Westward shift of tropical cyclogenesis over the southern Indian Ocean

Tropical cyclones (TCs), commonly called cyclones in the southern Indian Ocean (SIO), represent one of the most devastating disasters in the oceanfront regions of Africa. The present study explores the long-term tendency of annual mean TC genesis location in the SIO. A notable westward shift is detected in the SIO TC genesis longitude since 1979, which is linked to an increase in the TC genesis frequency in the southwestern SIO and a decrease in the TC genesis frequency in the northeastern SIO. The dipole trend pattern of the TC genesis frequency in the SIO is intimately linked to the weakening of the westerly vertical wind shear over the western SIO and the strengthening of the easterly vertical wind shear over the eastern SIO, resulting from a reduced meridional temperature gradient. The weakened meridional temperature gradient is attributed to the enhanced warming of the subtropical troposphere that is a response of atmospheric temperature to global warming. Our study implies a potential increase in the risks faced by coastal and island countries in eastern Africa.


Introduction
The South Indian Ocean (SIO) is recognized as one of the zones of vigorous tropical cyclogenesis around the globe (Schreck et al 2014).The tropical cyclones (TCs) over the SIO pose significant risks to island countries such as Madagascar, Mauritius, and the Mascarene Islands as well as the coastal communities along East Africa, leading to casualties and substantial socioeconomic losses (Mendelsohn et al 2012, Vidya et al 2021).The potential effects of humaninduced global warming on TCs have been thoroughly reviewed recently, including the observational trends and the future projection across the globe (Knutson et al 2019(Knutson et al , 2020)).However, compared to changes in TC activities over the northern Pacific and Atlantic Oceans (Cao et al 2018a(Cao et al , 2019(Cao et al , 2023)), relatively limited investigations have been conducted on changes in TCs over the SIO.
Earlier investigations have presented observed trends using various TC metrics over the SIO.These metrics encompass TC genesis, occurrence, intensity, latitude of TC maximum intensity, and TC destruction potential (Kuleshov et al 2008, 2010, Kossin et al 2014, 2016, Murakami et al 2020, Vidya et al 2021, Murakami 2022).Kossin et al (2014) demonstrated a statistically substantial trend of poleward shift in the mean latitude of TC lifetime maximum intensity globally, including the SIO.According to Murakami et al (2020), a clear spatial pattern of trends in TC occurrence frequency have been observed on a global scale since 1980, with a significant decrease over the SIO.They proposed that external forcing like volcanic eruptions, aerosols, and greenhouse gases are likely to play a crucial role in this trend.Murakami (2022) uncovered that the decrease in aerosol emissions across the United States and Europe has significantly aided in the marked decrease in TC occurrence over the SIO.There is no conspicuous trend in the whole number of TCs achieving a minimum central pressure of either 995 hPa or 970 hPa (Kuleshov et al 2008(Kuleshov et al , 2010)).However, Kuleshov et al (2010) identified a significantly rising tendency in TC numbers with a minimum central pressure of 945 hPa and 950 hPa over the SIO. Recently, Vidya et al (2021) examined the effects of SIO warming on the power dissipation index (PDI) for two distinct periods : 1980-1998 and 1999-2016.They found that the PDI within the SIO has doubled from 1999 to 2016 in comparison to the period from 1980 to 1998, primarily attributed to an escalation in TC intensity and duration.Oceanic processes are identified as major contributors to the rise in PDI during recent decades (Vidya et al 2021).
Prior studies have primarily analyzed the observed trends in TC number and intensity over the SIO.There has been a lack of investigation into the potential variation in the TC genesis location within the SIO region during the past 40 years.Some previous studies have explored the latitudinal and longitudinal change in the TC genesis over the western North Pacific and the North Indian Ocean (Wu et al 2015, Daloz and Camargo 2018, Sharmila and Walsh 2018, Fan et al 2019, Shan and Yu 2020).One question is whether the TC genesis region has experienced a significant tendency over the SIO.In this study, we reveal a robust westward migration in the longitude of SIO TC genesis since 1979.The primary aim of our research is to delve into the underlying physical mechanisms responsible for this westward shift in the SIO TC genesis.An improved understanding of the long-term trend of the SIO TC genesis and its mechanisms can help policymakers and communities to better prepare for and adapt to changing TC patterns, mitigating the potential impacts in vulnerable coastal regions and enhancing disaster resilience in the affected areas.
The subsequent sections of the current study are organized as below.Section 2 affords a simple introduction of the data and methodologies within this study.Section 3 presents the observed trend in TC genesis longitude over the SIO.In section 4, we examine the physical drivers accounting for the westward migration in the SIO TC genesis.In conclusion, section 5 wraps up with a summary and a discussion.

