Editing IPCC:AR6/WGI/Chapter-11 (section)

== 11.7 Extreme Storms ==

<div id="h1-8-siblings" class="h1-siblings"></div>

Extreme storms, such as tropical cyclones (TCs), extratropical cyclones (ETCs), and severe convective storms often have substantial societal impacts. Quantifying the effect of climate change on extreme storms is challenging, partly because extreme storms are rare, short-lived, and local, and individual events are largely influenced by stochastic variability. The high degree of random variability makes detection and attribution of extreme storm trends more uncertain than detection and attribution of trends in other aspects of the environment in which the storms evolve (e.g., larger-scale temperature trends). Projecting changes in extreme storms is also challenging because of constraints in the models’ ability to accurately represent the small-scale physical processes that can drive these changes. Despite the challenges, progress has been made since AR5.

The SREX (Chapter 3) concluded that there is ''low confidence'' in observed long-term (40 years or more) trends in TC intensity, frequency, and duration, and any observed trends in phenomena such as tornadoes and hail; it is ''likely'' that extratropical storm tracks have shifted poleward in both the Northern and Southern Hemispheres, and that heavy rainfalls and mean maximum wind speeds associated with TCs will increase with continued greenhouse gas warming; it is ''likely'' that the global frequency of TCs will either decrease or remain essentially unchanged, while it is ''more likely than not'' that the frequency of the most intense storms will increase substantially in some ocean basins; there is ''low confidence'' in projections of small-scale phenomena such as tornadoes and hail storms; and there is ''medium confidence'' that there will be a reduced frequency and a poleward shift of mid-latitude cyclones due to future anthropogenic climate change.

<div id="_idContainer067" class="Basic-Text-Frame _idGenObjectStyleOverride-1"></div>

[[File:a6e6b4f9c35dc9ff1daaa3873f346b6c IPCC_AR6_WGI_Figure_11_19.png]]

'''Figure 11.19 |''' '''Projected changes in (a–c) the number of consecutive dry days (CDD), (d–f) annual mean soil moisture over the total column, and (g–l) the frequency and intensity of 1-i''' n '''-10-year soil moisture drought for the June-to-August and December-to-February seasons at 1.5°C, 2°C, and 4°C of global warming compared to the 18''' ''50–1900 baseline.'' The unit for soil moisture change is the standard deviation of interannual variability in soil moisture during 1850–1900. Standard deviation is a widely used metric in characterizing drought severity. A projected reduction in mean soil moisture by one standard deviation corresponds to soil moisture conditions typical of about 1-in-6-year droughts during 1850–1900 becoming the norm in the future. Results are based on simulations from the Coupled Model Intercomparison Project Phase 6 (CMIP6) multi-model ensemble under the Shared Socio-economic Pathway (SSP), SSP1-1.9, SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5 scenarios. The numbers in the top right indicate the number of simulations included. Uncertainty is represented using the simple approach: no overlay indicates regions with high model agreement, where ≥80% of models agree on the sign of change; diagonal lines indicate regions with low model agreement, where &lt;80% of models agree on the sign of change. For more information on the simple approach, please refer to the Cross-Chapter Box ( [[IPCC:Wg1:Chapter:Atlas|Atlas]] 1. For details on the methods see Supplementary Material 11.SM.2. Changes in CDDs are also displayed in the Interactive Atlas. Further details on data sources and processing are available in the chapter data table (Table 11.SM.9).

Since SREX, several IPCC Reports also assessed storms. The AR5 (Chapter 2, [[#Hartmann--2013|Hartmann et al., 2013]] ) assessment observed with ''low confidence'' long-term trends in TC metrics, but revised the statement from SREX to state that it is ''virtually certain'' that there are increasing trends in North Atlantic TC activity since the 1970s, with ''medium confidence'' that anthropogenic aerosol forcing has contributed to these trends. The AR5 concluded that it is ''likely'' that TC precipitation and mean intensity will increase and ''more likely than not'' that the frequency of the strongest storms will increase with continued greenhouse gas warming. ''confidence'' in projected trends in overall TC frequency remained ''low'' . ''confidence'' in observed and projected trends in hail storm and tornado events also remained ''low'' . The SROCC (Chapter 6, [[#Collins--2019|Collins et al., 2019]] ) assessed past and projected TCs and ETCs, supporting the AR5 conclusions with some additional detail. Literature subsequent to AR5 adds support to the likelihood of increasing trends in TC intensity, precipitation, and frequency of the most intense storms, while some newer studies have added uncertainty to projected trends in overall frequency. A growing body of literature since AR5 on the poleward migration of TCs led to a new assessment in SROCC of ''low confidence'' that the migration in the western North Pacific represents a detectable climate change contribution from anthropogenic forcing. The SR1.5 (Chapter 3, [[#Hoegh-Guldberg--2018|Hoegh-Guldberg et al., 2018]] ) essentially confirmed the AR5 assessment of TCs and ETCs, adding that heavy precipitation associated with TCs is projected to be higher at 2°C compared to 1.5°C global warming ( ''medi'' ''um confidence'' ).

The SREX, AR5, SROCC, and SR1.5, do not provide assessments of the atmospheric rivers, and SROCC and SR1.5 do not assess severe convective storms and extreme winds. This section assesses the state of knowledge on the four phenomena of TCs, ETCs, severe convective storms, and extreme winds. Atmospheric rivers are addressed in Chapter 8. In this respect, this assessment closely mirrors the SROCC assessment of TCs and ETCs, while updating SREX and AR5 assessments of severe convective storms and extreme winds.

<div id="11.7.1" class="h2-container"></div>

<span id="tropical-cyclones"></span>
=== 11.7.1 Tropical Cyclones ===

<div id="h2-44-siblings" class="h2-siblings"></div>

<div id="11.7.1.1" class="h3-container"></div>

<span id="mechanisms-and-drivers-4"></span>
==== 11.7.1.1 Mechanisms and Drivers ====

<div id="h3-30-siblings" class="h3-siblings"></div>

The genesis, development, and tracks of TCs depend on conditions of the larger-scale circulations of the atmosphere and ocean ( [[#Christensen--2013|Christensen et al., 2013]] ). Large-scale atmospheric circulations, such as the Hadley and Walker circulations and the monsoon circulations can significantly affect TCs, as can internal variability acting on various time scales (Annex IV), from intra-seasonal (e.g., the Madden–Julian and Boreal Summer Intraseasonal oscillations and equatorial waves) and interannual (e.g., the El Niño–Southern Oscillation and Pacific and Atlantic Meridional Modes), to inter-decadal (e.g., Atlantic Multidecadal Variability and Pacific Decadal Variability). This broad range of natural variability makes detection of anthropogenic effects difficult, and uncertainties in the projected changes of these modes of variability increase uncertainty in the projected changes in TC activity. Aerosol forcing also affects sea surface temperature (SST) patterns and cloud microphysics, and it is ''likely'' that observed changes in TC activity are partly caused by changes in aerosol forcing ( [[#Evan--2011|Evan et al., 2011]] ; [[#Ting--2015|Ting et al., 2015]] ; [[#Sobel--2016|Sobel et al., 2016]] , 2019; [[#Takahashi--2017|Takahashi et al., 2017]] ; [[#Zhao--2018|Zhao et al., 2018]] ; [[#Reed--2019|Reed et al., 2019]] ). Among possible changes from these drivers, there is ''medium confidence'' that the Hadley cell has widened and will continue to widen in the future (Sections 2.3, 3.3 and 4.5). This ''likely'' causes latitudinal shifts of TC tracks ( [[#Sharmila--2018|Sharmila and Walsh, 2018]] ). Regional TC activity changes are also strongly affected by projected changes in SST warming patterns ( [[#Yoshida--2017|Yoshida et al., 2017]] ), which are highly uncertain (Chapters 4 and 9).

<div id="11.7.1.2" class="h3-container"></div>

<span id="observed-trends-4"></span>
==== 11.7.1.2 Observed Trends ====

<div id="h3-31-siblings" class="h3-siblings"></div>

Identifying past trends in TC metrics remains a challenge due to the heterogeneous character of the historical instrumental data, which are known as ‘best-track’ data ( [[#Schreck--2014|Schreck et al., 2014]] ). There is ''low confidence'' in most reported long-term (multi-decadal to centennial) trends in TC frequency- or intensity-based metrics due to changes in the technology used to collect the best-track data. This should not be interpreted as implying that no physical (real) trends exist, but rather as indicating that either the quality or the temporal length of the data is not adequate to provide robust trend detection statements, particularly in the presence of multi-decadal variability.

There are previous and ongoing efforts to homogenize the best-track data ( [[#Elsner--2008|Elsner et al., 2008]] ; [[#Kossin--2013|Kossin et al., 2013]] , 2020; [[#Choy--2015|Choy et al., 2015]] ; [[#Landsea--2015|Landsea, 2015]] ; [[#Emanuel--2018|Emanuel et al., 2018]] ) and there is substantial literature that finds positive trends in intensity-related metrics in the best-track during the ‘satellite period’, which is generally limited to around the past 40 years ( [[#Kang--2012|Kang and Elsner, 2012]] ; [[#Kishtawal--2012|Kishtawal et al., 2012]] ; [[#Kossin--2013|Kossin et al., 2013]] , 2020; [[#Mei--2016|Mei and Xie, 2016]] ; [[#Zhao--2018|Zhao et al., 2018]] ; [[#Tauvale--2019|Tauvale and Tsuboki, 2019]] ). When best-track trends are tested using homogenized data, the intensity trends generally remain positive, but are smaller in amplitude ( [[#Kossin--2013|Kossin et al., 2013]] ; [[#Holland--2014|Holland and Bruyère, 2014]] ). [[#Kossin--2020|Kossin et al. (2020)]] extended the homogenized TC intensity record to the period 1979–2017 and identified significant global increases in major TC exceedance probability of about 6% per decade. In addition to trends in TC intensity, there is evidence that TC intensification rates and the frequency of rapid intensification events have increased within the satellite era ( [[#Kishtawal--2012|Kishtawal et al., 2012]] ; [[#Balaguru--2018|Balaguru et al., 2018]] ; [[#Bhatia--2018|Bhatia et al., 2018]] ). The increase in intensification rates is found in the best-track and the homogenized intensity data.

A subset of the best-track data corresponding to hurricanes that have directly impacted the USA since 1900 is considered to be reliable, and shows no trend in the frequency of USA landfall events ( [[#Knutson--2019|Knutson et al., 2019]] ). However, an increasing trend in normalized USA hurricane damage, which accounts for temporal changes in exposed wealth ( [[#Grinsted--2019|Grinsted et al., 2019]] ), and a decreasing trend in TC translation speed over the USA (Kossin, 2019) have also been identified in this period. A similarly reliable subset of the data representing TC landfall frequency over Australia shows a decreasing trend in Eastern Australia since the 1800s ( [[#Callaghan--2011|Callaghan and Power, 2011]] ), as well as in other parts of Australia since 1982 ( [[#Chand--2019|Chand et al., 2019]] ; [[#Knutson--2019|Knutson et al., 2019]] ). A paleoclimate proxy reconstruction shows that recent levels of TC interactions along parts of the Australian coastline are the lowest in the past 550–1500 years ( [[#Haig--2014|Haig et al., 2014]] ). Existing TC datasets show substantial inter-decadal variations in basin-wide TC frequency and intensity in the western North Pacific, but a statistically significant north-westward shift in the western North Pacific TC tracks since the 1980s ( [[#Lee--2020|]] [[#Lee--2020|T.-C. Lee et al., 2020]] ). Inthe case of the North Indian Ocean, analyses of trends are highly dependent on the details of each analysis (e.g., pre- and/or post-monsoon season period, or Bay of Bengal and/or Arabian Sea region). The most consistent trends are an increase in the occurrence of the most intense TCs, and a decrease in the overall TC frequency, in particular in the Bay of Bengal ( [[#Sahoo--2016|Sahoo and Bhaskaran, 2016]] ; [[#Balaji--2018|Balaji et al., 2018]] ; [[#Singh--2019|Singh et al., 2019]] ; [[#Baburaj--2020|Baburaj et al., 2020]] ). In the South Indian Ocean (SIO), an increase in the occurrence of the most intense TCs has been noted; however, there are well-known data quality issues there ( [[#Kuleshov--2010|Kuleshov et al., 2010]] ; [[#Fitchett--2018|Fitchett, 2018]] ). When the SIO data are homogenized, a significant increase is found in the fractional proportion of global Category 3–5 TC instances (6-hourly intensity estimates during the lifetime of each TC) to all Category 1–5 instances ( [[#Kossin--2020|Kossin et al., 2020]] ).

