OCTOBER 2020 L E E E T A L . 4061

Diurnal Variation of the Convective Area and Size Associated with the Rapid Intensification of Tropical

JAE-DEOK LEE Department of Atmospheric Sciences, National University, Taipei, Taiwan, and Department of Physics and Earth Sciences, University of the Ryukyus, Okinawa, Japan

CHUN-CHIEH WU Department of Atmospheric Sciences, National Taiwan University, Taipei, Taiwan

KOSUKE ITO Department of Physics and Earth Sciences, University of the Ryukyus, Okinawa, Japan

(Manuscript received 15 October 2019, in final form 27 July 2020)

ABSTRACT

This study examines the diurnal variation of the convective area and eye size of 30 rapidly intensifying tropical cyclones (RI TCs) that occurred in the western North Pacific from 2015 to 2017 utilizing Himawari-8 satellite imagery. The convective area can be divided into the active convective area (ACA), mixed phase, and inactive convective area (IACA) based on specific thresholds of brightness temperature. In general, ACA tends to develop vigorously from late afternoon to early the next morning, while mixed phase and IACA develop during the day. This diurnal pattern indicates the potential for ACA to evolve into mixed phase or IACA over time. From the 30 samples, RI TCs tend to have at least a single-completed diurnal signal of ACA inside the radius of maximum wind (RMW) during the rapidly intensifying period. In the same period, the RMW also contracts significantly. Meanwhile, more intense storms such as those of category 4 or 5 hurricane intensity are apt to have continuous ACA inside the RMW and maintain eyewall convective clouds. These diurnal patterns of the ACA could vary depending on the impact of large-scale environments such as vertical wind shear, ocean heat content, environmental mesoscale convection, and terrain. The linear regression analysis shows that from the tropical storm stage, RI commences after a slow intensification period, which enhances both the primary circulation and eyewall convective cloud. Finally, after the eye structure appears in satellite imagery, its size changes inversely to the diurnal variation of the convective activity (e.g., the eye size becomes larger during the daytime). KEYWORDS: Diurnal effects; Tropical cyclones; Satellite observations

1. Introduction Jiang 2012; Kieper and Jiang 2012; Monette et al. 2012). In general, since the IR window channel around 11 mm Satellite imagery can provide massive amounts of in- is not significantly absorbed by atmospheric gases, formation over extensive areas, and multiple previous it has been widely used to monitor convective or studies have used infrared (IR) brightness temperature stratiform clouds. Previous studies have shown that to examine convective clouds related to tropical cy- cold brightness temperature suitable for indicating the clones (TCs) (Browner et al. 1977; Muramatsu 1983; convective area could be used as one of the indicators Steranka et al. 1986; Harnos and Nesbitt 2011, 2016; of TC intensification (Gentry et al. 1980; Jiang 2012; Monette et al. 2012; Fischer et al. 2018). For example, Supplemental information related to this paper is available Gentry et al. (1980) demonstrated that a future (124 h) at the Journals Online website: https://doi.org/10.1175/MWR-D- TC intensity change is strongly correlated with the 19-0345.s1. mean brightness temperature at a correlation coeffi- cient of 20.781. This relationship was also similarly Corresponding author: Dr. Chun-Chieh Wu, [email protected] confirmed by Monette et al. (2012) using a tropical

DOI: 10.1175/MWR-D-19-0345.1 Ó 2020 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses). Unauthenticated | Downloaded 10/07/21 11:25 AM UTC 4062 MONTHLY WEATHER REVIEW VOLUME 148 overshooting tops algorithm for rapid intensification environment, TC frequently fails to intensify due to (RI) prediction. significant precession motion or decoupling between the In total, 85 GHz microwave satellite imagery may lower troposphere circulation and the upper-troposphere have an advantage in detecting deep convective cells circulation, but the above mechanism is not as clear for 2 related to ice scattering (Jiang 2012; Fischer et al. 2018), moderate VWS (5–10 m s 1). and these deep convective cells may release abundant Meanwhile, deep convective cells are recognized as a latent heat above the freezing level. Jiang (2012) dem- primary source of the cirrus shield, which is also referred onstrated that RI TCs have the lowest 11-mm brightness to as the or cirrus canopy in the temperature ranging from 174 to 223 K and the tallest upper troposphere (Malkus et al. 1961; Sadler 1964; 20-dBZ echo height compared to other TC phases such Merritt and Wexler 1967; Weickmann et al. 1977; Gray as slow intensification (SI), weakening (W), and neutral and Jacobson 1977; Dunion et al. 2014). According to (N). Fischer et al. (2018) also showed that TCs in the RI Merritt and Wexler (1967), the maximum areal extent of stage exhibit more concentrated cold brightness tem- these cirrus clouds appears 12 to 18 h after deep con- peratures inside the 100-km radius than in other stages. vection initiation, that is, the cirrus cloud coverage is out These results support the statement that convective of phase approximately 12 h from the diurnal signal of bursts (CBs) can contribute to TC intensification by convective clouds (Browner et al. 1977; Muramatsu releasing substantial latent heat, as much as 6.6 3 1017 J 1983; Steranka et al. 1984; Kossin 2002; Dunion et al. in a 12-h period inside the inner-core area (Kelley and 2014; Leppert and Cecil 2016). Browner et al. (1977) Halverson 2011). This additional latent heat may in- discussed that the maximum 253-K area decreases as TC crease TC intensity at a rate that satisfies the general RI intensity increases. For example, the average magnitude threshold of Kaplan and DeMaria (2003) in a 24-h pe- of the diurnal oscillation in the area of the cloud canopy riod. The increase in convective cells can also contribute computed by comparing the maximum area with the to a significant contraction of the radius of maximum minimum area in the tropical depression and hurricane wind (RMW) through radially varying diabatic heating is 3.03 and 2.09, respectively. This result indicates that (Schubert and Hack 1982; Willoughby 1990). This con- the diurnal area oscillation appears to be more signifi- tracted RMW could be an efficient configuration for TC cant in tropical depressions than in hurricanes. spinup by concentrating diabatic heating in the high- Gray and Jacobson (1977) showed that the upper inertial stability area, e.g., inside the RMW excluding disturbance area could experience a significant net ra- the eye region (Vigh and Schubert 2009). diational warming and cooling for one day. However, in RI is primarily initiated at the tropical storm and the disturbance area, e.g., thick cloud area, the tem- category 1 hurricane intensity stage (Kaplan and DeMaria perature remains warm in the midtroposphere and 2003; Hendricks et al. 2010; Jiang 2012). Previous studies lower troposphere despite temperature changes induced have shown that right before the onset of RI, CBs are by the net radiation. It may be due to longwave emission- frequently observed near the RMW and also in downshear reabsorption and condensate heat from the vapor. It may quadrants (Braun et al. 2006; Braun and Wu 2007; Reasor be due to longwave emission-reabsorption and condensate et al. 2009; Rogers 2010; Guimond et al. 2010, 2016; Rogers heat from the vapor. Therefore, the net radiational con- et al. 2013, 2015; DeHart et al. 2014; Chang and Wu 2017; trast between daytime and nighttime may significantly Hazelton et al. 2017a,b; Fischer et al. 2018; Lee and Wu contribute to the stability between the upper and lower 2018). However, these deep convective cells could be clouds. Navarro and Hakim (2016) showed contrasting suppressed and tilted outward in upshear quadrants as a daytime and nighttime vertical flow patterns (see their result of significant convective-scale subsidence or the Figs. 8 and 9). Specifically, the vertical circulation related vertical wind shear (VWS) (Chen and Gopalakrishnan to the net radiative tendency appears as cyclonic circula- 2015; Lee and Wu 2018). In the moderate or strong VWS tion in the afternoon and changes into anticyclonic circu- environments, the vortex structure is typically tilted lation in the early morning, which could explain certain according to the VWS direction, but intensifying TCs physical mechanisms concerning the convective activity could overcome such tilted structure as a result of suf- invigorated between late afternoon and midnight. Tang ficient inner-core convective cells that serve to reduce et al. (2019) discussed that the diurnal radiation contrast the precession motion of the vortex (Gray 1968; Jones was significant for changes in the RMW. For example, 1995; DeMaria 1996; Frank and Ritchie 1999, 2001; the RMW contraction tends to be more accelerated Reasor et al. 2004; Braun and Wu 2007; Rios-Berrios et al. during the nighttime due to the radiative destabilization 2018; Lee and Wu 2018). These processes in sequence and moistening in the lower troposphere compared to may be how a TC reintensifies in a moderate or strong the daytime. Muramatsu (1983) discussed the diurnal 2 VWS environment. Under a strong VWS (.10 m s 1) variation of the maximum extent of convective clouds