Data and methodologies
The long-term trend of TC genesis over the SIO is delineated by the historical TC best-track dataset of the U.S. warning agencies, archived in the International Best Track Archive for Climate Stewardship (IBTrACS) version 4 (Knapp et al 2010).The TC dataset encompasses the 6 h latitude and longitude of the TC location as well as the maximum sustained wind speed, which is obtained from the National Centers for Environmental Information, National Oceanic and Atmospheric Administration.In accordance with prior research, the TC genesis is identified as the initial time that the TC wind speed surpasses 34 kts (approximately 17 m s −1 ) within the SIO region (Zhao et al 2019, Cao et al 2020, Murakami 2022).The current study concentrates on the region spanning from 30 • E to 135 • E and from 0 • to 30 • S where nearly 95% of TCs over the SIO are generated.The TC season within the SIO spans from 1 July of the preceding year to 30 June of the subsequent year (Klotzbach et al 2022).For example, the TC season denoted as '1979' represents the period from July 1978 to June 1979.We also examine the TC genesis change during the calendar year as well as during the active TC season spanning from November to April and the results are similar (figures not shown).
The monthly mean atmospheric parameters are derived from the ERA5 dataset with a grid resolution of 2.5 • × 2.5 • (Hersbach et al 2020).The worldwide monthly mean sea surface temperature (SST) dataset is derived from the Extended Reconstructed Sea Surface Temperature version 5 dataset, featuring a grid resolution of 2 • × 2 • and commencing from 1854 (Huang et al 2017).
The dynamic genesis potential index (DGPI) is an empirical and comprehensive metric utilized to evaluate the collective influence of various large-scale dynamical parameters on the TC genesis (Wang and Murakami 2020).The index is computed as follows: where 11.8 , V shear denotes the magnitude of the vertical wind vector difference between 200 hPa and 850 hPa, ∂u ∂y is the meridional gradient of zonal wind at 500 hPa, ω is the 500 hPa vertical p-velocity, and η represents the 850 hPa absolute vorticity.The contribution of the four dynamical terms to the DGPI anomalies on the interdecadal time scale are assessed in the subsequent manner (Li et al 2013, Cao et al 2023): (2) In equation ( 2), the entirety of the DGPI anomalies are decomposed into four linear components, each linked with the interdecadal anomalies of one factor and the climatological mean values of the remaining three factors.The four components on the right-hand side of equation (2) denote the impacts of vertical wind shear, mid-level meridional zonal wind gradient, mid-level vertical motion, and lower-level absolute vorticity, respectively.
Presently, a seeding-transition approach is formulated to elucidate the TC genesis (Hsieh et al 2020, 2022, Yang et al 2021).This approach divides the TC genesis process into two discernible phases: an initial seeding stage, succeeded by a subsequent transition stage.Hsieh et al (2020) introduced a seed propensity index (SPI) as a means to quantify the likelihood of seed formation within background circulation, which is calculated below: where f denotes the Coriolis parameter, β is the meridional gradient of f, ζ is the 850 hPa relative vorticity, and U is a constant of 20 m s −1 and α represents a fitting parameter set to 0.69 (Hsieh et al 2020).
The likelihood of a seed transforming into a TC is represented by the transition probability (TP).This TP is represented using a sigmoid function of the ventilation index (Tang and Emanuel 2012, Hsieh et al 2020, 2022), which is calculated as follows: where Λ represents the non-dimensional ventilation index, UWS is the vertical zonal wind shear between 200 hPa and 850 hPa, χ represents the moist saturation deficit, PI denotes the potential intensity (Bister and Emanuel 2002), Λ 0 = 0.014 and γ = 0.89 according to Tang and Emanuel (2012).
In the non-dimensional ventilation index, the moist saturation deficit (χ ) parameter is computed as below (Emanuel et al 2008, Emanuel 2013): where s * and s * 0 represent the moist entropies saturated at 600 hPa and the sea surface, respectively, and s m represents the moist entropy at 600 hPa, T is the temperature, r v is the mixing ratio, p d represents the partial pressure of dry air, RH denotes the relative humidity, and the remaining parameters are kept constant.A higher value of χ indicates a more significant deficit in the mid-tropospheric moisture content, which implies an environment less conducive to TC genesis.
The present study uses the monthly mean outputs of 19 atmospheric, general circulation model runs, that participate in the Coupled Model Intercomparison Project phase 6 (CMIP6) archive (CMIP6/AMIP for brevity) (Eyring et al 2016).The name of the models and the model centers are presented in table 1.All the model outputs are converted to a horizontal resolution of 2.5 • × 2.5 • for consistency.The CMIP6/AMIP runs are driven by observed SST and sea ice concentration spanning the time period from 1979 to 2014.
The least-square linear regression analysis is employed for assessing the long-term trend.The significance of the linear trend is determined using the nonparametric Mann-Kendall method.The Student's t test is employed to assess the significance of the correlation coefficient, regression coefficient, and composite analysis.