<div id="_idContainer069" class="Basic-Text-Frame"></div>

[[File:0adffd6e83a0f7f63ea9c6dad9ad8006 IPCC_AR6_WGI_Figure_11_20.png]]

'''Figure 11.20 |''' '''Summary schematic of past and projected changes in tropical cyclone (TC), extratropical cyclone (ETC), atmospheric river (AR), and severe convective storm (SCS) behaviour.''' Global changes (blue shading) from top to bottom: ''(i)'' Increased mean and maximum rain rates in TCs, ETCs, and ARs [past ( ''low confidence'' due to lack of reliable data) and projected ( ''high confidence'' )]; ''(ii)'' Increased proportion of stronger TCs [past ( ''medium confidence'' ) and projected ( ''high confidence'' )]; ''(iii)'' Decrease or no change in global frequency of TC genesis [past ( ''low confidence'' due to lack of reliable data) and projected ( ''medium confidence'' )]; and (iv) Increased and decreased ETC wind speed, depending on the region, as storm tracks change [past ( ''low confidence'' due to lack of reliable data) and projected ( ''medium confidence'' )]. Regional changes, from left to right: ''(i)'' Poleward TC migration in the western North Pacific and subsequent changes in TC exposure [past ( ''medium confidence'' ) and projected ( ''medium'' ''confidence'' )]; ''(ii)'' Slowdown of TC forward translation speed over the contiguous USA and subsequent increase in TC rainfall [past ( ''medium confidence'' ) and projected ( ''low'' ''confidence'' due to lack of directed studies)]; and ''(iii)'' Increase in mean and maximum SCS rain rate and increase in spring SCS frequency and season length over the contiguous USA [past ( ''low confidence'' due to lack of reliable data) and projected ( ''medium confidence'' )].

As with all confined regional analyses of TC frequency, it is generally unclear whether any identified changes are due to a basin-wide change in TC frequency, or to systematic track shifts (or both). From an impacts perspective, however, these changes over land are highly relevant and emphasize that large-scale modifications in TC behaviour can have a broad spectrum of impacts on a regional scale.

Subsequent to AR5, two metrics have been analysed that are argued to be comparatively less sensitive to data issues than frequency- and intensity-based metrics. Trends in these metrics have been identified over the past 70 years or more ( [[#Knutson--2019|Knutson et al., 2019]] ). The first metric – the mean latitude where TCs reach their peak intensity – exhibits a global and regional poleward migration during the satellite period ( [[#Kossin--2014|Kossin et al., 2014]] ). The poleward migration can influence TC hazard exposure and risk ( [[#Kossin--2016a|Kossin et al., 2016a]] ) and is consistent with the independently observed expansion of the tropics ( [[#Lucas--2014|Lucas et al., 2014]] ). The migration has been linked to changes in the Hadley circulation ( [[#Altman--2018|Altman et al., 2018]] ; [[#Sharmila--2018|Sharmila and Walsh, 2018]] ; [[#Studholme--2018|Studholme and Gulev, 2018]] ). The migration is also apparent in the mean locations where TCs exhibit eyes ( [[#Knapp--2018|Knapp et al., 2018]] ), which is when TCs are most intense. Part of the Northern Hemisphere poleward migration is due to basin-wide changes in TC frequency ( [[#Kossin--2014|Kossin et al., 2014]] , 2016b; [[#Moon--2015|Moon et al., 2015]] , 2016) and the trends, as expected, can be sensitive to the time period chosen ( [[#Tennille--2017|Tennille and Ellis, 2017]] ; [[#Kossin--2018|Kossin, 2018]] ; [[#Song--2018|Song and Klotzbach, 2018]] ) and to subsetting of the data by intensity ( [[#Zhan--2017|Zhan and Wang, 2017]] ). The poleward migration is particularly pronounced and well-documented in the western North Pacific basin ( [[#Kossin--2016a|Kossin et al., 2016a]] ; [[#Oey--2016|Oey and Chou, 2016]] ; [[#Liang--2017|Liang et al., 2017]] ; [[#Nakamura--2017|Nakamura et al., 2017]] ; [[#Altman--2018|Altman et al., 2018]] ; [[#Daloz--2018|Daloz and Camargo, 2018]] ; J. [[#Sun--2019|]] [[#Sun--2019|]] [[#Sun--2019|]] [[#Sun--2019|Sun et al., 2019]] ; [[#Lee--2020|]] [[#Lee--2020|T.-C. Lee et al., 2020]] ; [[#Yamaguchi--2020a|Yamaguchi and Maeda, 2020a]] ; [[#Kubota--2021|Kubota et al., 2021]] ).

A second metric that is argued to be comparatively less sensitive to data issues than frequency- and intensity-based metrics is TC translation speed ( [[#Kossin--2018|Kossin, 2018]] ), which exhibits a global slowdown in the best-track data over the period 1949–2016. TC translation speed is a measure of the speed at which TCs move across the Earth’s surface, and is very closely related to local rainfall amounts (i.e., a slower translation speed causes greater local rainfall). TC translation speed also affects structural wind damage and coastal storm surge by changing the hazard event duration. The slowdown is observed in the best-track data from all basins except the Northern Indian Ocean, and is also found in a number of regions where TCs interact directly with land. The slowing trends identified in the best-track data by [[#Kossin--2018|Kossin (2018)]] have been argued to be largely due to data heterogeneity. [[#Moon--2019|Moon et al. (2019)]] and [[#Lanzante--2019|Lanzante (2019)]] provide evidence that meridional TC track shifts project onto the slowing trends, and argue that these shifts are due to the introduction of satellite data. Kossin (2019) provides evidence that the slowing trend is real by focusing on Atlantic TC track data over the contiguous USA in the 118-year period 1900–2017, which are generally considered reliable. In this period, mean TC translation speed has decreased by 17%. The slowing TC translation speed is expected to increase local rainfall amounts, which would increase coastal and inland flooding. In combination with slowing translation speed, abrupt TC track direction changes – that can be associated with track ‘meanders’ or ‘stalls’ – have become increasingly common along the North American coast since the mid-20th century, leading to more rainfall in the region ( [[#Hall--2019|Hall and Kossin, 2019]] ).

In summary, there is mounting evidence that a variety of TC characteristics have changed over various time periods. It is ''likely'' that the global proportion of Category 3–5 tropical cyclone instances and the frequency of rapid intensification events have increased globally over the past 40 years. It is ''very likely'' that the average location where TCs reach their peak wind intensity has migrated poleward in the western North Pacific Ocean since the 1940s. It is ''likely'' that TC translation speed has slowed over the USA since 1900.

<div id="11.7.1.3" class="h3-container"></div>

<span id="model-evaluation-4"></span>
==== 11.7.1.3 Model Evaluation ====

<div id="h3-32-siblings" class="h3-siblings"></div>

Accurate projections of future TC activity have two principal requirements: accurate representation of changes in the relevant environmental factors (e.g., SSTs) that can affect TC activity, and accurate representation of actual TC activity in given environmental conditions.In particular, models’ capacity to reproduce historical trends or interannual variabilities of TC activity is relevant to the confidence in future projections. One test of the models is to evaluate their ability to reproduce the dependency of the TC statistics in the different basins in the real world, in addition to their capability of reproducing atmospheric and ocean environmental conditions. For the evaluation of projections of TC-relevant environmental variables, AR5 confidence statements were based on global surface temperature and moisture, but not on the detailed regional structure of SST and atmospheric circulation changes such as steering flows and vertical shear, which affect characteristics of TCs (genesis, intensity, tracks, etc.). Various aspects of TC metrics are used to evaluate how capable models are of simulating present-day TC climatologies and variability (e.g., TC frequency, wind intensity, precipitation, size, tracks, and their seasonal and interannual changes) ( [[#Walsh--2015|Walsh et al., 2015]] ; [[#Camargo--2016|Camargo and Wing, 2016]] ; [[#Knutson--2019|Knutson et al., 2019]] , 2020). Other examples of TC climatology/variability metrics are spatial distributions of TC occurrence and genesis ( [[#Walsh--2015|Walsh et al., 2015]] ), seasonal cycles and interannual variability of basin-wide activity ( [[#Zhao--2009|Zhao et al., 2009]] ; [[#Shaevitz--2014|Shaevitz et al., 2014]] ; [[#Kodama--2015|Kodama et al., 2015]] ; [[#Murakami--2015|Murakami et al., 2015]] ; [[#Yamada--2017|Yamada et al., 2017]] ) or landfalling activity ( [[#Lok--2018|Lok and Chan, 2018]] ), as well as newly developed process-diagnostics designed specifically for TCs in climate models (D. [[#Kim--2018|]] [[#Kim--2018|]] [[#Kim--2018|Kim et al., 2018]] ; [[#Wing--2019|Wing et al., 2019]] ; [[#Moon--2020|Moon et al., 2020]] ).

Confidence in the projection of intense TCs, such as those of Category 4–5, generally becomes higher as the resolution of the models becomes higher. The Coupled Model Intercomparison Project Phases 5 and 6 (CMIP5/6) class climate models (around 100–200 km grid spacing) cannot simulate TCs of Category 4–5 intensity. They do simulate storms of relatively high vorticity that are at best described as ‘TC-like’, but metrics such as storm counts are highly dependent on tracking algorithms ( [[#Camargo--2013|Camargo, 2013]] ; [[#Wehner--2015|Wehner et al., 2015]] ; [[#Zarzycki--2017|Zarzycki and Ullrich, 2017]] ; [[#Roberts--2020a|Roberts et al., 2020a]] ). High-resolution GCMs (around 10–60 km grid spacing), as used in HighResMIP ( [[#Haarsma--2016|Haarsma et al., 2016]] ; [[#Roberts--2020a|Roberts et al., 2020a]] ), begin to capture some structures of TCs more realistically, as well as produce intense TCs of Category 4–5 despite the effects of parametrized deep cumulus convection processes ( [[#Murakami--2015|Murakami et al., 2015]] ; [[#Wehner--2015|Wehner et al., 2015]] ; [[#Yamada--2017|Yamada et al., 2017]] ; [[#Roberts--2018|Roberts et al., 2018]] ; [[#Moon--2020|Moon et al., 2020]] ). Convection-permitting models (around 1–10 km grid-spacing), such as used in some dynamical downscaling studies, provide further realism with capturing TC eye-wall structures ( [[#Tsuboki--2015|Tsuboki et al., 2015]] ). Model characteristics besides resolution, especially details of convective parametrization, can influence a model’s ability to simulate intense TCs ( [[#Reed--2011|Reed and Jablonowski, 2011]] ; [[#Zhao--2012|Zhao et al., 2012]] ; [[#He--2015|He and Posselt, 2015]] ; D. [[#Kim--2018|]] [[#Kim--2018|]] [[#Kim--2018|Kim et al., 2018]] ; [[#Zhang--2018|Zhang and Wang, 2018]] ; [[#Camargo--2020|Camargo et al., 2020]] ). However, models’ dynamical cores and other physics also affect simulated TC properties ( [[#Reed--2015|Reed et al., 2015]] ; [[#Vidale--2021|Vidale et al., 2021]] ). Both wide-area regional and global convection-permitting models without the need for parameterized convection are becoming more useful for TC regional model projection studies ( [[#Tsuboki--2015|Tsuboki et al., 2015]] ; [[#Kanada--2017a|Kanada et al., 2017a]] ; [[#Gutmann--2018|Gutmann et al., 2018]] ) and global model projection studies ( [[#Satoh--2015|Satoh et al., 2015]] , 2017; [[#Yamada--2017|Yamada et al., 2017]] ), as they capture more realistic TC eye wall structures ( [[#Kinter%20III--2013|Kinter III et al., 2013]] ) and are becoming more useful for investigating changes in TC structures ( [[#Kanada--2013|Kanada et al., 2013]] ; [[#Yamada--2017|Yamada et al., 2017]] ). Large ensemble simulations of GCMs with 60 km grid spacing provide TC statistics that allow more reliable detection of changes in the projections, which are not well captured in any single experiment ( [[#Yoshida--2017|Yoshida et al., 2017]] ; [[#Yamaguchi--2020|Yamaguchi et al., 2020]] ). Variable resolution global models offer an alternative to regional models for individual TC or basin-wide simulations ( [[#Yanase--2012|Yanase et al., 2012]] ; [[#Zarzycki--2014|Zarzycki et al., 2014]] ; [[#Harris--2016|Harris et al., 2016]] ; [[#Reed--2020|Reed et al., 2020]] ; [[#Stansfield--2020|Stansfield et al., 2020]] ). Computationally less intense than equivalent uniform resolution global models, they also do not require lateral boundary conditions, thus reducing this source of error ( [[#Hashimoto--2016|Hashimoto et al., 2016]] ). Confidence in the projection of TC statistics and properties is increased by the use of higher-resolution models with more realistic simulations.

Operational forecasting models also reproduce TCs, and their use for climate projection studies shows promise. However, there is limited application for future projections as they are specifically developed for operational purposes, and TC climatology is not necessarily well evaluated. Intercomparison of operational models indicates that enhancement of horizontal resolution can provide more credible projections of TCs ( [[#Nakano--2017|Nakano et al., 2017]] ). Likewise, high-resolution climate models show promise as TC forecast tools ( [[#Zarzycki--2015|Zarzycki and Jablonowski, 2015]] ; [[#Reed--2020|Reed et al., 2020]] ), further narrowing the continuum of weather and climate models, and increasing confidence in projections of future TC behaviour. However, higher horizontal resolution does not necessarily lead to an improved TC climatology ( [[#Camargo--2020|Camargo et al., 2020]] ).