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC OCTOBER 2020 L E E E T A L . 4063 and the eye diameter in mature TCs and stated that the channels between 10 and 12 mm are not significantly convective area corresponding to a 2708C brightness absorbed by atmospheric gases, these wavelengths have temperature could reach its maximum size between been widely used to monitor severe weather systems the afternoon-evening and early morning, whereas the like supercells or intensifying TCs. Olander and Velden minimum occurs in the afternoon. In addition, Muramatsu (2009) proposed a methodology for the detection of in- (1983) stated that the maximum eye diameter occurs tense convective clouds based on the difference between during the daytime due to the dissipation of cirrus clouds the atmospheric window channel and WV absorption from the eye region. Dunion et al. (2014) similarly channel, herein referred to as IRWV, details of which confirmed that mature TCs exhibit diurnal variation are addressed in the next subsection. In this study, the and found that TC diurnal pulse speed ranges from solar zenith angle provided by Himawari-8 is used to 2 5to10ms 1. Elsberry and Park (2017) also com- distinguish between day and night. For example, when mented on the diurnal variation of the convective the averaged solar zenith angle within the area of a maximum and minimum that could affect VWS. TC (e.g., approximately 800 km 3 800 km) is higher These previous studies focused mainly on mature TCs than 808, it is defined as nighttime. In addition to the (category 2 and higher); however, since convective Himawari-8 satellite dataset, ERA-Interim reanalysis activity is one of the predominant indicators of TC data with 0.258 spatial resolution and 6-h temporal res- intensification or weakening, the diurnal variation in olution is used to compute VWS between 850 and the convective area related to RI needs to be examined 200 hPa within a 500 km radius. In this study, the vortex from before the onset of RI to the end of RI. Therefore, scale of the reanalysis data is filtered out by using a high- the primary purpose of this study is to characterize the order filter equation (Cheong et al. 2004), which allows diurnal variation of both the convective area and eye VWS to be computed thoroughly on a large-scale wind size related to RI TCs in the western North Pacific from field. Also, the Hybrid Coordinate Ocean Model data 2015 to 2017 based on Himawari-8 satellite imagery. are used to investigate the ocean condition, and the Joint The remainder of this study is composed as follows: Warning Center (JTWC) best track data are section 2 introduces the data and methodology; a com- used as the standard reference for TCs in this study. prehensive summary of RI TCs from 2015 to 2017 is b. Modification of the IRWV calculation using a given in section 3; sections 4, 5, and 6 elaborate the re- natural logarithm sults concerning the diurnal variation of the convective area depending on TC intensity categories, statistical The Himawari-8 geostationary satellite provides more analysis between TC intensification stages and eyewall segmentalized wavelength bands than previous geosta- convective cloud estimated by the normalized convec- tionary satellites such as the MTSAT series. These tive area between 2 and 3 times the RMW, and the di- segmentalized bands are more useful for monitoring urnal variation of the eye size, respectively; and section 7 detailed atmospheric characteristics such as humidity, summarizes the findings of this study. thin ice clouds, and other atmospheric gases (Bessho et al. 2016). The preexisting IRWV methodology, which is useful for diagnosing intense convective clouds (Olander 2. Data and methodology and Velden 2009), was used by Kurino (1997) and Schmetz et al. (1997). The atmospheric window channels generally a. Himawari-8 satellite, ERA-Interim reanalysis, and have warmer brightness temperature than that of the WV best track dataset channels in the clear-sky and a colder brightness temper- The next generation Japan Meteorological Agency ature than that in the convective area, which is caused by (JMA) geostationary meteorological satellite, the water vapor reemitting the absorbed radiation from the Himawari-8 satellite, was launched successfully in 2014, upper troposphere or lower stratosphere (Schmetz et al. replacing the Multifunctional Transport Satellites 1997). As a result, IRWV predominantly appears negative (MTSATs) previously used for monitoring the area in the convective area. However, this IRWV could be less between 608S and 608N and between 808E and 1608W. accurate in representing deep convection related to ice Compared to MTSATs, the Himawari-8 satellite can scattering compared with microwave satellite imagery, provide several specific bands, such as water vapor even though it can still provide a better temporal reso- (WV) absorption channels (6.2–7.3 mm) and atmospheric lution. For this reason, if IRWV can represent the window channels (10–12 mm), with higher spatial and intense convective area similar to that of microwave temporal resolution (Bessho et al. 2016). The Himawari-8 satellite imagery, it would be extremely useful in satellite provides three visible channels, three near-IR monitoring TC structure and intensity changes at a much channels, and 10 IR channels. Since the IR window higher temporal resolution for examining severe weather

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC 4064 MONTHLY WEATHER REVIEW VOLUME 148

FIG. 1. AHI vertical weighting functions for (a) 6.2, (b) 6.9, and (c) 11.2 mm from the CIMSS website (http://cimss.ssec.wisc.edu/goes/wf/ examples/AHI/). The maximum spectral response of each wavelength corresponds to 329, 391, and 1000 hPa, respectively. The gray area in (a),(b) indicates a critical level by which convective cells are determined to be shallow or deep. These profiles are obtainable when the standard tropical atmosphere, 108 zenith angle, 100% column moisture, and 10 K skin temperature adjustment are set. systems such as supercells or TCs occurring over the open and synthetic products of IRWV and IRWVln, re- ocean, making it possible to examine the diurnal variation spectively. A colder brightness temperature relative to of TC convective clouds over the open ocean. Therefore, the periphery generally represents the convective in this study, the preexisting IRWV has been modified as area, but the brightness temperature range seems to follows: be ambiguous. Therefore, the question ‘‘What is an ! exact brightness temperature to indicate the convec- T 5 3 6:9mm tive area?’’ is raised. The IRWV outputs could repre- IRWVln T11:2mm ln , (1) T6:2mm sent the convective clouds; however, compared to microwave satellite imagery, the IRWV results still where Tb is the brightness temperature of each wave- exaggerate the deep convective area (Figs. 2a–d). As length obtained from 6.2, 6.9, and 11.2 mm. The mean mentioned above, because IRWV shows the differ- brightness temperature of 11.2 mm can be obtained by ence between the atmospheric window and water va- summing the water vapor term that becomes zero in the por channels, it may include all information between bracket. The natural logarithm is taken to reduce the the low and midtroposphere. Meanwhile, since the scale of IRWVln. Figure 1 shows the weighting functions IRWVln indicates the difference between water vapor relative to 6.2, 6.9, and 11.2 mm Himawari-8 Advanced channels, it seems better to represent deep convection Himawari Imager (AHI), and each wavelength shows penetrating the critical level. Overall, the IRWVln a different peak spectral response depending on the output is very similar to the microwave satellite imagery pressure level. For instance, under the given settings (Figs. 2a,b,e), even though there are limitations in shown in Fig. 1, the peak spectral response of the 6.2 and identification of detailed structures such as the small 6.9 mm water vapor channels appear at 329 and 391 hPa, inner core with a moat and the outer band under the respectively (Figs. 1a,b); however, the peak spectral deep convective clouds. Nevertheless, IRWVln derived response of the atmospheric window channel (11.2 mm) from Himawari-8 satellite imagery may be useful for ex- is observed at 1000 hPa (Fig. 1c). These results show that amination of the diurnal variation of convective clouds although the water vapor channels have similar wave- related to TCs through the improved temporal and spatial lengths, their differences share the same characteristics resolution. as the IRWV. From Eq. (1), the negative IRWVln appears c. Definitions of the active convective area, mixed when the convective cells penetrate the layer between 329 phase, and inactive convective area and 391 hPa, which is referred to as the critical level in this study. Therefore, if IRWV primarily represents the char- The existence of cold brightness temperature inside acteristics of convective clouds below the midtroposphere, the TC inner-core area correlates with TC intensification IRWVlncanbeusedtodeterminewhethertheconvective (Gentry et al. 1980; Muramatsu 1983; Ebert and Holland clouds are deep or shallow. Figure 2 shows examples of 1992; Jiang 2012; Monette et al. 2012). Convective clouds (2016) observed by Global Change in supercells or TCs are normally colder than the 225 K Observation Mission Water (GCOM-W1) Advanced brightness temperature (Bedka et al. 2010; Monette et al. Microwave Scanning Radiometer 2 (AMSR2) micro- 2012). Since the Himawari-8 satellite full-disk resolution wave satellite imagery, Himawari-8 satellite imagery, has a 2-km spatial resolution, each grid point represents a

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC OCTOBER 2020 L E E E T A L . 4065

FIG. 2. Various satellite images of Typhoon Meranti (2016): (a) GCOM-W1 AMSR2 minimum 85-GHz polar- ization corrected brightness temperature and (b) 85-GHz at 0347 UTC 13 Sep (https://www.nrlmry.navy.mil/tc-bin/ tc_home2.cgi); (c) IRWV (IR11.2–IR6.2 mm), (d) IRWV (IR11.2–IR6.9 mm),(e)IRWVln,and(f)IRWVlndisplayed with three colors at 0350 UTC, respectively. In (f), the white and black areas represent ACA and mixed phase, respectively. The definitions of ACA, mixed phase, and IACA can be found in section 2c. All figures are in temperature (K).

4-km2 area; therefore, the active convective area (ACA), (Figs. 2e,f). The mixed phase may represent a change in mixed phase, and inactive convective area (IACA) can be state from ACA to IACA. ACA typically indicates defined by the following three thresholds: IRWVln # 21K, deep convection that could penetrate the critical level 21 , IRWVln , 1 K, and IRWVln $ 1 K, respectively (Figs. 1a,b), while the mixed phase and IACA could

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC 4066 MONTHLY WEATHER REVIEW VOLUME 148

TABLE 1. The sea level pressure (hPa) forecast error caused by intense storms occurring in the western North Pacific. The infor- mation is adopted from JMA annual reports from 2012 to 2017. Here, an intense storm is defined when the CI index is greater than or equal to 6.5. The storms migrated from the eastern North Pacific are excluded.

Year 2012 2013 2014 2015 2016 2017 24 h 17.5 19.5 17.7 13.8 19.0 12.4 48 h 22.3 29.7 25.9 18.9 28.0 20.2 72 h 25.7 30.7 27.9 21.2 29.8 19.9 6.5 # CI , 7.0231711 CI $ 7.0244351 primarily represent moderate or shallow convection that develops under the critical level.