Westward shift of the SIO TC genesis
The basin-wide average location of the SIO TC genesis is ascertained by calculating the mean longitude of annual mean TCs that achieve the initial intensity of 34 kts. Figure 1(a) displays a robust westward migration of the TC genesis location with a linear trend of 0.16 • longitude/year since 1979.Shan and Yu (2020) showed that the rate of poleward shift of the yearly averaged latitude where TCs form in the Northern Hemisphere is markedly impacted by the regional fluctuations in the number of the TC genesis over the past few decades.Thus, we further divide the entire SIO region into three smaller regions using the longitudes of 70 • E and 115 • E. The westward shift tendency of the TC genesis over the SIO is in accordance with a linear increasing trend of the TC genesis number in the western sector of the SIO (30 • -70 • E) and a linear decreasing trend of the TC genesis number in the eastern sector of the SIO (70 • -115 • E) (figures 1(b) and (c)).However, the linear trend of the TC genesis frequency is rather weak in the region of 115 Wu et al (2019) put forward that the factors influencing TC genesis variation are not uniformly distributed throughout the total region.This was consistent with our previous studies (Cao et al 2018b) that discovered a distinct difference in the correlation between TC variation to the north and south of 15 • N over the western North Pacific and the El Niño Modoki phenomenon.Thus, we further divide the SIO (0 • -30 • S, 30 • -115 • E) into four portions with boundaries along 15 • S and 70 • E, ensuring almost equivalent regions for each portion.

