Editing IPCC:AR6/SROCC/Chapter-6 (section)

== 6.3 Changes in Tracks, Intensity, and Frequency of Tropical and Extratropical Cyclones and Associated Sea Surface Dynamics ==

<div id="article-6-3changes-in-tracks-intensity-and-frequency-of-tropical-and-extratropical-cyclones-and-associated-sea-surface-dynamics-block-1"></div>

This section addresses new literature on TCs and ETCs and their effects on the ocean in the context of understanding how the changing nature of extreme events can cause compound hazards, risk and cascading impacts (discussed in Section 6.8). These topics are also discussed in Chapter 4 in the context of changes to ESLs (see Section 4.2.3.4).

<span id="changes-in-storms-and-associated-sea-surface-dynamics"></span>
=== 6.3.1 Changes in Storms and Associated Sea Surface Dynamics ===

<div id="section-6-3-1-1tropical-cyclones"></div>

<span id="tropical-cyclones"></span>
==== 6.3.1.1 Tropical Cyclones ====

<div id="section-6-3-1-1tropical-cyclones-block-1"></div>

IPCC AR5 concluded that there was ''low confidence'' in any long-term increases in TC activity globally and in attribution of global changes to any particular cause (Bindoff et al., 2013 <sup>[[#fn:r149|149]]</sup> ; Hartmann et al., 2013 <sup>[[#fn:r150|150]]</sup> ). Based on process understanding and agreement in 21st century projections, it is ''likely'' that the global TC frequency will either decrease or remain essentially unchanged, while global mean TC maximum wind speed and precipitation rates will ''likely'' increase although there is ''low confidence'' in region-specific projections of frequency and intensity (Christensen et al., 2013 <sup>[[#fn:r151|151]]</sup> ). The AR5 concluded that circulation features have moved poleward since the 1970s, associated with a widening of the tropical belt, a poleward shift of storm tracks and jet streams, and contractions of the northern polar vortex and the Southern Ocean westerly wind belts. However it is noted that natural modes of variability on interannual to decadal time scales prevent the detection of a clear climate change signal (Hartmann et al., 2013 <sup>[[#fn:r152|152]]</sup> ).

Since the AR5 and Knutson et al. (2010), palaeoclimatic surveys of coastal overwash sediments and stalagmites have provided further evidence of historical TC variability over the past several millennia. Patterns of storm activity across TC basins show variations through time that appear to be correlated with El Niño-Southern Oscillation (ENSO), North Atlantic Oscillation (NAO), and changes in atmospheric dynamics related to changes in precession of the sun (Toomey et al., 2013 <sup>[[#fn:r153|153]]</sup> ; Denommee et al., 2014 <sup>[[#fn:r154|154]]</sup> ; Denniston et al., 2015 <sup>[[#fn:r155|155]]</sup> ).

Further studies have investigated the dynamics of TCs. A modelling study investigated a series of low-frequency increases and decreases in TC activity over the North Atlantic over the 20th century (Dunstone et al., 2013 <sup>[[#fn:r156|156]]</sup> ). These variations, culminating in a recent rise in activity, are thought to be due in part to atmospheric aerosol forcing variations (aerosol forcing), which exerts a cooling effect (Booth et al., 2012 <sup>[[#fn:r157|157]]</sup> ; Dunstone et al., 2013 <sup>[[#fn:r158|158]]</sup> ). However, the relative importance of internal variability vs. radiative forcing for multidecadal variability in the Atlantic basin, including TC variability, remains uncertain (Weinkle et al., 2012 <sup>[[#fn:r159|159]]</sup> ; Zhang et al., 2013 <sup>[[#fn:r160|160]]</sup> ; Vecchi et al., 2017 <sup>[[#fn:r161|161]]</sup> ; Yan et al., 2017 <sup>[[#fn:r162|162]]</sup> ). Although the aerosol cooling effect has largely cancelled the increases in potential intensity over the observational period, according to Coupled Model Intercomparison Project Phase 5 (CMIP5) model historical runs, further anthropogenic warming in the future is expected to dominate the aerosol cooling effect leading to increasing TC intensities (Sobel et al., 2016 <sup>[[#fn:r163|163]]</sup> ).

TCs amplify wave heights along the tracks of rapidly moving cyclones (e.g., Moon et al., 2015a) and can therefore increase mixing to the surface of cooler subsurface water. Several studies found that TCs reduce the projected thermal stratification of the upper ocean in CMIP5 models under global warming, thereby slightly offsetting the simulated TC-intensity increases under climate warming conditions (Emanuel, 2015 <sup>[[#fn:r164|164]]</sup> ; Huang et al., 2015b <sup>[[#fn:r165|165]]</sup> ; Tuleya et al., 2016 <sup>[[#fn:r166|166]]</sup> ). On the other hand, freshening of the upper ocean by TC rainfall enhances density stratification by reducing near-surface salinity and this reduces the ability of TCs to cool the upper ocean, thereby having an influence opposite to the thermal stratification effect (Balaguru et al., 2015). In the late 21st century, increased salinity stratification was found to offset about 50% of the suppressive effects that TC mixing has on temperature stratification (Balaguru et al., 2015 <sup>[[#fn:r167|167]]</sup> ). Coupled ocean-atmosphere models still robustly project an increase of TC intensity with climate warming, and particularly for new TC-permitting coupled climate model simulations that compute internally consistent estimates of thermal stratification change (e.g., Kim et al., 2014a; Bhatia et al., 2018 <sup>[[#fn:r169|169]]</sup> ). Higher TC intensities in turn may further aggravate the impacts of SLR on TC-related coastal inundation extremes (Timmermans et al., 2017 <sup>[[#fn:r170|170]]</sup> ).