3. Summary of RI TCs from 2015 to 2017 in the western North Pacific Similar to the climatologically normal frequency1 of 25.6 TCs, the average TC genesis frequency in the western North Pacific is 26.6 TCs from 2015 to 2017, during which the number of TCs with RI is 10, 11, and 9 in 2015, 2016, and 2017, respectively. This RI occurrence frequency seems to be significantly higher than the cli- matological RI frequency of approximately 5.2 TCs per year from 1979 to 2015 (Fudeyasu et al. 2018); however, this climatological RI number may vary slightly de- pending on the definition of maximum surface wind speed. Considering that intense category 4 or 5 TCs generally undergo RI at least once during their lifetime (Kaplan and DeMaria 2003), this could reduce the TC intensity forecast accuracy. Table 1 shows the sea level forecast errors induced by intense storms, for example, over 6.5 current intensity (CI; Dvorak 1975) index value, adopted from the JMA annual reports from 2012 to 2017. It turns out that the sea level pressure forecast tends to deteriorate when the number of intense TCs increases. Figure 3 represents the track and intensity of RI TCs from 2015 to 2017 based on JTWC best track data. In 2 this study, the Saffir–Simpson hurricane wind scale is FIG. 3. The track and intensity of JTWC best track for RI storms adopted to classify the TC intensity. From the 30 sam- over the western North Pacific in (a) 2015, (b) 2016, and (c) 2017. ples, RI frequently occurs as the TCs move westward or The storm intensities are drawn based on the Saffir–Simpson hur- northwestward. This is consistent with Kaplan and DeMaria ricane wind scale. The dotted box in (a),(b) indicates the area with the most frequent formation of RI TC in that year. (2003). Some TCs experienced RI near the recurvature point; however, after passing the recurvature point, these TCs are prone to weaken significantly. Meanwhile, the genesis of RI TCs between 2015 and 2016 is slightly different. For example, RI TCs in 2015 were concentrated over the area between 1478 and 1648E and between 108 and 8 1 http://www.jma.go.jp/jma/jma-eng/jma-center/rsmc-hp-pub-eg/ 20 N, while in 2016, the RI concentration shifted approx- AnnualReport/2017/Text/Text2017.pdf. imately 108 westward compared to 2015 (Figs. 3a,b), which 2 https://www.nhc.noaa.gov/aboutsshws.php. may be attributed to El Niño–Southern Oscillation

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC OCTOBER 2020 L E E E T A L . 4067

TABLE 2. A summary of RI TCs occurring in 2015. The D represents the change of pressure, maximum surface wind speed, category, and VWS computed between 850 and 200 hPa from the onset of RI to the mature stage. The category refers to the Saffir–Simpson hurricane wind scale. The OHC represents the average value over the RI period. To eliminate cold wake effects induced by TC, OHC is computed under free ocean conditions (e.g., before TC genesis).

The period from the onset of RI to the mature stage

Analysis period DVWS OHC Pmin Vmax 2 2 Case name (start date–end date) Onset end DP (hPa) DV (kt) DCategory (m s 1) (GJ m 2) (hPa) (kt) Soudelor (13th) 0600 UTC 30 Jul 0400 UTC 1 Aug 286 115 TS/Cat 5 0.7 9.8 907 155 0600 UTC 9 Aug 1800 UTC 3 Aug Goni (15th) 0000 UTC 14 Aug 1600 UTC 15 Aug 250 66.6 TS/Cat 4 1.2 8.0 933 120 0000 UTC 26 Aug 0000 UTC 17 Aug 230 40 Cat 1/4 2.1 4.8 1200 UTC 22 Aug 0600 UTC 24 Aug Atsani (16th) 1200 UTC 14 Aug 0900 UTC 16 Aug 262 82.5 TS/Cat 5 0.7 9.6 918 140 0600 UTC 25 Aug 1800 UTC 19 Aug Krovanh (20th) 1800 UTC 14 Sep 1800 UTC 15 Sep 245 60 TS/Cat 3 3.8 6.5 948 100 0000 UTC 21 Sep 1800 UTC 17 Sep Dujuan (21st) 0000 UTC 22 Sep 1400 UTC 25 Sep 235.6 48.3 Cat 1/4 22.7 6.8 926 130 1800 UTC 29 Sep 1800 UTC 27 Sep Mujigae (22nd) 1200 UTC 30 Sep 1800 UTC 2 Oct 241 55 TS/Cat 3 20.4 5.2 944 105 0600 UTC 5 Oct 0600 UTC 4 Oct Koppu (24th) 0600 UTC 12 Oct 1800 UTC 14 Oct 267 90 TS/Cat 4 23.0 10.9 926 130 0000 UTC 20 Oct 1200 UTC 17 Oct Champi (25th) 0600 UTC 13 Oct 0600 UTC 16 Oct 253 70 TS/Cat 4 29.7 8.5 929 125 0600 UTC 25 Oct 1200 UTC 18 Oct In-Fa (27th) 1800 UTC 16 Nov 0600 UTC 17 Nov 222 30 TS/Cat 1 21.0 7.1 933 120 1200 UTC 26 Nov 0600 UTC 18 Nov 249 65 TS/Cat 4 5.9 0000 UTC 19 Nov 0000 UTC 21 Nov Melor (28th) 0000 UTC 11 Dec 0000 UTC 12 Dec 256 75 TS / Cat 4 20.5 6.0 933 120 1200 UTC 16 Dec 0600 UTC 14 Dec

(Kim et al. 2011). The oceanic Niñoindex3 shows a stronger in 24 h based on the 1-min average maximum surface El Niño year in 2015 than in 2016 and 2017, and that the wind. During the RI period, both TCs passed through TC track appeared quite erratic (Fig. 3c), although the very warm oceanic heat content (OHC) regions of 9.4 and 2 ocean had returned to normal conditions in 2017. 10.0 GJ m 2,respectively(Table 3). This intensification Furthermore, there were no category 5 storms in 2017. rate corresponds to the lowest 24-h intensity change fre- As a result, the TC intensity predictability seems to be quency (Kaplan and DeMaria 2003). Consequently, both temporarily improved compared to other years (Table 1). storms explosively intensified from the tropical storm A comprehensive summary of the current RI TCs is stage to the category 5 hurricane wind intensity at the end presented in Tables 2–4, showing that RI TCs normally of RI. Interestingly, both TCs had similar initial locations concentrate between August and October, during and tracks (Fig. 3b), but underwent different VWS which22RITCsoccurred.ThisRIactivityisalso evolutions (Figs. 4b,d). During the RI period, Nepartak consistent with the climatological seasonal-TC dis- primarily experienced weak VWS, whereas Meranti tribution (Holliday and Thompson 1979)andRIoc- experienced a significant change in VWS (Figs. 4b,d). currence (Fudeyasu et al. 2018). In addition, RI tends This significant change in the magnitude of VWS was to commence within 72-h after a tropical depression also observed for Champi (2015), Malakas develops into a tropical storm (Tables 2–4). Although (2016), and Chaba (2016) (Figs. 4a,e,f), and Typhoon RI typically represents an increase in the maximum Megi (2010) (Lee and Wu 2018). This dramatic weak- 2 surface wind of more than 30 kt (1 kt ’ 0.51 m s 1)in ening of the VWS may be explained by either of the 24 h, Typhoons Nepartak (2016) and Meranti (2016) following two scenarios: 1) when the TC moves into a significantly surpassed this general RI definition with a region with reduced VWS, and 2) when deep convective maximum surface wind increase of approximately 70-kt updrafts partially cancel the VWS caused by the enhanced outflow. Typhoons Namtheun (2016), Malakas (2016), Songda (2016), and Talim (2017) showed a recurving and 3 https://www.esrl.noaa.gov/psd/data/correlation/oni.data. abnormal track and experienced an increase in VWS

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC 4068 MONTHLY WEATHER REVIEW VOLUME 148

TABLE 3. As in Table 2, but for RI TCs in 2016.

The period from the onset of RI to the mature stage

Analysis period DVWS OHC Pmin Vmax 2 2 Case name (start date–end date) Onset end DP (hPa) DV (kt) DCategory (m s 1) (GJ m 2) (hPa) (kt) Nepartak (1st) 0000 UTC 3 Jul 0130 UTC 4 Jul 278 105 TS/Cat 5 1.5 9.4 907 155 1200 UTC 9 Jul 1200 UTC 6 Jul Lionrock (10th) 1800 UTC 17 Aug 0200 UTC 23 Aug 241 55 TS/Cat 4 2.0 5.3 933 120 1200 UTC 30 Aug 1800 UTC 24 Aug Namtheun (12th) 0000 UTC 31 Aug 1700 UTC 31 Aug 248.6 65.8 TS/Cat. 3 3.7 7.2 948 100 0000 UTC 5 Sep 1200 UTC 2 Sep Meranti (14th) 0000 UTC 9 Sep 0300 UTC 10 Sep 2101.5 132.5 TS/Cat 5 23.7 10 895 170 1200 UTC 15 Sep 1200 UTC 13 Sep Malakas (16th) 1800 UTC 11 Sep 0600 UTC 15 Sep 230 40 Cat 1/4 26.3 9.5 937 115 1200 UTC 20 Sep 1800 UTC 16 Sep 226 35 Cat 1/3 7.0 4.8 1800 UTC 17 Sep 0600 UTC 19 Sep Megi (17th) 1800 UTC 22 Sep 1100 UTC 23 Sep 259.6 80.8 TS/Cat 4 21.2 7.5 937 115 1200 UTC 28 Sep 0000 UTC 27 Sep Chaba (18th) 0000 UTC 27 Sep 1400 UTC 30 Sep 274 100 TS/Cat 5 21.4 6.5 911 150 0600 UTC 5 Oct 0600 UTC 3 Oct Songda (20th) 0000 UTC 7 Oct 0130 UTC 9 Oct 260 68.7 TS/Cat 4 11.7 5.0 926 130 0600 UTC 13 Oct 1800 UTC 10 Oct Sarika (21st) 0600 UTC 13 Oct 1700 UTC 13 Oct 251.6 80.9 TS/Cat 4 20.1 8.8 944 115 0000 UTC 19 Oct 1800 UTC 15 Oct Haima (22nd) 1800 UTC 14 Oct 1200 UTC 15 Oct 271.5 95 TS/Cat 5 1.8 8.1 914 145 1800 UTC 21 Oct 1800 UTC 18 Oct Nock-Ten (26th) 1800 UTC 20 Dec 1500 UTC 22 Dec 270.5 87.5 TS/Cat 5 22.5 7.5 915 140 0000 UTC 28 Dec 0000 UTC 25 Dec

