Revisiting the response of western North Pacific tropical cyclone intensity change to vertical wind shear in different directions

Xiuji Ling , Qingqing Li , b , *

a Key Laboratory of Meteorological Disaster of the Ministry of Education, Nanjing University of Information Science and Technology, Nanjing, China

b State Key Laboratory of Severe Weather, Chinese Academy of Meteorological Sciences, Beijing, China

Keywords:Tropical cyclone Intensity change Vertical wind shear direction Large-scale pattern

ABSTRACT The effect of vertical wind shear (VWS) directions on the change in western North Pacific tropical cyclone (TC)intensity is revisited in this study. Results show that the differences in the correlations between VWS in different orientations and the change in TC nondimensional intensity highly diminish, although slight differences are still present. The subtle differences in the correlations are likely associated with different synoptic-scale patterns at upper and lower levels. This result suggests that, in addition to thermodynamic effects, dynamic roles of the synoptic-scale patterns associated with the VWS should also be taken into account when the authors examine how VWS in different directions affects TC intensity change.

1. Introduction

Strong vertical wind shear (VWS) has been indicated to be generally detrimental to the intensification ( DeMaria and Kaplan, 1999 ;Zeng et al., 2010 ; Riemer et al., 2010 ; Wang et al., 2015 ) of tropical cyclones (TCs). Several mechanisms have been proposed to underpin the weakening of TCs due to VWS. It can tilt the vertical axis of the TC toward the downshear-left direction and weaken the storm by modulating the eddy momentum flux across the eyewall ( Wu and Braun, 2004 ).So-called ventilation effects caused by VWS, including those at the upper( Fu et al., 2019 ), middle ( Tang and Emanuel, 2010 ), and lower levels( Riemer et al., 2010 ), are also documented to play a role in TC intensity change. More recently, Gu et al. (2015) argued that TCs could be weakened by a reduction of the radial entropy gradient across the eyewall.

A growing body of studies has shown that, in addition to VWS magnitude, the height, depth, and direction of shear have striking effects on the change in TC intensity ( Ritchie and Frank, 2007 ; Zeng et al., 2010 ;Nolan and McGauley, 2012 ; Shu et al., 2013 , 2014 ; Wang et al., 2015 ;Finocchio and Majumdar, 2016 ; Finocchio et al., 2016 ; Wei et al., 2018 ;Fu et al., 2019 ). In particular, several physical processes have been postulated to map how the direction of shear affects the change in TC intensity. Ritchie and Frank (2007) and Zeng et al. (2010) suggested that easterly shear can offset the intrinsic northwesterly shear generated by the

β

gyre within the TC circulation, and it is thus less inimical to TC intensification than westerly VWS. Shu et al. (2014) argued that high-entropy air from the south and southeast quadrants of the TC that is in an easterly VWS environment fuels TC intensification. Wei et al. (2018) recently pointed out that easterly VWS is statistically less negative to TC intensification than westerly VWS because sea surface temperatures increase as the magnitude of easterly VWS increases. Additionally, different orientations of the shear relative to the near-surface wind direction can also yield different changes in TC intensity by modulating the vortexscale dynamics. Regardless of the role of thermodynamic environmental conditions, Rappin and Nolan (2012) found that VWS that is counteraligned with the mean surface wind can limit the tilt magnitude of the sheared vortex and thus aids in storm intensification.

The mechanism revealed in Rappin and Nolan (2012) differs from the results in Shu et al. (2014) and Wei et al. (2018) . On the other hand,the mechanisms for interpreting the effect of the VWS direction on the change in TC intensity are also distinct even in Shu et al. (2014) and Wei et al. (2018) . Therefore, the dependence of TC intensity change on the VWS direction remains to be further elucidated.

The objective of this study is to revisit the relationship between the VWS direction and the change in TC intensity. In particular, we will examine the possible effect of the orientation of VWS relative to the surface wind on the observed intensity change of TCs. The rest of the paper is organized as follows. Data and methods will be described briefly in Section 2 . Section 3 will analyze the results, and a summary will be provided in section 4 .

2. Data and methods

The best-track data containing TC location and intensity (measured as the maximum 1-min mean sustained 10-m wind speed;

V

max ) at sixhourly intervals are from the IBTrACs (International Best Track Archive for Climate Stewardship) dataset ( Knapp et al., 2010 ). Only those TCs with maximum sustained wind speed greater than 15 msin the region (0°—40°N, 90°—180°E) during 1980—2018 are analyzed. The samples within 300 km of the coast are specifically screened out to filter out the TCs that are possibly influenced by the land.