Kossin et al. (2014) <sup>[[#fn:r171|171]]</sup> identified a poleward expansion of the latitudes of maximum TC intensity in recent decades, which has been linked to an anthropogenically-forced tropical expansion (Sharmila and Walsh, 2018 <sup>[[#fn:r172|172]]</sup> ) and a continued poleward shift of cyclones projected over the western North Pacific in a warmer climate (Kossin et al., 2016 <sup>[[#fn:r173|173]]</sup> ). A 10% slowdown in translation speed of TCs over the 1949–2016 period has been linked to the weakening of the tropical summertime circulation associated with tropical expansion and a more pronounced slowdown in the range 16–22% was found over land areas affected by TCs in the western North Pacific, North Atlantic and Australian regions (Kossin, 2018 <sup>[[#fn:r174|174]]</sup> ). Slow-moving TCs together with higher moisture carrying capacity can cause significantly greater flood hazards (Emanuel, 2017 <sup>[[#fn:r175|175]]</sup> ; Risser and Wehner, 2017 <sup>[[#fn:r176|176]]</sup> ; van Oldenborgh et al., 2017 <sup>[[#fn:r177|177]]</sup> ; see also Table 6.2 and Box 6.1).

Trends in TCs over decades to a century or more have been investigated in several new studies. Key findings include: i) decreasing frequency of severe TCs that make landfall in eastern Australia since the late 1800s (Callaghan and Power, 2011 <sup>[[#fn:r178|178]]</sup> ); ii) increase in frequency of moderately large US storm surge events since 1923 (Grinsted et al., 2012 <sup>[[#fn:r179|179]]</sup> ); iii) recent increase of extremely severe cyclonic storms over the Arabian Sea in the post-monsoon season (Murakami et al., 2017 <sup>[[#fn:r180|180]]</sup> ); iv) intense TCs that make landfall in East and Southeast Asia in recent decades (Mei and Xie, 2016 <sup>[[#fn:r181|181]]</sup> ; Li et al., 2017 <sup>[[#fn:r182|182]]</sup> ); and v) an increase in annual global proportion of hurricanes reaching Category 4 or 5 intensity in recent decades (Holland and Bruyère, 2014 <sup>[[#fn:r183|183]]</sup> ).

Rapid intensification of tropical cyclones (RITCs) poses forecast challenges and increased risks for coastal communities (Emanuel, 2017 <sup>[[#fn:r184|184]]</sup> ). Warming of the upper ocean in the central and eastern tropical Atlantic associated with the positive phase of the Atlantic Multidecadal Oscillation (AMO) (Balaguru et al., 2018 <sup>[[#fn:r185|185]]</sup> ) and in the western North Pacific in recent decades due to a La Niña-like pattern (Zhao et al., 2018 <sup>[[#fn:r186|186]]</sup> ) has favoured RITCs in these regions. One new modelling study suggests there has been a detectable increase in RITC occurrence in the Atlantic basin in recent decades, with a positive contribution from anthropogenic forcing (Bhatia et al., 2019 <sup>[[#fn:r187|187]]</sup> ). Nonetheless, the background conditions that favour RITCs across the Atlantic basin as a whole tend to be associated with less favourable conditions for TC occurrence along the US east coast (Kossin, 2017 <sup>[[#fn:r188|188]]</sup> ).

New studies have used event attribution to explore attribution of certain individual TC events or anomalous seasonal cyclone activity events to anthropogenic forcing (Lackmann, 2015 <sup>[[#fn:r189|189]]</sup> ; Murakami et al., 2015 <sup>[[#fn:r190|190]]</sup> ; Takayabu et al., 2015 <sup>[[#fn:r191|191]]</sup> ; Zhang et al., 2016 <sup>[[#fn:r192|192]]</sup> ; Emanuel, 2017 <sup>[[#fn:r193|193]]</sup> ; see also Table 6.2 and Box 6.1). Risser and Wehner (2017) and van Oldenborgh et al. (2017) concluded that for the Hurricane Harvey event, there is a detectable human influence on extreme precipitation in the Houston area, although their detection analysis is for extreme precipitation in general and not specifically for TC-related precipitation.

