Editing IPCC:AR6/WGII/Chapter-4 (section)

== 4.2 Observed Changes in the Hydrological Cycle Due to Climate Change ==

<div id="h1-3-siblings" class="h1-siblings"></div>

All components of the global water cycle have been modified due to climate change in recent decades ( ''high confidence'' ) ( [[#Douville--2021|Douville et al., 2021]] ), with hundreds of millions of people now regularly experiencing hydrological conditions that were previously unfamiliar (Sections 4.2.1.1, 4.2.4, 4.2.5). Extensive records from weather stations, satellites and radar clearly show that precipitation patterns have shifted worldwide. Three major shifts documented are (a) some regions receiving more annual or seasonal precipitation and others less, (b) many regions have seen increased heavy precipitation, and many have seen either increases or decreases in dry spells and (c) some regions have seen shifts towards heavier precipitation events separated by more prolonged dry spells ( [[#4.2.1.1|Section 4.2.1.1]] ). Observationally based calculations suggest that ET has changed in response to changes in precipitation and increasing temperatures, resulting in changing patterns of soil moisture worldwide which are now detectable by satellite remote sensing (Sections 4.2.1.2, 4.2.1.3). Rising temperatures have caused profound and extensive changes in the global cryosphere, with mountain glaciers, land ice and snow cover shrinking, causing substantial, permanent impacts on the ways of life of people in these regions, particularly Indigenous Peoples with strong cultural links to long-term or seasonally frozen environments (Sections 4.2.2, 4.3.8). Groundwater recharge in spring may have been reduced due to shorter snowmelt seasons, although the dominant impact on groundwater has been non-climatic and through intensification of irrigation ( [[#4.2.6|Section 4.2.6]] ). The global-scale pattern of streamflow changes is now attributable to observed historical climate change, with human land and water use insufficient by themselves to explain the observed streamflow changes at global scales ( [[#4.2.3|Section 4.2.3]] ). Numerous examples of extreme hydrometeorological events, including heavy precipitation, flooding, drought and wildfire events causing deaths, high levels of economic damage and extensive ecological impacts, have been shown to have been made more ''likely'' by human influence on climate through increased GHG concentrations in the atmosphere (Sections 4.2.1.1, 4.2.4, 4.2.5). Overall, there is a clear picture of human alteration of the global water cycle, which is now affecting societies and ecosystems across the world. This section describes changes in the hydrological cycle through a lens of societal impacts.

<div id="4.2.1" class="h2-container"></div>

<span id="observed-changes-in-precipitation-evapotranspiration-and-soil-moisture"></span>
=== 4.2.1 Observed Changes in Precipitation, Evapotranspiration and Soil Moisture ===

<div id="h2-3-siblings" class="h2-siblings"></div>

<div id="4.2.1.1" class="h3-container"></div>

<span id="observed-changes-in-precipitation"></span>
==== 4.2.1.1 Observed Changes in Precipitation ====

<div id="h3-1-siblings" class="h3-siblings"></div>

AR6 WGI ( [[#Douville--2021|Douville et al., 2021]] ) concluded that GHG forcing has driven increased contrasts in precipitation amounts between wet and dry seasons and weather regimes over tropical land areas ( ''medium confidence'' ), with a detectable precipitation increase in the northern high latitudes ( ''high confidence'' ). GHG forcing has also contributed to drying in dry summer climates, including the Mediterranean, southwestern Australia, southwestern South America, South Africa and western North America ( ''medium to high confidence'' ) (Figure 4.3). AR6 WGI ( [[#Seneviratne--2021|Seneviratne et al., 2021]] ) also concluded that the frequency and intensity of heavy precipitation events have ''likely'' increased at the global scale over most land regions with good observational coverage. Heavy precipitation has ''likely'' increased on the continental scale over North America, Europe and Asia. Regional increases in heavy precipitation frequency and (or) intensity have been observed with at least ''medium confidence'' for nearly half of the AR6 WGI climatic regions (Figure 4.3). Human influence, in particular GHG emissions, is ''likely'' the main driver of the observed global-scale intensification of heavy precipitation in land regions

<div id="_idContainer028" class="Figure"></div>

'''[[File:0c19bfccc496c614c6eb3417eaa9c58e IPCC_AR6_WGII_Figure_4_003.png]] Figure 4.3 |''' '''Observed mean and extreme precipitation changes and people experiencing the emergence of historically unfamiliar precipitation and changes in extreme precipitation.'''

'''(a)''' Percentage changes in annual mean precipitation over land (1891–2019) per °C global warming in the Global Precipitation Climatology Centre (GPCC) v2020 data set ( [[#Schneider--2017|Schneider et al., 2017]] ; [[#Schneider--2020|Schneider et al., 2020]] ). Green shows increasing precipitation; orange shows decreasing precipitation.

'''(b)''' Levels of unfamiliarity of wetter and drier climates, classified in terms of the ratio of the signal S of change to the noise N of variability, where the latter is defined as one standard deviation in annual data with the trend removed, that is, occurs approximately one in 6 years. Grey regions are either unobserved (oceans) or deserts (&lt;250 mm year –1 ). Stippling indicates where the signal of change is not significant. See [[#Hawkins--2020|Hawkins et al. (2020)]] for further details.

'''(c)''' Population densities in regions with annual precipitation classified as “emerging”.

'''(d)''' Precipitation trends from the GPCC data set in December, January and February (mm day –1 decade –1 ).

'''(e)''' As (d) for June-July-August.

'''(f)''' Changes in annual maximum 1-day precipitation (Rx1day) in the HadEX3 data set ( [[#Dunn--2020|Dunn et al., 2020]] ).

'''(g)''' Trend in annual mean consecutive dry days (CDD), 1950–2018, in HadEX3.

'''(h)''' Population densities per grid box where the trend in Rx1day is significantly different from zero.

'''(i)''' Population densities per grid box where the trend in CDD is significantly different from zero. Stipples in (h) and (i) show where HadEX3 data is available. Population data in (c), (h) and (i) are for 2020 from CIESIN (2018a; 2018b).

Large numbers of people live in regions where the annual mean precipitation is now ‘unfamiliar’ compared to the mean and variability between 1891 and 2016 (Figure 4.3c). “Unfamiliar” is defined as the long-term change being greater than one standard deviation in the annual data (Figure 4.3b). In 2020, approximately 498 million people lived in unfamiliarly wet areas, where the long-term average precipitation is as high as previously seen in only about one in 6 years ( ''medium confidence'' ) (Figure 4.3c). These areas are primarily in mid and high latitudes ( [[#Hawkins--2020|Hawkins et al., 2020]] ). On the other hand, approximately 163 million people lived in unfamiliarly dry areas, mostly in low latitudes ( ''medium confidence'' ). Due to high variability over time, the signal of long-term change in annual mean precipitation is not distinguishable from the noise of variability in many areas ( [[#Hawkins--2020|Hawkins et al., 2020]] ), implying that the local annual precipitation cannot yet be defined ‘unfamiliar’ by the above definition.

Notably, many regions have seen increased precipitation for part of the year and decreased precipitation at other times ( ''high confidence'' ) (Figure 4.3d,e), leading to small changes in the annual mean precipitation. Therefore, the numbers of people seeing unfamiliar seasonal precipitation levels are expected to be higher than those quoted above for unfamiliar annual precipitation changes ( ''medium confidence'' ). Still, quantified analysis of this is not yet available.

The intensity of heavy precipitation has increased in many regions ( ''high confidence'' ), including much of North America, most of Europe, most of the Indian sub-continent, parts of northern and southeastern Asia, much of southern South America, parts of southern Africa and parts of central, northern and western Australia (Figure 4.3 f) ( [[#Dunn--2020|Dunn et al., 2020]] ; [[#Sun--2020|Sun et al., 2020]] ). Conversely, heavy precipitation has decreased in some regions, including eastern Australia, northeastern South America and western Africa. The length of dry spells has also changed, with increases in annual mean consecutive dry days (CDD) in large areas of western, eastern and southern Africa, eastern and southwestern South America, and Southeast Asia, and decreases across much of North America (Figure 4.3g). Precipitation extremes have changed in some places where annual precipitation shows no trend. Some regions such as southern Africa and parts of southern South America are seeing increased heavy precipitation and longer dry spells. Many regions with changing extremes are highly populated, such as the Indian sub-continent, Southeast Asia, Europe and parts of North America, South America and southern Africa (Figure 4.3h,i). Substantially more people (~709 million) live in regions where annual maximum one-day precipitation has increased than in regions where it has decreased (~86 million) ( ''medium confidence'' ). However, more people are experiencing longer dry spells than shorter dry spells: approximately 711 million people live in places where annual mean CDD is longer than in the 1950s, and ~404 million in places with shorter CDD ( ''medium confidence'' ) (Figure 4.3i).

In summary, annual mean precipitation is increasing in many regions worldwide and decreasing over a smaller area, particularly in the tropics. Nearly half a billion people live in areas with historically unfamiliar wet conditions, and over 160 million in areas with historically unfamiliar dry conditions ( ''medium confidence'' ). Over 700 million people experience heavy precipitation significantly more intense than in the 1950s, but less than 90 million experience decreased heavy precipitation. Compared to the 1950s, 711 million people now experience longer dry spells and 404 million experience shorter dry spells.

<div id="4.2.1.2" class="h3-container"></div>

<span id="observed-and-reconstructed-changes-in-evapotranspiration"></span>
==== 4.2.1.2 Observed and Reconstructed Changes in Evapotranspiration ====

<div id="h3-2-siblings" class="h3-siblings"></div>

WGI ( [[#Douville--2021|Douville et al., 2021]] ) conclude with ''high confidence'' that global terrestrial annual ET has increased since the early 1980s, driven by both increasing atmospheric water demand and vegetation greening ( ''medium confidence'' ), and can be partly attributed to anthropogenic forcing ( ''high confidence)'' .

Regional changes in ET depend on changes in both the climate and the properties of the land surface and ecosystems. The latter also responds to changes in climate and atmospheric composition. For example, a warming climate increases evaporative demand (Huang M et al., 2015; [[#Berg--2016|Berg et al., 2016]] ), although seasonal rainfall totals ( [[#Hovenden--2014|Hovenden et al., 2014]] ) affect the amount of soil moisture available for evaporation. Since transpiration accounts for much of the land-atmosphere water flux ( [[#Good--2015|Good et al., 2015]] ), vegetation changes also play a significant role in overall changes in ET.