during the RI period (Figs. 3b,c and 4c,e,g,h). The in- category 3 or below, category 4, and category 5, based on the creased VWS appears to be partially related to a mid- maximum intensity according to the Saffir–Simpson hurri- latitude trough that can enhance the TC’s outflow (Fig. S1 cane wind scale. In this study, the exact diurnal and semidi- in the online supplemental material; Rogers et al. 2015; urnal signals of ACA are obtained by applying an inverse Elsberry and Park 2017; Fischer et al. 2019). Molinari and Fourier transform to the ACA computed inside the RMW. Vollaro (2010) indicated that a sheared tropical storm For example, if a TC has a 5-day lifespan, the diurnal and could undergo RI by means of intense convection when it semidiurnal pulses of ACA could be obtained by synthe- interacts with a midlatitude trough. In addition to the sizing wavenumbers from 0 to 5, and 0 to 10, respectively. VWS magnitude, an abrupt change in the VWS direction a. Category 3 or below may disrupt the development of an ACA. For example, Meranti had maintained an ACA with a consistent VWS Figures 6 and 7 show a time series of the ACA, mixed direction during the RI period (Fig. 5a), while Malakas and phase, IACA, diurnal and semidiurnal signals, and the Talim experienced a dissipation of ACA after an abrupt eyewall convective cloud for nine RI TCs. In this study, changeintheVWSdirectionattheonsetofRI(Fig. 5c,not the eyewall convective cloud is determined by the nor- shown for Malakas). In addition, in the tropical depression malized IRWVln between 2 and 3 times the RMW, stage, Chaba experienced considerable dissipation of the called the eyewall area. The normalization is completed ACA after a sudden change in VWS direction (Fig. 5b). by the following procedure: first, the region is divided Compared to Talim, Chaba experienced significant shear into positive and negative areas; second, the positive impact on the development of ACA at an early stage due area is divided by its maximum while the negative area is to the absence of a stable structure, e.g., an eye structure divided by its minimum. Therefore, if the average value established from the lower to upper troposphere. taken from the eyewall area is negative, the eyewall is considered to be solid. On the contrary, a broken or disorganized eyewall is manifested by a positive value. 4. The diurnal variation of ACA In the time series figures, ACA tends to develop from To compare the diurnal variation of ACA based on TC late afternoon to midnight, while it shrinks significantly intensity, 30 RI TCs have been grouped into three categories: during the day. Meanwhile, the maximum mixed phase

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC OCTOBER 2020 L E E E T A L . 4069

TABLE 4. As in Table 2, but for RI TCs in 2017.

The period from the onset of RI to the mature stage

Analysis period DVWS OHC Pmin Vmax 2 2 Case name (start date–end date) Onset end DP (hPa) DV (kt) DCategory (m s 1) (GJ m 2) (hPa) (kt) Noru (5th) 0000 UTC 21 Jul 1500 UTC 22 Jul 227.5 40 TS/Cat 2 4.6 2.5 922 135 0000 UTC 8 Aug 0000 UTC 24 Jul 253.5 70 Cat 1/4 2.0 2.6 0730 UTC 29 Jul 1800 UTC 30 Jul Banyan (12th) 1800 UTC 10 Aug 0900 UTC 11 Aug 252 62.5 TS/Cat 3 2.2 7.7 948 100 0600 UTC 17 Aug 0000 UTC 13 Aug Hato (13th) 0000 UTC 21 Aug 1200 UTC 21 Aug 233 45 TS/Cat 2 4.3 8.1 956 90 0600 UTC 24 Aug 0000 UTC 23 Aug Sanvu (15th) 0000 UTC 29 Aug 0000 UTC 31 Aug 225 30 TS/Cat 2 1.5 2.4 953 90 0600 UTC 3 Sep 0000 UTC 1 Sep Talim (18th) 1200 UTC 9 Sep 1200 UTC 12 Sep 232 42.5 Cat 1/ 4 6.0 6.4 933 120 1200 UTC 17 Sep 0000 UTC 14 Sep Doksuri (19th) 1200 UTC 11 Sep 0600 UTC 13 Sep 239 50 TS/Cat 2 0.02 5.3 960 95 0000 UTC 16 Sep 0000 UTC 15 Sep Khanun (20th) 0600 UTC 12 Oct 2100 UTC 13 Oct 227.5 35 TS/Cat 2 22.7 5.6 959 85 0600 UTC 16 Oct 0600 UTC 15 Oct Lan (21st) 1200 UTC 15 Oct 1200 UTC 19 Oct 237 50 Cat 2/4 2.6 7.3 922 135 0000 UTC 23 Oct 1200 UTC 21 Oct Damrey (23rd) 1200 UTC 1 Nov 2300 UTC 1 Nov 230.1 55.9 TS/Cat 2 9.9 3.9 967 90 1800 UTC 4 Nov 1800 UTC 3 Nov

and IACA occur during the day. This pattern indicates the small number of convective cells inside the RMW of that ACA could potentially evolve into mixed phase or midget typhoons Namtheun (2016) and Sanvu (2017), they IACA over time (Merritt and Wexler 1967; Muramatsu underwent RI (Figs. 6c,f). Although midget TCs may not 1983; Dunion et al. 2014; Leppert and Cecil 2016). include sufficient CBs inside the RMW, small CBs could During the RI period, most RI storms of this group facilitate a significant contraction of the RMW of a midget exhibited a single diurnal variation of ACA, and formed TC (Figs. 7c,f). After the RMW contraction, they quickly a solid eyewall, although they fluctuated slightly (Figs. 6 formed an eye structure during the RI period (not shown). and 7). If the ACA can be treated the same as CBs, a Namtheun had a greater maximum intensity than Sanvu, single-completed diurnal variation of the ACA inside whichappearstobecausedbythe different oceanic con- the RMW may be adequate to trigger RI. Note that CBs ditions during the RI period. For example, Namtheun 2 typically release extra latent heat of approximately passed through high OHC region (7.2 GJ m 2), due to the 6.6 3 1017 J in 12 h, which may enhance the maximum Kuroshio, while Sanvu passed through a low OHC 2 2 surface wind from 9 to 16 m s 1 (Kelley and Halverson region (2.4 GJ m 2) (see Tables 3 and 4). Among the 2011). However, Typhoons Mujigae (2016) and Hato 9 RI TCs, Typhoon Banyan (2017) seems to have a (2017) experienced a sudden dissipation of ACA in the relatively more favorable environment for the develop- form of significant fluctuation in the eyewall convective ment of ACA because it did not interact with terrain 2 clouds inside the RMW and within 3 times the RMW and passed through 7.7 GJ m 2 OHC region (Fig. 3c and (Figs. 6, 7b,e). This can result in partially broken or Table 4). Consequently, ACA seems to be maintained disorganized eyewalls that allow interactions between longer without any significant dissipation from the onset of the TC inner-core and outer environments (Tang and RI, and the diurnal signal can more accurately describe the Emanuel 2010). Once low equivalent potential tem- time series of ACA inside the RMW (Fig. 6d). If thick perature air flows into the TC inner-core area, it could convective clouds are covered throughout the day, net stabilize the convective activity, which could eventually radiational cooling could be significantly reduced in the weaken the TC intensity. This abrupt ACA disappearance mid- or lower troposphere, possibly caused by longwave may be due to the complex flows caused by the land as the emission-reabsorption and condensate heat from the va- TCs travel into the or the direct/indirect por (Gray and Jacobson 1977). As a result, destabilization interaction of terrain (Figs. 3b,c).OtherRIstormsofthis caused by net radiative forcing (net radiational contrast) category also experienced fluctuations in the ACA devel- between the upper clouds and mid- or lower clouds may opment, however to a lesser degree. Meanwhile, despite promote convective activity at night. Therefore, unless the

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC 4070 MONTHLY WEATHER REVIEW VOLUME 148

FIG. 4. The VWS magnitude (on the left axis) and direction (on the right axis) computed from different pressure levels denoted by solid lines and pentagrams of different colors. The magnitude and direction of 850–200 hPa VWS is highlighted by a thick solid line and large pentagram. The VWS direction in the right ordinate indicates a downshear direction; for example, the W represents a shear direction from East to West. The abscissa denotes the time in the month–day–hour format. The RI period is indicated by two vertical-dashed lines.