The European Center for Medium-Range Weather Forecasts interim reanalysis (ERA-Interim) ( Dee et al., 2011 ) is employed to investigate the VWS and environmental patterns associated with the TCs. ERAInterim has a horizontal grid spacing of 0.75°. Because TC centers identified from the reanalysis sometimes subtly deviate from those in the best-track data, the TC center in the ERA-Interim data is estimated to be the 850-hPa vorticity centroid within a 500 km ×500 km box centered on the best-track TC center.

To estimate the environmental VWS associated with TCs, we first remove TC circulation from the wind fields in the reanalysis data by subtracting the rotational and divergent winds over a vortex-centered,500-km-long domain, according to the method of Davis et al. (2008) .VWS is thus computed as the difference in the area-weighted averaged winds within 500 km from the identified TC center between 200 and 1000 hPa. Fig. 1 (a) shows that the most frequent VWS has a magnitude between 3 and 9 ms.

In addition to the TC intensity derived from the best-track data,nondimensional TC intensity ( Peirano et al., 2016 ) is also used to characterize the change in TC intensity. Nondimensional intensity (

I

) is defined as the best-track storm intensity normalized by the maximum potential intensity (MPI) ( Emanuel, 1986 ); namely,

I

=

V

/ MPI. MPI is estimated here with the formula proposed in Zeng et al. (2007) , based on the ERA-Interim dataset. Fig. 1 (b) shows that the 24-h nondimensional intensity change is mostly between ? 0.04 and 0.07, whereas the 24-h intensity change is mostly between ? 10 and 1 ms. The use of the nondimensional intensity metric can, to some extent, control for changes in the thermodynamic environment of the TC, providing a tangible way to examine the effect of the VWS orientations on the change in TC intensity argued in Rappin and Nolan (2012) .

3. Results

Fig. 1 (c) depicts the correlation coefficients between the magnitude of VWS in different directions and the 24-h intensity change of TCs. Akin to the results in previous studies ( Zeng et al., 2010 ; Shu et al., 2014 ;Wei et al., 2018 ), westerly and southerly VWS is much more strongly negatively correlated with the TC intensity change measured with the best-track intensity than the easterly and northerly VWS ( Fig. 1 (c)). If the dependence of TC intensity change on VWS directions is mainly controlled by different properties of sea surface temperatures as proposed by Wei et al. (2018) , it is expected that there are few differences in the relationship between the VWS in different orientations and the change in nondimensional TC intensity. Indeed, Fig. 1 (c) indicates that the correlation coefficients between the VWS in different directions and the nondimensional intensity change of TCs are comparable. Nevertheless,the easterly and southerly VWS is slightly more unfavorable to TC intensification than the westerly and northerly VWS based on TC nondimensional intensity.

Why, then, in comparison to the westerly and northerly VWS, is the easterly and southerly VWS more strongly negatively correlated with the 24-h change in nondimensional intensity of western North Pacific typhoons? The findings in Rappin and Nolan (2012) are one possible explanation, which showed that the VWS aligned (counter-aligned) with the mean surface wind is less (more) conducive to TC intensification.This result seems to be verified by the TC cases in northerly VWS.Fig. 1 (d) shows that the northerly VWS that is counter-aligned (aligned)with the mean surface wind is less (more) unfavorable to the intensification of the TCs, and more than 90% of the TCs in northerly VWS show the VWS that is counter-aligned with the mean surface wind. The nondimensional intensity change of the TCs in northerly VWS is thus less negatively correlated with the VWS ( Fig. 1 (d)). However, Rappin and Nolan’s( 2012 ) argument is not in agreement with the relationship between the easterly VWS and nondimensional intensity change. The correlation coefficients between the easterly VWS that is aligned and counter-aligned with the mean surface wind and the nondimensional intensity change of TCs are both around ? 0.3 at 99% confidence level ( Fig. 1 (d)). Additionally, the southerly VWS counter-aligned with the mean surface wind is also more strongly negatively correlated with the change in the nondimensional intensity of the TCs than the northerly VWS counter-aligned with the mean surface wind ( Fig. 1 (d)). As a result, the southerly VWS is slightly more strongly negatively correlated with the change in the nondimensional intensity of the TCs than the northerly VWS. Furthermore, the westerly VWS counter-aligned with the mean surface wind is marginally less negatively correlated with the nondimensional intensity change of the TCs than the easterly and southerly VWS ( Fig. 1 (d)).

The above discussion indicates that the results of the idealized numerical experiments by Rappin and Nolan (2012) cannot explain the relationship between the VWS in different orientations and the nondimensional intensity change of western North Pacific TCs. Large-scale environmental patterns have been indicated to significantly influence the TC intensity change ( Hanley et al., 2001 ; Chen et al., 2015 ). To examine the possible effects of ambient environments associated with the VWS on the change in TC intensity, we investigate the characteristics of large-scale environments associated with the VWS through composite analysis.

