental VWS between any two non-RI groups are insignificant (Figs. 9a,c). This suggests that the future intensity changes of TCs under moderate or strong VWS may be more correlated with the local VWS than the environmental VWS. To further confirm this hypothesis, the middle and right columns of Fig. 10 show the scatterplots of the environmental/local VWS versus 24-h future intensity change. Note that the results in these panels are obtained using the cases of all intensity change categories including WK, NT, SI and RI. For both the moderate-shear and the strong-shear cases, the environmental shear shows a weak correlation with the 24-h future intensity change, with the correlation coefficients of −0.13 and −0.15, respectively (Figs. 10b,e). In contrast, the correlation coefficients between the local shear and 24-h future intensity change exhibit a moderate increase, being −0.22 for the moderate-shear cases and −0.23 for the strong-shear cases (Figs. 10c,f), respectively. Hence, the local VWS can act as a better potential predictor than the environmental VWS for future intensity change of TCs. This result is consistent with Dai et al. (2021), although their correlation analysis focused on the TCs under strong shear greater than 10 m s−1.

The distributions of composite T IR for the four intensity change categories are compared in Fig. 11. Consistent with previous studies (Kaplan et al. 2010; Jiang 2012; Zagrodnik and Jiang 2014), the RI TCs (Figs. 11d,h) have stronger convection and larger convection coverage in the inner-core region than the non-RI TCs (Figs. 11a–c,e–g). Moreover, the distribution of T IR becomes more symmetric as the intensification rate increases, as indicated by the increasing coverage of T IR < −30°C the in the upshear flank, consistent with the finding of Fischer et al. (2018). Figure 12 further shows the T IR deviations of the non-RI categories from that of the RI category. The most prominent differences in T IR between RI and three non-RI categories are all situated in the upshear flank within the 200-km radius, under either the moderate or the strong environmental VWS. This further confirms the crucial role of the upshear convection in the occurrence of RI. In short, the above analysis suggests that TCs with weaker convection and thus weaker outflow in upshear flank are more vulnerable to VWS, supporting the idea that the outflow blocking of upper-level environmental flow is an important mechanism by which TCs can overcome the detrimental impact of the moderate-to-strong environmental VWS and commence RI.

Fig. 11.

Fig. 11.

Fig. 11.

Composite T IR (°C) for the (a),(e) WK; (b),(f) NT; (c),(g) SI; and (d),(h) RI cases under (top) moderate VWS and (bottom) strong VWS. The dashed circles mark the 200- and 400-km radii. The environmental shear direction is denoted by the vector in the bottom right of the figure.

Citation: Monthly Weather Review 149, 12; 10.1175/MWR-D-21-0141.1

Fig. 12.

Fig. 12.

Fig. 12.

Deviations of the composite T IR (°C) of the (a),(d) WK; (b),(e) NT; and (c),(f) SI cases from that of the RI cases under (top) moderate VWS and (bottom) strong VWS. The black dots denote the deviation is significant at the 95% confidence level. The dashed circles mark the radii of 200 and 400 km. The environmental shear direction is denoted by the vector in the bottom right of the figure.

Citation: Monthly Weather Review 149, 12; 10.1175/MWR-D-21-0141.1

b. Possible causes for the different strengths of upshear convection

Although the above comparisons of the outflow and convective structures prove the importance of the outflow blocking mechanism on RI under moderate-to-strong VWS, it is still unclear what is responsible for the different strengths of the upshear convection and outflow among intensity change cases. Addressing this problem is important for answering why some TCs can become resilient via the outflow blocking mechanism while others cannot. Previous studies suggested that the environmental thermodynamic conditions have a great impact on the strength and spatial distribution of the inner-core convection in sheared TCs (Ge et al. 2013; Rios-Berrios and Torn 2017; Alvey et al. 2020). Therefore, we compare several relevant parameters among different intensity change categories (Figs. 13 and 14) in the shear-relative framework, including the midlevel relative humidity (RH; evaluated at 700 hPa), SST and surface latent heat flux (LHF). Note that all parameters being examined are averaged over the 24-h period prior to the initial time of the cases, since the mean thermodynamic conditions before the cases influence the development or maintenance of upshear convection and thus the strength of upshear convection at the initial time of the cases. Following Rios-Berrios and Torn (2017), the LHF is calculated through the bulk aerodynamic formula:

LHF = L υ c k ρ d U 10 ⁡ ( q * − q 2 ) ,

where L υ is the latent heat of vaporization (2.5 × 106 J kg)−1, c k is the exchange coefficients for latent heat (taken as 1.2 × 10−3; Nguyen et al. 2019), ρ d is the dry air density (taken as 1.2 kg m−3), U 10 is the 10-m wind speed, q* is the saturated specific humidity at SST, and q 2 is the 2-m specific humidity. Note that the surface sensible heat flux is not examined here because the total surface enthalpy flux is generally dominated by the LHF (Nguyen et al. 2019).

Fig. 13.

Fig. 13.

Fig. 13.

(a),(e),(i) Composite thermodynamic parameters of all cases under moderate environmental VWS, and the deviations of the parameter of the (b),(f),(j) WK; (c),(g),(k) NT; and (d),(h),(l) SI cases from that of the RI cases: (top) 700-hPa RH (%), (middle) SST (°C), and (bottom) surface LHF (W m−2). All parameters are averaged from t = −24 to 0 h relative to the initial time of the cases. The black dots denote the deviation is significant at the 95% confidence level. The environmental shear direction is denoted by the vector in the bottom right of the figure.

Citation: Monthly Weather Review 149, 12; 10.1175/MWR-D-21-0141.1

Consistent with the climatological analysis of Rios-Berrios and Torn (2017), the composite mean RH field for the moderate-shear cases exhibits a clear wavenumber-1 asymmetry, with the larger RH downshear and the smaller RH upshear (Fig. 13a). Composite differences between the groups reveal that the RI cases have significantly higher 700-hPa RH than the WK and NT cases outside the 200-km radius, with the largest differences observed in the left-of-shear to upshear-left quadrant (Figs. 13b,c). This indicates that the RI cases are less susceptible to the intrusions of environmental dry air than the WK and NT cases, which mitigates the detrimental influence of the midlevel ventilation on the upshear convection (Ge et al. 2013; Alvey et al. 2020; Alland et al. 2021b). By comparison, the RH differences between the SI and RI cases are not significant (Fig. 13d), indicating other factors may be responsible the different strengths of upshear convection in these two groups. An examination of the midlevel RH fields for the strong-shear cases (Figs. 14a–d) shows similar results with that for the moderate-shear cases. The RI cases under strong shear also have higher environmental RH than the WK and NT cases and the significant differences are more concentrated in the upshear-left quadrant. The RH differences between the RI and SI cases are again insignificant.

Compared with the RH, the distributions of the mean SST and the SST differences are quite uniform (Figs. 13e–h,14e–h), with weak contrasts between different shear-relative quadrants. For both the moderate-shear and strong-shear regimes, the TCs with greater intensification is over the higher SST, which agrees with previous studies (Kaplan and DeMaria 2003; Rios-Berrios and Torn 2017; Nguyen et al. 2019).

The composite mean LHF for the moderate-shear TCs shows a wavenumber-1 asymmetry with the largest LHF in the left-of-shear flank (Fig. 13i). The RI cases have significantly higher LHF at all quadrants than the non-RI cases, and the difference in LHF increases with the difference in intensification rate (Figs. 13j–l). The largest LHF differences from the RI cases appear in the upshear flank within the 200-km radius for each non-RI group, consistent with the previous composite results using the reanalysis (Rios-Berrios and Torn 2017) and dropsonde observations (Nguyen et al. 2019). Previous studies have suggested that the high surface enthalpy flux in the left-of-shear and upshear quadrants can help recover the lower-entropy air flushed into the boundary layer by the shear-induced downdrafts, and thus favor the axisymmetrization of the convection (Rappin and Nolan 2012; Chen et al. 2021; Wadler et al. 2021). Therefore, the higher LHF may play an important role in the stronger upshear convection in the RI cases.