With higher CO 2 , the increase in evaporative demand can, to some extent, be counteracted by reduced stomatal conductance (‘physiological effect’), which reduces transpiration and increases leaf-level water use efficiency (WUE), but is highly species-specific. There is evidence for recent increases in leaf-scale WUE from tree rings (14 ± 10%, broadleaf to 22 ± 6%, evergreen over the 20th century: ( [[#Frank--2015|Frank et al., 2015]] )), carbon isotopes (30 to 35% increase in 150 years: ( [[#van%20der%20Sleen--2014|van der Sleen et al., 2014]] )), and satellite-based measurements (1982–2008) combined with data-driven models (Huang M et al., 2015). WUE is also affected by aerodynamic conductance ( [[#Knauer--2017|Knauer et al., 2017]] ), nutrient limitation ( [[#Medlyn--2015|Medlyn et al., 2015]] ; [[#Donohue--2017|Donohue et al., 2017]] ), soil moisture availability ( [[#Bernacchi--2015|Bernacchi and VanLoocke, 2015]] ; [[#Medlyn--2015|Medlyn et al., 2015]] ), and ozone pollution ( [[#King--2013|King et al., 2013]] ; [[#Frank--2015|Frank et al., 2015]] ).

Higher CO 2 also increases photosynthesis rates, though this may not be maintained in the longer term ( [[#Warren--2015|Warren et al., 2015]] ; [[#Adams--2020|Adams et al., 2020]] ), particularly where temperatures exceed the thermal maxima for photosynthesis (Duffy et al., 2021). Higher photosynthesis increases leaf area index (LAI) (‘structural effect’) and therefore transpiration; 55 ± 25% of observed increases in ET (1980–2011) have been attributed to LAI change (Zeng Z. et al., 2018). Increases in ET driven by increased LAI (from satellite observations 1982–2012) are estimated at 0.32 ± 0.07 mm month –1 per decade, generating a climate forcing of −0.31 Wm– 2 per decade ( [[#Zeng--2017|Zeng et al., 2017]] ).

Overall regional transpiration change depends on the balance between the physiological and structural effects (e.g., [[#Tor-ngern--2015|Tor-ngern et al., 2015]] ; [[#Ukkola--2015|Ukkola et al., 2015]] ). In dry regions, ET may increase due to increasing LAI (Huang M et al., 2015), but in some densely vegetated regions, the stomatal effect dominates ( [[#Mao--2015|Mao et al., 2015]] ). Reductions in transpiration due to rising CO 2 concentrations may also be offset by a longer growing season ( [[#Frank--2015|Frank et al., 2015]] ; [[#Mankin--2019|Mankin et al., 2019]] ). Other factors modulate the transpiration effect both temporally and spatially, for example, additional vegetation structural changes ( [[#Kim--2015|Kim et al., 2015]] ; [[#Domec--2017|Domec et al., 2017]] ), vegetation disturbance and age ( [[#Donohue--2017|Donohue et al., 2017]] ) and species ( [[#Bernacchi--2015|Bernacchi and VanLoocke, 2015]] ).

Recent studies report global ET increases from the early 1980s to 2009 and 2013 (Table 4.1). Calculations informed by observations suggest that ET has increased in most regions, with statistically significant (p&lt;0.05) trends of up to 10 mm yr -2 observed in large parts of North America and northern Eurasia. Larger increases in ET are also observed in several regions, including northeast Brazil, western central Africa, southern Africa, southern India, southern China, and northern Australia. Decreases of around 10 mm yr -2 are reported for western Amazonia and central Africa ( [[#Miralles--2014|Miralles et al., 2014]] ), although not across all data sets ( [[#Zeng--2018|Zeng et al., 2018]] ). In estimates of past changes in long-term drying or wetting of the land surface driven by climate, uncertainties in ET observations or reconstructions make a more substantial contribution to the overall uncertainty than observed changes in precipitation ( [[#Greve--2014|Greve et al., 2014]] ). Other changes in ET are also driven strongly by land cover changes and irrigation ( [[#Bosmans--2017|Bosmans et al., 2017]] ).

'''Table 4.1 |''' Trends in global evapotranspiration for different periods between 1981–1982 and 2009–2013.

{| class="wikitable"
|-
! Trend (mm yr -2 )

! Period

! Data source

! Author(s)

|-
| +0.54

| 1981 to 2012

| Observations

| (Zhang Y. et al., 2016)

|-
| +1.18

| 1982 to 2010

| Observations

| ( [[#Mao--2015|Mao et al., 2015]] )

|-
| +0.93 ± 0.31

| 1982 to 2010

| LSMs

| ( [[#Mao--2015|Mao et al., 2015]] )

|-
| +0.88

| 1982 to 2013

| Remote-sensing data

| (Zhang K. et al., 2015)

|-
| +1.5

| 1982 to 2009

| Remote-sensing and surface observations

| ( [[#Zeng--2014|Zeng et al., 2014]] )

|}

The contribution of changes in WUE to observed changes in ET is a key knowledge gap. WGI assigned ''low confidence'' to this contribution. Estimating large-scale transpiration response to increased CO 2 based on leaf-level responses of WUE is not straightforward ( [[#Bernacchi--2015|Bernacchi and VanLoocke, 2015]] ; [[#Medlyn--2015|Medlyn et al., 2015]] ; [[#Tor-ngern--2015|Tor-ngern et al., 2015]] ; [[#Walker--2015|Walker et al., 2015]] ; [[#Kala--2016|Kala et al., 2016]] ) and new methodological approaches are needed.

In summary, there is ''high confidence'' that ET increased by between approximately 0.5 and 1.5 mm yr -2 between the 1980s and early 2010s due to warming-induced increased atmospheric demand worldwide and greening of vegetation in many regions. Increases in many areas are 10 mm yr –2 or more, but in some tropical land areas, ET has decreased by 10 mm yr –2 . Plant stomatal responses to rising CO 2 concentrations may play a role, but there is ''low confidence'' in quantifying this. Changes in land cover and irrigation have also changed regional ET ( ''medium confidence'' ).

<div id="4.2.1.3" class="h3-container"></div>

<span id="observed-and-estimated-past-changes-in-soil-moisture-and-aridity"></span>
==== 4.2.1.3 Observed and Estimated Past Changes in Soil Moisture and Aridity ====

<div id="h3-3-siblings" class="h3-siblings"></div>

AR6 WGI ( [[#Douville--2021|Douville et al., 2021]] ) find that a global trend in soil moisture is detectable in a reanalysis and is attributable to GHG forcing, and conclude that it is ''very likely'' that anthropogenic climate change affected global patterns of soil moisture over the 20th century.

Changes in soil moisture and land surface aridity are due to changes in the relative balance of precipitation and ET. Soil moisture is also affected by irrigation. Regional trends derived from satellite remote sensing products show increases and decreases in annual surface soil moisture of up to 20% or more between the late 1970s and late 2010s (Figure 4.4). For example, using the ESA CCI SM v03.2 COMBINED products ( [[#van%20der%20Schalie--2021|van der Schalie et al., 2021]] ), approximately 0.9 billion people live in regions with decreasing surface soil moisture, and 2.1 billion people live in regions with increasing surface soil moisture (Figure 4.4, b). However, there are disagreements between data sets on the direction of change in some regions ( [[#Seneviratne--2010|Seneviratne et al., 2010]] ; [[#Feng--2015|Feng and Zhang, 2015]] ; [[#Feng--2016|Feng, 2016]] ), so quantification is subject to ''low confidence'' .

<div id="_idContainer032" class="Figure"></div>

[[File:8628d0bedc684eb085010919d5df685c IPCC_AR6_WGII_Figure_4_004.png]]

'''Figure 4.4 |''' '''Global patterns of changes in surface soil moisture and people in regions with significant changes.'''

'''(a)''' Percentage changes in annual mean surface soil moisture (0–5 cm) for 1978–2018 from satellite remote sensing: the “COMBINED” product of European Space Agency Climate Change Initiative Soil Moisture (ESA CCI SM v03.2), which blends data products from two microwave instruments, a scatterometer measuring radar backscattering and a radiometer measuring brightness temperature ( [[#van%20der%20Schalie--2021|van der Schalie et al., 2021]] ).

'''(b)''' The population density in 0.25° grid boxes with trends of significantly increasing and decreasing soil moisture from (a). Stippling indicates where changes are not significant.

Analysis of changes in P–ET estimates for 1948–2005 ( [[#Greve--2014|Greve et al., 2014]] ) suggests that geographical variations in soil moisture trends are more complex than the ‘wet get wetter, dry get drier’ (WGWDGD) paradigm. This is also supported by remote sensing data, with ESA CCI data for 1979–2013 showing only 15% of land following the WGWDGD paradigm for soil moisture ( [[#Feng--2015|Feng and Zhang, 2015]] ). Defining arid, humid and transitional areas according to precipitation and temperature regimes, all three classes of regions see more widespread trends of declining soil moisture than increasing soil moisture ( [[#Feng--2015|Feng and Zhang, 2015]] ). In the ESA CCI product, increasing soil moisture trends are mainly seen in humid or transitional areas and are rare in arid regions (Table 4.2)

'''Table 4.2 |''' Proportions of arid, transitional and humid areas with drying and wetting trends in surface soil moisture from remote sensing, 1979–2013 ( [[#Feng--2015|Feng and Zhang, 2015]] ).

{| class="wikitable"
|-
! Areas

! % of the area with a drying trend

! % of the area with a wetting trend

|-
| Arid

| 38.4

| 2.9

|-
| Transitional

| 13.0

| 10.5

|-
| Humid

| 16.3

| 8.1

|}

Reconstructions of historical soil moisture trends with data-driven models and process-based land surface models indicate drier dry seasons predominantly in extratropical latitudes, including Europe, western North America, northern Asia, southern South America, Australia and eastern Africa, consistent with climate model simulations of changes due to human-induced climate change ( [[#Padrón--2020|Padrón et al., 2020]] ). Furthermore, reduced water availability in the dry season is generally a consequence of increasing ET rather than decreasing precipitation ( [[#Padrón--2020|Padrón et al., 2020]] ).

While observationally based data for soil moisture are now more widely available, regional trends remain uncertain due to disagreements between data sets, so confident assessments of soil moisture changes remain a knowledge gap.

In summary, global mean soil moisture has slightly decreased, but regional changes vary, with both increases and decreases of 20% or more in some regions ( ''medium confidence'' ). Drying soil moisture trends are more widespread than wetting trends, not only in arid areas but also in humid and transitional areas ( ''medium confidence'' ). Reduced dry-season water availability is driven mainly by increasing transpiration ( ''medium confidence'' )

<div id="4.2.2" class="h2-container"></div>

<span id="observed-changes-in-the-cryosphere-snow-glaciers-and-permafrost"></span>
=== 4.2.2 Observed Changes in the Cryosphere (Snow, Glaciers and Permafrost) ===

<div id="h2-4-siblings" class="h2-siblings"></div>

AR5 reported a decrease in snow cover over most of the Northern Hemisphere, decreases in the extent of permafrost and increases in its average temperature, and glacier mass loss in most parts of the world ( [[#Jiménez%20Cisneros--2014|Jiménez Cisneros et al., 2014]] ). SROCC ( [[#IPCC--2019c|IPCC, 2019c]] ) stated with ''very high'' or ''high confidence'' (a) reduction in seasonal snow cover (snow cover extent decreased by 13.4% per decade for 1967–2018); (b) glacier mass budget of all mountain regions (excluding the Canadian and Russian Arctic, Svalbard, Antarctica, Greenland) was 490 ± 100 kg m –2 yr –1 in 2006–2015; (c) warming of permafrost (e.g., permafrost temperatures increased by 0.39°C in the Arctic for 2007–2017). Tourism and recreation activities have been negatively impacted by declining snow cover, glaciers and permafrost in high mountains ( ''medium confidence'' ).