Atmosphere–ocean interaction is an important process in TC evolution. Atmosphere–ocean coupled models are generally better than atmosphere-only models at capturing realistic processes related to TCs ( [[#Murakami--2015|Murakami et al., 2015]] ; [[#Ogata--2015|Ogata et al., 2015]] , 2016; [[#Zarzycki--2016|Zarzycki, 2016]] ; [[#Kanada--2017b|Kanada et al., 2017b]] ; [[#Scoccimarro--2017|Scoccimarro et al., 2017]] ). However, the basin-scale SST biases commonly found in atmosphere–ocean models can introduce substantial errors in the simulated TC number ( [[#Hsu--2019|Hsu et al., 2019]] ). Higher-resolution ocean models improve the simulation of TCs by reducing the SST climatology bias ( [[#Li--2018|Li and Sriver, 2018]] ; [[#Roberts--2020a|Roberts et al., 2020a]] ). Coarse resolution atmospheric models may degrade coupled model performance as well. For example, in a case study of Hurricane Harvey, [[#Trenberth--2018|Trenberth et al. (2018)]] suggested that the lack of realistic hurricane frequency and intensity within coupled climate models hampers the models’ ability to simulate SST and ocean heat content and their changes.

Even with higher-resolution atmosphere–ocean coupled models, TC projection studies still rely on assumptions in experimental design that introduce uncertainties. Computational constraints often limit the number of simulations, resulting in relatively small ensemble sizes and incomplete analyses of possible future SST magnitude and pattern changes ( [[#Zhao--2011|Zhao and Held, 2011]] ; [[#Knutson--2013|Knutson et al., 2013]] ). Uncertainties in aerosol forcing also are reflected in TC projection uncertainty ( [[#Wang--2014|Wang et al., 2014]] ).

Regional climate models (RCM) with grid spacing around 15–50 km can be used to study the projection of TCs. RCMs are run with lateral and surface boundary conditions, which are specified by the atmospheric state and SSTs simulated by GCMs. Various combinations of the lateral and surface boundary conditions can be chosen for RCM studies, and uncertainties in the projection can be further examined in general. They are used for studying changes in TC characteristics in a specific area, such as Vietnam ( [[#Redmond--2015|Redmond et al., 2015]] ) and the Philippines ( [[#Gallo--2019|Gallo et al., 2019]] ).

Less computationally expensive downscaling approaches that allow larger ensembles and long-term studies are also used in the projection of TCs ( [[#Emanuel--2006|Emanuel et al., 2006]] ; C.Y. [[#Lee--2018|]] [[#Lee--2018|]] [[#Lee--2018|Lee et al., 2018]] ). A statistical–dynamical TC downscaling method requires assumptions of the rate of seeding of random initial disturbances, which are generally assumed to not change with climate change ( [[#Emanuel--2008|Emanuel et al., 2008]] ; [[#Emanuel--2013|Emanuel, 2013]] ). The results with the downscaling approach might depend on the assumptions, which are required for the simplification of the methods.

In summary, various types of models are useful to study how TCs change in response to climate changes, and there is no unique solution for choosing a model type. However, higher-resolution models generally capture TC properties more realistically ( ''high confidence'' ). In particular, models with horizontal resolutions of 10–60 km are capable of reproducing strong TCs with Category 4–5 and those of 1–10 km are capable of the eye wall structure of TCs. Uncertainties in TC simulations come from details of the model configuration of both dynamical and physical processes. Models with realistic atmosphere–ocean interactions are generally better than atmosphere-only models at reproducing realistic TC evolutions ( ''hi'' ''gh confidence'' ).

<div id="11.7.1.4" class="h3-container"></div>

<span id="detection-and-attribution-event-attribution-4"></span>
==== 11.7.1.4 Detection and Attribution, Event Attribution ====

<div id="h3-33-siblings" class="h3-siblings"></div>

There is general agreement in the literature that anthropogenic greenhouse gases and aerosols have measurably affected observed oceanic and atmospheric variability in TC-prone regions (see Chapter 3). This underpinned the SROCC assessment of ''medium confidence'' that humans have contributed to the observed increase in Atlantic hurricane activity since the 1970s (Chapter 5, [[#Bindoff--2013|Bindoff et al., 2013]] ). Literature subsequent to AR5 lends further support to this statement ( [[#Knutson--2019|Knutson et al., 2019]] ). However, there is still no consensus on the relative magnitude of human and natural influences on past changes in Atlantic hurricane activity, and particularly on which factor has dominated the observed increase ( [[#Ting--2015|Ting et al., 2015]] ) and it remains uncertain whether past changes in Atlantic TC activity are outside the range of natural variability. A recent result using high-resolution dynamical model experiments suggested that the observed spatial contrast in TC trends cannot be explained only by multi-decadal natural variability, and that external forcing plays an important role ( [[#Murakami--2020|Murakami et al., 2020]] ).Observational evidence for significant global increases in the proportion of major TC intensities ( [[#Kossin--2020|Kossin et al., 2020]] ) is consistent with both theory and numerical modelling simulations, which generally indicate an increase in mean TC peak intensity and the proportion of very intense TCs in a warming world ( [[#Knutson--2015|Knutson et al., 2015]] , 2020; [[#Walsh--2015|Walsh et al., 2015]] , 2016). In addition, high-resolution coupled model simulations provide support that natural variability alone is ''unlikely'' to explain the magnitude of the observed increase in TC intensification rates and upward TC intensity trend in the Atlantic basin since the early 1980s ( [[#Bhatia--2019|Bhatia et al., 2019]] ; [[#Murakami--2020|Murakami et al., 2020]] ).

The cause of the observed slowdown in TC translation speed is not yet clear. [[#Yamaguchi--2020|Yamaguchi et al. (2020)]] used large ensemble simulations to argue that part of the slowdown is due to actual latitudinal shifts of TC tracks, rather than data artefacts, in addition to atmospheric circulation changes. G. [[#Zhang--2020|]] [[#Zhang--2020|]] [[#Zhang--2020|]] [[#Zhang--2020|Zhang et al. (2020)]] used large ensemble simulations to show that anthropogenic forcing can lead to a robust slowdown, particularly outside of the tropics at higher latitudes. [[#Yamaguchi--2020b|Yamaguchi and Maeda (2020b)]] found a significant slowdown in the western North Pacific over the past 40 years and attributed the slowdown to a combination of natural variability and global warming. The slowing trend since 1900 over the USA is robust and significant after removing multi-decadal variability from the time series (Kossin, 2019). Among the hypotheses discussed is the physical linkage between warming and slowing circulation ( [[#Held--2006|Held and Soden, 2006]] ; see also [[IPCC:Wg1:Chapter:Chapter-8#8.2.2.2|Section 8.2.2.2]] ), with expectations of Arctic amplification and weakening circulation patterns through weakening meridional temperature gradients ( [[#Coumou--2018|Coumou et al., 2018]] ; see also Cross-Chapter Box 10.1), or through changes in planetary wave dynamics ( [[#Mann--2017|Mann et al., 2017]] ). The tropics expansion and the poleward shift of the mid-latitude westerlies associated with warming is also suggested as the reason of the slowdown (G. [[#Zhang--2020|]] [[#Zhang--2020|]] [[#Zhang--2020|]] [[#Zhang--2020|Zhang et al., 2020]] ). However, the connection of these mechanisms to the slowdown has not been robustly shown. Furthermore, slowing trends have not been unambiguously observed in circulation patterns that steer TCs, such as the Walker and Hadley circulations ( [[IPCC:Wg1:Chapter:Chapter-2#2.3.1.4|Section 2.3.1.4]] ), although these circulations generally slow down in numerical simulations under global warming (Sections 4.5.1.6 and 8.4.2.2).

The observed poleward trend in western North Pacific TCs remains significant after accounting for the known modes of dominant interannual to decadal variability in the region ( [[#Kossin--2016a|Kossin et al., 2016a]] ), and is also found in CMIP5 model-simulated TCs (in the recent historical period 1980–2005), although it is weaker than observed and is not statistically significant ( [[#Kossin--2016a|Kossin et al., 2016a]] ). However, the trend is significant in 21st-century CMIP5 projections under the RCP8.5 scenario, with a similar spatial pattern and magnitude to the past observed changes in that basin over the period 1945–2016, supporting a possible anthropogenic greenhouse gas contribution to the observed trends ( [[#Kossin--2016a|Kossin et al., 2016a]] ; [[#Knutson--2019|Knutson et al., 2019]] ).

The recent active TC seasons in some basins have been studied to determine whether there is anthropogenic influence. For 2015, [[#Murakami--2017b|Murakami et al. (2017b)]] explored the unusually high TC frequency near Hawaii and in the eastern Pacific basin. W. [[#Zhang--2016b|Zhang et al. (2016b)]] considered unusually high Accumulated Cyclone Energy (ACE) in the western North Pacific; and S.-H. [[#Yang--2018|]] [[#Yang--2018|]] [[#Yang--2018|]] [[#Yang--2018|]] [[#Yang--2018|Yang et al. (2018)]] and [[#Yamada--2019|Yamada et al. (2019)]] looked at TC intensification in the western North Pacific. These studies suggest that the anomalous TC activity in 2015 was not solely explained by the effect of an extreme El Niño (see Box 11.4) and that there was also an anthropogenic contribution, mainly through the effects of SSTs in subtropical regions. In the post-monsoon seasons of 2014 and 2015, tropical storms with lifetime maximum winds greater than 46 m s <sup>−1</sup> were first observed over the Arabian Sea, and [[#Murakami--2017a|Murakami et al. (2017a)]] showed that the probability of late-season severe tropical storms is increased by anthropogenic forcing compared to the preindustrial era. [[#Murakami--2018|Murakami et al. (2018)]] concluded that the active 2017 Atlantic hurricane season was mainly caused by pronounced SSTs in the tropical North Atlantic and that these types of seasonal events will intensify with projected anthropogenic forcing. The trans-basin SST change, which might be driven by anthropogenic aerosol forcing, also affects TC activity. [[#Takahashi--2017|Takahashi et al. (2017)]] suggested that a decrease in sulphate aerosol emissions caused about half of the observed decreasing trends in TC genesis frequency in the south-eastern region of the western North Pacific during 1992–2011.

Event attribution is used in TC case studies to test whether the severities of recent intense TCs are explained without anthropogenic effects. In a case study of Hurricane Sandy (2012), [[#Lackmann--2015|Lackmann (2015)]] found no statistically significant impact of anthropogenic climate change on storm intensity, while projections in a warmer world showed significant strengthening. However, [[#Magnusson--2014|Magnusson et al. (2014)]] found that, in European Centre for Medium-Range Weather Forecast (ECMWF) simulations, the simulated cyclone depth and intensity, as well as precipitation, were larger when the model was driven by the warmer actual SSTs than the climatological average SSTs. In Super Typhoon Haiyan, which struck the Philippines on 8 November 2013, [[#Takayabu--2015|Takayabu et al. (2015)]] took an event attribution approach with cloud system-resolving (around 1 km) downscaling ensemble experiments to evaluate the anthropogenic effect on typhoons, and showed that the intensity of the simulated worst-case storm in the actual conditions was stronger than that in a hypothetical condition without historical anthropogenic forcing in the model. However, in a similar approach with two coarser parametrized convection models, Wehner et al. (2019) found conflicting human influences on Haiyan’s intensity. [[#Patricola--2018|Patricola and Wehner (2018)]] found little evidence of an attributable change in intensity of hurricanes Katrina (2005), Irma (2017), and Maria (2017) using a regional climate model configured between 3 km and 4.5 km resolution. They did, however, find attributable increases in heavy precipitation totals. These results imply that higher resolution, such as in a convective permitting 5 km or less mesh model, is required to obtain a robust anthropogenic intensification of a strong TC by simulating realistic rapid intensification ( [[#Kanada--2016|Kanada and Wada, 2016]] ; [[#Kanada--2017a|Kanada et al., 2017a]] ), and that whether the TC intensification can be attributed to the recent warming depends on the case.

The dominant factor in the extreme rainfall amounts during Hurricane Harvey’s passage onto the USA in 2017 was its slow translation speed. But studies published after the event have argued that anthropogenic climate change contributed to an increase in rain rate, which compounded the extreme local rainfall caused by the slow translation. [[#Emanuel--2017|Emanuel (2017)]] used a large set of synthetically-generated storms and concluded that the occurrence of extreme rainfall as observed in Harvey was substantially enhanced by anthropogenic changes to the larger-scale ocean and atmosphere characteristics; [[#Trenberth--2018|Trenberth et al. (2018)]] linked Harvey’s rainfall totals to the anomalously large ocean heat content from the Gulf of Mexico; and [[#van%20Oldenborgh--2017|van Oldenborgh et al. (2017)]] and [[#Risser--2017|Risser and Wehner (2017)]] applied extreme value analysis to extreme rainfall records in the Houston, Texas region, both attributing large increases to climate change. Large precipitation increases during Harvey due to global warming were also found using climate models ( [[#van%20Oldenborgh--2017|van Oldenborgh et al., 2017]] ; S.-Y.S. [[#Wang--2018|]] [[#Wang--2018|Wang et al., 2018]] ). Harvey precipitation totals were estimated in these papers to be three to 10 times more probable due to climate change. A best estimate from a regional climate and flood model is that urbanization increased the risk of the Harvey flooding by a factor of 21 (W. [[#Zhang--2018|]] [[#Zhang--2018|]] [[#Zhang--2018|Zhang et al., 2018]] ), using a regional climate and flood model, found that surface roughness from urbanization increased the risk of the Harvey flooding by a factor of 21. Anthropogenic effects on precipitation increases were also predicted in advance from a forecast model for Hurricane Florence in 2018 ( [[#Reed--2020|Reed et al., 2020]] ).