In the strong-shear TCs, the RI cases also have significant higher LHF than the non-RI cases, but the largest LHF differences are not confined in the upshear flank as found in the moderate-shear TCs (Figs. 14j–l). Specifically, the LHF differences between the RI and the non-RI TCs are generally larger in the downshear flank than in the upshear flank. This suggests that the distribution of the LHF differences among the intensity change categories may influenced by the environmental VWS strength, whereas the detailed cause for this discrepancy is beyond the scope of this study and awaits to be explored in the future. Nevertheless, since the significant LHF differences between the RI and non-RI cases are identified at all azimuths, the RI cases are still expected to be less susceptible to the detrimental effect of downdraft ventilation and thus have stronger upshear convection.

Since the LHF is a function of the 10-m wind speed U 10 and the air–sea moisture contrast q* − q 2, it is informative to examine how these two parameters differ among different intensity change groups. Figure 15 shows the box-and-whisker plots of the U 10 and q* − q 2 averaged between −24 and 0 h and within the 200-km radius for different cases. The results are qualitatively similar for the moderate-shear and the strong-shear TCs. Note that the differences in U 10 can be largely attributed to the differences in TC intensities as revealed by the distribution of the maximum sustained wind speed (Figs. 15c,f). It is found that the WK cases have the strongest U 10 among the four intensity change groups (Figs. 15a,d), which is due to that 90% WK cases occur after TC lifetime peak intensities (not shown). The differences in U 10 between the NT and RI cases are generally insignificant. Therefore, the lower LHF in the WK and NT cases than the RI cases is not attributed to the differences in near-surface wind but the differences in air–sea moisture contrast. As shown in Figs. 15b and 15e, the WK and NT cases have significantly lower q* − q 2 than the RI cases. It is worth mentioning that the q* − q 2 grows larger with increasing intensification rate, which is likely attributed to the variations in the SST (Figs. 13f–h and 14f–h). By comparison, the SI cases have both significantly lower U 10 and q* − q 2 than the RI cases, suggesting the lower LHF in the SI cases than in the RI cases is a combined result of the weaker near-surface wind and the weaker air–sea moisture contrast.

Fig. 15.

Fig. 15.

Fig. 15.

Box-and-whisker plots of the mean from −24 to 0 h of (a),(d) 10-m wind speed U 10 (m s−1); (b),(e) air–sea moisture contrast q* − q 2 (g kg−1); and (c),(f) maximum sustained wind speed (kt) in the four intensity change categories under (top) moderate and (bottom) strong environmental VWS. The U 10 and q* − q 2 are averaged within a 200-km circle relative to the storm center. The maximum sustain wind speed is derived from the best track data. Squares, triangles and dots represent means, where the square (triangle) denotes the mean in an intensity change category is significantly lower (higher) than that in the RI category at the 95% confidence level.

Citation: Monthly Weather Review 149, 12; 10.1175/MWR-D-21-0141.1

Overall, the above results suggest that the different strengths of the upshear convection and outflow may stem from the differences in the environmental thermodynamic conditions. The environment for the RI cases is characterized by higher midlevel RH, higher SST and higher LHF than the non-RI cases, which can mitigate the detrimental effects of the dry-air intrusion and downdraft ventilation on the development of upshear convection and thus promotes the outflow blocking of the upper-level environmental wind.

5. Summary

Based on the idealized numerical simulations and observations, previous studies (Ryglicki et al. 2019, 2020; Dai et al. 2021) proposed that the TC-induced outflow acts to withstand the upper-level environmental flow and thus favors RI under moderate-to-strong VWS. As an extension of the previous studies, the current study presents a climatological analysis on the role of outflow blocking in TC RI using the ERA5 reanalysis, GridSat IR brightness temperature, and TC best track datasets. Following Tao et al. (2017), TC intensity change cases in the best track data are classified into four categories (WK, NT, SI, and RI) according to 24-h future intensity change, whereas RI events are defined as a continuous period of one or multiple consecutive RI cases. Based on the strength of environmental VWS, the intensity change cases and RI events are further divided into the weak-shear group, the moderate-shear group and the strong-shear group.