Recent studies confirmed with ''high confidence'' that snow cover extent continues to decrease across the Northern Hemisphere in all months of the year (see [[#Douville--2021|Douville et al. (2021)]] ; [[#Eyring--2021|Eyring et al. (2021)]] ; [[#Fox-Kemper--2021|Fox-Kemper et al. (2021)]] for more details). From 1922 to 2018, snow cover extent in the Northern Hemisphere peaked in the 1950s to 1970s ( [[#Mudryk--2020|Mudryk et al., 2020]] ) and has consistently reduced since the end of the 20th century ( [[#Hernández-Henríquez--2015|Hernández-Henríquez et al., 2015]] ; [[#Thackeray--2016|Thackeray et al., 2016]] ; [[#Mudryk--2017|Mudryk et al., 2017]] ; [[#Beniston--2018|Beniston et al., 2018]] ; [[#Hammond--2018|Hammond et al., 2018]] ; [[#Thackeray--2019|Thackeray et al., 2019]] ; [[#Mudryk--2020|Mudryk et al., 2020]] ). The consistently negative snow-mass trend of approximately 5 Gt yr −1 in 1981–2018 for all winter-spring months ( [[#Mudryk--2020|Mudryk et al., 2020]] ), including 4.6 Gt yr −1 decrease of snow mass across North America and a negligible trend across Eurasia, has been observed ( [[#Pulliainen--2020|Pulliainen et al., 2020]] ). Negative trends in snow-dominated period duration of 2.0–6.5 weeks per decade was detected from surface and satellite observations during 1971–2014 ( [[#Allchin--2017|Allchin and Déry, 2017]] ), mainly owing to earlier seasonal snowmelt ( [[#Fox-Kemper--2021|Fox-Kemper et al., 2021]] ). The observed decrease of snow cover metrics (extent, mass, duration) led to changes in runoff seasonality and has impacted water supply infrastructure ( [[#Blöschl--2017|Blöschl et al., 2017]] ; [[#Huss--2017|Huss et al., 2017]] ), particularly in southwestern Russia, western USA and central Asia. In these regions, snowmelt runoff accounts for more than 30% of irrigated water supplies ( [[#Qin--2020|Qin et al., 2020]] ). Negative impacts on hydropower production due to changes in the seasonality of snowmelt have also been documented ( [[#Kopytkovskiy--2015|Kopytkovskiy et al., 2015]] ).

During the last two decades, the global glacier mass loss rate exceeded 0.5-meter water equivalent (m w.e.) per year compared to an average of 0.33 m w.e. yr –1 in 1950–2000. This volume of mass loss is the highest since the start of the entire observation period ( ''very high confidence'' ) ( [[#Zemp--2015|Zemp et al., 2015]] ; [[#Zemp--2019|Zemp et al., 2019]] ; [[#Hugonnet--2021|Hugonnet et al., 2021]] ) (also see Douville et al. 2021; Fox-Kemper et al. 2021; [[#Gulev--2021|Gulev et al. (2021)]] for more details). Regional estimates of glacier mass balance are also mostly negative ( [[#Dussaillant--2019|Dussaillant et al., 2019]] ; [[#Menounos--2019|Menounos et al., 2019]] ; [[#Zemp--2019|Zemp et al., 2019]] ; [[#Douville--2021|Douville et al., 2021]] ; [[#Fox-Kemper--2021|Fox-Kemper et al., 2021]] ; [[#Hugonnet--2021|Hugonnet et al., 2021]] ), except for West Kunlun, eastern Pamir and northern Karakoram ( [[#Brun--2017|Brun et al., 2017]] ; [[#Lin--2017|Lin et al., 2017]] ; [[#Berthier--2019|Berthier and Brun, 2019]] ). Changes in glacier metrics estimated in post-SROCC publications are summarised in Figure 4.5.

<div id="_idContainer034" class="Figure"></div>

[[File:e46a1e704a6a9f39cd5b3aa0ddd41c6b IPCC_AR6_WGII_Figure_4_005.png]]

'''Figure 4.5 |''' '''Global and regional estimates of changes in glacier characteristics (elevation, m yr''' '''–1''' '''; mass Gt yr''' '''–1''' ''', mass balance, m.''' '''w.e. yr''' '''–1''' ''') and 95% confidence intervals of the estimates.''' Results are taken from the post-SROCC publications, which are labelled in the chart titles as 1 – ( [[#Hugonnet--2021|Hugonnet et al., 2021]] ); 2 – ( [[#Yang--2020|Yang et al., 2020]] ); 3 – ( [[#Dussaillant--2019|Dussaillant et al., 2019]] ); 4 – ( [[#Davaze--2020|Davaze et al., 2020]] ); 5 – ( [[#Sommer--2020|Sommer et al., 2020]] ); 6 – ( [[#Schuler--2020|Schuler et al., 2020]] ).

Regional and global decreasing trends in glacier mass loss are about linear until 1990, after which they accelerated, especially in western Canada, the USA, and the southern Andes ( [[#WGMS--2017|WGMS, 2017]] ). There is a worldwide growth in the number, total area and total volume of glacial lakes by around 50% between 1990 to 2018 due to the global increase in glacier melt rate ( [[#Shugar--2020|Shugar et al., 2020]] ) ( [[#Shugar--2020|Shugar et al., 2020]] ) that can potentially increase risks of glacial lake outburst floods (GLOFs) with significant negative societal impacts ( [[#Ikeda--2016|Ikeda et al., 2016]] ). A drop in glacier runoff has happened in the regions where the glaciers have already passed their peak water stage, for example, in the Canadian Rocky Mountains, European Alps, tropical Andes and North Caucasus ( [[#Bard--2015|Bard et al., 2015]] ; [[#Hock--2019b|Hock et al., 2019b]] ; [[#Rets--2020|Rets et al., 2020]] ). There is ''medium confidence'' that the accelerated melting of glaciers has negatively impacted glacier-supported irrigation systems worldwide ( [[#Buytaert--2017|Buytaert et al., 2017]] ; [[#Nüsser--2017|Nüsser and Schmidt, 2017]] ; [[#Xenarios--2019|Xenarios et al., 2019]] ). Varying impacts on hydropower production ( [[#Schaefli--2019|Schaefli et al., 2019]] ) and tourism industry in some places due to cryospheric changes have also been documented ( [[#Hoy--2016|Hoy et al., 2016]] ; [[#Steiger--2019|Steiger et al., 2019]] ).

Permafrost changes mainly refer to changes in temperature and active layer thickness (ALT) ( [[#Hock--2019b|Hock et al., 2019b]] ; [[#Fox-Kemper--2021|Fox-Kemper et al., 2021]] ; [[#Gulev--2021|Gulev et al., 2021]] ). Permafrost temperature near the depth of zero annual temperature amplitude increased globally by 0.29 ± 0.12°C during 2007–2016, by 0.39 ± 0.15°C in the continuous permafrost and by 0.20 ± 0.10°C in the discontinuous permafrost ( [[#Biskaborn--2019|Biskaborn et al., 2019]] ). Thus, permafrost has been warming during the last 3–4 decades ( [[#Romanovsky--2017|Romanovsky et al., 2017]] ) with a rate of 0.4°C–1.4°C per decade throughout the Russian Arctic, 0.1°C–0.8°C per decade in Alaska and Arctic Canada during 2007–2016 ( [[#Biskaborn--2019|Biskaborn et al., 2019]] ) and 0.1°C–0.24°C per decade in the Tibetan plateau ( [[#Wu--2015|Wu et al., 2015]] ). The ALT has also been increasing in the European and Russian Arctic and high-mountain areas of Eurasia since the mid-1990s ( [[#Hock--2019b|Hock et al., 2019b]] ; [[#Fox-Kemper--2021|Fox-Kemper et al., 2021]] ; [[#Gulev--2021|Gulev et al., 2021]] ). Unfortunately, unlike glaciers and snow, the lack of ''in situ'' observations on permafrost still cannot be compensated for by remote sensing. Still, some methodological progress on this front has been happening recently ( [[#Nitze--2018|Nitze et al., 2018]] ).

There is ''high confidence'' that degradation of the cryospheric components is negatively affecting terrestrial ecosystems, infrastructure and settlements in the high-latitude and high-altitude areas ( [[#Fritz--2017|Fritz et al., 2017]] ; [[#Oliva--2018|Oliva and Fritz, 2018]] ; [[#Streletskiy--2019|Streletskiy et al., 2019]] ). Similarly, communities in the north polar regions and the ecosystems on which they depend for their livelihoods are at risk ( [[#Mustonen--2015|Mustonen, 2015]] ; [[#Pecl--2017|Pecl et al., 2017]] ; [[#Mustonen--2020|Mustonen and Lehtinen, 2020]] ) (Figure 4.6).

<div id="_idContainer036" class="Figure"></div>

[[File:8c95a164c7dac128be1225e29f014658 IPCC_AR6_WGII_Figure_4_006.png]]

'''Figure 4.6 |''' '''Map of selected observed impacts on cultural water uses of Indigenous Peoples of the cryosphere.''' Map location is approximate; text boxes provide names of the Indigenous Peoples whose cultural water uses have been impacted by climate change; changed climate variable; impact on water; and specific climate impact on cultural water use ( [[#4.3.7|Section 4.3.7]] ).

In summary, the cryosphere is one of the most sensitive indicators of climate change. There is ''high confidence'' that cryospheric components (glaciers, snow, permafrost) are melting or thawing since the end of the 20th and beginning of the 21st century. Widespread cryospheric changes are affecting humans and ecosystems in mid-to-high latitudes and the high-mountain regions ( ''high confidence'' ). These changes are already impacting irrigation, hydropower, water supply, cultural and other services provided by the cryosphere, and populations depending on ice, snow and permafrost.

<div id="4.2.3" class="h2-container"></div>

<span id="observed-changes-in-streamflow"></span>
=== 4.2.3 Observed Changes in Streamflow ===

<div id="h2-5-siblings" class="h2-siblings"></div>

AR5 ( [[#Jiménez%20Cisneros--2014|Jiménez Cisneros et al., 2014]] ) concluded with ''medium evidence'' and ''high agreement'' that trends in annual streamflow have generally followed observed changes in regional precipitation and temperature since the 1950s. AR6 WGI ( [[#Eyring--2021|Eyring et al., 2021]] ; [[#Gulev--2021|Gulev et al., 2021]] ) (12.4.5) conclude with ''medium confidence'' that anthropogenic climate change has altered local and regional streamflow in various parts of the world, but with no clear signal in the global mean.