In summary, it is ''very likely'' that the recent active TC seasons in the North Atlantic, the North Pacific, and Arabian basins cannot be explained without an anthropogenic influence. The anthropogenic influence on these changes is principally associated to aerosol forcing, with stronger contributions to the response in the North Atlantic. It is ''more likely than not'' that the slowdown of TC translation speed over the USA has contributions from anthropogenic forcing. It is ''likely'' that the poleward migration of TCs in the western North Pacific and the global increase in TC intensity rates cannot be explained entirely by natural variability. Event attribution studies of specific strong TCs provide ''limited evidence'' for anthropogenic effects on TC intensifications so far, but ''high confidence'' for increases in TC heavy precipitation. There is ''high confidence'' that anthropogenic climate change contributed to extreme rainfall amounts during Hurricane Harvey (2017) and other intense TCs.

<div id="11.7.1.5" class="h3-container"></div>

<span id="projections-3"></span>
==== 11.7.1.5 Projections ====

<div id="h3-34-siblings" class="h3-siblings"></div>

A summary of studies on TC projections for the late 21st century, particularly studies since AR5, is given by [[#Knutson--2020|Knutson et al. (2020)]] , which is an assessment report mandated by the World Meteorological Organization (WMO). Studies subsequent to [[#Knutson--2020|Knutson et al. (2020)]] are generally consistent, and the confidence assessments here closely follow theirs ( [[#Cha--2020|Cha et al., 2020]] ), although there are some differences due to the varying confidence calibrations between the IPCC and WMO reports.

There is not an established theory for the drivers of future changes in the frequency of TCs. Most, but not all, high-resolution global simulations project significant reductions in the total number of TCs, with the bulk of the reduction at the weaker end of the intensity spectrum as the climate warms ( [[#Knutson--2020|Knutson et al., 2020]] ). Recent exceptions based on high-resolution coupled model results arenoted in [[#Bhatia--2018|Bhatia et al. (2018)]] and [[#Vecchi--2019|Vecchi et al. (2019)]] . [[#Vecchi--2019|Vecchi et al. (2019)]] showed that the representation of synoptic-scale seeds for TC genesis in their high-resolution model causes different projections of global TC frequency, and there is evidence for a decrease in cyclone seeds in some projected TCsimulations ( [[#Sugi--2020|Sugi et al., 2020]] ; Yamada et al., 2011). However, other research indicates that TC seeds are not an independent control on climatological TC frequency, rather the seeds covary with the large-scale controls on TCs ( [[#Patricola--2018|Patricola et al., 2018]] ). While empirical genesis indices derived from observations and reanalysis describe well the observed subseasonal and interannual variability of current TC frequency ( [[#Camargo--2007|Camargo et al., 2007]] , 2009; [[#Tippett--2011|Tippett et al., 2011]] ; [[#Menkes--2012|Menkes et al., 2012]] ), they fail to predict the decreased TC frequency found in most high-resolution model simulations ( [[#Zhang--2010|Zhang et al., 2010]] ; [[#Camargo--2013|Camargo, 2013]] ; [[#Wehner--2015|Wehner et al., 2015]] ), as they generally project an increase as the climate warms. This suggests a limitation of the use of the empirical genesis indices for projections of TC genesis, in particular due to their sensitivity to the humidity variable considered in the genesis index for these projections ( [[#Camargo--2014|Camargo et al., 2014]] ). In a different approach, a statistical–dynamical downscaling framework assuming a constant seeding rate with warming ( [[#Emanuel--2013|Emanuel, 2013]] , 2021) exhibits increases in TC frequency consistent with genesis indices-based projections, while downscaling with a different model leads to two different scenarios depending on the humidity variable considered (C.-Y. [[#Lee--2020|]] [[#Lee--2020|Lee et al., 2020]] ). This disparity in the sign of the projected change in global TC frequency, and the difficulty in explaining the mechanisms behind the different signed responses, further emphasize the lack of process understanding of future changes in tropical cyclogenesis ( [[#Walsh--2015|Walsh et al., 2015]] ; [[#Hoogewind--2020|Hoogewind et al., 2020]] ). Even within a single model, uncertainty in the pattern of future SST changes leads to large uncertainties (including the sign) in the projected change in TC frequency in individual ocean basins, although global TCs would appear to be less sensitive ( [[#Yoshida--2017|Yoshida et al., 2017]] ; [[#Bacmeister--2018|Bacmeister et al., 2018]] ).

Changes in SST and atmospheric temperature and moisture play a role in tropical cyclogenesis ( [[#Walsh--2015|Walsh et al., 2015]] ). Reductions in vertical convective mass flux due to increased tropical stability have been associated with a reduction in cyclogenesis ( [[#Held--2011|Held and Zhao, 2011]] ; [[#Sugi--2012|Sugi et al., 2012]] ). [[#Satoh--2015|Satoh et al. (2015)]] further posit that the robust simulated increase in the number of intense TCs, and hence increased vertical mass flux associated with intense TCs, must lead to a decrease in overall TC frequency because of this association. The Genesis Potential Index can be modified to mimic the TC frequency decreases of a model by altering the treatment of humidity ( [[#Camargo--2014|Camargo et al., 2014]] ). This supports the idea that increased mid-tropospheric saturation deficit ( [[#Emanuel--2008|Emanuel et al., 2008]] ) controls TC frequency, but the approach remains empirical. Other possible controlling factors, such as a decline in the number of seeds (held constant in Emanuel’s downscaling approach, or dependent on the genesis index formulation in the approach proposed by C.-Y. [[#Lee--2020|]] [[#Lee--2020|Lee et al., 2020]] ) caused by increased atmospheric stability have been proposed, but questioned as an important factor ( [[#Patricola--2018|Patricola et al., 2018]] ). The resolution of atmospheric models affects the number of seeds, hence TC genesis frequency ( [[#Vecchi--2019|Vecchi et al., 2019]] ; [[#Sugi--2020|Sugi et al., 2020]] ; [[#Yamada--2021|Yamada et al., 2021]] ). The diverse and sometimes inconsistent projected changes in global TC frequency by high-resolution models indicate that better process understanding and improvement of the models are needed to raise confidence in these changes.

Most TC-permitting model simulations (10–60 km or finer grid spacing) are consistent in their projection of increases in the proportion of intense TCs (Category 4–5), as well as an increase in the intensity of the strongest TCs defined by maximum wind speed or central pressure fall ( [[#Murakami--2012|Murakami et al., 2012]] ; [[#Tsuboki--2015|Tsuboki et al., 2015]] ; [[#Wehner--2018a|Wehner et al., 2018a]] ; [[#Knutson--2020|Knutson et al., 2020]] ). The general reduction in the total number of TCs, which is concentrated in storms weaker than or equal to Category 1, contributes to this increase. The models are somewhat less consistent in projecting an increase in the frequency of Category 4–5TCs (Wehner et al., 2018a; [[#Knutson--2020|Knutson et al., 2020]] ). The projected increase in the intensity of the strongest TCs is consistent with theoretical understanding (e.g., [[#Emanuel--1987|Emanuel, 1987]] ) and observations (e.g., [[#Kossin--2020|Kossin et al., 2020]] ). For a 2°C global warming, the median proportion of Category 4–5 TCs increases by 13%, while the median global TC frequency decreases by 14%, which implies that the median of the global Category 4–5 TC frequency is slightly reduced by 1% or almost unchanged ( [[#Knutson--2020|Knutson et al., 2020]] ). [[#Murakami--2020|Murakami et al. (2020)]] projected a decrease in TC frequency over the coming century in the North Atlantic due to greenhouse warming, as consistent with [[#Dunstone--2013|Dunstone et al. (2013)]] , and a reduction in TC frequency almost everywhere in the tropics in response to +1% CO <sub>2</sub> forcing. Exceptions include the central North Pacific (Hawaii region), east of the Philippines in the North Pacific, and two relatively small regions in the northern Arabian Sea and Bay of Bengal. These projections can vary substantially between ocean basins, possibly due to differences in regional SST warming and warming patterns ( [[#Sugi--2017|Sugi et al., 2017]] ; [[#Yoshida--2017|Yoshida et al., 2017]] ; [[#Bacmeister--2018|Bacmeister et al., 2018]] ). A summary of projections of TC characteristics is schematically shown by Figure 11.20.

The increase in global TC maximum surface wind speeds is about 5% for a 2°C global warming across a number of high-resolution multi-decadal studies ( [[#Knutson--2020|Knutson et al., 2020]] ). This indicates the deepening in global TC minimum surface pressure under the global warming conditions. A regional cloud-permitting model study shows that the strongest TC in the western North Pacific can be as strong as 857 hPa in minimum surface pressure with a wind speed of 88 m s <sup>–1</sup> under warming conditions in 2074–2087 ( [[#Tsuboki--2015|Tsuboki et al., 2015]] ). TCs are also measured by quantities such as ACE and the power dissipation index (PDI), which conflate TC intensity, frequency, and duration ( [[#Murakami--2014|Murakami et al., 2014]] ). Several TC modelling studies ( [[#Yamada--2010|Yamada et al., 2010]] ; H.S. [[#Kim--2014|]] [[#Kim--2014|Kim et al., 2014]] ; [[#Knutson--2015|Knutson et al., 2015]] ) project little change or decreases in the globally accumulated value of PDI or ACE, which is due to the decrease in the total number of TCs.

A projected increase in global average TC rain rates of about 12% for a 2°C global warming is consistent with the Clausius–Clapeyron scaling of saturation-specific humidity ( [[#Knutson--2020|Knutson et al., 2020]] ). Increases substantially greater than Clausius–Clapeyron scaling are projected in some regions, which is caused by increased low-level moisture convergence due to projected TC intensity increases in those regions ( [[#Knutson--2015|Knutson et al., 2015]] ; [[#Phibbs--2016|Phibbs and Toumi, 2016]] ; [[#Patricola--2018|Patricola and Wehner, 2018]] ; M. [[#Liu--2019|Liu et al., 2019]] a). Projections of TC precipitation using large-ensemble experiments ( [[#Kitoh--2019|Kitoh and Endo, 2019]] ) show that the annual maximum one-day precipitation total is projected to increase, except for the western North Pacific where only a small change (or even a reduction) is projected, mainly due to a projected decrease of TC frequency. They also show that the 10-year return value of extreme Rx1day associated with TCs will greatly increase in a region extending from Hawaii to the south of Japan. TC tracks and the location of topography relative to TCs significantly affect precipitation, thus, in general, areas on the eastern and southern faces of mountains have more impacts of TC precipitation changes ( [[#Hatsuzuka--2020|Hatsuzuka et al., 2020]] ). Projection studies using variable-resolution models in the North Atlantic ( [[#Stansfield--2020|Stansfield et al., 2020]] ) indicate that TC-related precipitation rates within North Atlantic TCs and the amount of hourly precipitation due to TC are projected to increase by the end of the century compared to a historical simulation. However, the annual average TC-related Rx5day over the eastern USA is projected to decrease because of a reduction in landfalling TCs. RCM studies with around 25–50 km grid spacing are used to study projected changes in TCs. The projected changes of TCs in South East Asia simulated by RCMs are consistent with those of most GCMs, showing a decrease in TC frequency and an increase in the amount of TC-associated precipitation or an increase in the frequency of intense TCs ( [[#Redmond--2015|Redmond et al., 2015]] ; [[#Gallo--2019|Gallo et al., 2019]] ).

Projected changes in TC tracks or TC areas of occurrence in the late 21st century vary considerably among available studies, although there is better agreement in the western North Pacific. Several studies project either poleward or eastward expansion of TC occurrence over the western North Pacific region, and more TC occurrence in the central North Pacific ( [[#Yamada--2017|Yamada et al., 2017]] ; [[#Yoshida--2017|Yoshida et al., 2017]] ; [[#Wehner--2018a|Wehner et al., 2018a]] ; [[#Roberts--2020b|Roberts et al., 2020b]] ). The observed poleward expansion of the latitude of maximum TC intensity in the western North Pacific is consistently reproduced by the CMIP5 models and downscaled models, and these models show further poleward expansion in the future; the projected mean migration rate of the mean latitude where TCs reach their lifetime-maximum intensity is 0.2±0.1° from CMIP5 model results, while it is 0.13±0.04° from downscaled models in the western North Pacific ( [[#Kossin--2014|Kossin et al., 2014]] , 2016a). In the North Atlantic, while the location of TC maximum intensity does not show clear poleward migration observationally ( [[#Kossin--2014|Kossin et al., 2014]] ), it tends to migrate poleward in projections ( [[#Garner--2017|Garner et al., 2017]] ). The poleward migration is less robust among models and observations in the Indian Ocean, eastern North Pacific, and South Pacific (e.g., [[#Tauvale--2019|Tauvale and Tsuboki, 2019]] ; Ramsay et al. 2018; Cattiaux et al. 2020). There is presently no clear consensus in projected changes in TC translation speed ( [[#Knutson--2020|Knutson et al., 2020]] ), although recent studies suggest a slowdown outside of the tropics (Kossin, 2019; [[#Yamaguchi--2020|Yamaguchi et al., 2020]] ; G. [[#Zhang--2020|]] [[#Zhang--2020|]] [[#Zhang--2020|]] [[#Zhang--2020|Zhang et al., 2020]] ), but regionally there can even be an acceleration of the storms ( [[#Hassanzadeh--2020|Hassanzadeh et al., 2020]] ).