The TC samples are classified into several groupings in terms of VWS orientations, and Fig. 2 shows the kernel density estimations of TC locations for the groupings. The majority of western North Pacific TCs in easterly and southerly VWS ( Fig. 2 (a, c)) are located south of those in westerly and northerly VWS ( Fig. 2 (b, d)). There are no striking differences in the locations of TCs in VWS that is aligned and counter-aligned with the mean surface wind ( Fig. 2 (e—i)). We examine the composites of large-scale patterns associated with the TCs within the regions marked by the red boxes in Fig. 2 in which most samples of each TC grouping are observed.

The composite center of the TCs in easterly VWS aligned with the mean surface wind tends to lie south of the zonally distributed subtropical high at 200 hPa ( Fig. 3 (a)) and near the base of an inverted trough at 350 hPa ( Fig. 3 (e)). Shieh et al. (2013) showed that the upper-tropospheric inverted trough aids in TC intensification. At lower levels, the composite TC center is located immediately downwind of the monsoon trough ( Fig. 3 (i, m)). Chen et al. (2015) and Qiu et al. (2020) showed that the well-developed monsoon trough could provide enhanced water vapor, thus partly offsetting the adverse effect of VWS on TC intensification. By contrast, the composite center of the TCs in northerly VWS aligned with the mean surface wind is generally observed southeast of the subtropical high ( Fig. 3 (d, h)), in the absence of an evident upper-level inverted trough ( Fig. 3 (h)) and the monsoon trough ( Fig. 3 (i, p)) near the TC. As a result, the easterly VWS aligned with the mean surface wind is less detrimental to the change in TC nondimensional intensity than the northerly VWS aligned with the mean surface wind ( Fig. 1 (d)).

Fig. 1. (a) Frequency distribution of VWS in different directions with respect to VWS magnitudes. (b) Frequency distribution of 24-h TC intensity and nondimensional intensity changes. (c) Correlation coefficients between the easterly, southerly, westerly, and northerly VWS and the changes in 24-h best-track intensity (cyan dots)and the changes in 24-h nondimensional intensity (red dots). Sample numbers of VWS in different directions are indicated by numbers in cyan. (d) Correlation coefficients of the VWS with different directions and the changes in 24-h nondimensional intensity (red dots), together with the correlation coefficients between the VWS aligned (counter-aligned) with the mean surface wind and the nondimensional intensity shown by orange (blue) dots. Solid (open) points indicate that the correlations are (are not) statistically significant at 99% confidence level. Black values indicate the correlation coefficients referenced. The proportion of the VWS counter-aligned with the mean surface wind out of the VWS in a given orientation is also shown as a percentage in blue.

The composite center of the TCs in westerly VWS that is aligned with the mean surface wind is located immediately downstream of an upper-level trough ( Fig. 3 (b, f)). Hanley et al. (2001) defined the trough located within 400 km upstream of a TC as the superposition trough.They showed that 78% of such TCs could intensify due to an uppertropospheric potential vorticity maximum approaching the TC center and enhanced divergence downstream of that trough. A trough also exists nearly 900 km upstream of the composite center of the TCs in southerly VWS that is aligned with the mean surface wind ( Fig. 3 (c, g)).That trough fits into the distant trough defined by Hanley et al. (2001) ,which may favor TC development due to strengthened divergence downstream of the trough ( Fig. 3 (g)). There is also a monsoon trough around the composite center of the TCs in southerly VWS aligned with the mean surface wind ( Fig. 3 (k, o)), which is likely conducive to TC intensification ( Qiu et al., 2020 ). As a result, Fig. 1 (d) indicates that the negative effect of the westerly and southerly VWS aligned with the mean surface wind on the nondimensional intensity change is statistically insignificant.

Fig. 2. Kernel density estimations of TC locations for the groupings of (a, e, i) easterly, (b, f, j) westerly, (c, g, k) southerly, and (d, h, l) northerly VWS, with (a—d) for TCs in the VWS in each direction, (e—h) for TCs in the VWS aligned with the mean surface wind, and (i—l) for TCs in the VWS counter-aligned with the mean surface wind. Red squares indicate the regions within which the TC samples in the VWS groupings are selected to conduct the composite analysis. The sample number of each grouping is shown in the top-right corner of the panel.