Between the 1950s and 2010s, stream flows showed decreasing trends in parts of western and central Africa, eastern Asia, southern Europe, western North America and eastern Australia, and increasing trends in northern Asia, northern Europe, and northern and eastern North America ( [[#Dai--2016|Dai, 2016]] ; [[#Gudmundsson--2017|Gudmundsson et al., 2017]] ; [[#Gudmundsson--2019|Gudmundsson et al., 2019]] ; [[#Li--2020b|Li et al., 2020b]] ; [[#Masseroni--2020|Masseroni et al., 2020]] ). Significant spatial heterogeneity is also found in streamflow changes at the regional scale. Significant declines occurred at 11% of stations and significant increases at 4% of stations, with most decreases occurring in southern Canada ( [[#Bonsal--2019|Bonsal et al., 2019]] ). An increasing trend (1950–2010) is found in the northern region, mainly due to climate warming. Mixed trends are found in other regions.

The spatial differences in annual mean streamflow trends around the world are influenced by climatic factors, particularly changes in precipitation and evaporation ( [[#Zang--2013|Zang and Liu, 2013]] ; [[#Greve--2014|Greve et al., 2014]] ; [[#Hannaford--2015|Hannaford, 2015]] ; [[#Ficklin--2018|Ficklin et al., 2018]] ), as well as by anthropogenic forcing ( [[#Gudmundsson--2016|Gudmundsson et al., 2016]] ; 2017; 2021).. Other factors (e.g., land use change and CO 2 effects on vegetation) dominate in some areas, especially dryland regions ( [[#Berghuijs--2017b|Berghuijs et al., 2017b]] ). Human activities can reduce runoff through water withdrawal and land use changes ( [[#Zaherpour--2018|Zaherpour et al., 2018]] ; [[#Sun--2019a|Sun et al., 2019a]] ; [[#Vicente-Serrano--2019|Vicente-Serrano et al., 2019]] ), and human regulation of streamflows via impounding reservoirs can also play a major role ( [[#Hodgkins--2019|Hodgkins et al., 2019]] ).

Streamflow trends are attributed to varying combinations of climate change and direct human influence through water and land use in different basins worldwide, with conclusions on the relative contribution of climatic and anthropogenic factors sometimes depending on the methodology ( [[#Dey--2017|Dey and Mishra, 2017]] ). Precipitation explains over 80% of the changes in discharge of large rivers from 1950 to 2010 in northern Asia and northern Europe, where the impact of human activities is relatively limited ( [[#Li--2020b|Li et al., 2020b]] ). In northwest Europe, precipitation and evaporation changes explain many observed trends in streamflow ( [[#Vicente-Serrano--2019|Vicente-Serrano et al., 2019]] ). In several polar areas in northern Europe (e.g., Finland), North America (e.g., British Columbia in Canada) and Siberia, many studies reported increased winter streamflow primarily due to climate warming, for instance, more rainfall instead of snowfall and more glacier runoff in the winter period (e.g., [[#Bonsal--2020|Bonsal et al., 2020]] ) ( [[#4.2.2|Section 4.2.2]] ). A similar phenomenon of the earlier snowmelt runoff is also found in North America during 1960–2014 ( [[#Dudley--2017|Dudley et al., 2017]] ). Thus, climate drivers largely explain changes in the average and maximum runoff of predominantly snow-fed rivers ( [[#Yang--2015|Yang et al., 2015]] a; [[#Bring--2016|Bring et al., 2016]] ; [[#Tananaev--2016|Tananaev et al., 2016]] ; [[#Frolova--2017b|Frolova et al., 2017b]] ; [[#Ficklin--2018|Ficklin et al., 2018]] ; [[#Magritsky--2018|Magritsky et al., 2018]] ; [[#Rets--2018|Rets et al., 2018]] ).

In contrast, in southwestern Europe, land cover changes and increased water demands by irrigation are the main drivers of streamflow reduction ( [[#Vicente-Serrano--2019|Vicente-Serrano et al., 2019]] ) ( [[#4.3.1|Section 4.3.1]] ). In addition, the human intervention also contributed to the increase of the winter streamflow due to the release of water in the winter season for hydropower generation in large rivers in the northern regions ( [[#Rawlins--2021|Rawlins et al., 2021]] ). In some regions, the impact of human activities on runoff and streamflow outplays the climate factors, for example, in some typical catchments with area near to or less than 15000 km 2 in China ( [[#Zhai--2017|Zhai and Tao, 2017]] ).

[[#Shi--2019|Shi et al. (2019)]] found that in 40 major basins worldwide, both climatic and direct human impact contribute to observed flow changes to varying degrees. Climate change or variability is the main contributor to changes in basin-scale trends for 75% of rivers, while direct human effects on streamflow dominate for 25%. However, this does not consider attribution of the climate drivers to anthropogenic forcing. Using time series of low, mean and high river flows from 7250 observatories around the world (1971–2010) and global hydrological models (GHMs) driven by Earth System Model (ESM) simulations with and without anthropogenic forcing of climate change, [[#Gudmundsson--2021|Gudmundsson et al. (2021)]] also found direct human influence to have a relatively small impact on global patterns of streamflow trends. [[#Gudmundsson--2021|Gudmundsson et al. (2021)]] further identified anthropogenic climate change as a causal driver of the global pattern of recent trends in mean and extreme river flow (Figure 4.7). Overall, the sign of observed trends and simulations accounting for human influence on the climate system was found to be consistent for decreased mean flows in western and eastern North America, southern Europe, northeast South America and the Indian sub-continent, and increased flows in northern Europe. Similar conclusions were drawn for low and high flows, except for the Indian sub-continent. However, in some regions, the observed trend was opposite to that simulated with anthropogenic climate forcing. Thus, human water and land use alone did not explain the observed pattern of trends.

<div id="_idContainer038" class="Figure"></div>

[[File:985c07655e1d8baf026a9380d6ef56d8 IPCC_AR6_WGII_Figure_4_007.png]]

'''Figure 4.7 |''' '''Observed changes in river flows and attribution to externally forced climate change.'''

'''(a)''' Percentage changes in flow in individual rivers 1971 to 2010. Black box outlines show climatic regions with at least 80 gauging stations with almost complete daily observations over 1971–2010, using the SREX ( [[#Seneviratne--2012|Seneviratne et al., 2012]] ) regions.

'''(b)''' Left column: observed regional median trends from 1971 to 2010 in SREX regions with at least 80 gauging stations with almost complete daily observations over that period. Middle column: trends simulated by eight global hydrological models driven by four CMIP5 Earth System Models, with human water and land use from 1971 to 2020 and the pre-industrial control climate state. Right column: same as the middle column but with ESM-simulated climates from 1971 to 2010 with both anthropogenic forcings (greenhouse gases, aerosols and land use) and natural external forcings (solar variability and volcanic eruptions). Top row: low flows (annual 10th percentile). Middle row: mean flows. Bottom row: high flows (annual 90th percentile). Reproduced from [[#Gudmundsson--2021|Gudmundsson et al. (2021)]] .

Although there are different observational and simulated runoff and streamflow data sets (e.g., Global Runoff Data Centre, GRDC), it is still challenging to obtain and update long-term river discharge records in several regions, particularly Africa, South and East Asia ( [[#Dai--2016|Dai, 2016]] ). When observed data are scarce, hydrological models are used to detect trends in runoff and streamflow. However, simulations of streamflow can differ between models depending on their structures and parametrisations, contributing to uncertainties for trend detection, especially when considering human intervention (e.g., [[#Caillouet--2017|Caillouet et al., 2017]] ; [[#Hattermann--2017|Hattermann et al., 2017]] ; [[#Smith--2019b|Smith et al., 2019b]] ; [[#Telteu--2021|Telteu et al., 2021]] ).

In summary, both climate change and human activities influence the magnitude and direction of change in runoff and streamflow. There are no clear trends of changing streamflow on the global level. However, trends emerge on a regional level (a general increasing trend in the northern higher latitude region and mixed trend in the rest of the word) ( ''high confidence'' ). Climatic factors contribute to these trends in most basins ( ''high confidence'' ). They are more important than direct human influence in a larger share of major global basins ( ''medium confidence'' ), although direct human influence dominates in some ( ''medium confidence'' ). Overall, anthropogenic climate change is attributed as a driver to the global pattern of change in streamflow ( ''medium confidence'' ).

<div id="4.2.4" class="h2-container"></div>

<span id="observed-changes-in-floods"></span>
=== 4.2.4 Observed Changes in Floods ===

<div id="h2-6-siblings" class="h2-siblings"></div>

AR6 WGI [[IPCC:Wg2:Chapter:Chapter-11|Chapter 11]] ( [[#Seneviratne--2021|Seneviratne et al., 2021]] ) assessed with ''high confidence'' the increase in the extreme precipitation and associated increase in the frequency and magnitude of river floods. However, there is ''low confidence'' in changes in the river flooding regionally, which is strongly dependent upon complex catchment characteristics and land use patterns. SROCC ( [[#Hock--2019b|Hock et al., 2019b]] ) summarised with ''high confidence'' that changes in the cryosphere have led to changes in frequency, magnitude and location of rain-on-snow floods, snowmelt floods and glacier-related floods.

There is ''high confidence'' that the frequency and magnitude of river floods have changed in the past several decades in some regions mentioned below (and in WGI 11.5.2; SM4.1) with impacts across human and natural systems ( [[#4.3|Section 4.3]] ). A global flood database based on ''in situ'' measurement and satellite remote-sensing during 1985–2015 show that floods have increased 4-fold and 2.5-fold in the tropics and northern mid-latitudes, respectively ( [[#Najibi--2018|Najibi and Devineni, 2018]] ). Estimates of flood exposure using satellite-derived inundation area and high-resolution population data showed a 20–24% increase during 2000–2018 ( [[#Tellman--2021|Tellman et al., 2021]] ). Analyses of ''in situ'' streamflow measurement showed both increases and decreases in the frequency of river floods for 1960–2010 in Europe ( [[#Berghuijs--2017a|Berghuijs et al., 2017a]] ; [[#Blöschl--2019a|Blöschl et al., 2019a]] ) and the USA ( [[#Berghuijs--2017a|Berghuijs et al., 2017a]] ), an overall increase in China, Brazil and Australia ( [[#Berghuijs--2017a|Berghuijs et al., 2017a]] ) but decrease in some areas in the Mediterranean ( [[#Tramblay--2019|Tramblay et al., 2019]] ) and southern Australia ( [[#Ishak--2013|Ishak et al., 2013]] ; [[#Do--2017|Do et al., 2017]] ). Warming in the last 40–60 years has led to a 1–10-d earlier per decade spring flood occurrence depending on the location (the most frequent being 2–4 d per decade) ( ''high confidence'' ) (Yang L. et al., 2015; [[#Blöschl--2017|Blöschl et al., 2017]] ; [[#Dudley--2017|Dudley et al., 2017]] ; [[#Solander--2017|Solander et al., 2017]] ; [[#Rokaya--2018|Rokaya et al., 2018]] ; [[#Kireeva--2020|Kireeva et al., 2020]] ).