There have been more TC dynamical or statistical/dynamical downscaling studies and higher resolution General Circulation Model (GCM) experiments (e.g., Emanuel, 2013; Manganello et al., 2014 <sup>[[#fn:r196|196]]</sup> ; Knutson et al., 2015 <sup>[[#fn:r197|197]]</sup> ; Murakami et al., 2015 <sup>[[#fn:r198|198]]</sup> ; Roberts et al., 2015 <sup>[[#fn:r199|199]]</sup> ; Wehner et al., 2015 <sup>[[#fn:r200|200]]</sup> ; Yamada et al., 2017 <sup>[[#fn:r201|201]]</sup> ). The findings of these studies generally support the AR5 projections of a general increase in intensity of the most intense TCs and a decline in TC frequency overall. However, the projected increase in global TC frequency by Emanuel (2013) <sup>[[#fn:r202|202]]</sup> and Bhatia et al. (2018) <sup>[[#fn:r203|203]]</sup> differed from most other TC frequency projections and previous assessments. For studies into future track changes of TCs under climate warming scenarios (Li et al., 2010 <sup>[[#fn:r204|204]]</sup> ; Kim and Cai, 2014 <sup>[[#fn:r205|205]]</sup> ; Manganello et al., 2014 <sup>[[#fn:r206|206]]</sup> ; Knutson et al., 2015 <sup>[[#fn:r207|207]]</sup> ; Murakami et al., 2015 <sup>[[#fn:r208|208]]</sup> ; Roberts et al., 2015 <sup>[[#fn:r209|209]]</sup> ; Wehner et al., 2015 <sup>[[#fn:r210|210]]</sup> ; Nakamura et al., 2017 <sup>[[#fn:r211|211]]</sup> ; Park et al., 2017 <sup>[[#fn:r212|212]]</sup> ; Sugi et al., 2017 <sup>[[#fn:r213|213]]</sup> ; Yamada et al., 2017 <sup>[[#fn:r214|214]]</sup> ; Yoshida et al., 2017 <sup>[[#fn:r215|215]]</sup> ; Zhang et al., 2017a <sup>[[#fn:r216|216]]</sup> ), it is difficult to identify a robust consensus of projected change in TC tracks, although several of the studies found either poleward or eastward expansion of TC occurrence over the North Pacific region resulting in greater storm occurrence in the central North Pacific. There have been new studies on storm size (Kim et al., 2014a <sup>[[#fn:r217|217]]</sup> ; Knutson et al., 2015 <sup>[[#fn:r218|218]]</sup> ; Yamada et al., 2017 <sup>[[#fn:r219|219]]</sup> ) under climate warming scenarios. These project TC size changes of up to ±10% between basins and studies and provide preliminary findings on this issue that future studies will continue to investigate. Several studies of TC storm surge (e.g., Lin et al., 2012; Garner et al., 2017 <sup>[[#fn:r220|220]]</sup> ) suggest that SLR will dominate the increased height of storm surge due to TCs under climate change.

Taking the above into account, the following is a summary assessment of TC detection and attribution. The observed poleward migration of the latitude of maximum TC intensity in the western North Pacific appears to be unusual compared to expected natural variability and therefore there is ''low to medium confidence'' that this change represents a detectable climate change, though with only ''low confidence'' that the observed shift has a discernible positive contribution from anthropogenic forcing. Anthropogenic forcing is believed to be producing some poleward expansion of the tropical circulation with climate warming. Additional studies of observed long-term TC changes such as: an increase in annual global proportion of Category 4 or 5 TCs in recent decades, severe TCs occurring in the Arabian Sea and making landfall in East and Southeast Asia, the increasing frequency of moderately large US storm surge events since 1923 and the decreasing frequency of severe TCs that make landfall in eastern Australia since the late 1800s, may each represent emerging anthropogenic signals, but still with ''low confidence (limited evidence)'' . The lack of confident climate change detection for most TC metrics continues to limit confidence in both future projections and in the attribution of past changes and TC events, since TC event attribution in most published studies is generally being inferred without support from a confident climate change detection of a long-term trend in TC activity.

TCs projections for the late 21st century are summarised as follows: 1) there is ''medium confidence'' that the proportion of TCs that reach Category 4–5 levels will increase, that the average intensity of TCs will increase (by roughly 1–10%, assuming a 2 ° C global temperature rise), and that average TCs precipitation rates (for a given storm) will increase by at least 7% per degree Celsius SST warming, owing to higher atmospheric water vapour content, 2) there is ''low confidence (low agreement, medium evidence)'' in how global TC frequency will change, although most modelling studies project some decrease in global TC frequency and 3) SLR will lead to higher storm surge levels for the TCs that do occur, assuming all other factors are unchanged ( ''very'' ''high confidence'' ).