TC experiences substantial ACA dissipation during the day, as a result of downdraft cooling with substantial pre- it could significantly influence the next convective activity by cipitation. If these low equivalent potential temperature increasing destabilization. This feature is more apparent in air parcels flow into the storm’s inner-core area along Category 4 and 5 hurricanes as described below. with the radial inflow in the boundary layer, it could potentially affect TC intensification as well as the dis- b. Category 4 organization of convective cells. As a result, the ACA Fourteen RI TCs reached category 4 hurricane in- appears to fluctuate temporarily (Fig. 8n). In this case, tensity after RI. Compared to the previous group, cat- the semidiurnal signal synthesized with the diurnal egory 4 RI TCs typically display sufficient and extended signal could better describe the development of the development of the ACA inside the RMW and within ACA. However, these category 4 RI storms rarely 3 times the RMW (Figs. 6 and 8). In addition, the diurnal experienced a complete loss of ACA inside the RMW signal better replicates the development of the ACA or within 3 times the RMW compared to storms of inside the RMW, except for Typhoons Melor (2015), category 3 or below. As stated in the previous subsection and Songda (2016), Sarika (2016), Talim (2017), and Lan the Introduction, thick clouds can reduce net radiational (2017). These 5 RI storms were specifically affected by cooling in the mid- or lower troposphere regardless if unfavorable environmental conditions such as terrain it is daytime or nighttime. In contrast, the upper tropo- (Melor and Sarika, Figs. 3a,b), significant changes in sphere may experience a significant net radiational OHC (Melor and Songda) and VWS (Songda and warmingandcoolingduringdaytimeandnighttime. Talim, Figs. 4g,h), and mesoscale convective clouds This strong radiational contrast may enhance desta- (Lan). In particular, mesoscale convective clouds in bilization that can enhance convective activity at night, the TC’s moving direction could foster the low equiva- and this sequential process may result in a positive lent potential temperature environment in the low levels convective activity feedback cycle. In addition, the onset

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC OCTOBER 2020 L E E E T A L . 4071

FIG. 5. The sequential IRWVln images at 12-h intervals of (a) Meranti, (b) Chaba, and (c) Talim. The innermost blue concentric circle indicates the RMW adopted from JTWC best track data, and the second and third circles represent 2 or 3 times the RMW, respectively. The black arrow denotes VWS computed between 850 and 200 hPa based on ERA-Interim re- analysis data (see more Fig. 4). of RI normally commences with an increase of the ACA Willoughby 1982; Jorgensen 1984; Weatherford and inside the RMW and a significant contraction of the Gray 1988; Vigh and Schubert 2009; Vigh et al. 2012). If RMW (Figs. 8 and 9). Meanwhile, most TCs passed TC adequately satisfies thermal wind balance, the through a moderate or high OHC region during the RI momentum field could be balanced with the thermo- period (Tables 2–4). When (2017) un- dynamic field at the levels at which the friction suffi- derwent RI, it showed a sustainable ACA inside the ciently decreases (Schubert and Hack 1982; Shapiro RMW (Fig. 8c), while being located in a high OHC re- and Willoughby 1982). Subsequently, the maximum 2 gion (10.9 GJ m 2) and under moderate VWS (Table 2). gradient of the inertial stability, which generally co- By contrast, despite insufficient ACA inside the RMW incides with the RMW, could move quickly toward the and unfavorable environments such as strong VWS and maximum heating area, and the heating induced by low OHC, Songda underwent RI. This unusual intensi- the ACA inside the RMW could possibly be responsible fication may be related to the substantial RMW con- for this considerable RMW contraction. This could ex- traction, the enhanced ventilation flow in the upper plain why convective cells, like CBs inside the RMW, troposphere due to a midtrough interaction, and quick are considered necessary for TC intensification (Rogers development of a tiny eye structure inside the RMW et al. 2013). Therefore, the differences in the ACA during the RI period (Figs. 8j and 9j). Except for the sustainability inside the RMW and eyewall convective VWS, characteristics on the amount of ACA inside the cloud may be important to TC intensification. Overall, RMW and substantial RMW contraction are similar to category 4 RI TCs showed sufficient development of the Namtheun (2016) and Sanvu (2017) (Figs. 6c,f and 7c,f). ACA inside the RMW and maintain eyewall convec- Previous studies demonstrated that TC eye formation is tive cloud as compared with the storms of category 3 or indicative of both stabilization and intensification of the below (Figs. 8 and 9). As a result, the diurnal signal of vortex structure (Schubert and Hack 1982; Shapiro and the ACA can better describe the ACA development

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC 4072 MONTHLY WEATHER REVIEW VOLUME 148

FIG. 6. The time series analysis of the ACA, mixed phase, and IACA for category 3 hurricane wind intensity or below. The blue solid line (scaled by a one-third ratio) indicates the ACA com- puted inside the RMW, and black and gray dashed lines repre- sent the ACA and mixed phase1IACA calculated within 3 times the RMW. The solid and dotted red lines denote the diurnal and semidiurnal signals obtained by applying the inverse Fourier transform to ACA inside the RMW. The yellow line overlaid with the maximum surface wind speed line represents the period from the onset of RI to the mature stage. inside the RMW, although some cases have exhibited maintain eyewall convective clouds by means of persis- many semidiurnal variations in ACA development tent or long-lived convective clouds within 3 times the and fluctuations in the eyewall convective cloud RMW since the onset of RI (Figs. 10 and 11). In addi- (Figs. 8 and 9). tion, they passed through a moderate or high OHC re- gion and favorable VWS environment during the RI c. Category 5 period (Tables 2–4). This solid eyewall structure could From the 30 RI samples, seven storms reached a cat- prevent unfavorable interactions such as dry air intru- egory 5 intensity (Tables 2 and 3 and Fig. 10), which sion from the outer environment and help maintain the represent a higher frequency in 2016, while there were favorable internal structure of the TC. Most of the cat- no category 5 storms in 2017 (Tables 3 and 4). As seen in egory 5 RI storms maintained a solid eyewall until Fig. 3c, the track of the RI TCs in 2017 appears much making landfall or encountering the region with low more erratic compared to the other two years. These OHC. For example, significant fluctuations in the eyewall erratic tracks may show that largescale currents are convective cloud of Typhoons Soudelor (2015) and complex, which may be unfavorable to TC intensifica- Atsani (2015) were observed at the end of RI (Figs. 11a,b), tion. For example, category 5 storms tend to have con- after which both storms had considerably weakened sistent westward or northwestward tracks (Figs. 3a,b) (Figs. 10a,b). These characteristics seem to be highly and the OHC in 2017 was lower than in 2015 and 2016 related to OHC. For instance, after RI, Soudelor en- (Tables 2–4). Figure 10 shows the time series of the countered the locally developed lower OHC, which also ACA of category 5 RI TCs. During the RI period, they included cold wakes induced by the storm (Figs. 12a,c), have exhibited a persistent ACA inside the RMW and whereas Atsani experienced a gradual decrease in OHC within 3 times the RMW without significant fluctuations as it moved northwestward (Figs. 12b,d). During this until the TC’s eye forms, which is distinct compared to period, both storms experienced significant ACA dissi- the other previous groups. In this case, the ACA’s di- pation within 3 times the RMW and collapse of the urnal signal tends to better account for their changes eyewall convective cloud (Figs. 10 and 11a,b), while si- inside the RMW. These very intense storms tend to multaneously, the TC intensity had weakened significantly

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC OCTOBER 2020 L E E E T A L . 4073

FIG. 7. The time series of the RMW in gray and the eyewall convective cloud in black for category 3 hurricane intensity or below. The eyewall convective cloud represents the averaged value of the normalized IRWVln between 2 and 3 times the RMW. For example, if the value is negative (positive), it shows that the eyewall is well organized (partially disorganized or disorganized). The two vertical- dashed lines indicate the period from the onset of RI to the mature stage.

(Figs. 10a,b). These results may represent a good example intensity change. Figure 13 shows the results of the of the relationship between OHC and ACA. scatterplots, which illustrate the relationship between To summarize this section, these category 5 RI storms the normalized IRWVln and four TC intensity changes tend to have a more persistent ACA inside the RMW such as SI, RI, N, and W, in Table 5. Each scatterplot and maintain eyewall convective cloud as compared includes both an instantaneous IRWVln normalized in with two previous RI groups. This sustainable ACA may the eyewall area and a successive TC intensity change be possible under moderate or high OHC, weak or mod- at 6-h intervals, which explains the dependency of TC erate VWS, no apparent mesoscale convection close to the intensity changes on the normalized IRWVln in the storm, and no terrain interaction. Also, these category 5 eyewall. The concentration of scatter points from the storms rarely have accompanied mesoscale convective first two columns from the left shows that RI is likely to clouds in front of the TC’s heading direction during the begin after the SI phase (Figs. 13a,b,e,f,i,j). This result intensifying period. indicates that the onset of RI requires improvements in the eyewall convective activity and TC’s primary circulation during the SI phase. This characteristic is 5. The relationship between eyewall convective more apparent in intense storms such as those of cat- cloud and TC intensity change egories 4 and 5. After SI, the TC intensity almost As previously discussed, the eyewall convective cloud reached the category 1 hurricane intensity, which may is believed to be associated with TC intensity change. explain why the onset of RI is normally highly con- For example, convective clouds in the eyewall region of centrated around the category 1 hurricane intensity category 5 TCs tends to persist without significant dis- (Kaplan and DeMaria 2003; Hendricks et al. 2010; sipation, as compared to relatively weak TCs (Figs. 7, 9, Jiang 2012). Meanwhile, during the RI period, TC and 11). In this section, a simple linear regression anal- tends to rapidly form a solid eyewall structure ysis has been carried out to understand the relation- (Figs. 13b,f,j). In this case, the TC is able to withstand un- ship between the eyewall convective cloud and TC favorable interactions such as dry air intrusion. However,

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC 4074 MONTHLY WEATHER REVIEW VOLUME 148

FIG.8.AsinFig. 6, but for category 4 storms. for category 3 or below, since TCs are prone to have a upper troposphere (Fig. 14). This vast convective cloud single diurnal variation, eyewall convective clouds are is typically referred to as the central cold cover, which is usually short lived (Figs. 6 and 7). As a result, the typically associated with steady intensity (Dvorak 1984; scatter points are somewhat less concentrated as com- Lander 1999). Despite substantial convective clouds, paredwithcategory4or5RITCs(Figs. 13a,b,e,f,i,j). In these typhoons did not intensify at all during this period. other words, it appears that the correlation between the Such nonintensification may be associated with the weak normalized IRWVln in the eyewall area and maximum primary circulation or the absence of eye structure in the wind speed is low. early TC development (Fig. 14). Finally, TC weakening The N phase represents the changes in the eyewall generally occurs after the mature stage. During this convective cloud without any significant maximum sur- period, both the eyewall convective cloud and the pri- face wind enhancement (Fig. 13c). In other words, it mary circulation tend to weaken (Figs. 13d,h,l). Of the manifests a temporary cessation of TC intensification or four intensity changes, the W phase shows the highest weakening. For example, both typhoons Dujuan and correlation coefficient, indicating that a weakening of Chaba showed a substantial ACA in the tropical storm the eyewall convective clouds can lead to a weakening of stage, but they did not form any eye structure in the TC intensity. Since most scatterplots show moderate or