Between 1970 to 2019, 44% of all disasters and 31% of all economic losses were flood related ( [[#WMO--2021|WMO, 2021]] ). Observed flood risks changes in recent decades are often caused by human factors such as increased urbanisation and population growth rather than climate change alone ( [[#Tramblay--2019|Tramblay et al., 2019]] ). There is ''medium confidence'' that flood vulnerability varies among various regions and countries ( [[#Jongman--2012|Jongman et al., 2012]] ; [[#Scussolini--2016|Scussolini et al., 2016]] ; [[#Tanoue--2016|Tanoue et al., 2016]] ) (Figure 4.8), reflecting differences in GDP, severity and characteristics of hazard and political and social conditions ( [[#Rufat--2015|Rufat et al., 2015]] ). Flood vulnerability has decreased with economic development in many regions, while increased exposure has elevated risk in some places ( [[#Mechler--2016|Mechler, 2016]] ; [[#Tanoue--2016|Tanoue et al., 2016]] ). Global annual mean expected damage considering the current flood protection standard is estimated to be USD 54 million under the climate of 1976–2005 and unevenly distributed ( [[#Alfieri--2017|Alfieri et al., 2017]] ). Similar estimation using different models shows an increase of flood exposure in the past (USD 31 million for 1971–1990 and USD 45 million for 1991–2010 without population change as fixed in 2010) ( [[#Tanoue--2016|Tanoue et al., 2016]] ) ( [[#4.7.5|Section 4.7.5]] ).

<div id="_idContainer041" class="Figure"></div>

[[File:6ae64e0f771d823153bf40630200917f IPCC_AR6_WGII_Figure_4_008.png]]

'''Figure 4.8 |'''

'''(a)''' Modelled mean global fluvial flood water depth ( [[#Tanoue--2016|Tanoue et al., 2016]] ; [[#Tanoue--2021|Tanoue et al., 2021]] ) based on a land surface model and a river and inundation model driven by reanalysis climate forcing of five CMIP5 GCMs (metres). The annual maximum daily river water was allocated along elevations, and inundation depth was calculated for each year and averaged for the target period.

'''(b)''' Local flood protection standard (return period) at sub-country scale ( [[#Scussolini--2016|Scussolini et al., 2016]] ) based on published reports and documents, websites and personal communications with experts. Note that the vulnerability of this map reflects local flood protection such as complex infrastructure and does not fully reflect the other source of vulnerabilities, including exposure.

'''(c)''' Population distribution per 30 arc second grid cell ( [[#Klein%20Goldewijk--2010|Klein Goldewijk et al., 2010]] ; [[#Klein%20Goldewijk--2011|Klein Goldewijk et al., 2011]] ).

'''(d)''' Population exposed to flood (number of people where inundation occurs) per 30 arc-second grid cell. Population under inundation depth &gt; 0 m (a) was counted when the return period of annual maximum daily river water exceeds the flood protection standard (c) calculated by the authors. All values are averages for the period 1958–2010 for the past and 2050–2070 for the future.

The link between rainfall and flooding is complex. While observed increases in extreme precipitation have increased the frequency and magnitude of pluvial floods and river floods in some regions, floods could decrease in some regions due to other factors. These factors could include soil wetness condition, cryospheric change, land cover change and river system management, adaptation measures or water usage within the river basin (WGI FAQ8.2). For example, in the USA and Europe, a study indicated that major (e.g., 25–100-year return period) floods did not show significant long-term trends ( [[#Hodgkins--2019|Hodgkins et al., 2019]] ). Nevertheless, anthropogenic climate change increased the likelihood of a number of major heavy precipitation events and floods that resulted in disastrous impacts in southern and eastern Asia, Europe, North America and South America (Table 4.3) ( ''high confidence'' ). Davenport et al. (2021) demonstrated that anthropogenic changes in precipitation extremes had contributed one third of the cost of flood damages (from 1988 to 2017) in the USA. Anthropogenic climate change has altered 64% (eight out of 22 events increased, eight decreased) of floods events with significant losses and damages during 2010–2013 ( [[#Hirabayashi--2021a|Hirabayashi et al., 2021a]] ). [[#Gudmundsson--2021|Gudmundsson et al. (2021)]] attributed observed change in extreme river flow trends to anthropogenic climate change ( [[#4.2.3|Section 4.2.3]] ). Although there is growing evidence on the effects of anthropogenic climate change on each event, given the relatively poor regional coverage and high model uncertainty, there is ''low confidence'' in the attribution of human-induced climate change to flood change on the global scale.

'''Table 4.3 |''' Selected major heavy-precipitation events from 2014 to 2021 that led to flooding and their impacts. Studies were selected for presentation based on the availability of scientific literature with impacts information and do not necessarily represent the most severe events. Impactful events are included even if not found to have a component attributable to climate change. This is not a systematic assessment of event attributions studies and their physical science conclusions. ‘Sign of influence’ indicates whether anthropogenic climate change was found to have made the event ''more or less likely'' , and ‘mechanism/magnitude of influence’ quantifies the change in likelihood and the processes or quantities involved.

{| class="wikitable"
|-
! rowspan="2"| Year

! rowspan="2"| Country/region

! rowspan="2"| Impact

! colspan="2"| Anthropogenic climate change influence on the likelihood of an event

! rowspan="2"| Reference

|-
! Sign of influence

! Mechanism/magnitude of influence

|-
| 2021

| Germany, Belgium, Luxembourg and neighbouring countries

| At least 222 fatalities, substantial damage to transport and communications infrastructure and houses, severe disruption to businesses and livelihoods.,

| Increase

| One-day rainfall intensity increased by 3–19%, the likelihood of event increased by a factor between 1.2 and 9.

| [[#Kreienkamp--2021|Kreienkamp et al. (2021)]]

|-
| rowspan="2"| 2019

| Canada (Ottawa)

| Thousands of people evacuated, extended states of emergency, and about $200 million in insured losses

| Increase

| Spring maximum 30-d rainfall accumulation in 2019 was three times as likely with anthropogenic forcing.

| Kirchmeier-Young et al. (2021)

|-
| Southern China

| Over 6 million people across several southern China provinces were affected by heavy rains, floods and landslides. These extremes caused at least 91 deaths, collapsed over 19,000 houses, damaged around 83,000 houses and affected 419,400 ha of crops (China Ministry of Emergency Management 2020). The direct economic loss was estimated to be more than 20 billion RMB (equivalent to 3 billion USD)

| Decrease

| Anthropogenic forcings have reduced the likelihood of heavy precipitation in southern China like the 2019 March–July event by about 60%.

| [[#Li--2021b|Li et al. (2021b)]]

|-
| rowspan="5"| 2018

| USA (Mid-Atlantic)

| One fatality, $12 million damages

| Increase

| 1.1 to 2.3 times more likely

| [[#Winter--2020|Winter et al. (2020)]]

|-
| Central western China

| Persistent heavy rain led to floods, landslides and house collapse affecting 2.9 million people. The direct economic loss of over USD 1.3 billion.

| Decrease

| ~47% reduction in the probability

| [[#Zhang--2020b|Zhang et al. (2020b)]]

|-
| Northwestern China

| Extreme flooding in the Upper Yellow River basin affected about 1.4 million people and led to 30 deaths and disappearances.

| Decrease

| 34% reduction in the probability

| [[#Ji--2020|Ji et al. (2020)]]

|-
| Japan

| 237 fatalities, more than 6000 buildings destroyed by floods and landslides

| Increase

| 7% increase in total precipitation

| [[#Kawase--2020|Kawase et al. (2020)]]

|-
| Australia (Tasmania)

| $100 million in insurance claims

| Unknown

| Unknown

| [[#Tozer--2020|Tozer et al. (2020)]]

|-
| rowspan="4"| 2017

| Peru

| Widespread flooding and landslides affected 1.7 million people, 177 fatalities, estimated total damage of $3.1 billion

| Increase

| At least 1.5 times more likely

| Christidis et al. (2019)

|-
| Uruguay and Brazil

| Direct economic loss in Brazil of USD 102 million, displacement of more than 3500 people in Uruguay

| Increase

| At least double, with a most likely increase of about fivefold

| [[#de%20Abreu--2019|de Abreu et al. (2019)]]

|-
| North-East Bangladesh

| Flash flood affected ~850,000 households, ~220,000 ha of nearly harvestable Boro rice damaged. Crop failure contributed to a record 30% rice price hike compared to the previous year.

| Increase

| Doubled the likelihood of the 2017 pre-monsoon extreme 6-d rainfall event

| [[#Rimi--2019|Rimi et al. (2019)]]

|-
| China

| 7.8 million people affected 34 fatalities, about 0.8 million people displaced, 605,000 hectares of crops affected, 116,000 hectares without harvest. 32,000 houses collapsed, 41,000 were severely damaged. Direct economic loss 24.12 billion Chinese Yuan (~ USD 3.6 billion)

| Increase

| Doubled the probability from 0.6% to 1.2%

| [[#Sun--2019b|Sun et al. (2019b)]]

|-
| rowspan="4"| 2016

| South China

| Widespread severe flooding, waterlogging, and landslides in the Yangtze–Huai region.

| Increase

| 1.5-fold (0.6 to 4.7) increase in the probability

| [[#Sun--2018|Sun and Miao (2018)]]

|-
| China (Wuhan)

| 237 fatalities, 93 people missing, at least USD 22 billion in damage

| Increase

| Approximately 60% of the risk

| Zhou et al. (2018a)

|-
| China (Yangtze River)

| Direct economic loss of about USD 10 billion

| Increase

| Increased probability by 38% (± 21%)

| Yuan et al. (2018)

|-
| Australia

| Flooding and wild weather impacted some agriculture and power generation.

| None

| Minimal

| [[#Hope--2018|Hope et al. (2018)]]

|-
| 2015

| India (Chennai)

| City declared a disaster area. Damages estimated as $3 billion.

| None

| None

| [[#van%20Oldenborgh--2017a|van Oldenborgh et al. (2017a)]]

|-
| 2014

| Indonesia (Jakarta)

| 26 reported deaths, thousands of buildings flooded, much infrastructure damaged. Losses up to USD 384 million

| Unclear

| 2-d rain event approximately 2.4 times more likely compared to 1900, but cause not established

| [[#Siswanto--2015|Siswanto et al. (2015)]]

|}

In snow-dominated regions, 1~10 d earlier spring floods per decade due to warmer temperature are reported for the last decades ( ''high confidence'' ), such as in Europe ( [[#Morán-Tejeda--2014|Morán-Tejeda et al., 2014]] ; [[#Kormann--2015|Kormann et al., 2015]] ; [[#Matti--2016|Matti et al., 2016]] ; [[#Vormoor--2016|Vormoor et al., 2016]] ; [[#Blöschl--2017|Blöschl et al., 2017]] ), the European part of Russia ( [[#Frolova--2017a|Frolova et al., 2017a]] ; [[#Frolova--2017b|Frolova et al., 2017b]] ; [[#Kireeva--2020|Kireeva et al., 2020]] ), Canada (Yang L. et al., 2015; [[#Burn--2016|Burn et al., 2016]] ; [[#Rokaya--2018|Rokaya et al., 2018]] ) and the USA ( [[#Mallakpour--2015|Mallakpour and Villarini, 2015]] ; [[#Solander--2017|Solander et al., 2017]] ).