<div id="section-6-3-1-2extratropical-cyclones-and-blocking"></div>

<span id="extratropical-cyclones-and-blocking"></span>
==== 6.3.1.2 Extratropical Cyclones and Blocking ====

<div id="section-6-3-1-2extratropical-cyclones-and-blocking-block-1"></div>

ETCs form in the mid-latitudes of the North Atlantic, North Pacific and Southern Oceans, and the Mediterranean Sea. The storm track regions are characterised by large surface equator-to-pole temperature gradients and baroclinic instability, and jet streams influence the direction and speed of movement of ETCs in this region. The thermodynamic response of the atmosphere to CO 2 tends to have opposing influences on storm tracks; surface shortwave cloud radiative changes increase the equator-to-pole temperature gradient whereas longwave cloud radiative changes reduce it (Shaw et al., 2016 <sup>[[#fn:r221|221]]</sup> ). AR5 concluded that the global number of ETCs is not expected to decrease by more than a few percent due to anthropogenic change. The Southern Hemisphere (SH) storm track is projected to have a small poleward shift, but the magnitude is model dependent (Christensen et al., 2013). AR5 also found a ''low confidence'' in the magnitude of regional storm track changes and the impact of such changes on regional surface climate (Christensen et al., 2013 <sup>[[#fn:r223|223]]</sup> ).

A ‘blocking’ event is an extratropical weather system in which the anticyclone (region of high pressure) becomes quasi-stationary and interrupts the usual westerly flow and/or storm tracks for up to a week or more (Woollings et al., 2018 <sup>[[#fn:r224|224]]</sup> ). Recent attention has focused on whether Arctic warming is linked to increased blocking and mid-latitude weather extremes (Barnes and Screen, 2015 <sup>[[#fn:r225|225]]</sup> ; Francis and Skific, 2015 <sup>[[#fn:r226|226]]</sup> ; Francis and Vavrus, 2015 <sup>[[#fn:r227|227]]</sup> ; Kretschmer et al., 2016 <sup>[[#fn:r228|228]]</sup> ), such as drought in California due to sea ice changes that cause a reorganisation of tropical convection (Cvijanovic et al., 2017 <sup>[[#fn:r229|229]]</sup> ), cold and snowy winters over Europe and North America (Liu et al., 2012 <sup>[[#fn:r230|230]]</sup> ; Cohen et al., 2018 <sup>[[#fn:r231|231]]</sup> ), extreme summer weather (Tang et al., 2013 <sup>[[#fn:r232|232]]</sup> ; Coumou et al., 2014 <sup>[[#fn:r233|233]]</sup> ) and Balkan flooding (Stadtherr et al., 2016 <sup>[[#fn:r234|234]]</sup> ). Studies suggest how blocking may influence arctic sea ice extent (Gong and Luo, 2017 <sup>[[#fn:r235|235]]</sup> ) and various pathways whereby Arctic warming could influence extreme weather (Barnes and Screen, 2015 <sup>[[#fn:r236|236]]</sup> ) such as reducing the equator to pole temperature gradient, slowing the jet stream thereby increasing its meandering behaviour (Röthlisberger et al., 2016 <sup>[[#fn:r237|237]]</sup> ; Mann et al., 2017 <sup>[[#fn:r238|238]]</sup> ) or causing it to split (Coumou et al., 2014 <sup>[[#fn:r239|239]]</sup> ), changing local dynamics in the vicinity of the sea ice edge (Screen and Simmonds, 2013 <sup>[[#fn:r240|240]]</sup> ) or weakening the stratospheric polar vortex (Cohen et al., 2014 <sup>[[#fn:r241|241]]</sup> ). However, sensitivity to choice of methodology (Screen and Simmonds, 2013 <sup>[[#fn:r242|242]]</sup> ) and large internal atmospheric variability masks the detection of such links in past records, and climate change can lead to opposing effects on the mid-latitude jet stream response leading to large uncertainty in future changes (Barnes and Polvani, 2015 <sup>[[#fn:r243|243]]</sup> ; Barnes and Screen, 2015 <sup>[[#fn:r244|244]]</sup> ).

New studies of future storm track behaviour in the NH, include Harvey et al. (2014) <sup>[[#fn:r245|245]]</sup> who find that the future changes to upper and lower tropospheric equator-to-pole temperature differences by the end of the century in a CMIP5 multi-model RCP8.5 ensemble are not well correlated and the lower temperature gradient dominates the summer storm track response whereas both upper and lower temperature gradients play a role in winter. In the northern North Atlantic storm track region, projected changes are found to be more strongly associated with changes in the lower rather than upper tropospheric equator-to-pole temperature difference (Harvey et al., 2015). In the SH, Harvey et al. (2014) find equator-to-pole temperature differences in the upper and lower troposphere in the future climate across a multi-model ensemble are well correlated with a general strengthening of the storm track. The total number of ETCs in a CMIP5 GCM multi-model ensemble decreased in the future climate, whereas the number of strong ETCs increased in most models and in the ensemble mean (Grieger et al., 2014 <sup>[[#fn:r246|246]]</sup> ). This was associated with a general poleward shift related to both tropical upper tropospheric warming and shifting meridional SST gradients in the Southern Ocean. The poleward movement of baroclinic instability and associated storm formation over the observational period due to external radiative forcing, is projected to continue, with associated declining rainfall trends in the mid-latitudes and positive trends further polewards (Frederiksen et al., 2017 <sup>[[#fn:r247|247]]</sup> ).