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC OCTOBER 2020 L E E E T A L . 4075

FIG.9.AsinFig. 7, but for category 4 storms. high linear regression correlation coefficients, these re- decrease with TC intensity. Muramatsu (1983) noted sults may account for that the close association of TC that the maximum and minimum eye sizes could be intensification with the convective clouds in the eyewall observed typically during daytime and early morning, region (Fig. 13). respectively, due to cloud dissipation or development from the eye area. To further investigate the diurnal variation of TC eye size, four category 5 storms, which 6. The diurnal variation of TC eye diameter maintain eyewall convective clouds until the mature The formation of the eye structure typically repre- stage without any interruption such as dry air intrusion, sents both stabilization and intensification of the TC are selected. (Schubert and Hack 1982; Shapiro and Willoughby Figure 15 shows the TC eye size estimated by pos- 1982; Jorgensen 1984; Weatherford and Gray 1988; itive IRWVln within 2 times the RMW. From satellite Vigh and Schubert 2009; Vigh et al. 2012). Weatherford imagery, it was confirmed that all four category 5 and Gray (1988) showed that TC eye size tends to storms formed an eye structure during the RI period.

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC 4076 MONTHLY WEATHER REVIEW VOLUME 148

FIG. 10. As in Fig. 6, but for category 5 storms.

The estimated TC eye size generally expands con- in TC eye size estimated from satellite imagery exhibits siderably during the day, while remaining constant or an opposite pattern compared to the convective activity shrinking at night, which is in concurrence with the (Figs. 10a,c,e and 15). findings of Muramatsu (1983). Since the eye is always Figure 16 shows schematic diagrams that summarize an open area, it may experience a significant radiational the diurnal variation of the convective cloud and eye cooling, and is very stable, that is, a nonconvective area, diameter identified through satellite imagery. Thick as compared with the eyewall. While the convective convective clouds can reduce the radiational cooling and activity is suppressed during the day, significant radia- simultaneously enhance radiational warming (Gray and tional cooling in the eye region may contribute to the Jacobson 1977); therefore, solar radiation can suppress removal of clouds that encroach into the eye because of convective clouds because of radiatively stabilized up- the convective activity. As a result, this diurnal variation per clouds during the day (Fig. 16a). If these convective

FIG. 11. As in Fig. 7, but for category 5 storms.

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC OCTOBER 2020 L E E E T A L . 4077

22 FIG. 12. The OHC (GJ m ) overlaid with TC tracks of (a),(c) Soudelor and (b),(d) Atsani. The status of OHC (top) before RI and (bottom) after RI. The RI period is displayed in Table 2. The black arrow indicates the TC 2 position of each date, and the contour is drawn at 4 GJ m 2 intervals. The OHC has been explicitly calculated by using the Global Hybrid Coordinate Ocean Model data. clouds do not dissipate significantly during the day, during the day and shrinks at night (Fig. 15). Since the destabilization could increase, caused by the contrast TC eye is represented as the cloud-free area, the radi- of net radiation between the upper and mid- or lower ational cooling could always be significant; therefore, clouds at night. Navarro and Hakim (2016) elucidated some of the clouds that encroach from the innermost this diurnal contrast through idealized numerical re- eyewall could dissipate considerably during the day, sults. From their Figs. 8 and 9, the radial-vertical while the convective clouds could again encroach into anomalous vectors are manifested as cyclonic circula- the eye region at night. This eventually contributes to tion during the daytime, while it is anticyclonic circu- contraction of the TC eye. lation in the nighttime. This anticyclonic circulation in the midtroposphere may enhance the convective ac- tivity at night. Zhang et al. (2020) recently discussed 7. Summary that the boundary layer inflow is deeper and stronger Based on Himawari-8 satellite imagery, this study ex- during the nighttime than in the daytime. During amines the diurnal variation of the convective area and eye nighttime and in the boundary layer, high equivalent size of RI TCs occurring in the western North Pacific from potential temperature and relative humidity were ob- 2015 to 2017. The findings are summarized as follows: served in the inner and outer areas, respectively. These previous studies may support the theory that kinetic d The ACA generally develops at night, while shrink- and thermodynamic structures of the storm can be al- ingsignificantlyduringtheday(Figs. 6, 8,and10). tered by the diurnal variation. These favorable inner- Generally, the RI TCs experience at least a single core variations caused by the diurnal variation may diurnal variation of the ACA inside the RMW explain why the TC’s convective activity becomes ac- during the RI period (Figs. 6, 8,and10). For intense tive mainly at night. Once the TC eye forms inside the TCs, the ACA is maintained longer both inside the RMW, the eye size tends to change inversely to the RMW and within 3 times the RMW, in which case convective activity, which is evidenced by the satellite the diurnal signal appears continuous (Figs. 8 and imagery that shows a large eye is usually identified 10). During the RI period, eyewall convective clouds

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC 4078 MONTHLY WEATHER REVIEW VOLUME 148

FIG. 13. The scatterplots that account for a relationship between the normalized IRWVln and TC intensity change at 30-min intervals of satellite imagery. Four TC intensity changes such as SI, RI, N, and W, are divided by the definitions shown in Table 5. Each scatter point includes both the current maximum surface wind and the normalized IRWVln in the eyewall area. The colored vertical lines denote the hurricane intensity. The cyan line indicates a linear regression between these two variables.

tend to be further organized (Figs. 9 and 11). However, highest correlation coefficient (Figs. 13d,h,l), indicating when the TC encounters low OHC, it could cause that TC intensification and weakening could be highly weakening of TC intensity as a result of the weakening related to TC eyewall convective clouds. of the original eyewall convective clouds (disorganiza- d From satellite imagery, the eye structure is normally tion of the ACA) (Figs. 10a,b, 11a,b,and12). formed during the RI period (Fig. 15). Once the TC d From the results of the linear regression analysis, the eye forms inside the RMW, the eye size tends to scatterplots demonstrate that RI tends to commence change inversely with the convective activity. For after SI stage (Figs. 13a,b,e,f,i,j), which indicates that RI example, a large eye is usually identified during the may require enhancement of the eyewall convective day, while contracting at night. Since the TC eye is cloud and the primary circulation during the SI phase. located in the cloud-free area, the radiational cooling The N phase simply represents a change of the eyewall is always significant, and thereby causing the clouds in convective cloud without any significant enhancement the eye region to dissipate over time, which may ex- of the primary circulation (Figs. 13c,g,k)andindicatesa plain the appearance of a large eye during the day. In temporary cessation of TC intensification or weakening. contrast, since the convective activity is invigorated at Of the four intensity changes, the W phase shows the night, the convective clouds could limit the eye size.

TABLE 5. Four phases of TC intensity change. The threshold indicates a 6-h intensity change, and the amount represents the number of scatterplots for each phase depending on categories.

Amount Phase Threshold Category 3 or below Category 4 Category 5

RI DVmax $ 7.5 kt 386 939 543 SI 2.5 ,DVmax , 7.5 kt 562 1437 693 N 22.5 # DVmax # 2.5 kt 362 1919 483 W DVmax ,22.5 kt 751 1838 988

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC OCTOBER 2020 L E E E T A L . 4079

FIG. 14. Snapshots of the field of IRWVln (K) for tropical storms (a) Dujuan (2015) at 2300 UTC 22 Sep and (b) Chaba (2016) at 1400 UTC 29 Sep. Both typhoons show an expansive area of ACA, mixed phase, and IACA.

In future work, we plan to quantitatively examine the also shallow or moderate convection in conjunction with difference in the ACA derived from IRWVln and con- other observations such as microwave satellite. In ad- vective area observed from microwave satellite imagery. dition, three brightness temperature definitions used to In particular, we will think about how this IRWVln distinguish the convective area remain to be thoroughly technique could address not only deep convection, but verified with in situ aircraft observation data. Finally, to

FIG. 15. The estimated TC’s eye area calculated within 2 times the RMW: (a) Soudelor, (b) Nepartak, (c) Meranti, and (d) Chaba, respectively. The two vertical-dashed lines indicate the RI period. The blue line in- dicates the percentage of ACA computed within 3 times the RMW (Figs. 10a,c–e).