There is a knowledge gap in how ice-related floods, including glacier-related and ice-jam floods, respond to ongoing climate change. Despite the increase in the number of glacial lake studies ( [[#Wang--2017|Wang and Zhou, 2017]] ; [[#Harrison--2018|Harrison et al., 2018]] ; [[#Begam--2019|Begam and Sen, 2019]] ; [[#Bolch--2019|Bolch et al., 2019]] ), changes in the frequency of occurrence of glacier-related floods associated with climate change remain unclear ( ''medium confidence'' ). Studies show that the compound occurrence of high surges and high river discharge has increased in some regions (WGI Chapter 11), but few studies quantify changes and impacts. Increases in precipitation from tropical cyclones (WGI Chapter 11) and associated high tide are expected to exacerbate coastal flooding. However, more studies are required to quantify their impacts. In addition, limitations in the duration of data hinder the assessment of trends in low-likelihood high-impact flooding (WGI BOX 11.2).

In summary, the frequency and magnitude of river floods have changed in the past several decades with high regional variations ( ''high confidence'' ). Anthropogenic climate change has increased the likelihood of extreme precipitation events and the associated increase in the frequency and magnitude of river floods ( ''high confidence'' ). There is ''high confidence'' that the warming in the last 40–60 years has led to a maximum of 10 days earlier spring floods per decade, shifts in timing and magnitude of ice-jam floods and changes in frequency and magnitude of snowmelt floods.

<div id="4.2.5" class="h2-container"></div>

<span id="observed-changes-in-droughts"></span>
=== 4.2.5 Observed Changes in Droughts ===

<div id="h2-7-siblings" class="h2-siblings"></div>

There are different types of droughts, and they are interconnected in terms of processes ( [[#Douville--2021|Douville et al., 2021]] ). ''Meteorological droughts'' (periods of persistent low precipitation) propagate over time into deficits in soil moisture, streamflow and water storage, leading to a reduction in water supply ( ''hydrological drought'' ). Increased atmospheric evaporative demand increases plant water stress, leading to ''agricultural and ecological drought.''

Hydrological drought can result in shortages of drinking water and cause substantial economic damages. Agricultural drought threatens food production through crop damage and yield decreases (e.g., [[#4.3.1|Section 4.3.1]] ) ( ''high confidence'' ) and consequent economic impacts (Table 4.4). For example, drought in India in 2014 was reported to have led to an estimated USD 30 billion in losses ( [[#Ward--2018|Ward and Makhija, 2018]] ). Ecological drought increases the risks of wildfire (Table 4.4). Cascading effects of droughts can include health issues triggered by a lack of sanitation ( [[#4.3.3|Section 4.3.3]] ); can cause human displacements and loss of social ties, sense of place and cultural identity; and migration to unsafe settlements ( ''medium confidence'' ) ( [[#Serdeczny--2017|Serdeczny et al., 2017]] ) ( [[#4.3.7|Section 4.3.7]] ). Between 1970 and 2019, only 7% of all disaster events were drought-related, yet they contributed disproportionately to 34% of disaster-related death, mostly in Africa ( [[#WMO--2021|WMO, 2021]] ). Nevertheless, IK, TK and LK have increased drought resilience among crop and livestock farmers, for example, in South Africa ( [[#Muyambo--2017|Muyambo et al., 2017]] ), Uganda ( [[#Mfitumukiza--2020|Mfitumukiza et al., 2020]] ) and India ( [[#Patel--2020|Patel et al., 2020]] ) ( [[#4.8.4|Section 4.8.4]] ).

'''Table 4.4 |''' Selected major drought events from 2013 to 2020 and their societal impact. Studies were selected for presentation based on the availability scientific literature impacts information and do not necessarily represent the most severe events.Impactful events are included even if not found to have a component attributable to climate change. This is not a systematic assessment of event attributions studies and their physical science conclusions. ‘Sign of influence’ indicates whether anthropogenic climate change was found to have made the event more or less likely , and ‘mechanism/magnitude of influence’ quantifies the change in likelihood and the processes or quantities involved.

{| class="wikitable"
|-
! rowspan="2"| Year

! rowspan="2"| Country/region

! rowspan="2"| Impact

! colspan="2"| Influence of anthropogenic climate change on the likelihood of an event

! Reference

|-
! ''Sign of influence''

! ''Mechanism/magnitude of influence''

!
|-
| 2019/2020

| Australia

| Wildfires burning ~97,000 km 2 across southern and eastern Australia; 34 human fatalities; 5900 buildings destroyed; millions of people affected by hazardous air quality; between 0.5 and 1.5 billion wild animals and tens of thousands of livestock killed; at least 30% of habitat affected for seventy taxa, including 21 already listed as threatened with extinction, over USD 110 billion financial loss

| Increase

| Extreme high temperatures causing drying of fuel. The likelihood of extreme heat at least doubled due to the long-term warming trend, and the likelihood of Fire Weather Index as severe or worse as observed in 2019/2020 by at least 30%, despite no attributable increase in meteorological (precipitation) drought.

| [[#van%20Oldenborgh--2020|van Oldenborgh et al. (2020)]] ; [[#Ward--2020|Ward et al. (2020)]] ; [[#Haque--2021|Haque et al. (2021)]]

|-
| rowspan="5"| 2019

| Western Cape, South Africa

| Water supply was reduced to 20% of capacity in January 2018. Agricultural yields in 2019 declined by 25%.

| Increase

| Anthropogenic greenhouse forcing at least doubled the likelihood of drought levels seen in 2015–2019, offsetting anthropogenic aerosol forcing.

| [[#Kam--2021|Kam et al. (2021)]]

|-
| Yunnan, southwestern China

| Water scarcity affected nearly 7 million residents and resulted in crop failure over at least 1.35 × 10 4 km 2 cropland. More than 94% of the total area in the province was drought-stricken, and around 2 million people faced drinking water shortages, with a direct economic loss of about 6.56 billion RMB.

| Increase

| Anthropogenic influence increased the risk of 2019 March–June hot and dry extremes over Yunnan province in southwestern China by 123–157% and 13–23%, respectively.

| Wang et al. (2021b)

|-
| Southwestern China

| Over 640,100 hectares of crops with rice, corn and potatoes were extensively damaged. Over 100 rivers and 180 reservoirs dried out. Over 824,000 people and 566,000 head of livestock experienced a severe lack of drinking water, with a direct economic loss of 2.81 billion Chinese yuan (USD 400 million).

| Increase

| Anthropogenic forcing has likely increased the likelihood of the May–June 2019 severe low-precipitation event in southwestern China by approximately 1.4 to 6 times.

| [[#Lu--2021|Lu et al. (2021)]]

|-
| South China

| A lightning-caused forest fire in Muli County killed 31 firefighters and burned about 30 ha of forest.

| Increase

| Anthropogenic global warming increased the weather-related risk of extreme wildfire by 7.2 times. In addition, the El Niño event increased risk by 3.6 times.

| Du et al. (2021)

|-
| Middle and lower reaches of the Yangtze River, China

| Reduced agriculture productivity and increased load on power system supplies and transportations, and on human health

| Decrease

| Anthropogenic forcing reduced the probability of rainfall amount in the extended rainy winter of 2018/2019 by ~19%, but exerted no influence on the excessive rainy days.

| Hu et al. (2021)

|-
| 2018

| South China

| Shrinking reservoirs, water shortages. Area and yield for early rice reduced by 350 thousand hectares and 1.28 million tons relative to 2017

| Increase

| Likelihood increased by 17 times in the HadGEM3-A model. However, the event did not occur without human influence in the CAM5 model.

| [[#Zhang--2020|Zhang et al. (2020)]]

|-
|
| China (Beijing)

| A record 145 consecutive dry days (CDD), severe drought, increased risk of wildfires

| Increase

| The likelihood of the record 145 CDD was increased by between 1.29 and 2.09 times by anthropogenic climate change and between 1.43 and 4.59 times by combining the La Niña event and a weak Arctic polar vortex.

| Du et al. (2021)

|-
| rowspan="2"| 2017

| USA (Northern Great Plains)

| “billion-dollar disaster”; widespread wildfires (one of Montana’s worst wildfire seasons on record) compromised water resources, destruction of property, livestock sell-offs, reduced agricultural production, agricultural losses of USD 2.5 billion

| Increase

| 1.5 times more likely due to increased ET (minimal anthropogenic impact on precipitation)

| [[#Hoell--2019|Hoell et al. (2019)]]

|-
| East Africa

| Extensive drought across Tanzania, Ethiopia, Kenya and Somalia contributed to extreme food insecurity approaching near-famine conditions.

| Increase

| Likelihood doubled

| [[#Funk--2019|Funk et al. (2019)]]

|-
| 2016

| Southern Africa

| Millions of people were affected by famine, disease and water shortages. In addition, a 9-million-tonne cereal deficit resulted in 26 million people in need of humanitarian assistance.

| Increase

| Anthropogenic climate change ''likely'' increased the intensity of the 2015/2016 El Niño, and a drought of this severity would have been very unlikely (probability ~9%) in the pre-industrial climate.

| [[#Funk--2018|Funk et al. (2018)]]

|-
| 2016

| Brazil

| Três Marias, Sobradinho, and Itaparica reservoirs reached 5% of volume capacity. Ceará registered 39 (of 153) reservoirs empty. Another 42 reached inactive volume; 96 (of 184) Ceará municipalities experienced water supply interruption.

| Not found

| Not found

| [[#Martins--2018|Martins et al. (2018)]]

|-
| 2016

| Thailand

| Severe drought affected 41 Thai provinces, had devastating effects on major crops, such as rice and sugar cane, and incurred a total loss in the agricultural production of about half a billion USD.