Physical mechanisms for the westward shift of SIO TC genesis
Why does the TC genesis increase over the southwestern region of the SIO and decrease over the northeastern region of the SIO?To address this question, we investigate changes in the environmental parameters that are intricately tied to the TC genesis (Chia and Ropelewski 2002, Wang and Chan 2002, Cao et al 2018a, 2019, 2023, Xu et al 2023).In order to quantitatively evaluate the impacts of various large-scale environmental parameters to the TC genesis over the SIO, we employ the DGPI formulated by Wang and Murakami (2020).The study period is separated into two equal periods: 2001-2022 and 1979-2000.Figure 3 illustrates the differences in the annualmean DGPI and the individual factors between these two periods.Notably, positive DGPI anomalies are observed in the western sector of the SIO, coinciding with the increase in the TC genesis in that region, while negative DGPI anomalies are present in the eastern sector, corresponding to the decrease in the TC genesis in that area (figure 3  However, the linear increasing trend of the TC genesis in this tropical region is insignificant (figure 2(a)).The dominant term promoting the dipole pattern of the TC genesis trend between the southwestern region and the northeastern region of the SIO is the vertical wind shear (figure 3(f)).The remaining three terms make a fairly minor contribution to the changes in the TC genesis across the respective regions (figures 3(c)-(e)).It is interesting to note that the vertical wind shear term exhibits a nearly perfect match with the observed dipole change in the TC genesis.Specifically, the vertical wind shear promotes the increase of the TC genesis over the southwestern region of the SIO and the decrease of the TC genesis over the northeastern region of the SIO, respectively (figures 3(f) and 2(b), (c)).Furthermore, the significance level for vertical wind shear is found to be stronger than that for the DGPI (figures 3(a) and (f)).It suggests that other terms like absolute vorticity and meridional shear of zonal wind may counteract the contribution of the vertical wind shear to some extent.Note that the seasonal mean (from November to April) differences show the results similar to those based on the annual mean (figure not shown).
The analysis presented above highlights a pivotal role of the vertical wind shear in the observed changes in TC genesis across the southwestern and northeastern regions of the SIO.In the following, we examine the trends of vertical zonal wind shear between 200 hPa and 850 hPa and tropospheric temperature averaged from 850 hPa to 200 hPa, as depicted in figures 4 and 5.This is because total vertical wind shear is dominantly contributed by changes in zonal wind.A remarkable uniform negative trend of vertical zonal wind shear is observed across the entire SIO (figure 4(a)).Concurrently, the tropospheric temperature exhibits a consistent warming trend across the SIO, with more pronounced warming in the subtropical region (figure 4(d)).Physically, the negative vertical wind shear is intimately connected to the reduced temperature gradient according to the thermal wind balance.Specifically, the temperature gradient in the western region of the SIO is characterized by a contrast in mean temperature between the tropics (5 • S-10 • N, 30 • -70 • E) and the subtropics (15 • S-30 • S, 30 • -70 • E) at altitudes between 850 hPa and 200 hPa.Likewise, the temperature gradient in the eastern region of the SIO is defined by a contrast in mean temperatures between the tropical region (5 • S-10 • N, 70 • -115 • E) and the subtropical region (10 • S-25 • S, 70 • -115 • E).It is seen that the negative vertical zonal wind shear is intimately linked to the decreased temperature gradient in the western and eastern regions of the SIO (figures 4(b), (c) and (e), (f)), with a high correlation coefficient of 0.95.
Figures 5(a  wind at upper-levels and easterly wind at lower-levels (figure 5(a)).The decreased temperature gradient across the troposphere corresponds to the warming trends observed in the upper levels of the subtropical troposphere (figure 5(c)), consequently resulting in a decline in the vertical westerly wind shear.This decrease in vertical wind shear creates a beneficial setting for the heightened TC genesis over the western region of the SIO (figure 1(b)).
In contrast, the eastern region of the SIO is characterized by the mean vertical easterly wind shear, composed of upper-level stronger easterly wind and lower-level weaker easterly wind (figure 5(b)).The negative trend of vertical wind shear, induced by the warming trends in the subtropical troposphere at the upper level (figure 5(d)), amplifies the vertical easterly wind shear.The reinforced vertical easterly wind shear creates an environment that is less conducive to the TC genesis over the eastern SIO (figure 1(c)).
We use the thermal wind equation to estimate quantitatively the trend of upper-level wind.The thermal wind equation is given as follows: where R represents the gas constant, P 0 and P 1 represent 850 hPa and 200 hPa, respectively, and T is tropospheric temperature averaged from 850 to 200 hPa.The trend of upper-level wind derived from the thermal wind equation closely resembles that obtained directly from the ERA5 reanalysis data except for the equatorial region where the thermal X Cao et al wind equation is not applicable (figure 6).It indicates that the trends in the upper-level wind and the corresponding vertical zonal wind shear are primarily governed by the thermal wind balance.
Recent findings have proposed that the TC genesis is the product of the TC seed frequency and the seed transition probability (Vecchi et al 2019, Hsieh et al 2020, 2022, Sugi et al 2020, Yang et al 2021, Emanuel 2022).Here, we examine how the TC seed and seed transition probability contribute to the long-term tendency of the TC genesis location over the SIO based on the SPI and TP indices.Resembling the changes observed in the TC genesis number, the seed transition probability displays a significant increasing tendency across the southwestern SIO, while a decreasing tendency is observed across the northeastern SIO (figure 7(b)).In sharp contrast, the TC seed index does not exhibit a significant trend over the SIO (figure 7(a)).These findings suggest that the westward migration in the TC genesis location is primarily dependent on the seed transition probability.The analysis of the TP index reveals that the vertical zonal wind shear is the dominant factor contributing to the long-term trend of seed transition probability, whereas moist saturation deficit and potential intensity have a minor impact on the dipole pattern observed in seed transition probability (figures 7(d)-(f)).Consequently, there is a pronounced link between the dipole pattern of the TC genesis over the SIO and the seed transition probability, primarily mediated by the influence of vertical zonal wind shear.
The previous analysis has shown that the trend of vertical zonal wind shear is physically linked to the warming trend in the subtropical troposphere at the upper level.The question arises as to why there is a warming trend in the subtropical troposphere over the SIO. Figure 8 illustrates the observed SST trend  from 1979 to 2022.There is a robust warming trend over the whole SIO, with the most substantial warming occurring in the subtropical region.Prior research has detected a substantial rise in the upper ocean heat content and SST in the SIO during recent periods, which is predominantly attributed to the escalated transport of warm upper-ocean waters from the western Pacific region via the Indonesian Throughflow  the subtropical sector of the SIO is likely to have a significant influence on triggering the enhanced warming in the troposphere.This warming effect intensifies with altitude, reaching its zenith around 300 hPa (as depicted in figures 5(c) and (d)), which is driven by the moist adiabatic adjustment process (Stone and Carlson 1979, Su et al 2003, Held and Soden 2006).
In order to confirm the linkage between the SST warming pattern and the enhanced upper tropospheric warming over the SIO, we examine the vertical profiles of the climatological mean temperature and ensemble-mean temperature trend from the 19 CMIP6/AMIP runs (figure 9), similar to those in observational analysis (figures 5(c) and (d)).The time period of analysis spans from 1980 to 2014.It is seen that the upper-level warming patterns also occur in the subtropical region (figure 9).Compared to the observational results (figures 5(c) and (d)), the upper-level warming pattern from the CMIP6/AMIP runs are slightly stronger and wider.