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC 4080 MONTHLY WEATHER REVIEW VOLUME 148

FIG. 16. Schematic diagrams for the diurnal variation of convection and eye diameter during the (a) daytime and (b) nighttime. The long solid black arrows indicate vertical flows. During the day, the convective area could shrink due to solar radiation that makes the upper clouds stabilized. If thick clouds do not dissipate considerably during the day, it could act to reduce the radiational cooling in the mid- and lower troposphere. In this case, the radiational contrast between the upper clouds and mid- or lower clouds becomes significant at night. As a result, destabilization caused by this radiational contrast could enhance the convective activity. The arrows in the cylinder indicate the changes in TC eye size related to the diurnal cycle. The eye size is sensitive to the convective activity. For example, convective clouds developed from the innermost eyewalls may invade the eye area. In the early morning, the eye could seem small in the satellite imagery (Fig. 15). In contrast, since the radiational cooling is always significant in the eye region, some clouds could naturally dissipate from that region. As a result, the eye could become larger and more distinct during the day. elaborate the impact of diurnal variation on RI in terms eyewall vertical motion. J. Atmos. Sci., 63, 19–42, https:// of kinetic or thermodynamic perspectives, numerical doi.org/10.1175/JAS3598.1. simulations would also be conducted. Browner, S. P., W. L. Woodley, and C. G. Griffith, 1977: Diurnal oscillation of the area of cloudiness associated with tropical storms. Mon. Wea. Rev., 105, 856–864, https://doi.org/10.1175/ Acknowledgments. This work is supported by the 1520-0493(1977)105,0856:DOOTAO.2.0.CO;2. Ministry of Science and Technology of Taiwan under Chang, C.-C., and C.-C. Wu, 2017: On the processes leading Grants MOST 106-2111-M-002-013-MY3, MOST 107- to the rapid intensification of (2010). 2111-M-002-016-MY3, and by the Office of Naval J. Atmos. Sci., 74, 1169–1200, https://doi.org/10.1175/JAS- D-16-0075.1. Research through Grant N62909-16-1-2169. We thank Chen, H., and S. G. Gopalakrishnan, 2015: A study on the asym- Dr. Russell L. Elsberry for an excellent discussion metric rapid intensification of Hurricane Earl (2010) using the about the diurnal variation of the tropical . HWRF system. J. Atmos. Sci., 72, 531–550, https://doi.org/ Also, we thank Japan Aerospace Exploration Agency 10.1175/JAS-D-14-0097.1. for providing a full-disk dataset of Himawari-8/AHI Cheong, H.-B., I.-H. Kwon, and T.-Y. Goo, 2004: Further study on the high-order double-Fourier-series spectral filtering on a operated by JMA through the P-Tree System. Last, sphere. J. Comput. Phys., 193, 180–197, https://doi.org/10.1016/ helpful comments from Anna Vaughan, Yi-Hsuan Hwang, j.jcp.2003.07.029. and two anonymous reviewers are also highly appreciated. DeHart, J. C., R. A. Houze Jr., and R. F. Rogers, 2014: Quadrant distribution of inner-core kinematics in rela- REFERENCES tion to environmental shear. J. Atmos. Sci., 71, 2713–2732, https://doi.org/10.1175/JAS-D-13-0298.1. Bedka, K., J. Brunner, R. Dworak, W. Feltz, J. Otkin, and DeMaria, M., 1996: The effect of vertical shear on tropical cyclone T. Greenwald, 2010: Objective satellite-based detection of intensity change. J. Atmos. Sci., 53, 2076–2088, https://doi.org/ overshooting tops using infrared window channel brightness 10.1175/1520-0469(1996)053,2076:TEOVSO.2.0.CO;2. temperature gradients. J. Appl. Meteor. Climatol., 49, 181–202, Dunion, J. P., C. D. Thorncroft, and C. S. Velden, 2014: The https://doi.org/10.1175/2009JAMC2286.1. tropical cyclone diurnal cycle of mature hurricanes. Mon. Bessho, K., and Coauthors, 2016: An introduction to Himawari- Wea. Rev., 142, 3900–3919, https://doi.org/10.1175/MWR- 8/9—Japan’s new-generation geostationary meteorological D-13-00191.1. satellites. J. Meteor. Soc. Japan, 94, 151–183, https://doi.org/ Dvorak, V. F., 1975: Tropical cyclone intensity analysis and 10.2151/jmsj.2016-009. forecasting from satellite imagery. Mon. Wea. Rev., 103, Braun, S. A., and L. Wu, 2007: A numerical study of Hurricane 420–430, https://doi.org/10.1175/1520-0493(1975)103,0420: Erin (2001). Part II: Shear and the organization of eyewall TCIAAF.2.0.CO;2. vertical motion. Mon. Wea. Rev., 135, 1179–1194, https:// ——, 1984: Tropical cyclone intensity analysis using satellite doi.org/10.1175/MWR3336.1. data.NOAATech.Rep.NESDIS11,NOAA,46pp.,http:// ——, M. T. Montgomery, and Z. Pu, 2006: High-resolution simu- satepsanone.nesdis.noaa.gov/pub/Publications/Tropical/ lation of Hurricane Bonnie (1998). Part I: The organization of Dvorak_1984.pdf.

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC OCTOBER 2020 L E E E T A L . 4081

Ebert, E. E., and G. J. Holland, 1992: Observations of record cold due to bursts. Mon. Wea. Rev., 145, 3095–3117, https:// cloud-top temperatures in Tropical Cyclone Hilda (1990). doi.org/10.1175/MWR-D-16-0268.1. Mon. Wea. Rev., 120, 2240–2251, https://doi.org/10.1175/1520- Hendricks, E. A., M. S. Peng, B. Fu, and T. Li, 2010: Quantifying 0493(1992)120,2240:OORCCT.2.0.CO;2. environmental control on tropical cyclone intensity change. Elsberry, R. L., and M.-S. Park, 2017: Comments on ‘‘Multiscale Mon. Wea. Rev., 138, 3243–3271, https://doi.org/10.1175/ structure and evolution of Hurricane Earl (2010) during rapid 2010MWR3185.1. intensification.’’ Mon. Wea. Rev., 145, 1565–1571, https:// Holliday, C. R., and A. H. Thompson, 1979: Climatological char- doi.org/10.1175/MWR-D-16-0301.1. acteristics of rapidly intensifying typhoons. Mon. Wea. Rev., Fischer, M. S., B. H. Tang, K. L. Corbosiero, and C. M. Rozoff, 107, 1022–1034, https://doi.org/10.1175/1520-0493(1979)107,1022: 2018: Normalized convective characteristics of tropical cy- CCORIT.2.0.CO;2. clone rapid intensification events in the North Atlantic and Jiang, H., 2012: The relationship between tropical cyclone in- eastern North Pacific. Mon. Wea. Rev., 146, 1133–1155, https:// tensity change and the strength of inner-core convection. doi.org/10.1175/MWR-D-17-0239.1. Mon. Wea. Rev., 140, 1164–1176, https://doi.org/10.1175/ ——, ——, and ——, 2019: A climatological analysis of tropical MWR-D-11-00134.1. cyclone rapid intensification in environments of upper- Jones, S. C., 1995: The evolution of vortices in vertical shear. I: tropospheric troughs. Mon. Wea. Rev., 147, 3693–3719, https:// Initially barotropic vortices. Quart. J. Roy. Meteor. Soc., 121, doi.org/10.1175/MWR-D-19-0013.1. 821–851, https://doi.org/10.1002/qj.49712152406. Frank, W. M., and E. A. Ritchie, 1999: Effects of environmental Jorgensen, D. P., 1984: Mesoscale and convective-scale characteristics flow upon tropical cyclone structure. Mon. Wea. Rev., 127, of mature hurricanes. Part II: Inner core structure of Hurricane 2044–2061, https://doi.org/10.1175/1520-0493(1999)127,2044: Allen (1980). J. Atmos. Sci., 41, 1287–1311, https://doi.org/ EOEFUT.2.0.CO;2. 10.1175/1520-0469(1984)041,1287:MACSCO.2.0.CO;2. ——, and ——, 2001: Effects of vertical wind shear on the intensity Kaplan, J., and M. DeMaria, 2003: Large-scale characteristics of and structure of numerically simulated hurricanes. Mon. Wea. rapidly intensifying tropical cyclones in the North Atlantic ba- Rev., 129, 2249–2269, https://doi.org/10.1175/1520-0493(2001) sin. Wea. Forecasting, 18, 1093–1108, https://doi.org/10.1175/ 129,2249:EOVWSO.2.0.CO;2. 1520-0434(2003)018,1093:LCORIT.2.0.CO;2. Fudeyasu, H., K. Ito, and Y. Miyamoto, 2018: Characteristics of Kelley, O. A., and J. B. Halverson, 2011: How much tropical cy- tropical cyclone rapid intensification over the western North clone intensification can result from the energy released inside Pacific. J. Climate, 31, 8917–8930, https://doi.org/10.1175/ of a convective burst? J. Geophys. Res., 116, D20118, https:// JCLI-D-17-0653.1. doi.org/10.1029/2011JD015954. Gentry, R. C., E. Rodgers, J. Steranka, and W. E. Shenk, 1980: Kieper, M., and H. Jiang, 2012: Predicting tropical cyclone rapid Predicting tropical cyclone intensity using satellite-measured intensification using the 37 GHz ring pattern identified from equivalent blackbody temperatures of cloud tops. Mon. Wea. passive microwave measurements. Geophys. Res. Lett., 39, Rev., 108, 445–455, https://doi.org/10.1175/1520-0493(1980) L13804, https://doi.org/10.1029/2012GL052115. 108,0445:PTCIUS.2.0.CO;2. Kim, H.-M., P. J. Webster, and J. A. Curry, 2011: Modulation of Gray, W. M., 1968: Global view of the origin of tropical distur- North Pacific tropical cyclone activity by three phases of bances and storms. Mon. Wea. Rev., 96, 669–700, https://doi.org/ ENSO. J. Climate, 24, 1839–1849, https://doi.org/10.1175/ 10.1175/1520-0493(1968)096,0669:GVOTOO.2.0.CO;2. 2010JCLI3939.1. ——, and R. W. Jacobson Jr., 1977: Diurnal variation of deep Kossin, J. P., 2002: Daily hurricane variability inferred from GOES cumulus convection. Mon. Wea. Rev., 105, 1171–1188, https:// infrared imagery. Mon. Wea. Rev., 130, 2260–2270, https://doi.org/ doi.org/10.1175/1520-0493(1977)105,1171:DVODCC.2.0.CO;2. 10.1175/1520-0493(2002)130,2260:DHVIFG.2.0.CO;2. Guimond, S. R., G. M. Heymsfield, and F. J. Turk, 2010: Multiscale Kurino, T., 1997: A satellite infrared technique for estimating observations of Hurricane Dennis (2005): The effects of hot ‘‘deep/shallow’’ precipitation. Adv. Space Res., 19, 511–514, towers on rapid intensification. J. Atmos. Sci., 67, 633–654, https://doi.org/10.1016/S0273-1177(97)00063-X. https://doi.org/10.1175/2009JAS3119.1. Lander, M. A., 1999: A tropical cyclone with an enormous central ——, ——, P. D. Reasor, and A. C. Didlake Jr., 2016: The rapid cold cover. Mon. Wea. Rev., 127, 132–136, https://doi.org/ intensification of Hurricane Karl (2010): New remote sensing 10.1175/1520-0493(1999)127,0132:ATCWAE.2.0.CO;2. observations of convective bursts from the Global Hawk Lee, J.-D., and C.-C. Wu, 2018: The role of polygonal eyewalls in platform. J. Atmos. Sci., 73, 3617–3639, https://doi.org/10.1175/ rapid intensification of Typhoon Megi (2010). J. Atmos. Sci., JAS-D-16-0026.1. 75, 4175–4199, https://doi.org/10.1175/JAS-D-18-0100.1. Harnos, D. S., and S. W. Nesbitt, 2011: Convective structure in Leppert, K. D., and D. J. Cecil, 2016: Tropical cyclone diurnal cycle rapidly intensifying tropical cyclones as depicted by passive as observed by TRMM. Mon. Wea. Rev., 144, 2793–2808, microwave measurements. Geophys. Res. Lett., 38, L07805, https://doi.org/10.1175/MWR-D-15-0358.1. https://doi.org/10.1029/2011GL047010. Malkus, J. S., C. Ronne, and M. Chafee, 1961: Cloud patterns in ——, and ——, 2016: Passive microwave quantification of tropical Hurricane Daisy, 1958. Tellus, 13,8–30,https://doi.org/10.3402/ cyclone inner-core cloud populations relative to subsequent tellusa.v13i1.9439. intensity change. Mon. Wea. Rev., 144, 4461–4482, https:// Merritt, E. S., and R. Wexler, 1967: Cirrus canopies in tropical doi.org/10.1175/MWR-D-15-0090.1. storms. Mon. Wea. Rev., 95, 111–120, https://doi.org/10.1175/ Hazelton, A. T., R. Rogers, and R. E. Hart, 2017a: Analyzing 1520-0493(1967)095,0111:CCITS.2.3.CO;2. simulated convective bursts in two Atlantic hurricanes. Part I: Molinari, J., and D. Vollaro, 2010: Rapid intensification of a Burst formation and development. Mon. Wea. Rev., 145, 3073– sheared tropical storm. Mon. Wea. Rev., 138, 3869–3885, 3094, https://doi.org/10.1175/MWR-D-16-0267.1. https://doi.org/10.1175/2010MWR3378.1. ——, ——, and ——, 2017b: Analyzing simulated convective Monette, S. A., C. S. Velden, K. S. Griffin, and C. Rozoff, 2012: bursts in two Atlantic hurricanes. Part II: Intensity change Examining trends in satellite-detected tropical overshooting