| Increase

| The record temperature of April 2016 in Thailand would not have occurred without the influence of both anthropogenic forcings and El Niño. Anthropogenic forcing has contributed to drier Aprils, but El Niño was the dominant cause of low rainfall.

| [[#Christidis--2018|Christidis et al. (2018)]]

|-
| 2015

| Washington state, USA

| USD 335 million loss for the agricultural industry

| Increase

| Snowpack drought resulted from exceedingly high temperatures despite normal precipitation

| Fosu et al. (2016)

|-
| 2014

| São Paulo, Brazil

| In January 2015, the largest water supply system used for Sao Paulo, Cantareira, sank to a water volume of just 5% of capacity, and the number of people supplied fell from 8.8 million people to 5.3 million people, with other systems taking over supplies for the remainder.

| No impact

| Anthropogenic climate change is not found to be a major influence on the hazard, whereas increasing population and water consumption increased vulnerability.

| [[#Otto--2015|Otto et al. (2015)]]

|-
| 2014

| Southern Levant, Syria

| While the extent to which the 2007/2008 drought in the Levant region destabilised the Syrian government was not clear, ‘there is no questioning the enormous toll this extreme event took on the region’s population. The movement of refugees from both the drought and war-affected regions into Jordan and Lebanon ensured that the anomalously low precipitation in the winter of 2013/2014 amplified impacts on already complex water and food provisions.’

| Increase

| The persistent drought in the 2014 rainy season

was unprecedented for the critical January–February period in the observational

record, and was made ~45% more likely by anthropogenic climate change.

| [[#Bergaoui--2015|Bergaoui et al. (2015)]]

|-
| 2013–2014

| Mediterranean coastal Middle East, northward through Turkey and eastward through Kazakhstan, Uzbekistan and Kyrgyzstan

| The eastern (main) basin of the Aral Sea dried up for the first time in modern history.

| Unclear

| High western Pacific sea surface temperatures (SSTs) linked to drought in the Middle East and central-southwest Asia, and the SSTs in that region showed a strong warming trend.

| [[#Barlow--2015|Barlow and Hoell (2015)]]

|-
| 2014

| East Africa

| Some isolated food security crises

| Increase

| Anthropogenic warming contributed to the 2014 East African drought by increasing East African and west Pacific temperatures, and increasing the gradient between standardised western and central Pacific SST, causing reduced rainfall, ET and soil moisture.

| [[#Funk--2018|Funk et al. (2018)]]

|}

When hazard, vulnerability and exposure are considered together, drought risk is lower for sparsely populated regions, such as tundra and tropical forests, and higher for populated areas and intensive crop and livestock farming regions, such as southern and central Asia, southeastern South America, central Europe and the southeastern USA (Figure 4.9). Dynamics in exposure and vulnerability are rarely addressed ( [[#Jurgilevich--2017|Jurgilevich et al., 2017]] ; [[#Hagenlocher--2019|Hagenlocher et al., 2019]] ). Quantifying economic vulnerability to drought in terms of damages as a percentage of exposed GDP, [[#Formetta--2019|Formetta and Feyen (2019)]] show a disproportionate burden of drought impact on low-income countries, but with a clear decrease in global economic drought vulnerability between 1980–1989 and 2007–2016, including a convergence between lower-income and higher-income countries due to stronger vulnerability reduction in less-developed countries. Nevertheless, during 2007–2016, economic vulnerability to drought was twice as high in lower-income countries compared to higher-income countries ( [[#Formetta--2019|Formetta and Feyen, 2019]] ).

<div id="_idContainer043" class="Figure"></div>

[[File:81039aa792ccfc2cbcb954f48dfc8b65 IPCC_AR6_WGII_Figure_4_009.png]]

'''Figure 4.9 |''' '''Current global drought risk and its components.'''

'''(a)''' Drought hazard computed for the events between 1901 and 2010 by the probability of exceedance the median of global severe precipitation deficits, using precipitation data from the Global Precipitation Climatology Centre (GPCC) for 1901–2010.

'''(b)''' Drought vulnerability is derived from an arithmetic composite model combining social, economic and infrastructural factors proposed by the United Nations International Strategy for Disaster Risk Reduction ( [[#UNISDR--2004|UNISDR, 2004]] ).

'''(c)''' Drought exposure computed at the sub-national level with the non-compensatory Data Envelopment Analysis (DEA) model ( [[#Cook--2014|Cook et al., 2014]] ).

'''(d)''' Drought risk based on the above components of hazard, vulnerability and exposure, scored on a scale of 0 (lowest risk) to 1 (highest risk) with the lowest and highest hazard, exposure and vulnerability ( [[#Carrão--2016|Carrão et al., 2016]] ).

AR6 WGI ( [[#Douville--2021|Douville et al., 2021]] ; [[#Seneviratne--2021|Seneviratne et al., 2021]] ) found that increasing agricultural and ecological droughts trends are more evident than increasing trends in meteorological drought in several regions due to increased evaporative demand. Therefore, WGI concluded with ''high confidence'' that the increased frequency and the severity of agricultural/ecological droughts over the last decades in the Mediterranean and western North America can be attributed to anthropogenic warming.

In addition, there is ''high confidence'' in anthropogenic influence on increased meteorological drought in southwestern Australia and ''medium confidence'' that recent drying and severe droughts in southern Africa and southwestern South America can be attributed to human influence. Increased agricultural/ecological and (or) meteorological and (or) hydrological drought is also seen with either ''medium confidence'' or ''high confidence'' in the trend but with ''low confidence'' on attribution to anthropogenic climate change in western, northeastern and central Africa; central, eastern and southern Asia; eastern Australia; southern and northeastern South America and the South American monsoon region; and western and central Europe. Finally, decreased drought in one or more categories is seen with ''medium confidence'' in western and eastern Siberia; northern and central Australia; southeastern South America; central North America and northern Europe, but with ''low confidence'' in attribution to anthropogenic influence, except in northern Europe, where anthropogenic influence on decreased meteorological drought is assessed with ''medium confidence'' .

Major drought events worldwide have had substantial societal and ecological impacts, including reduced crop yields, shortages of drinking water, wildfires causing deaths of people and very large numbers of animals, impacting the habitats of threatened species, and widespread economic losses (Table 4.4, Cross-Chapter Box DISASTER in Chapter 4). In addition, anthropogenic climate change was found to have increased the likelihood or severity of most such events examined in event attribution studies.

Although long-term drought trends are clearer for agricultural or ecological drought compared to meteorological droughts ( [[#Douville--2021|Douville et al., 2021]] ; [[#Seneviratne--2021|Seneviratne et al., 2021]] ), most attribution studies for individual extreme events focus on meteorological (precipitation) drought and sometimes also consider temperature anomalies. A complete examination of drought relevant to societal impacts often requires consideration of hydrological and agricultural drought, so extreme event attribution conclusions relating to precipitation alone may not fully capture the processes leading to societal effects. There is, therefore, a critical knowledge gap in the attribution of changes in drought indicators more closely related to societal impacts such as soil moisture and the availability of fresh water supplies.

In summary, droughts can have substantial societal impacts ( ''virtually certain'' ), and agricultural and ecological drought conditions in particular have become more frequent and severe in many parts of the world but less frequent and severe in some others ( ''high confidence'' ). Drought-induced economic losses relative to GDP are approximately twice as high in lower-income countries compared to higher-income countries, although the gap has narrowed since the 1980s, and at the global scale there is a decreasing trend of economic vulnerability to drought ( ''medium confidence'' ). Nevertheless, anthropogenic climate change has contributed to the increased likelihood or severity of drought events in many parts of the world, causing reduced agricultural yields, drinking water shortages for millions of people, increased wildfire risk, loss of lives of humans and other species and loss of billions of dollars of economic damages ( ''medium confidence'' ).

<div id="4.2.6" class="h2-container"></div>

<span id="observed-changes-in-groundwater"></span>
=== 4.2.6 Observed Changes in Groundwater ===

<div id="h2-8-siblings" class="h2-siblings"></div>

AR5 concluded that the extent to which groundwater abstractions are affected by climate change is not well known due to the lack of long-term observational data ( [[#Jiménez%20Cisneros--2014|Jiménez Cisneros et al., 2014]] ). AR6 ( [[#Douville--2021|Douville et al., 2021]] ) confirmed that, despite considerable progress since AR5, limitations in the spatio-temporal coverage of groundwater monitoring networks, abstraction data and numerical representations of groundwater recharge processes continue to constrain understanding of climate change impacts on groundwater.

Globally, groundwater use has societal and economic benefits, providing a critical buffer against precipitation variability. Groundwater irrigation has ensured food security, livelihood support and poverty alleviation, for example, in India ( [[#Sekhri--2014|Sekhri, 2014]] ), Bangladesh ( [[#Salem--2018|Salem et al., 2018]] ) and sub-Saharan Africa ( [[#Taylor--2013a|Taylor et al., 2013a]] ; [[#Cuthbert--2019b|Cuthbert et al., 2019b]] ). Groundwater is a safe drinking water source during natural hazard-induced disasters ( [[#Richts--2016|Richts and Vrba, 2016]] ). However, groundwater over-exploitation leads to the attenuation of societal benefits, including reduced agricultural production ( [[#Asoka--2020|Asoka and Mishra, 2020]] ; [[#Jain--2021|Jain et al., 2021]] ), decrease in adaptive capacity of communities ( [[#Blakeslee--2020|Blakeslee et al., 2020]] ) and water quality deterioration ( [[#Mas-Pla--2019|Mas-Pla and Menció, 2019]] ). Loss of traditional water systems based on groundwater, such as ''foggara'' in Tunisia ( [[#Mokadem--2018|Mokadem et al., 2018]] ), ''qanat'' in Pakistan ( [[#Mustafa--2008|Mustafa and Usman Qazi, 2008]] ), ''aflaj'' in Oman ( [[#Remmington--2018|Remmington, 2018]] ) and spring boxes in the Himalayas ( [[#Kumar--2018|Kumar and Sen, 2018]] ), also leads to loss of cultural values for local communities.

Even though global groundwater abstraction (789 ± 30 km3 yr−1) is just about 6% of the annual recharge (~13,466 km 3 ) ( [[#Hanasaki--2018|Hanasaki et al., 2018]] ), a few hotspots of groundwater depletion have emerged at local to regional scales since the end of 20th century to the beginning of the 21st century due to intensive groundwater use for irrigation. The variability in groundwater storage is a function of human abstraction and natural recharge, which is in turn controlled by local geology ( [[#Green--2016|Green, 2016]] ). In humid regions, precipitation influences recharge, and linear associations between precipitation and recharge are often observed ( [[#Kotchoni--2019|Kotchoni et al., 2019]] ); for example, over humid locations in sub-Saharan Africa ( [[#Cuthbert--2019b|Cuthbert et al., 2019b]] ).