A number of new studies have found links between Arctic amplification, blocking events and various types of weather extremes in NH mid-latitudes in recent decades. However, the sensitivity of results to analysis technique and the generally short record with respect to internal variability means that at this stage there is ''low confidence'' in these connections. Consistent with the AR5, projected changes to NH storm tracks exhibit large differences between responses, causal mechanisms and ocean basins and so there remains ''low confidence'' in future changes in blocking and storm tracks in the NH. The storm track projections for the SH remain consistent with previous studies in indicating an observed poleward contraction and a continued strengthening and southward contraction of storm tracks in the future ( ''medium confidence'' ).

<div id="section-6-3-1-3waves-and-extreme-sea-levels"></div>

<span id="waves-and-extreme-sea-levels"></span>
==== 6.3.1.3 Waves and Extreme Sea Levels ====

<div id="section-6-3-1-3waves-and-extreme-sea-levels-block-1"></div>

AR5 also concluded that there is ''medium confidence'' that mean significant wave height has increased in the North Atlantic north of 45°N based on ship observations and reanalysis-forced wave model hindcasts. ESL events have increased since 1970, mainly due to a rise in mean sea levels (MSLs) over this period (Rhein et al., 2013 <sup>[[#fn:r248|248]]</sup> ). There is ''medium confidence'' that mid-latitude jets will move 1–2 degrees further poleward by the end of the 21st century under RCP8.5 in both hemispheres with weaker shifts in the NH. In the SH during austral summer, the poleward movement of the mid-latitude westerlies under climate change is projected to be partially offset by stratospheric ozone recovery. There is ''low confidence'' in projections of NH storm tracks particularly in the North Atlantic. Tropical expansion is ''likely'' to continue causing wider tropical regions and poleward movement of the subtropical dry zones (Collins et al., 2013 <sup>[[#fn:r249|249]]</sup> ). In the SH, it is ''likely'' that enhanced wind speeds will cause an increase in annual mean significant wave heights. Wave swells generated in the Southern Ocean may also affect wave heights, periods and directions in adjacent ocean basins. The projected reduction in sea ice extent in the Arctic Ocean (Holland et al., 2006 <sup>[[#fn:r250|250]]</sup> ) will increase wave heights and wave season length (Church et al., 2013 <sup>[[#fn:r251|251]]</sup> ).

Since AR5, new studies have shown observed changes in wave climate. Satellite observations from 1985 – 2018, showed small increases in significant wave height (+0.3 cm/year) and larger increases in extreme wave heights (90th percentiles), especially in the Southern (+1 cm/year) and North Atlantic (+0.8 cm/year) Oceans (Young and Ribal, 2019 <sup>[[#fn:r252|252]]</sup> ) as well as positive trends in wave height in the Arctic over 1992–2014 due to sea ice loss (Stopa et al., 2016 <sup>[[#fn:r253|253]]</sup> ; Thomson et al., 2016 <sup>[[#fn:r254|254]]</sup> ). Based on a wave reanalysis and satellite observations, Reguero et al. (2019) <sup>[[#fn:r255|255]]</sup> found that the global wave power, which represents the transport of the energy transferred from the wind into the sea surface motion, therefore including wave height, period and direction, has increased globally at a rate of 0.41% yr -1 between 1948 and 2008, with large variations across oceans. Long-term correlations are found between the increase in wave power and SSTs, particularly between the tropical Atlantic temperatures and the wave power in high southern latitudes, the most energetic region globally.

The results of several new global wave climate projection studies are consistent with those presented in IPCC AR5. Mentaschi et al. (2017) find up to a 30% increase in 100-year return level wave energy flux (the rate of transfer of wave energy) for the majority of coastal areas in the southern temperate zone, and a projected decrease in wave energy flux for most NH coastal areas at the end of the century in wave model simulations forced by six CMIP5 RCP8.5 simulations. The most significant long-term trends in extreme wave energy flux are explained by their relationship to modelled climate indices (Arctic Oscillation, ENSO and NAO). Wang et al. (2014b) assessed the climate change signal and uncertainty in a 20-member ensemble of wave height simulations, and found model uncertainty (inter-model variability) is significant globally, being about 10 times as large as the variability between RCP4.5 and RCP8.5 scenarios. In a study focussing on the western north Pacific wave climate, Shimura et al. (2015) <sup>[[#fn:r256|256]]</sup> associate projected regions of future change in wave climate with spatial variation of SSTs in the tropical Pacific Ocean. A review of 91 published global and regional scale wind-wave climate projection studies found a consensus on a projected increase in significant wave height over the Southern Ocean, tropical eastern Pacific ( ''high confidence'' ) and Baltic Sea ( ''medium confidence'' ), and decrease over the North Atlantic and Mediterranean Sea. They found little agreement between studies of projected changes over the Atlantic Ocean, southern Indian and eastern North Pacific Ocean and no regional agreement of projected changes to extreme wave height. It was noted that few studies focussed on wave direction change, which is important for shoreline response (Morim et al., 2018 <sup>[[#fn:r257|257]]</sup> ).