A comparison of the large-scale composite patterns shows that the upper-level, large-scale environments pertaining to the VWS counteraligned with the mean surface wind ( Fig. 4 (a—g)) resemble those of the VWS aligned with the mean surface wind ( Fig. 3 (a—g)). However,the composite patterns at lower levels are highly different ( Fig. 4 (i—p)).There is a pronounced monsoon trough near the composite center of the TCs in westerly VWS ( Fig. 4 (j, n) and northerly VWS ( Fig. 4 (l, p))that is counter-aligned with the mean surface wind. By contrast, the composite center of the TCs in easterly VWS counter-aligned with the mean surface wind is away from the monsoon trough ( Fig. 4 (i, m)), and even no monsoon trough exists near the composite center ( Fig. 4 (k, o)).The westerly and northerly VWS counter-aligned with the mean surface wind is hence less negatively correlated with the nondimensional intensity change than the easterly and southerly VWS counter-aligned with the mean surface wind ( Fig. 1 (d)).

Because the easterly shear aligned with the mean surface wind is predominant in the easterly shear samples, the composites of large-scale synoptic systems associated with the easterly shear are akin to those associated with the easterly shear aligned with the mean surface wind (not shown). Therefore, the correlation between the easterly shear and the change in nondimensional intensity is highly dependent on the correlation between the easterly shear aligned with the mean surface wind and the intensity change ( Fig. 1 (d)). Also, the composites of large-scale synoptic systems of the southerly, westerly, and northerly VWS are similar to those of their respective counterparts counter-aligned with the mean surface wind (not shown). The correlations between the southerly, westerly, and northerly VWS counter-aligned with the mean surface wind and the change in nondimensional intensity hence shape the correlations between the VWS of these orientations and the nondimensional intensity change ( Fig. 1 (d)).

4. Summary

Previous studies have argued that the VWS direction has a marked effect on the change in TC intensity, and easterly VWS is documented to be regularly less detrimental to TC development than westerly VWS.Several mechanisms have been proposed to elucidate such an impact,such as easterly VWS partly counteracting the shear effect induced by the βgyre, an environment of easterly VWS conducive to bringing highentropy air into the TC, and sea surface temperatures increasing with an increase in the magnitude of easterly VWS. In this study, we revisited the effect of VWS directions on the intensity change of western North Pacific TCs. Different from previous studies, the nondimensional intensity that is defined as the best-track TC intensity normalized by MPI is used here to measure the TC intensity change, which controls for changes in the thermodynamic environment of the TC.

The differences in correlations between the VWS in different directions and the 24-h change in TC nondimensional intensity indeed diminish. This result indicates that certain thermodynamic conditions associated with the VWS in different directions act upon the change in TC intensity. We did not investigate here these thermodynamic roles,which are beyond the scope of the present study. Nevertheless, the easterly and southerly VWS is subtly more unfavorable to TC intensification than the westerly and northerly VWS based on the nondimensional intensity. The mechanisms in Rappin and Nolan (2012) do not contribute to such slight differences in the correlations between the VWS in different directions and the change in TC nondimensional intensity. In addition to the effects of VWS, large-scale synoptic patterns associated with the VWS likely play a significant role in the change in nondimensional intensity.

Fig. 3. Composites of (a—d) 200-hPa, (e—h) 350-hPa, (i—l) 850-hPa, and (m—p) 1000-hPa wind fields for the (a, e, i, m) easterly, (b, f, j, n) westerly, (c, g, k, o)southerly, and (d, h, i, p) northerly VWS that is aligned with the mean surface wind. Divergence (shading; units: s ? 1 ) is also plotted at 200 and 350 hPa. Red curves mark the troughs noted in the text. The red dot indicates the composite center of the TCs in each VWS grouping. The numbers in the top-right corner of (a—d) indicate the sample numbers of each grouping.

Funding

This work was jointly supported by the National Key Research and Development Program of China [grant numbers 2018YFC1507103 and 2017YFC1501601], the Key Program of the Ministry of Science and Technology of China [grant number 2017YFE0107700], and the National Natural Science Foundation of China [grant numbers 41875054,41730961, 41730960, and 41775065].

Fig. 4. Composites of (a—d) 200-hPa, (e—h) 350-hPa, (i—l) 850-hPa, and (m—p) 1000-hPa wind fields for the (a, e, i, m) easterly, (b, f, j, n) westerly, (c, g, k, o)southerly, and (d, h, i, p) northerly VWS that is counter-aligned with the mean surface wind. Divergence (shading; units: s?1 ) is also plotted at 200 and 350 hPa. Red curves mark the troughs noted in the text. The red dot indicates the composite center of the TCs in each VWS grouping. The numbers in the top-right corner of (a—d)indicate the sample numbers of each grouping.

Acknowledgments

The calculations in this paper were made on the high-performance computing system in the High-Performance Computing Center, Nanjing University of Information Science and Technology.