The spatial extent, or ‘size’, of the TC wind field is an important determinant of storm surge and damage. No detectable anthropogenic influences on TC size have been identified to date, because TCs in observations vary in size substantially ( [[#Chan--2015|Chan and Chan, 2015]] ) and there is no definite theory on what controls TC size, although this is an area of active research ( [[#Chavas--2014|Chavas and Emanuel, 2014]] ; [[#Chan--2018|Chan and Chan, 2018]] ). However, projections by high-resolution models indicate future broadening of TC wind fields when compared to TCs of the same categories ( [[#Yamada--2017|Yamada et al., 2017]] ), while [[#Knutson--2015|Knutson et al. (2015)]] simulate a reasonable interbasin distribution of TC size climatology, but project no statistically significant change in global average TC size. A plausible mechanism is that, as the tropopause height becomes higher with global warming, the eye wall areas become wider because the eye walls are inclined outward with height to the tropopause. This effect is only reproduced in high-resolution convection-permitting models capturing eye walls, and such modelling studies are not common. Moreover, the projected TC size changes are generally on the order of 10% or less, and these size changes are still highly variable between basins and studies. Thus, the projected change in both magnitude and sign of TC size is uncertain.

The coastal effects of TCs depend on TC intensity, size, track, and translation speed. Projected increases in sea level, average TC intensity, and TC rainfall rates each generally act to further elevate future storm surge and fresh-water flooding (see [[IPCC:Wg1:Chapter:Chapter-9#9.6.4.2|Section 9.6.4.2]] ). Changes in TC frequency could contribute toward increasing or decreasing future storm surge risk, depending on the net effects of changes in weaker vs stronger storms. Several studies ( [[#McInnes--2014|McInnes et al., 2014]] , 2016; [[#Little--2015|Little et al., 2015]] ; [[#Garner--2017|Garner et al., 2017]] ; [[#Timmermans--2017|Timmermans et al., 2017]] , 2018) have explored future projections of storm surge in the context of anthropogenic climate change with the influence of both sea level rise and future TC changes. [[#Garner--2017|Garner et al. (2017)]] investigated the near-future changes in the New York City coastal flood hazard, and suggested a small change in storm-surge height because effects of TC intensification are compensated by the offshore shifts in TC tracks, but concluded that the overall effect due to the rising sea levels would increase the flood hazard. Future projection studies of storm surge in East Asia, including China, Japan and Korea, also indicate that storm surges due to TCs become more severe ( [[#Yang--2018|]] [[#Yang--2018|]] [[#Yang--2018|]] [[#Yang--2018|]] [[#Yang--2018|J.A. Yang et al., 2018]] ; [[#Mori--2019|Mori et al., 2019]] , 2021; J. [[#Chen--2020|]] [[#Chen--2020|Chen et al., 2020]] b). For the Pacific Islands, [[#McInnes--2014|McInnes et al. (2014)]] found that the future projected increase in storm surge in Fiji is dominated by sea level rise, and projected TC changes make only a minor contribution. Among various storm surge factors, there is ''high confidence'' that sea level rise will lead to a higher possibility of extreme coastal water levels in most regions, with all other factors assumed equal.

In the North Atlantic, vertical wind shear, which inhibits TC genesis and intensification, varies in a quasi-dipole pattern, with one centre of action in the tropics and another along the south-east USA coast ( [[#Vimont--2007|Vimont and Kossin, 2007]] ). This pattern of variability creates a protective barrier of high shear along the USA coast during periods of heightened TC activity in the tropics ( [[#Kossin--2017|Kossin, 2017]] ), and appears to be a natural part of the Atlantic ocean–atmosphere climate system ( [[#Ting--2019|Ting et al., 2019]] ). Greenhouse gas forcing in CMIP5 and the Community Earth System Model Large Ensemble ( [[#Kay--2015|Kay et al., 2015]] ) simulations, however, erodes the pattern and degrades the natural shear barrier along the USA coast. Following the RCP8.5 emissions scenario, the magnitude of the erosion of the barrier equals the amplitude of past natural variability (time of emergence) by the mid-21st century ( [[#Ting--2019|Ting et al., 2019]] ). The projected reduction of shear along the USA East Coast with warming is consistent among studies (e.g., [[#Vecchi--2007|Vecchi and Soden, 2007]] ).

In summary, average peak TC wind speeds and the proportion of Category 4–5 TCs will ''very likely'' increase globally with warming. It is ''likely'' that the frequency of Category 4–5 TCs will increase in limited regions over the western North Pacific. It is ''very likely'' that average TC rain rates will increase with warming, and ''likely'' that the peak rain rates will increase at rate greater than the Clausius–Clapeyron scaling rate of 7% per 1°C of warming in some regions due to increased low-level moisture convergence caused by regional increases in TC wind intensity. It is ''likely'' that the average location where TCs reach their peak wind intensity will migrate poleward in the western North Pacific Ocean as the tropics expand with warming, and that the global frequency of TCs over all categories will decrease or remain unchanged.

<div id="11.7.2" class="h2-container"></div>

<span id="extratropical-storms"></span>
=== 11.7.2 Extratropical Storms ===

<div id="h2-45-siblings" class="h2-siblings"></div>

This section focuses on extratropical cyclones (ETCs) that are either classified as strong or extreme by using some measure of their intensity, or by being associated with the occurrence of extremes in variables such as precipitation or near-surface wind speed ( [[#Seneviratne--2012|Seneviratne et al., 2012]] ). Since AR5, the high relevance of ETCs for extreme precipitation events has been well established ( [[#Pfahl--2012|Pfahl and Wernli, 2012]] ; [[#Catto--2013|Catto and Pfahl, 2013]] ; [[#Utsumi--2017|Utsumi et al., 2017]] ), with 80% or more of hourly and daily precipitation extremes being associated with either ETCs or fronts over oceanic mid-latitude regions, and somewhat smaller, but still very large, proportions of events over mid-latitude land regions ( [[#Utsumi--2017|Utsumi et al., 2017]] ). The emphasis in this section is on individual ETCs that have been identified using some detection and tracking algorithms. Mid-latitude atmospheric rivers are assessed in [[IPCC:Wg1:Chapter:Chapter-8#8.3.2.8|Section 8.3.2.8]] .

<div id="11.7.2.1" class="h3-container"></div>

<span id="observed-trends-5"></span>
==== 11.7.2.1 Observed Trends ====

<div id="h3-35-siblings" class="h3-siblings"></div>

[[IPCC:Wg1:Chapter:Chapter-2#2.3.1.4.3|Section 2.3.1.4.3]] concluded that there is overall ''low confidence'' in recent changes in the total number of ETCs over both hemispheres, and that there is ''medium confidence'' in a poleward shift of the storm tracks over both hemispheres since the 1980s. Overall, there is also ''low confidence'' in past-century trends in the number and intensity of the strongest ETCs due to the large interannual and decadal variability ( [[#Feser--2015|Feser et al., 2015]] ; [[#Reboita--2015|Reboita et al., 2015]] ; [[#Wang--2016|Wang et al., 2016]] ; [[#Varino--2019|Varino et al., 2019]] ) and due to temporal and spatial heterogeneities in the number and type of assimilated data in reanalyses, particularly before the satellite era ( [[#Krueger--2013|Krueger et al., 2013]] ; [[#Tilinina--2013|Tilinina et al., 2013]] ; [[#Befort--2016|Befort et al., 2016]] ; [[#Chang--2016|Chang and Yau, 2016]] ; [[#Wang--2016|Wang et al., 2016]] ). There is ''medium confidence'' that the agreement among reanalyses and detection and tracking algorithms is higher when considering stronger cyclones ( [[#Neu--2013|Neu et al., 2013]] ; [[#Pepler--2015|Pepler et al., 2015]] ; [[#Wang--2016|Wang et al., 2016]] ). Over the Southern Hemisphere, there is ''high confidence'' that the total number of ETCs with low central pressures (&lt;980 hPa) has increased between 1979 and 2009, with all eight reanalyses considered by [[#Wang--2016|Wang et al. (2016)]] showing positive trends, and five of them showing statistically significant trends. Similar results were found by [[#Reboita--2015|Reboita et al. (2015)]] using a different detection and tracking algorithm and a single reanalysis product. Over the Northern Hemisphere, there is ''high agreement'' among reanalyses that the number of cyclones with low central pressures (&lt;970 hPa) has decreased in summer and winter during the period 1979–2010 ( [[#Tilinina--2013|Tilinina et al., 2013]] ; [[#Chang--2016|Chang et al., 2016]] ). However, changes exhibit substantial decadal variability and do not show monotonic trends since the 1980s. For example, over the Arctic and North Atlantic, [[#Tilinina--2013|Tilinina et al. (2013)]] showed that the number of cyclones with very low central pressure (&lt;960 hPa) increased from 1979 to 1990 and then declined until 2010 in all five reanalyses considered. Over the North Pacific, the number of cyclones with very low central pressure reached a peak around 2000 and then decreased until 2010 in the five reanalyses considered ( [[#Tilinina--2013|Tilinina et al., 2013]] ). Overall, however, it should be noted that characterising trends in the dynamical intensity of ETCs (e.g., wind speeds) using the absolute central pressure is problematic because the central pressure depends on the background mean sea level pressure, which varies seasonally and regionally (e.g., [[#Befort--2016|Befort et al., 2016]] ).

<div id="11.7.2.2" class="h3-container"></div>

<span id="model-evaluation-5"></span>
==== 11.7.2.2 Model Evaluation ====

<div id="h3-36-siblings" class="h3-siblings"></div>

There is ''high confidence'' that coarse-resolution climate models (e.g., CMIP5 and CMIP6) underestimate the dynamical intensity of ETCs, including the strongest ETCs, as measured using a variety of metrics, including mean pressure gradient, mean vorticity and near-surface wind speeds, over most regions ( [[#Colle--2013|Colle et al., 2013]] ; [[#Zappa--2013a|Zappa et al., 2013a]] ; [[#Govekar--2014|Govekar et al., 2014]] ; [[#Di%20Luca--2016|Di Luca et al., 2016]] ; [[#Trzeciak--2016|Trzeciak et al., 2016]] ; [[#Seiler--2018|Seiler et al., 2018]] ; [[#Priestley--2020|Priestley et al., 2020]] ). There is also ''high confidence'' that most current climate models underestimate the number of explosive systems (i.e., systems showing a decrease in mean sea level pressure of at least 24 hPa in 24 hours) over both hemispheres ( [[#Seiler--2016a|Seiler and Zwiers, 2016a]] ; [[#Gao--2020|Gao et al., 2020]] ; [[#Priestley--2020|Priestley et al., 2020]] ). There is ''high confidence'' that the underestimation of the intensity of ETCs is associated with the coarse horizontal resolution of climate models, with higher horizontal resolution models, including HighResMIP and CORDEX, usually showing better performance ( [[#Colle--2013|Colle et al., 2013]] ; [[#Zappa--2013a|Zappa et al., 2013a]] ; [[#Di%20Luca--2016|Di Luca et al., 2016]] ; [[#Trzeciak--2016|Trzeciak et al., 2016]] ; [[#Seiler--2018|Seiler et al., 2018]] ; [[#Gao--2020|Gao et al., 2020]] ; [[#Priestley--2020|Priestley et al., 2020]] ). The improvement by higher-resolution models is found, even when comparing models and reanalyses after post-processing data to a common resolution ( [[#Zappa--2013a|Zappa et al., 2013a]] ; [[#Di%20Luca--2016|Di Luca et al., 2016]] ; [[#Priestley--2020|Priestley et al., 2020]] ). The systematic bias in the intensity of ETCs has also been linked to the inability of current climate models to resolve diabatic processes, particularly those related to the release of latent heat ( [[#Willison--2013|Willison et al., 2013]] ; [[#Trzeciak--2016|Trzeciak et al., 2016]] ) and the formation of clouds ( [[#Govekar--2014|Govekar et al., 2014]] ). There is ''medium confidence'' that climate models simulate well the spatial distribution of precipitation associated with ETCs over the Northern Hemisphere, together with some of the main features of the ETC life cycle, including the maximum in precipitation occurring just before the peak in dynamical intensity (e.g., vorticity) as observed in a reanalysis and observations ( [[#Hawcroft--2018|Hawcroft et al., 2018]] ). There is, however, large observational uncertainty in ETC-associated precipitation ( [[#Hawcroft--2018|Hawcroft et al., 2018]] ) and limitations in the simulation of frontal precipitation, including overly low rainfall intensity over mid-latitude oceanic areas in both hemispheres ( [[#Catto--2015|Catto et al., 2015]] ).