Significant developments have taken place since the AR5 to model storm surges and tides at the global scale. An unstructured global hydrodynamic modelling system has been developed with maximum coastal resolution of 5 km (Verlaan et al., 2015 <sup>[[#fn:r258|258]]</sup> ) and used to develop a global climatology of ESLs due to the combination of storm surge and tide (Muis et al., 2016 <sup>[[#fn:r259|259]]</sup> ). A global modelling study finds that under SLR of 0.5–10 m, changes to astronomical tidal mean high water exceed the imposed SLR by 10% or more at around 10% of coastal cities when coastlines are held fixed. When coastal recession is permitted a reduction in tidal range occurs due to changes in the period of oscillation of the basin under the changed coastline morphology (Pickering et al., 2017 <sup>[[#fn:r260|260]]</sup> ). A recent study on global probabilistic projections of ESLs considering MSL, tides, wind-waves and storm surges shows that under RCP4.5 and RCP8.5, the global average 100-year ESL is ''very likely'' to increase by 34–76 cm and 58–172 cm, respectively between 2000 – 2100 (Vousdoukas et al., 2018 <sup>[[#fn:r261|261]]</sup> ). Despite the advancements in global tide and surge modelling, using CMIP GCM multi-model ensembles to examine the effects of future weather and circulation changes on storm surges in a globally consistent way is still a challenge because of the ''low confidence'' in GCMs being able to represent small scale weather systems such as TCs. To date only a small number of higher resolution GCMs are able to produce credible cyclone climatologies (e.g., Murakami et al., 2012) although this will probably improve with further GCM development and increases to GCM resolution (Walsh et al., 2016 <sup>[[#fn:r262|262]]</sup> ).

The role of austral winter swell waves on ESL have been investigated in the Gulf of Guinea (Melet et al., 2016 <sup>[[#fn:r263|263]]</sup> ) and the Maldives (Wadey et al., 2017 <sup>[[#fn:r264|264]]</sup> ). Multivariate statistical analysis and probabilistic modelling is used to show that flood risk in the northern Gulf of Mexico is higher than determined from short observational records (Wahl et al., 2016 <sup>[[#fn:r265|265]]</sup> ). In Australia, changes in ESLs were modelled using four CMIP5 RCP8.5 simulations (Colberg et al., 2019 <sup>[[#fn:r266|266]]</sup> ). On the southern mainland coast, the southward movement of the subtropical ridge in the climate models led to small reductions (up to 0.4 m) in the modelled 20-year (5% probability of occurring in a year) storm surge. Over the Gulf of Carpentaria in the north, changes were largest and positive during austral summer in two out of the four models in response to a possible eastward shift in the northwest monsoon. Synthetic cyclone modelling was used to evaluate probabilities, interannual variability and future changes of extreme water levels from tides and TC-induced storm surge (storm tide) along the coastlines of Fiji (McInnes et al., 2014) and Samoa (McInnes et al., 2016 <sup>[[#fn:r268|268]]</sup> ). Higher resolution modelling for Apia, Samoa incorporating waves highlights that although SLR reduces wave setup and wind setup by 10–20%, during storm surges it increases wave energy reaching the shore by up to 200% (Hoeke et al., 2015 <sup>[[#fn:r269|269]]</sup> ).

In the German Bight, Arns et al. (2015) show that under SLR, increases in extreme water levels occur due to a change in phase of tidal propagation; which more than compensates for a reduction in storm surge due to deeper coastal sea levels. Vousdoukas et al. (2017) <sup>[[#fn:r271|271]]</sup> develop ESL projections for Europe that account for changes in waves and storm surge. In 2100, increases of up to 0.35 m relative to the SLR projections occur towards the end of the century under RCP8.5 along the North Sea coasts of northern Germany and Denmark and the Baltic Sea coast, whereas little to negative change is found for the southern European coasts.

In the USA, Garner et al. (2017) combine downscaled TCs, storm surge models, and probabilistic SLR projections to assess flood hazard associated with changing storm characteristics and SLR in New York City from the pre-industrial era to 2300. Increased storm intensity was found to compensate for offshore shifts in storm tracks leading to minimal change in modelled storm surge heights through 2300. However, projected SLR leads to large increases in future overall flood heights associated with TCs in New York City. Consequently, flood height return periods that were ∼ 500y (0.2% probability of occurring in a given year) during the pre-industrial era have fallen to ∼ 25y (4% probability of occurring annually) at present and are projected to fall to ∼ 5y (20% probability of occurring annually) within the next three decades.