Summary and discussion
A significant westward migration in the annual mean TC genesis across the SIO has been identified since 1979.Our analysis shows that it is closely associated with a rise in the TC genesis in the southwestern SIO and a reduction in the TC genesis in the northeastern SIO.Analysis based on the DGPI decomposition and the seed-transition framework reveals that the seed transition probability has a crucial contribution in the TC genesis location change over the SIO and the vertical wind shear is a key driver behind the dipole pattern of the TC genesis location change.Notably, a significantly negative trend in vertical zonal wind shear is observed over the entire SIO, which is  physically tied to the weakened meridional temperature gradient in light of thermal wind balance.This weakened meridional temperature gradient is a consequence of the subtropical tropospheric warming at the upper level.As such, the negative trend in vertical shear of zonal wind reduces the vertical westerly wind shear, which favors the TC genesis in the western region of the SIO.Simultaneously, the negative trend amplifies the vertical easterly wind shear, which hinders TC genesis in the eastern region of the SIO.Their combined effects have resulted in a westward migration of the yearly TC genesis longitude over the SIO during the past 40 years.Our finding sheds new light on the interplay of atmospheric factors influencing TC genesis longitude and highlights the significance of the subtropical tropospheric warming at the upper level as a driver for the observed westward shift in the TC genesis over the SIO.Knapp and Kossin (2007) pointed out that geostationary satellite dataset across the Indian Ocean was not accessible until the deployment of Meteosat-5 in 1989.As a result, acquiring a precise estimation of TC intensity before 1990 could pose a challenge (Klotzbach et al 2022).Our study displays a significant westward shift of the TC genesis since 1979, which is linked to large-scale environmental dynamic and thermodynamical factors.This finding indirectly implies that TC genesis data may be more dependable compared to TC intensity data over the SIO.Additionally, our study demonstrates a considerable warming tendency at the upper level of subtropical troposphere over the SIO.The exact physical mechanisms behind the subtropical warming at the upper level over the SIO requires further in-depth investigation in future research endeavors.The complexity of the interactions between oceanic and atmospheric processes in this region necessitates additional studies to better understand and explain the observed warming tendency in the subtropical troposphere over the SIO.