<div id="11.7.2.3" class="h3-container"></div>

<span id="detection-and-attribution-event-attribution-5"></span>
==== 11.7.2.3 Detection and Attribution, Event Attribution ====

<div id="h3-37-siblings" class="h3-siblings"></div>

( [[IPCC:Wg1:Chapter:Chapter-3#3.3.3.3%20|Section 3.3.3.3]] concluded that there is ''low confidence'' in the attribution of observed changes in the number of ETCs in the Northern Hemisphere and ''high confidence'' that the poleward shift of storm tracks in the Southern Hemisphere is linked to human activity, mostly due to emissions of ozone-depleting substances. Specific studies attributing changes in the most extreme ETCs are not available. The human influence on individual extreme ETC events has been considered only a few times and there is overall ''low confidence'' in the attribution of these changes ( [[#NASEM--2016|NASEM, 2016]] ; [[#Vautard--2019|Vautard et al., 2019]] ).

<div id="11.7.2.4" class="h3-container"></div>

<span id="projections-4"></span>
==== 11.7.2.4 Projections ====

<div id="h3-38-siblings" class="h3-siblings"></div>

The frequency of ETCs is expected to change, primarily following a poleward shift of the storm tracks as discussed in [[IPCC:Wg1:Chapter:Chapter-4#4.5.1.6|Section 4.5.1.6]] (see also Figure 4.31) and [[IPCC:Wg1:Chapter:Chapter-8#8.4.2.8|Section 8.4.2.8]] . There is ''medium confidence'' that changes in the dynamical intensity (e.g., wind speeds) of ETCs will be small, although changes in the location of storm tracks can lead to substantial changes in local extreme wind speeds ( [[#Zappa--2013b|Zappa et al., 2013b]] ; [[#Chang--2014|Chang, 2014]] ; [[#Li--2014|Li et al., 2014]] ; [[#Seiler--2016b|Seiler and Zwiers, 2016b]] ; [[#Yettella--2017|Yettella and Kay, 2017]] ; [[#Barcikowska--2018|Barcikowska et al., 2018]] ; [[#Kar-Man%20Chang--2018|Kar-Man Chang, 2018]] ). [[#Yettella--2017|Yettella and Kay (2017)]] detected and tracked ETCs over both hemispheres in an ensemble of 30 Community Earth System Model Large Ensemble simulations, differing only in their initial conditions, and found that changes in mean wind speeds around ETC centres are often negligible between present (1986–2005) and future (2081–2100) periods. Using 19 CMIP5 models, [[#Zappa--2013b|Zappa et al. (2013b)]] found an overall reduction in the number of cyclones associated with low-troposphere (850-hPa) wind speeds larger than 25 m s <sup>–1</sup> over the North Atlantic and Europe with the number of the 10% strongest cyclones decreasing by about 8% and 6% in December–January–February and June–July–August according to the RCP4.5 scenario (2070–2099 vs. 1976–2005). Over the North Pacific, [[#Chang--2014|Chang (2014)]] showed that CMIP5 models project a decrease in the frequency of ETCs, with the largest central pressure perturbation (i.e., the depth, strongly related with low-level wind speeds) by the end of the century according to simulations using the RCP8.5 scenario. Using projections from CMIP5 GCMs under the RCP8.5 scenario (1981–2000 to 2081–2100), [[#Seiler--2016b|Seiler and Zwiers (2016b)]] projected a northward shift in the number of explosive ETCs in the northern Pacific, with fewer and weaker events south, and more frequent and stronger events north of 45°N. Using 19 CMIP5 GCMs under the RCP8.5 scenario, [[#Kar-Man%20Chang--2018|Kar-Man Chang (2018)]] found a significant decrease in the number of ETCs associated with extreme wind speeds (2081–2100 vs. 1980–99) over the Northern Hemisphere (average decrease of 17%) and over some smaller regions, including the Pacific and Atlantic regions.

Over the Southern Hemisphere, future changes (RCP8.5 scenario; 1980–1999 to 2081–2100) in extreme ETCs were studied by [[#Chang--2017|Chang (2017)]] using 26 CMIP5 models, and a variety of intensity metrics (850-hPa vorticity, 850-hPa wind speed, mean sea level pressure and near-surface wind speed). They found that the number of extreme cyclones is projected to increase by at least 20% and as much as 50%, depending on the specific metric used to define extreme ETCs. Increases in the number of strong cyclones appear to be robust across models and for most seasons, although they show strong regional variations, with increases occurring mostly over the southern flank of the storm track, consistent with a shift and intensification of the storm track. Overall, there is ''medium confidence'' that projected changes in the dynamical intensity of ETCs depend on the resolution and formulation (e.g., explicit or implicit representation of convection) of climate models ( [[#Booth--2013|Booth et al., 2013]] ; [[#Michaelis--2017|Michaelis et al., 2017]] ; [[#Zhang--2017|Zhang and Colle, 2017]] ).

As reported in AR5 and in [[IPCC:Wg1:Chapter:Chapter-8#8.4.2.8|Section 8.4.2.8]] , despite small changes in the dynamical intensity of ETCs, there is ''high confidence'' that the precipitation associated with ETCs will increase in the future ( [[#Zappa--2013b|Zappa et al., 2013b]] ; [[#Marciano--2015|Marciano et al., 2015]] ; [[#Pepler--2016|Pepler et al., 2016]] ; [[#Michaelis--2017|Michaelis et al., 2017]] ; [[#Yettella--2017|Yettella and Kay, 2017]] ; [[#Zhang--2017|Zhang and Colle, 2017]] ; [[#Barcikowska--2018|Barcikowska et al., 2018]] ; [[#Hawcroft--2018|Hawcroft et al., 2018]] ; [[#Zarzycki--2018|Zarzycki, 2018]] ; [[#Kodama--2019|Kodama et al., 2019]] ; [[#Bevacqua--2020a|Bevacqua et al., 2020a]] ; [[#Reboita--2021|Reboita et al., 2021]] ). There is ''high confidence'' that increases in precipitation will follow increases in low-level water vapour (i.e., about 7% per 1°C of surface warming; see Box 11.1) and will be larger for higher warming levels ( [[#Zhang--2017|Zhang and Colle, 2017]] ). There is ''medium confidence'' that precipitation changes will show regional and seasonal differences due to distinct changes in atmospheric humidity and dynamical conditions ( [[#Zappa--2015|Zappa et al., 2015]] ; [[#Hawcroft--2018|Hawcroft et al., 2018]] ), with decreases in some specific regions such as the Mediterranean ( [[#Zappa--2015|Zappa et al., 2015]] ; [[#Barcikowska--2018|Barcikowska et al., 2018]] ). There is ''high confidence'' that snowfall associated with winter ETCs will decrease in the future, because increases in tropospheric temperatures lead to a lower proportion of precipitation falling as snow ( [[#O’Gorman--2014|O’Gorman, 2014]] ; [[#Rhoades--2018|Rhoades et al., 2018]] ; [[#Zarzycki--2018|Zarzycki, 2018]] ). However, there is ''medium confidence'' that extreme snowfall events associated with winter ETCs will change little in regions where snowfall will be supported in the future ( [[#O’Gorman--2014|O’Gorman, 2014]] ; [[#Zarzycki--2018|Zarzycki, 2018]] ).

In summary, there is ''low confidence'' in past changes in the dynamical intensity (e.g., maximum wind speeds) of ETCs and ''medium confidence'' that, in the future, these changes will be small, although changes in the location of storm tracks could lead to substantial changes in local extreme wind speeds. There is ''high confidence'' that average and maximum ETC precipitation-rates will increase with warming, with the magnitude of the increases associated with increases in atmospheric water vapour. There is ''medium confidence'' that projected changes in the intensity of ETCs, including wind speeds and precipitation, depend on the resolution and formulation of climate models.

<div id="11.7.3" class="h2-container"></div>

<span id="severe-convective-storms"></span>
=== 11.7.3 Severe Convective Storms ===

<div id="h2-46-siblings" class="h2-siblings"></div>

Severe convective storms are convective systems that are associated with extreme phenomena such as tornadoes, hail, heavy precipitation (rain or snow), strong winds, and lightning. The assessment of changes in severe convective storms in SREX (Chapter 3, [[#Seneviratne--2012|Seneviratne et al., 2012]] ) and AR5 (Chapter 12, [[#Collins--2013|Collins et al., 2013]] ) is limited and focused mainly on tornadoes and hail storms. The SREX assessed that there is ''low confidence'' in observed trends in tornadoes and hail because of data inhomogeneities and inadequacies in monitoring systems. Subsequent literature assessed in the ''Climate Science Special Report'' ( [[#Kossin--2017|Kossin et al., 2017]] ) led to the assessment of the observed tornado activity over the 2000s in the USA, with a decrease in the number of days per year with tornadoes and an increase in the number of tornadoes on these days ( ''medium confidence'' ). However, there is ''low confidence'' in past trends for hail and severe thunderstorm winds. Climate models consistently project environmental changes that would support an increase in the frequency and intensity of severe thunderstorms that combine tornadoes, hail, and winds ( ''high confidence'' ), but there is ''low confidence'' in the details of the projected increase. Regional aspects of severe convective storms and details of the assessment of tornadoes and hail are also assessed in [[IPCC:Wg1:Chapter:Chapter-12#12.3.3.2|Section 12.3.3.2]] (tornadoes), [[IPCC:Wg1:Chapter:Chapter-12#12.3.4.5|Section 12.3.4.5]] (hail), [[IPCC:Wg1:Chapter:Chapter-12#12.4.5.3|Section 12.4.5.3]] (Europe), [[IPCC:Wg1:Chapter:Chapter-12#12.4.6.3|Section 12.4.6.3]] (North America), and [[IPCC:Wg1:Chapter:Chapter-12#12.7.2|Section 12.7.2]] (regional gaps and uncertainties).

<div id="11.7.3.1" class="h3-container"></div>

<span id="mechanisms-and-drivers-5"></span>
==== 11.7.3.1 Mechanisms and Drivers ====

<div id="h3-39-siblings" class="h3-siblings"></div>

Severe convective storms are sometimes embedded in synoptic-scale weather systems, such as TCs, ETCs, and fronts ( [[#Kunkel--2013|Kunkel et al., 2013]] ). They are also generated as individual events as mesoscale convective systems (MCSs) and mesoscale convective complexes (MCCs, a special type of a large, organized and long-lived MCS), without being clearly embedded within larger-scale weather systems. In addition to the general vigorousness of precipitation, hail, and winds associated with MCSs, characteristics of MCSs are viewed in new perspectives in recent years, probably because of both the development of dense mesoscale observing networks and advances in high-resolution mesoscale modelling (Sections 11.7.3.2 and 11.7.3.3). The horizontal scale of MCSs is discussed with their organization of the convective structure, and it is examined with a concept of ‘convective aggregation’ in recent years ( [[#Holloway--2017|Holloway et al., 2017]] ). MCSs sometimes take a linear shape and stay almost stationary with successive production of cumulonimbus on the upstream side (back-building type convection), and cause heavy rainfall ( [[#Schumacher--2005|Schumacher and Johnson, 2005]] ). Many of the recent severe rainfall events in Japan are associated with band-shaped precipitation systems ( [[#Kunii--2016|Kunii et al., 2016]] ; [[#Oizumi--2018|Oizumi et al., 2018]] ; [[#Tsuguti--2018|Tsuguti et al., 2018]] ; [[#Kato--2020|Kato, 2020]] ), suggesting common characteristics of severe precipitation, at least in East Asia. The convective modes of severe storms in the USA can be classified into rotating or linear modes and preferable environmental conditions for these modes, such as vertical shear, have been identified ( [[#Trapp--2005|Trapp et al., 2005]] ; [[#Smith--2013|Smith et al., 2013]] ; [[#Allen--2018|Allen, 2018]] ). Cloud microphysics characteristics of MCSs were examined and the roles of warm rain processes on extreme precipitation were emphasized recently ( [[#Sohn--2013|Sohn et al., 2013]] ; [[#Hamada--2015|Hamada et al., 2015]] ; [[#Hamada--2018|Hamada and Takayabu, 2018]] ). Idealized studies also suggest the importance of ice and mixed-phase processes of cloud microphysics on extreme precipitation ( [[#Sandvik--2018|Sandvik et al., 2018]] ; [[#Bao--2019|Bao and Sherwood, 2019]] ). However, it is unknown whether the types of MCS are changing in recent periods or observed ubiquitously all over the world.

Severe convective storms occur under conditions preferable for deep convection, that is, conditionally unstable stratification, sufficient moisture, both in lower and middle levels of the atmosphere, and a strong vertical shear. These large-scale environmental conditions are viewed as necessary conditions for the occurrence of severe convective systems, or the resulting tornadoes and lightning, and the relevance of these factors strongly depends on the region (e.g., [[#Antonescu--2016a|Antonescu et al., 2016a]] ; [[#Allen--2018|Allen, 2018]] ; [[#Tochimoto--2018|Tochimoto and Niino, 2018]] ). Frequently used metrics are atmospheric static stability, moisture content, convective available potential energy (CAPE) and convective inhibition, wind shear or helicity, including storm-relative environmental helicity ( [[#Tochimoto--2018|Tochimoto and Niino, 2018]] ; [[#Elsner--2019|Elsner et al., 2019]] ). These metrics, largely controlled by large-scale atmospheric circulations or synoptic weather systems, such as TCs and ETCs, are then generally used to examine severe convective systems. In particular, there is ''high confidence'' that CAPE in the tropics and the subtropics increases in response to global warming (M.S. [[#Singh--2017|]] [[#Singh--2017|Singh et al., 2017]] ), as supported by theoretical studies ( [[#Singh--2013|Singh and O’Gorman, 2013]] ; [[#Seeley--2015|Seeley and Romps, 2015]] ; [[#Romps--2016|Romps, 2016]] ; [[#Agard--2017|Agard and]] [[#Emanuel--2017|Emanuel, 2017]] ). The uncertainty, however, arises from the balance between factors affecting severe storm occurrence. For example, the warming of mid-tropospheric temperatures leads to an increase in the freezing level, which leads to increased melting of smaller hailstones, while there may be some offset by stronger updrafts driven by increasing CAPE, which would favour the growth of larger hailstones, leading to less melting when falling ( [[#Allen--2018|Allen, 2018]] ; [[#Mahoney--2020|Mahoney, 2020]] ).