In summary, new studies on observed wave climate change from 1985–2018 showed small increases in significant wave height of +0.3 cm/year and larger increases in 90th percentile wave heights of +1 cm/year in the Southern Ocean and +0.8 cm/year in the North Atlantic ocean ( ''medium confidence'' ). Sea ice loss in the Arctic has also increased wave heights over the period 1992–2014 ( ''medium confidence'' ). Global wave power has increased over the last six decades with differences across oceans related to long-term correlations with SST ( ''low confidence'' ). Future projections indicate an increase of the mean significant wave height across the Southern Ocean and tropical eastern Pacific ( ''high confidence'' ) and Baltic Sea ( ''medium confidence'' ) and decrease over the North Atlantic and Mediterranean Sea under RCP8.5 ( ''high confidence'' ). Extreme waves are projected to increase in the Southern Ocean and decrease in the North Atlantic and Mediterranean Sea under RCP4.5 and RCP8.5 ( ''high confidence'' ). There is still limited knowledge on projected wave period and direction. For coastal ESLs, new studies at the regional to global scale have generally had a greater focus on multiple contributing factors such as waves, tides, storm surges and SLR. At the global scale, probabilistic projections of extreme sea levels considering these factors projects the global average 100- year ESL is very likely to increase by 34–76 cm and 58–172 cm, under RCP4.5 and RCP8.5, respectively between 2000–2100.

<span id="impacts"></span>
=== 6.3.2 Impacts ===

<div id="section-6-3-2impacts-block-1"></div>

As shown in previous assessments, increasing exposure is a major driver of increased cyclone risk (wind damages), as well as flood risk associated with cyclone rainfall and surge, besides possible changes in hazard intensities from anthropogenic climate change (Handmer et al., 2012 <sup>[[#fn:r273|273]]</sup> ; Arent et al., 2014 <sup>[[#fn:r274|274]]</sup> ). Changes in TC trajectories are potentially a major source of increased risk, as the degree of vulnerability is typically much higher in locations that were previously not exposed to the hazard (Noy, 2016). Typhoon Haiyan’s move to the south of the usual trajectories of TCs in the western North Pacific basin (Yonson et al., 2018 <sup>[[#fn:r275|275]]</sup> ) made the evacuation more difficult as people were less willing to heed storm surge warnings they received.

Abrupt changes in impacts therefore are not only determined by changes in cyclone hazard, but also by the sensitivity or tipping points that are crossed in terms of flooding for instance, that can be driven by SLR but also by changes in local exposure. The frequency of nuisance flooding along the US east coast is expected to accelerate further in the future (Sweet and Park, 2014 <sup>[[#fn:r276|276]]</sup> ). The loss of coral reef cover and mangrove forests have also been shown to increase damages from storm surge events (e.g., Beck et al., 2018). Cyclones also affect marine life, habitats and fishing. There is some evidence that fish may evacuate storm areas or be redistributed by storm waves and currents (FAO, 2018; Sainsbury et al., 2018 <sup>[[#fn:r277|277]]</sup> ). Other examples of damage to fisheries from cyclones and storm surges can be found in FAO (2018: Chapter V, Table 1).

With regard to property losses, according to most projections, increasing losses from more intense cyclones are not offset by a possible reduction in frequency (Handmer et al., 2012 <sup>[[#fn:r278|278]]</sup> ). While the relation between aggregate damages and frequency may be linear, the relationship between intensity and damages is most probably highly nonlinear; with research suggesting a 10% increase in wind speed associated with a 30–40% increase in damages (e.g., Strobl, 2012). Although it is clear that direct damages from cyclones could increase, investigations into the economic impact of past cyclone events is less common, as these are much more difficult to identify. Examples of such work include Strobl (2012) <sup>[[#fn:r279|279]]</sup> on hurricane impacts in the Caribbean, Haque and Jahan (2016) on TC Sidr in Bangladesh, Jakobsen (2012) on Hurricane Mitch in Nicaragua, and Taupo and Noy (2017) on TC Pam in Tuvalu. The relation between changes in TCs and property losses is complex, and there are indications that wind shear changes may have larger impact than changes in global temperatures (Wang and Toumi, 2016 <sup>[[#fn:r280|280]]</sup> ). With regard to loss of life, total fatalities and mortality from cyclone-related coastal flooding is globally declining, probably as a result of improved forecasting and evacuation, although in some low-income countries mortality is still high (Paul, 2009 <sup>[[#fn:r281|281]]</sup> ; Lumbroso et al., 2017 <sup>[[#fn:r282|282]]</sup> ; Bouwer and Jonkman, 2018 <sup>[[#fn:r283|283]]</sup> ). A global analysis finds that despite adaptation efforts, further SLR could increase storm surge mortality in many parts of the developing world (Lloyd et al., 2016 <sup>[[#fn:r284|284]]</sup> ).

An assessment of future changes in coastal impacts based on direct downscaling of indicators of flooding such as total water level and number of hours per year with breakwater overtopping over a given threshold for port operability is provided by Camus et al. (2017) <sup>[[#fn:r285|285]]</sup> . These indicators are multivariable and include the combined effect of SLR, storm surge, astronomical tide and waves. Regional projected wave climate is downscaled from global multi-model projections from 30 CMIP5 model realisations. For example, projections by 2100 under the RCP8.5 scenario show a spatial variability along the coast of Chile with port operability loss between 600–800 h yr –1 and around 200 h yr –1 relative to present (1979–2005) conditions. Although wave changes are included in projected overtopping distributions, future changes of operability are mainly due to the SLR contribution.