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC 4082 MONTHLY WEATHER REVIEW VOLUME 148

tops as a potential predictor of tropical cyclone rapid inten- Schubert, W. H., and J. J. Hack, 1982: Inertial stability and tropical sification. J. Appl. Meteor. Climatol., 51, 1917–1930, https:// cyclone development. J. Atmos. Sci., 39, 1687–1697, https:// doi.org/10.1175/JAMC-D-11-0230.1. doi.org/10.1175/1520-0469(1982)039,1687:ISATCD.2.0.CO;2. Muramatsu, T., 1983: Diurnal variations of satellite-measured TBB Shapiro, L. J., and H. E. Willoughby, 1982: The response of balanced areal distribution and eye diameter of mature typhoons. J. Meteor. hurricanes to local sources of heat and momentum. J. Atmos. Soc. Japan, 61, 77–90, https://doi.org/10.2151/jmsj1965.61.1_77. Sci., 39, 378–394, https://doi.org/10.1175/1520-0469(1982) Navarro, E. L., and G. J. Hakim, 2016: Idealized numerical mod- 039,0378:TROBHT.2.0.CO;2. eling of the diurnal cycle of tropical cyclones. J. Atmos. Sci., Steranka, J., E. Rodgers, and R. Gentry, 1984: The diurnal varia- 73, 4189–4201, https://doi.org/10.1175/JAS-D-15-0349.1. tion of Atlantic Ocean tropical cyclone cloud distribution in- Olander, T. L., and C. S. Velden, 2009: Tropical cyclone convection ferred from geostationary satellite infrared measurements. and intensity analysis using differenced infrared and water va- Mon. Wea. Rev., 112, 2338–2344, https://doi.org/10.1175/1520- por imagery. Wea. Forecasting, 24, 1558–1572, https://doi.org/ 0493(1984)112,2338:TDVOAO.2.0.CO;2. 10.1175/2009WAF2222284.1. ——, ——, and ——, 1986: The relationship between satellite Reasor, P. D., M. T. Montgomery, and L. D. Grasso, 2004: A new measured convective bursts and tropical cyclone intensifica- look at the problem of tropical cyclones in vertical shear flow: tion. Mon. Wea. Rev., 114, 1539–1546, https://doi.org/10.1175/ Vortex resiliency. J. Atmos. Sci., 61, 3–22, https://doi.org/ 1520-0493(1986)114,1539:TRBSMC.2.0.CO;2. 10.1175/1520-0469(2004)061,0003:ANLATP.2.0.CO;2. Tang, B., and K. Emanuel, 2010: Midlevel ventilation’s constraint ——, M. D. Eastin, and J. F. Gamache, 2009: Rapidly intensifying on tropical cyclone intensity. J. Atmos. Sci., 67, 1817–1830, Hurricane Guillermo (1997). Part I: Low-wavenumber struc- https://doi.org/10.1175/2010JAS3318.1. ture and evolution. Mon. Wea. Rev., 137, 603–631, https:// Tang, X., Z.-M. Tan, J. Fang, E. B. Munsell, and F. Zhang, 2019: doi.org/10.1175/2008MWR2487.1. Impact of the diurnal radiation contrast on the contraction of Rios-Berrios, R., C. A. Davis, and R. D. Torn, 2018: A hypothesis radius of maximum wind during intensification of Hurricane for the intensification of tropical cyclones under moderate Edouard (2014). J. Atmos. Sci., 76, 421–432, https://doi.org/ vertical wind shear. J. Atmos. Sci., 75, 4149–4173, https:// 10.1175/JAS-D-18-0131.1. doi.org/10.1175/JAS-D-18-0070.1. Vigh, J. L., and W. H. Schubert, 2009: Rapid development of the Rogers, R., 2010: Convective-scale structure and evolution tropical cyclone warm core. J. Atmos. Sci., 66, 3335–3350, during a high-resolution simulation of tropical cyclone rapid https://doi.org/10.1175/2009JAS3092.1. intensification. J. Atmos. Sci., 67, 44–70, https://doi.org/10.1175/ ——, J. A. Knaff, and W. H. Schubert, 2012: A climatology of 2009JAS3122.1. hurricane eye formation. Mon. Wea. Rev., 140, 1405–1426, ——, P. Reasor, and S. Lorsolo, 2013: Airborne Doppler obser- https://doi.org/10.1175/MWR-D-11-00108.1. vations of the inner-core structural differences between in- Weatherford, C. L., and W. M. Gray, 1988: Typhoon structure as tensifying and steady-state tropical cyclones. Mon. Wea. Rev., revealed by aircraft reconnaissance. Part II: Structural vari- 141, 2970–2991, https://doi.org/10.1175/MWR-D-12-00357.1. ability. Mon. Wea. Rev., 116, 1044–1056, https://doi.org/10.1175/ ——, ——, and J. A. Zhang, 2015: Multiscale structure and evo- 1520-0493(1988)116,1044:TSARBA.2.0.CO;2. lution of Hurricane Earl (2010) during rapid intensification. Weickmann, H. K., A. B. Long, and L. R. Hoxit, 1977: Some ex- Mon. Wea. Rev., 143, 536–562, https://doi.org/10.1175/MWR- amples of rapidly growing oceanic cumulonimbus clouds. D-14-00175.1. Mon. Wea. Rev., 105, 469–476, https://doi.org/10.1175/1520- Sadler, J. C., 1964: Tropical cyclones of the eastern North Pacific 0493(1977)105,0469:SEORGO.2.0.CO;2. as revealed by TIROS observations. J. Appl. Meteor., 3, Willoughby, H. E., 1990: Temporal changes of the primary circulation 347–366, https://doi.org/10.1175/1520-0450(1964)003,0347: in tropical cyclones. J. Atmos. Sci., 47, 242–264, https://doi.org/ TCOTEN.2.0.CO;2. 10.1175/1520-0469(1990)047,0242:TCOTPC.2.0.CO;2. Schmetz, J., S. A. Tjemkes, M. Gube, and L. van de Berg, 1997: Zhang J. A., J. P. Dunion, and D. S. Nolan, 2020: In situ observa- Monitoring deep convection and convective overshooting tions of the diurnal variation in the boundary layer of mature with METEOSAT. Adv. Space Res., 19, 433–441, https:// hurricanes. Geophys. Res. Lett., 47, 2019GL086206, https:// doi.org/10.1016/S0273-1177(97)00051-3. doi.org/10.1029/2019GL086206.

Unauthenticated | Downloaded 10/07/21 11:25 AM UTC