Composite analyses were performed for the wind and T IR fields in a shear-relative framework with respect to the onset of RI events. Results show that, for the moderate-shear and strong-shear RI events, the TC upper-level outflow in the upshear flank tends to intensify significantly during the 24-h period preceding RI onset. Consistent with previous studies (Ryglicki et al. 2019; Dai et al. 2021), the enhanced outflow acts to block the upper-level environmental flow from the inner-core region of TC, which leads to a significant decrease in the local VWS and thus sets up a favorable environment for RI. Through visual inspection, we found that about 76% of moderate- and strong-shear RI events are in the vicinity of upper-level anticyclones, which supports the hypothesis of Ryglicki et al. (2019) that outflow-blocking is more effective to lead to RI when TCs are sheared by upper-level anticyclones.

The intensification of the upshear outflow in moderate- and strong-shear RI events is coincided with a decrease in T IR in the upshear flank within the 200-km radius, indicating the axisymmetrization of the inner-core convection is closely related to the outflow blocking process leading up to RI. By comparison, for the weak-shear RI events, the outflow has already developed in all shear-relative quadrants one day before RI onset. There are very weak changes in upshear outflow in the inner-core region during the 24-h period leading up to RI onset and the change in the local VWS is relatively small, suggesting that the RI under weak environmental VWS is less attributed the outflow blocking mechanism.

To fully verify the importance of the outflow blocking mechanism on RI under moderate-to-strong environmental shear, the outflow structures in the four intensity change categories are compared. On average, the RI cases under moderate-to-strong environmental shear have stronger outflow in the upshear flank than the WK, NT and SI cases, which can mitigate the detrimental effect of the upper-level environmental flow on TC intensification. Statistically significant differences are found in the local VWS between RI and non-RI cases. Furthermore, a correlation analysis shows that, for either the moderate-shear or strong-shear cases, the 24-h future intensity change of the TCs is more correlated with the local shear than with the environmental shear, suggesting that the local VWS may act as a better potential predictor than the environmental VWS for future intensity change of TCs. Furthermore, the most significant differences in the shear-relative distribution of convection between the RI cases and the three categories of non-RI cases all appear on the upshear side, confirming the crucial role of the upshear convection in the outflow blocking.

Further examinations show that the different strengths of upshear convection among the intensity change categories under moderate-to-strong environmental shear may be mainly ascribed to the differences in thermodynamic conditions. In particular, the RI cases have more favorable thermodynamic conditions characterized by higher midlevel RH, warmer SST and higher surface LHF than the non-RI cases, which can mitigate the detrimental effect of the midlevel and low-level ventilation on the upshear convection. Moreover, an evaluation of the 10-m wind and the air–sea moisture contrast indicates that the LHF differences between the WK, NT cases and the RI cases are attributed to the differences in air–sea moisture contrast, whereas the lower LHF in the SI cases than in the RI cases is a combined result of the weaker TC intensity and the weaker air–sea moisture contrast.

To summarize, the findings based on this climatological analysis support the idea that the TC outflow-related blocking of the upper-level environmental flow is an important mechanism by which TCs can overcome the detrimental impact of VWS and commence RI, which implies the potential utility of calculating upshear outflow and local shear to improve real-time intensity forecasting for the TCs under moderate-to-strong VWS.

Acknowledgments

The authors thank Dr. David Ryglicki and two anonymous reviewers for their valuable comments that greatly improved the current paper. This study is supported by the National Key Research and Development Program of China (Grant 2017YFC1501901) and the National Natural Science Foundation of China (Grants 41975071 and 42175073).

REFERENCES