<span id="risk-management-and-adaptation"></span>
=== 6.3.3 Risk Management and Adaptation ===

<div id="section-6-3-3risk-management-and-adaptation-block-1"></div>

The most effective risk management strategy in the last few decades has been the development of early warning systems for cyclones (Hallegatte, 2013 <sup>[[#fn:r286|286]]</sup> ). Generally, however, a lack of familiarity with the changed nature of storms prevails. Powerful storms often generate record storm surges (Needham et al., 2015 <sup>[[#fn:r287|287]]</sup> ), such as in the cases of Cyclone Nargis and Typhoon Haiyan but surge warnings had been less well understood and followed because they had tended to be new or rare to the locality (Lagmay et al., 2015 <sup>[[#fn:r288|288]]</sup> ). A US study on storm surge warnings highlights the issue of the right timing to warn, as well as the difficulty in delivering accurate surge maps (Morrow et al., 2015 <sup>[[#fn:r289|289]]</sup> ). Previous experience with warnings that were not followed by hazard events show the ‘crying wolf’ problem leading many to ignore future warnings (Bostrom et al., 2018 <sup>[[#fn:r290|290]]</sup> ).

There is scant literature on the management of storms that follow less common trajectories. The most recent and relatively well-studied ones are Superstorm Sandy in 2012 in the USA and Typhoon Haiyan in 2013 in the Philippines. These two storms were unexpected and having underestimated the levels of impact, people ignored warnings and evacuation directives. In the case of Typhoon Haiyan, the dissemination of warnings via scripted text messages were ineffective without an explanation of the difference between Haiyan’s accompanying storm surge and that of other ‘normal’ storms to which people were used to (Lejano et al., 2016 <sup>[[#fn:r291|291]]</sup> ). Negative experiences of previous evacuations also lead to the reluctance of authorities to issue mandatory evacuation orders, for example, during Superstorm Sandy (Kulkarni et al., 2017 <sup>[[#fn:r292|292]]</sup> ), and contributes to a preventable high number of casualties (Dalisay and De Guzman, 2016 <sup>[[#fn:r293|293]]</sup> ). These examples also show that saving lives and assets through warning and evacuation is limited. Providing biophysical protection measures as well as improving self-reliance during such events can complement warning and evacuation.

After the storms, retreat or rebuild options exist. Rebuilding options can depend on whether insurance is still affordable after the event. Buyout programs, a form of ‘managed retreat’ whereby government agencies pay people affected by extreme weather events to relocate to safer areas, gained traction in recent years as a potential solution to reduce exposure to changing storm surge and flood risk. The decision to retreat or rebuild ''in situ'' depends, at least partially, on how communities have recovered in the past and therefore on the perceived success of a future recovery (Binder, 2014 <sup>[[#fn:r294|294]]</sup> ). However, political and jurisdictional conflicts between local, regional, and national government over land management responsibilities, lack of coordinated nation-wide adaptation plans, and clashes between individual and community needs have led to some unpopular buyout programs after Hurricane Sandy (Boet-Whitaker, 2017 <sup>[[#fn:r295|295]]</sup> ). Relocation (i.e., managed retreat) is often very controversial, can incur significant political risk even when it is in principle voluntary (Gibbs et al., 2016 <sup>[[#fn:r296|296]]</sup> ), and is rarely implemented with much success at larger scales (Beine and Parsons, 2015 <sup>[[#fn:r297|297]]</sup> ; Hino et al., 2017 <sup>[[#fn:r298|298]]</sup> ). In addition, managed retreats are often fraught with legal, distributional and human rights issues, as seen in the case of resettlements after Typhoon Haiyan (Thomas, 2015 <sup>[[#fn:r299|299]]</sup> ; see also Cross-Chapter Box 5 in Chapter 1), and extend to loss of cultural heritage and indigenous qualities in the case of small island states.

If rebuilding ''in situ'' is pursued after catastrophic events and without decreased exposure, it is often accompanied by actions that aim to reduce vulnerability in order to adapt to the increasing risk (Harman et al., 2013 <sup>[[#fn:r300|300]]</sup> ). In many cases, resilient designs and sustainable urban plans integrating climate change concerns, that are inclusive of vegetation barriers as coastal defences and hybrid designs, are considered (Cheong et al., 2013 <sup>[[#fn:r301|301]]</sup> ; Saleh and Weinstein, 2016 <sup>[[#fn:r302|302]]</sup> ). However, often more physical structures that are known to be less sustainable in the long-term, but potentially more protective in the short-term, are constructed (Knowlton and Rotkin-Ellman, 2014 <sup>[[#fn:r303|303]]</sup> ; Rosenzweig and Solecki, 2014 <sup>[[#fn:r304|304]]</sup> ). Anticipatory planning approaches are under way to warn and enable decision making in time (Bloemen et al., 2018 <sup>[[#fn:r305|305]]</sup> ; Lawrence et al., 2018 <sup>[[#fn:r306|306]]</sup> ).

<span id="marine-heatwaves-and-their-implications"></span>