There are few studies on relations between changes in severe convective storms and those of the large-scale circulation patterns. Tornado outbreaks in the USA are usually associated with ETCs with their frontal systems and TCs ( [[#Fuhrmann--2014|Fuhrmann et al., 2014]] ; [[#Tochimoto--2016|Tochimoto and Niino, 2016]] ). In early June to late July in East Asia, associated with the Baiu/Changma/Mei-yu, severe precipitation events are frequently caused by MCSs. Severe precipitation events are also caused by remote effects of TCs, known as predecessor rain events ( [[#Galarneau--2010|Galarneau et al., 2010]] ). Atmospheric rivers and other coherent types of enhanced water vapour flux also have the potential to induce severe convective systems ( [[#Kamae--2017a|Kamae et al., 2017a]] ; [[#Waliser--2017|Waliser and Guan, 2017]] ; [[#Ralph--2018|Ralph et al., 2018]] ; see [[IPCC:Wg1:Chapter:Chapter-8#8.3.2.8.2|Section 8.3.2.8.2]] ). Combined with the above drivers, topographic effects also enhance the intensity and duration of severe convective systems and the associated precipitation ( [[#Ducrocq--2008|Ducrocq et al., 2008]] ; [[#Piaget--2015|Piaget et al., 2015]] ). However, the changes in these drivers are not generally significant, so their relations to severe convective storms are unclear.

In summary, severe convective storms are sometimes embedded in synoptic-scale weather systems, such as TCs, ETCs and fronts, and modulated by large-scale atmospheric circulation patterns. The occurrence of severe convective storms and the associated severe events, including tornadoes, hail, and lightning, is affected by environmental conditions of the atmosphere, such as CAPE and vertical shear. The uncertainty, however, arises from the balance between these environmental factors affecting severe storm occurrence.

<div id="11.7.3.2" class="h3-container"></div>

<span id="observed-trends-6"></span>
==== 11.7.3.2 Observed Trends ====

<div id="h3-40-siblings" class="h3-siblings"></div>

Observed trends in severe convective storms or MCSs are not well documented, but the climatology of MCSs has been analysed in specific regions (North America, South America, Europe, Asia; regional aspects of convective storms are separately assessed in Chapter 12). As the definition of severe convective storms varies depending on the literature, it is not straightforward to make a synthesizing view of observed trends in severe convective storms in different regions. However, analysis using satellite observations provides a global view of MCSs ( [[#Kossin--2017|Kossin et al., 2017]] ). The global distribution of thunderstorms is captured ( [[#Zipser--2006|Zipser et al., 2006]] ; [[#Liu--2015|Liu and Zipser, 2015]] ) by using the satellite precipitation measurements by the Tropical Rainfall Measuring Mission (TRMM) and Global Precipitation Mission (GPM) ( [[#Hou--2014|Hou et al., 2014]] ). The climatological characteristics of MCSs are provided by satellite analyses in South America ( [[#Durkee--2010|Durkee and Mote, 2010]] ; [[#Rasmussen--2011|Rasmussen and Houze, 2011]] ; [[#Rehbein--2018|Rehbein et al., 2018]] ) and those of MCCs in the Maritime Continent by [[#Trismidianto%20and%20H.%C2%A0Satyawardhana--2018|Trismidianto and Satyawardhana (2018)]] . Analysis of the environmental conditions favourable for severe convective events indirectly indicates the climatology and trends of severe convective events ( [[#Allen--2018|Allen et al., 2018]] ; [[#Taszarek--2018|Taszarek et al., 2018]] , 2019), though favourable conditions depend on the location, such as the difference for tornadoes associated with ETCs between the USA and Japan ( [[#Tochimoto--2018|Tochimoto and Niino, 2018]] ).

Observed trends in severe convective storms are highly regionally dependent. In the USA, it is indicated that there is no significant increase in convective storms, and hail and severe thunderstorms ( [[#Kunkel--2013|Kunkel et al., 2013]] ; [[#Kossin--2017|Kossin et al., 2017]] ). There is an upward trend in the frequency and intensity of extreme precipitation events in the USA ( ''high confidence'' ) ( [[#Kunkel--2013|Kunkel et al., 2013]] ; Easterling et al., 2017), and MCSs have increased in occurrence and precipitation amounts since 1979 ( ''limited evidence'' ) ( [[#Feng--2016|Feng et al., 2016]] ). Significant interannual variability of hailstone occurrences is found in the Southern Great Plains of the USA ( [[#Jeong--2020|Jeong et al., 2020]] ). The mean annual number of tornadoes has remained relatively constant, but their variability of occurrence has increased since the 1970s,particularly over the 2000s, with a decrease in the number of days per year, but an increase in the number of tornadoes on these days ( [[#Brooks--2014|Brooks et al., 2014]] ; [[#Elsner--2015|Elsner et al., 2015]] , 2019; [[#Kossin--2017|Kossin et al., 2017]] ; [[#Allen--2018|Allen, 2018]] ). There has been a shift in the distribution of tornadoes, with increases in the mid-south of the USA and decreases over the High Plains ( [[#Gensini--2018|Gensini and Brooks, 2018]] ). Trends in MCSs are relatively more visible for particular aspects of MCSs, such as lengthening of active seasons and dependency on duration. MCSs have increased in occurrence and precipitation amounts since 1979 (Easterling et al., 2017). [[#Feng--2016|Feng et al. (2016)]] analysed that the observed increases in spring total and extreme rainfall in the central USA are dominated by MCSs, with increased frequency and intensity of long-lasting MCSs.

Studies on trends in severe convective storms and their ingredients outside of the USA are limited. [[#Westra--2014|Westra et al. (2014)]] found that there is an increase in the intensity of short-duration convective events (minutes to hours) over many regions of the world, except eastern China. In Europe, a climatology of tornadoes shows an increase in detected tornadoes between 1800 and 2014, but this trend might be affected by the density of observations ( [[#Antonescu--2016a|Antonescu et al., 2016a]] , b). An increase in the trend in extreme daily rainfall is found in south-eastern France, where MCSs play a key role in this type of event ( [[#Blanchet--2018|Blanchet et al., 2018]] ; [[#Ribes--2019|Ribes et al., 2019]] ). Trend analysis of the mean annual number of days with thunderstorms since 1979 in Europe indicates an increase over the Alps and central, south-eastern, and eastern Europe, with a decrease over the south-west ( [[#Taszarek--2019|Taszarek et al., 2019]] ). In the Sahelian region, [[#Taylor--2017|Taylor et al. (2017)]] analysed MCSs using satellite observations since 1982 and showed an increase in the frequency of extreme storms. In Bangladesh, the annual number of propagating MCSs decreased significantly during 1998–2015 based on TRMM precipitation data ( [[#Habib--2019|Habib et al., 2019]] ). [[#Prein--2018|Prein and Holland (2018)]] estimated the hail hazard from large-scale environmental conditions using a statistical approach and showed increasing trends in the USA, Europe, and Australia. However, trends in hail on regional scales are difficult to validate because of an insufficient length of observations and inhomogeneous records ( [[#Allen--2018|Allen, 2018]] ). The high spatial variability of hail suggests it is reasonable that there would be local signals of both positive and negative trends, and the trends that are occurring in hail globally are uncertain. In China, the total number of days that have either a thunderstorm or hail have decreased by about 50% from 1961 to 2010, and the reduction in these severe weather occurrences correlates strongly with the weakening of the East Asian summer monsoon (Q. [[#Zhang--2017|]] [[#Zhang--2017|]] [[#Zhang--2017|Zhang et al., 2017]] ). More regional aspects of severe convective storms are detailed in Chapter 12.

In summary, because the definition of severe convective storms varies depending on the literature and the region, it is not straightforward to make a synthesizing view of observed trends in severe convective storms in different regions. In particular, observational trends in tornadoes, hail, and lightning associated with severe convective storms are not robustly detected due to insufficient coverage of the long-term observations. There is ''medium confidence'' that the mean annual number of tornadoes in the USA has remained relatively constant, but their variability of occurrence has increased since the 1970s, particularly over the 2000s, with a decrease in the number of days per year, and an increase in the number of tornadoes on these days ( ''high confidence'' ). Detected tornadoes have also increased in Europe, but the trend depends on the density of observations.

<div id="11.7.3.3" class="h3-container"></div>

<span id="model-evaluation-6"></span>
==== 11.7.3.3 Model Evaluation ====

<div id="h3-41-siblings" class="h3-siblings"></div>

The explicit representation of severe convective storms requires non-hydrostatic models with horizontal grid spacings finer than 4 km, denoted as convection-permitting models or storm-resolving models ( [[IPCC:Wg1:Chapter:Chapter-10#10.3.1|Section 10.3.1]] ). Convection-permitting models are becoming available to run over a wide domain, such as a continental scale or even over the global area, and show realistic climatological characteristics of MCSs ( [[#Prein--2015|Prein et al., 2015]] ; [[#Guichard--2017|Guichard and Couvreux, 2017]] ; [[#Satoh--2019|Satoh et al., 2019]] ). Such high-resolution simulations are computationally too expensive to perform at the larger domain and for long periods, and alternative methods by using an RCM with dynamical downscaling are generally used ( [[IPCC:Wg1:Chapter:Chapter-10#10.3.1|Section 10.3.1]] ). Convection-permitting models are used as the flagship project of CORDEX to particularly study projections of thunderstorms ( [[IPCC:Wg1:Chapter:Chapter-10#10.3.3|Section 10.3.3]] ). Simulations of North American MCSs by a convection-permitting model conducted by [[#Prein--2020|Prein et al. (2020)]] were able to capture the main characteristics of the observed MCSs, such as their size, precipitation rate, propagation speed, and lifetime. Cloud-permitting model simulations in Europe also showed sub-daily precipitation realistically ( [[#Ban--2014|Ban et al., 2014]] ; [[#Kendon--2014|Kendon et al., 2014]] ). Evaluation of precipitation conducted using convection-permitting simulations around Japan showed that finer resolution improves intense precipitation ( [[#Murata--2017|Murata et al., 2017]] ). MCSs over Africa simulated using convection-permitting models showed better extreme rainfall ( [[#Kendon--2019|Kendon et al., 2019]] ) and diurnal cycles and convective rainfall over land than the coarser-resolution RCMs or GCMs ( [[#Stratton--2018|Stratton et al., 2018]] ; [[#Crook--2019|Crook et al., 2019]] ).

The other modelling approach is the analysis of the environmental conditions that control characteristics of severe convective storms using the typical climate model results in CMIP5/6 ( [[#Allen--2018|Allen, 2018]] ). Severe convective storms are generally formed in environments with large CAPE and tornadic storms are, in particular, formed with a combination of large CAPE and strong vertical wind shear. As the processes associated with severe convective storms occur over a wide range of spatial and temporal scales, some of which are poorly understood and are inadequately sampled by observational networks, the model calibration approaches are generally difficult and insufficiently validated. Therefore, model simulations and their interpretations should be done with much caution.

In summary, there are typically two kinds of modelling approaches for studying changes in severe convective storms. One is to use convection-permitting models in wider regions or the global domain in time-sliced downscaling methods to directly simulate severe convective storms. The other is the analysis of the environmental conditions that control characteristics of severe convective storms by using coarse-resolution GCMs. Even in finer-resolution convection-permitting models, it is difficult to directly simulate tornadoes, hail storms, and lightning, so modelling studies of these changes are limited.

<div id="11.7.3.4" class="h3-container"></div>

<span id="detection-and-attribution-event-attribution-6"></span>
==== 11.7.3.4 Detection and Attribution, Event Attribution ====

<div id="h3-42-siblings" class="h3-siblings"></div>

It is extremely difficult to detect differences in time and space of severe convective storms ( [[#Kunkel--2013|Kunkel et al., 2013]] ). Although some ingredients that are favourable for severe thunderstorms have increased over the years, others have not; thus, overall, changes in the frequency of environments favourable for severe thunderstorms have not been statistically significant. Event attribution studies on severe convective events have now been undertaken for some cases. For the case of the heavy rain event of July 2018 in Japan (Box 11.4), [[#Kawase--2020|Kawase et al. (2020)]] took a storyline approach to show that the rainfall during this event in Japan was increased by approximately 7% due to recent rapid warming around Japan. For the case of the December 2015 extreme rainfall event in Chennai, India, the extremity of the event was equally caused by the warming trend in the Bay of Bengal SSTs and the strong El Niño conditions ( [[#van%20Oldenborgh--2016|van Oldenborgh et al., 2016]] ; [[#Boyaj--2018|Boyaj et al., 2018]] ). For hailstorms, such as those that caused disasters in the USA in 2018, detection of the role of climate change in changing hail storms is more difficult, because hail storms are not, in general, directly simulated by convection-permitting models and not adequately represented by the environmental parameters of coarse-resolution GCMs ( [[#Mahoney--2020|Mahoney, 2020]] ).