Figure 1 .
Figure 1.(a) The time series of the annual-mean longitude of TC genesis over the SIO.(b)-(d) The time series of the TC genesis number over the (b) 30 • -70 • E longitude, the (c) 70 • -115 • E longitude, and (d) 115 • -135 • E longitude of the SIO.The red line indicates the linear regression trend.The text on the top represents the significance level (p value) and linear trend.
Figure 2 displays that the increasing trend of the TC genesis in the western sector of the SIO (30 • -70 • E) is primarily contributed by the southwestern SIO, while the decreasing trend of TC genesis in the eastern sector of the SIO (70 • -115 • E) is predominantly influenced by the northeastern region as shown in figures 2(b) and (c).
(a)).It is noteworthy that significant positive DGPI anomalies are located in the tropical region of 0 • -15 • S and 50 • -70 • E, which is dominantly contributed by the enhanced convection (figures 3(a) and (e)).
) and (b) display the vertical profile of the zonal wind trend averaged across the latitudinal bands of 5 • S-25 • S and 0 • -15 • S, respectively, along with climatological mean zonal wind.Similarly, figures 5(c) and (d) display the vertical profile of the temperature trend averaged across the longitudinal bands of 30 • -70 • E and 70 • -115 • E, respectively, along with the climatological mean temperature.The western SIO is characterized by a mean vertical westerly wind shear, with westerly

Figure 3 .
Figure3.Composite decadal annual-mean differences of (a) DGPI anomalies, (b) the sum of all terms and the terms due to (c) absolute vorticity, (d) meridional shear of mid-level zonal wind, (e) vertical p-velocity at middle level, and (f) vertical wind shear between2001-2022 and 1979-2000.The red and blue boxes denote the western and eastern regions of SIO, respectively.The areas marked with black dot denote statistically significant anomalies at the 95% confidence level.

Figure 4 .
Figure 4.The linear trend of (a) vertical zonal wind shear and (d) tropospheric temperature averaged from 850 hPa to 200 hPa during the period from 1979 to 2022.(b)-(c) The time series of vertical zonal wind shear averaged in (b) red box and (c) blue box of figure 4(a).(e)-(f) The time series of tropospheric (200-850 hPa) temperature difference (e) between the tropical red-box region and the subtropical red-box region of figures 4(b) and (f) between the tropical blue-box region and the subtropical blue-box region of figure (b).The red line in (b)-(c) and (e)-(f) denotes the linear regression trend.The text on the top in (b)-(c) and (e)-(f) represents the significance level (p value) and linear trend.The areas marked with black dot in (a) and (d) denote statistically significant trend at the 95% confidence level.

Figure 6 .
Figure 6.The climatological mean 200 hPa zonal wind (contour, unit: m s −1 ) and the corresponding linear trend (shaded, unit: m s −1 yr −1 ) of 200 hPa zonal wind from (a) ERA5 data and calculated from (b) the thermal wind relationship.The areas marked with black dots denote statistically significant anomalies at the 95% confidence level.