In summary, it is extremely difficult to detect and attribute changes in severe convective storms. There is ''limited evidence'' that extreme precipitation associated with severe convective storms has increased in some cases.

<div id="11.7.3.5" class="h3-container"></div>

<span id="projections-5"></span>
==== 11.7.3.5 Projections ====

<div id="h3-43-siblings" class="h3-siblings"></div>

Future projections of severe convective storms are usually studied either by analysing the environmental conditions simulated by climate models, or by a time-slice approach with higher-resolution convection-permitting models by comparing simulations downscaled with climate model results under historical conditions and those under hypothesized future conditions ( [[#Kendon--2017|Kendon et al., 2017]] ; [[#Allen--2018|Allen, 2018]] ). Up to now, individual studies using convection-permitting models gave projections of extreme events associated with severe convective storms in local regions, and it is not generally possible to obtain global or general views of projected changes of severe convective storms. [[#Prein--2017|Prein et al. (2017)]] investigated future projections of North American MCS simulations and showed an increase in MCS frequency and an increase in total MCS precipitation volume by the combined effect of increases in maximum precipitation rates associated with MCSs and increases in their size. [[#Rasmussen--2020|Rasmussen et al. (2020)]] investigated future changes in the diurnal cycle of precipitation by capturing organized and propagating convection and showed that weak-to-moderate convection will decrease, and strong convection will increase in frequency in the future. [[#Ban--2015|Ban et al. (2015)]] found that the day-long and hour-long precipitation events in summer intensify in the European region covering the Alps. [[#Kendon--2019|Kendon et al. (2019)]] showed future increases in extreme three-hourly precipitation in Africa. [[#Murata--2015|Murata et al. (2015)]] investigated future projections of precipitation around Japan and showed a decrease in monthly mean precipitation in the eastern Japan Sea region in December, suggesting that convective clouds become shallower in the future in the winter over the Japan Sea.

The other approach is the projection of the environmental conditions that control characteristics of severe convective storms by analysing climate model results. There is ''high confidence'' that CAPE, particularly summer mean CAPE and high percentiles of the CAPE in the tropics and subtropics, increases in response to global warming in an ensemble of climate models including those of CMIP5, mainly from increased low-level specific humidity ( [[#Sobel--2011|Sobel and Camargo, 2011]] ; M.S. [[#Singh--2017|]] [[#Singh--2017|Singh et al., 2017]] ; J. [[#Chen--2020|]] [[#Chen--2020|Chen et al., 2020]] a). Convective inhibition becomes stronger over most land areas under global warming, resulting mainly from reduced low-level relative humidity over land (J. [[#Chen--2020|]] [[#Chen--2020|Chen et al., 2020]] a). However, there are large differences within the CMIP5 ensemble for environmental conditions, which contribute to some degree of uncertainty ( [[#Allen--2018|Allen, 2018]] ). Because the relation between simulated environments in models and the occurrence of severe convective storms are, in general, insufficiently validated, there is generally ''low confidence'' in the projection of severe convective storms with the approach of the environmental conditions.

In the USA, projected changes in the environmental conditions show an increase in CAPE and no changes or decreases in the vertical wind shear, suggesting favourable conditions for an increase in severe convective storms in the future, but the interpretation of how tornadoes or hail will change is an open question because of the strong dependence on shear ( [[#Brooks--2013|Brooks, 2013]] ). [[#Diffenbaugh--2013|Diffenbaugh et al. (2013)]] showed robust increases in the occurrence of the favourable environments for severe convective storms with increased CAPE and stronger low-level wind shear in response to future global warming. A downscaling approach showed that the variability of the occurrence of severe convective storms increases in spring in late 21st-century simulations ( [[#Gensini--2015|Gensini and Mote, 2015]] ). Future changes in hail occurrence in the USA examined through convection-permitting dynamical downscaling suggested that the hail season may begin earlier in the year and exhibit more interannual variability, with increases in the frequency of large hail in broad areas over the USA ( [[#Trapp--2019|Trapp et al., 2019]] ). There is ''medium confidence'' that the frequency and variability of the favourable environments for severe convective storms will increase in spring, and ''low confidence'' for summer and autumn ( [[#Diffenbaugh--2013|Diffenbaugh et al., 2013]] ; [[#Gensini--2015|Gensini and Mote, 2015]] ; [[#Hoogewind--2017|Hoogewind et al., 2017]] ). The occurrence of hail events in Colorado in the USA was examined by comparing both present-day and projected future climates using high-resolution model simulations capable of resolving hailstorms ( [[#Mahoney--2012|Mahoney et al., 2012]] ), which showed that hail is almost eliminated at the surface in the future in most of the simulations, despite more intense future storms and significantly larger amounts of hail generated in-cloud.

Future changes in severe convection environments show enhancement of instability with less robust changes in the frequency of strong vertical wind shear in Europe ( [[#Púčik--2017|Púčik et al., 2017]] ) and in Japan ( [[#Muramatsu--2016|Muramatsu et al., 2016]] ). In Japan, the frequency of conditions favourable for strong tornadoes increases in spring, and partly in summer.

In summary, the average and maximum rain rates associated with severe convective storms increase in a warming world in some regions, including the USA ( ''high confidence'' ). There is ''high confidence'' from climate models that CAPE increases in response to global warming in the tropics and subtropics, suggesting more favourable environments for severe convective storms. The frequency of severe convective storms in spring is projected to increase in the USA, leading to a lengthening of the severe convective storm season ( ''medium confidence'' ); evidence in other regions is limited. There is significant uncertainty about projected regional changes in tornadoes, hail, and lightning due to limited analysis of simulations using convection-permitting models ( ''hi'' ''gh confidence'' ).

<div id="11.7.4" class="h2-container"></div>

<span id="extreme-winds"></span>
=== 11.7.4 Extreme Winds ===

<div id="h2-47-siblings" class="h2-siblings"></div>

Extreme winds are defined here in terms of the strongest near-surface wind speeds that are generally associated with extreme storms, such as TCs, ETCs, and severe convective storms. In previous IPCC reports, near-surface wind speed (including extremes), has not been assessed as a variable in its own right, but rather in the context of other extreme atmospheric or oceanic phenomena. The exception was the SREX report ( [[#Seneviratne--2012|Seneviratne et al., 2012]] ), which specifically examined past changes and projections of mean and extreme near-surface wind speeds. A strong decline in extreme winds compared to mean winds was reported for the continental northern mid-latitudes. Due to the small number of studies and uncertainties in terrestrial-based surface wind measurements, the findings were assigned ''low confidence'' in SREX. The AR5 reported a weakening of mean and maximum winds from the 1960s or 1970s to the early 2000s in the tropics and mid-latitudes, and increases in high latitudes, but with ''low confidence'' in changes in the observed surface winds over land ( [[#Hartmann--2013|Hartmann et al., 2013]] ). Observed trends in mean wind speed over land and the ocean are assessed in [[IPCC:Wg1:Chapter:Chapter-2#2.3.1.4.4|Section 2.3.1.4.4]] . Aspects of climate impact-drivers for winds are addressed in Sections 12.3.3 and 12.5.2, and their regional changes are assessed in [[IPCC:Wg1:Chapter:Chapter-12#12.4|Section 12.4]] .

Observationally, although not specifically addressing extreme wind speed changes, negative surface wind speed trends (stilling) were found in the tropics and mid-latitudes of both hemispheres of –0.014 m s <sup>–1</sup> yr <sup>–1</sup> , while positive trends were reported at high latitudes poleward of 70 degrees, based on a review of 148 studies ( [[#McVicar--2012b|McVicar et al., 2012b]] ). An earlier study attributed the stilling to both changes in atmospheric circulation and an increase in surface roughness due to an overall increase in vegetation cover ( [[#Vautard--2010|Vautard et al., 2010]] ). Since then, a number of studies have mostly confirmed these general negative mean-wind trends based on anemometer data for Spain ( [[#Azorin-Molina--2017|Azorin-Molina et al., 2017]] ), Turkey, ( [[#Dadaser-Celik--2014|Dadaser-Celik and Cengiz, 2014]] ), the Netherlands, ( [[#Wever--2012|Wever, 2012]] ), Saudi Arabia, ( [[#Rehman--2013|Rehman, 2013]] ), Romania, ( [[#Marin--2014|Marin et al., 2014]] ), and China ( [[#Chen--2013|Chen et al., 2013]] ). [[#Lin--2013|Lin et al. (2013)]] note that wind speed variability over China is greater at high-elevation locations compared to those closer to mean sea level. [[#Hande--2012|Hande et al. (2012)]] , using radiosonde data, found an increase in surface wind speed on Macquarie Island of Australia.

A number of new studies have examined surface wind speeds over the ocean using ship-based measurements, satellite altimeters, and Special Sensor Microwave/Imagers ( [[#Tokinaga--2011|Tokinaga and Xie, 2011]] ; [[#Zieger--2014|Zieger et al., 2014]] ). It has been noted that wind speed trends tend to be stronger in altimeter measurements, although the spatial patterns of change are qualitatively similar in both instruments ( [[#Zieger--2014|Zieger et al., 2014]] ). Q. [[#Liu--2016|]] [[#Liu--2016|Liu et al. (2016)]] found positive trends in surface wind speeds over the Arctic Ocean in 20 years of satellite observations. Small positive trends in mean wind speed were found in 33 years of satellite data, together with larger trends in the 90th percentile values over global oceans ( [[#Ribal--2019|Ribal and Young, 2019]] ). These results were consistent with an earlier study that found a positive trend in 1-in-100-year wind speeds ( [[#Young--2012|Young et al., 2012]] ). A positive change in mean wind speeds was found for the Arabian Sea and the Bay of Bengal ( [[#Shanas--2015|Shanas and Kumar, 2015]] ) and [[#Zheng--2017|Zheng et al. (2017)]] found that positive wind speed trends over the ocean were larger during winter seasons than summer seasons.

Changes in extreme winds are associated with changes in the characteristics (locations, frequencies, and intensities) of extreme storms, including TCs, ETCs, and severe convective storms. For TCs, as assessed in [[#11.7.1.5|Section 11.7.1.5]] , it is projected that the average peak TC wind speeds will increase globally with warming, while the global frequency of TCs over all categories will decrease or remain unchanged; the average location where TCs reach their peak wind intensity will migrate poleward in the western North Pacific Ocean as the tropics expand with warming. Frequency, intensities, and geographical distributions of extreme wind events associated with TCs will change according to these TC changes. For ETCs, by the end of the century, CMIP5 models show that the number of ETCs associated with extreme winds will significantly decrease in the mid- and high latitudes of the Northern Hemisphere in winter, with the projected decrease being larger over the Atlantic ( [[#Kar-Man%20Chang--2018|Kar-Man Chang, 2018]] ), while it will significantly increase irrespective of the season in the Southern Hemisphere ( [[#11.7.2.4|Section 11.7.2.4]] ; [[#Chang--2017|Chang, 2017]] ). Over the ocean in the subtropics, a large ensemble of 60-km global model simulations indicated that extreme winds associated with storm surges will intensify over 15–35°N in the Northern Hemisphere ( [[#Mori--2019|Mori et al., 2019]] ). However, extreme surface wind speeds will mostly decrease due to decreases in the number and intensity of TCs over most tropical areas of the Southern Hemisphere ( [[#Mori--2019|Mori et al., 2019]] ). The projected changes in the frequency of extreme winds are associated with the future changes in TCs and ETCs.

Extreme cyclonic windstorms that share some characteristics with both TCs and ETCs occur regularly over the Mediterranean Sea and are often referred to as ‘medicanes’ (Ragone et al., 2018; [[#Miglietta--2019|Miglietta and Rotunno, 2019]] ; [[#Zhang--2021|Zhang et al., 2021]] ). Medicanes pose substantial threats to regional islands and coastal zones. A growing body of literature consistently found that the frequency of medicanes decreases under warming, while the strongest medicanes become stronger (Gaertner et al., 2007; Romero and [[#Emanuel--2013|Emanuel, 2013]] , 2017; [[#Cavicchia--2014|Cavicchia et al., 2014]] ; [[#Tous--2016|Tous et al., 2016]] ; [[#Romera--2017|Romera et al., 2017]] ; [[#González-Alemán--2019|González-Alemán et al., 2019]] ). This is also consistent with expected global changes in TCs under warming ( [[#11.7.1|Section 11.7.1]] ). Based on the consistency of these studies, it is ''likely'' that medicanes will decrease in frequency, while the strongest medicanes become stronger under warming scenario projections ( ''medi'' ''um confidence'' ).

In summary, the observed intensity of extreme winds is becoming less severe in the low to mid-latitudes, while becoming more severe in high latitudes poleward of 60 degrees ( ''low confidence'' ). Projected changes in the frequency and intensity of extreme winds are associated with projected changes in the frequency and intensity of TCs and ETCs ( ''medi'' ''um confidence'' ).

<div id="11.8" class="h1-container"></div>

<span id="compound-events"></span>