(
Zhang et al 2018, Vidya et al 2021).It has been consistently found that the tropospheric warming is amplified, compared to the surface, under climate change (Fu et al 2004, Christensen et al 2013, Steiner et al 2020).Fu et al (2006) specifically highlighted an intensified tropospheric warming within the 15 • -45 • latitude bands in both hemispheres by using satellite measurements.The pronounced warming of SST in

Figure 7 .
Figure 7.The linear regression trend of (a) seed propensity index (SPI), (b) transition probability (TP), (c) ventilation index, (d) moist saturation deficit (χ ), (e) the absolute value of vertical zonal wind shear (|UWS|), and (f) potential intensity (PI).The black line denotes statistical significance at the 95% confidence level for the trend.The red and blue boxes indicate the key regions of TC genesis change.

Figure 8 .
Figure 8.The linear regression trend of SST from the ERSST data.The black line denotes statistical significance at the 95% confidence level for the trend.
V, Bony S, Meehl G A, Senior C A, Stevens B, Stouffer R J and Taylor K E 2016 Overview of the coupled model intercomparison project phase 6 (CMIP6) experimental design and organization Geosci.Model Dev. 9 1937-58 Fan K, Wang X, Foltz G R and Balaguru K 2019 Meridional oscillation in genesis location of tropical cyclones in the postmonsoon Bay of Bengal Clim.Dyn.53 2103-18 Fu Q, Johanson C M, Wallace J M and Reichler T 2006 Enhanced mid-latitude tropospheric warming in satellite measurements Science 312 1179 Fu Q, Johanson C M, Warren S G and Seidel D J 2004 Contribution of stratospheric cooling to satelliteinferred tropospheric temperature trends Nature 429 55-58 Held I M and Soden B J 2006 Robust responses of the hydrological cycle to global warming J. Clim.19 5686-99 Hersbach H et al 2020 The ERA5 global reanalysis Q.J. R. Meteorol.Soc.146 1999-2049 Hsieh T-L, Vecchi G A, Yang W, Held I M and Garner S T 2020 Large-scale control on the frequency of tropical cyclones and seeds: a consistent relationship across a hierarchy of global atmospheric models Clim.Dyn.55 3177-96 Hsieh T-L, Yang W, Vecchi G A and Zhao M 2022 Model spread in the tropical cyclone frequency and seed propensity index across global warming and ENSO-like perturbations Geophys.Res.Lett.49 e2021GL097157 Huang B, Thorne P W, Banzon V F, Boyer T, Chepurin G, Lawrimore J H, Menne M J, Smith T M, Vose R S and Zhang H-M 2017 Extended reconstructed sea surface temperature, version 5 (ERSSTv5): upgrades, validations, and intercomparisons J. Clim.30 8179-205 Klotzbach P J, Wood K M, Schreck C J, Bowen S G, Patricola C M and Bell M M 2022 Trends in global tropical cyclone activity: 1990-2021 Geophys.Res.Lett.49 e2021GL095774 Knapp K R, Kruk M C, Levinson D H, Diamond H J and Neumann C J 2010 The international best track archive for climate stewardship (IBTrACS): unifying tropical cyclone data Bull.Am.Meteorol.Soc.91 363-76 Knapp K and Kossin J 2007 New global tropical cyclone data set from ISCCP B1 geostationary satellite observations J. Appl.Remote Sens. 1 013505 et al 2013 Climate phenomena and their relevance for future regional climate change Climate Change 2013 the Physical Science Basis: Working Group I Contribution to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (Cambridge University Press) pp 1217-308 Daloz A S and Camargo S J 2018 Is the poleward migration of tropical cyclone maximum intensity associated with a poleward migration of tropical cyclone genesis?Clim.Dyn.50 705-15 Emanuel K A 2013 Downscaling CMIP5 climate models shows increased tropical cyclone activity over the 21st century Proc.Natl Acad.Sci.USA 110 12219-24 Emanuel K 2022 Tropical cyclone seeds, transition probabilities, and genesis J. Clim.35 3557-66 Emanuel K, Sundararajan R and Williams J 2008 Hurricanes and global warming: results from downscaling IPCC AR4 simulations Bull.Am.Meteorol.Soc.89 347-68 Eyring