A global review ( [[#Bierkens--2019|Bierkens and Wada, 2019]] ) of groundwater storage changes highlights that estimates of depletion rates at the global scale are variable. These estimates range from approximately 113 to 510 km 3 yr −1 and variation in estimates is due to methods and spatio-temporal scales considered ( ''high confidence'' ). Global hydrological models ( [[#Herbert--2019|Herbert and Döll, 2019]] ) show that human-induced groundwater depletion at rates exceeding 20 mm yr –1 (2001–2010) is occurring in the major aquifers systems such as the High Plains and California Central Valley aquifers (USA), Arabian aquifer (Middle East), North-Western Sahara Aquifer System (North Africa), Indo-Gangetic Basin (India) and North China Plain (China) ( ''high confidence'' ). Groundwater depletion at lower rates (&lt;10 mm yr –1 ) is taking place in the Amazon Basin (Brazil) and Mekong River Basin (South East Asia), primarily due to climate variability and change ( ''high confidence'' ). A global-scale analysis ( [[#Shamsudduha--2020|Shamsudduha and Taylor, 2020]] ) of GRACE satellite measurements (2002–2016) for the 37 world’s large aquifer systems reveals that trends in groundwater storage are mostly nonlinear and declines are not secular ( ''high confidence'' ). There are strong statistical associations between changes in groundwater storage and extreme annual precipitation from 1901 to 2016 in the Great Artesian Basin (Australia) and the California Central Valley aquifer (USA). Groundwater recharge of high magnitudes can be generated from intensive precipitation events. On the other hand, recharge can become more episodic, mostly in arid to semiarid locations ( ''robust evidence, medium agreement'' ). For example, in central Tanzania, seven rainfall events between 1955 and 2010 generated 60% of total recharge ( [[#Taylor--2013b|Taylor et al., 2013b]] ). Similarly, in southern India ( [[#Asoka--2018|Asoka et al., 2018]] ) and the southwestern USA ( [[#Thomas--2016|Thomas et al., 2016]] ), focused recharge via losses from ephemeral river channels, overland flows, and floodwaters is documented ( [[#Cuthbert--2019b|Cuthbert et al., 2019b]] ).

In cold regions, where snowmelt dominates the local hydrological processes, Irannezhad et al. (2016) and Vincent et al. (2019) show high recharge to aquifers from glacial meltwater, while [[#Nygren--2020|Nygren et al. (2020)]] report a decrease in groundwater recharge due to a shift in main recharge period from spring (snowmelt) to winter (rainfall). In Finland, a sustained reduction (almost 100 mm in 100 years) of long-term snow accumulation combined with early snowmelt has reduced spring recharge ( [[#Irannezhad--2016|Irannezhad et al., 2016]] ) ( ''medium confidence'' ).

Data from ground-based long-term records in the Indo-Gangetic Basin reveals that sustainable groundwater supplies are constrained more by extensive contamination (e.g., arsenic, salinity) than depletion ( [[#MacDonald--2016|MacDonald et al., 2016]] ). Many low-lying coastal aquifers are contaminated with increased salinity due to land use change, rising sea levels, reduced stream flows and increased storm surge inundation ( [[#Lall--2020|Lall et al., 2020]] ). Nearly 26 million people are currently exposed to very high (&gt;1500 μ S cm –1 ) salinity in shallow groundwater in coastal Bangladesh ( [[#Shamsudduha--2020|Shamsudduha and Taylor, 2020]] ).

Groundwater-dependent ecosystems (GDEs), such as terrestrial wetlands, stream ecosystems and estuarine and marine ecosystems ( [[#Kløve--2014|Kløve et al., 2014]] ), support wetlands and biodiversity, provide water supply and baseflows to rivers, offer recreational services and help control floods ( [[#Rohde--2017|Rohde et al., 2017]] ). Globally, 10–23% of the watersheds have reached the environmental flow limits due to groundwater pumping ( [[#de%20Graaf--2019|de Graaf et al., 2019]] ). A recent study of 4.2 million wells across the USA shows that induced groundwater recharge in nearly two thirds of these wells could reduce stream discharges, thereby threatening GDEs ( [[#Jasechko--2021|Jasechko et al., 2021]] ). [[#Work--2020|Work (2020)]] found reduced spring flow due to increased groundwater abstraction in 26 out of 56 springs studied in Florida (USA). GDEs in semiarid and arid regions tend to have much longer groundwater response times and may be more resilient to climate change than those in humid areas where groundwater occurrence is mostly at shallow levels ( [[#Cuthbert--2019a|Cuthbert et al., 2019a]] ; [[#Opie--2020|Opie et al., 2020]] ). However, groundwater depletion impacts on the full range of ecosystem services remain understudied ( [[#Bierkens--2019|Bierkens and Wada, 2019]] ).

A better understanding of and incorporating subsurface storage dynamics into ESMs will improve climate–groundwater interactions under global warming ( [[#Condon--2020|Condon et al., 2020]] ). Long-term groundwater-level monitoring data are of critical importance ( [[#Famiglietti--2014|Famiglietti, 2014]] ) for understanding the sensitivity of recharge processes to climate variability and, more critically, calibration and validation of hydrological models ( [[#Goderniaux--2015|Goderniaux et al., 2015]] ). GRACE satellite-derived groundwater storage estimates provide important insights at a regional scale ( [[#Rodell--2018|Rodell et al., 2018]] ) but overlook more localised depletion or short-term storage gains. Low- and middle-income countries such as central Asia and sub-Saharan Africa lack such monitoring networks, which is a significant knowledge gap.

In summary, groundwater storage has declined in many parts of the world, most notably since the beginning of the 21st century, due to the intensification of groundwater-fed irrigation ( ''high confidence'' ). Groundwater in aquifers across the tropics appears to be more resilient to climate change as enhanced recharge is observed to occur mostly episodically from intense precipitation and flooding events ( ''robust evidence, medium agreement'' ). In higher altitudes, warmer climates have altered groundwater regimes and may have led to reduced spring recharge due to reduced duration and snowmelt discharges ( ''medium confidence'' ).

<div id="4.2.7" class="h2-container"></div>

<span id="observed-changes-in-water-quality"></span>
=== 4.2.7 Observed Changes in Water Quality ===

<div id="h2-9-siblings" class="h2-siblings"></div>

AR5 ( [[#Jiménez%20Cisneros--2014|Jiménez Cisneros et al., 2014]] ) concluded with ''medium evidence'' and ''high agreement'' that climate change affected water quality, posing additional risks to drinking water quality and human health ( [[#Field--2014b|Field et al., 2014b]] ), particularly due to increased eutrophication at higher temperatures or release of contaminants due to extreme floods ( [[#Jiménez%20Cisneros--2014|Jiménez Cisneros et al., 2014]] ). In addition, SROCC ( [[#Hock--2019b|Hock et al., 2019b]] ; [[#Meredith--2019|Meredith et al., 2019]] ) assessed that glacier decline and permafrost degradation impacts water quality through increases in legacy contaminants ( ''medium evidence, high agreement'' ).

Warming temperatures and extreme weather events can potentially impact water quality ( [[#Khan--2015|Khan et al., 2015]] ). Water quality can be compromised through algal blooms that affect the taste and odour of recreational and drinking water and can harbour toxins and pathogens ( [[#Khan--2015|Khan et al., 2015]] ). Warming directly affects thermal water regimes, promoting harmful algal blooms ( [[#Li--2018|Li et al., 2018]] ; [[#Noori--2018|Noori et al., 2018]] ) ( [[#4.3.5|Section 4.3.5]] ). Additionally, permafrost degradation leads to an increased flux of contaminants ( [[#MacMillan--2015|MacMillan et al., 2015]] ; [[#Roberts--2017|Roberts et al., 2017]] ; [[#Mu--2019|Mu et al., 2019]] ). The increased meltwater from glaciers ( [[#Zhang--2019|Zhang et al., 2019]] ) releases deposited contaminants and reduces water quality downstream ( [[#Zhang--2017|Zhang et al., 2017]] ; [[#Hock--2019b|Hock et al., 2019b]] ).

Floods intensify the mixing of floodwater with wastewater and the redistribution of pollutants ( [[#Andrade--2018|Andrade et al., 2018]] ). In addition, contaminated floodwaters pose an immediate health risk through waterborne diseases ( [[#Huang--2016b|Huang et al., 2016b]] ; [[#Paterson--2018|Paterson et al., 2018]] ; [[#Setty--2018|Setty et al., 2018]] ). Wildfires, along with heavy rainfalls and floods, can also affect turbidity, which increases drinking water treatment challenges and has been linked to increases in gastrointestinal illness ( [[#de%20Roos--2017|de Roos et al., 2017]] ). Droughts reduce river dilution capacities and groundwater levels (Wen et al., 2017) increasing the risk of groundwater contamination ( [[#Kløve--2014|Kløve et al., 2014]] ). More generally, contaminated water diminishes its aesthetic value, compromising recreational activities, reducing tourism and property values and creating challenges for management and drinking water treatment ( [[#Eves--2014|Eves and Wilkinson, 2014]] ; [[#Khan--2015|Khan et al., 2015]] ; [[#Walters--2015|Walters et al., 2015]] ).

Between 2000 and 2010, ~10% of the global population faced adverse water quality issues ( [[#van%20Vliet--2021|van Vliet et al., 2021]] ). Adverse drinking water quality has been associated with extreme weather events in countries located in Asia, Africa and South and North America ( [[#Jagai--2015|Jagai et al., 2015]] ; [[#Levy--2016|Levy et al., 2016]] ; [[#Huynh--2018|Huynh and Stringer, 2018]] ; [[#Leal%20Filho--2018|Leal Filho et al., 2018]] ; [[#Abedin--2019|Abedin et al., 2019]] ) ( ''medium evidence, high agreement'' ). Dilution factors in 635 of 1049 US streams fell extremely low during drought conditions. Additionally, the safety threshold for endocrine-disrupting compound concentration exceeded in roughly a third of streams studied ( [[#Rice--2017|Rice and Westerhoff, 2017]] ). Natural acid rock drainage, which can potentially release toxic substances, has experienced intensification in an alpine catchment of the Central Pyrenees due to climate change and severe droughts in the last decade. River length affected by natural acid drainage increased from 5 km in 1945 to 35 km in 2018 ( [[#Zarroca--2021|Zarroca et al., 2021]] ). Threefold increases in contaminants and fivefold increases in nutrients have been observed in water sources after wildfires ( [[#Khan--2015|Khan et al., 2015]] ). Due to permafrost thawing, the concentration of major ions, especially SO 4 2− in two high Arctic lakes, has rapidly increased up to 500% and 340% during 2006–2016 and 2008–2016, respectively ( [[#Roberts--2017|Roberts et al., 2017]] ). The exports of dissolved organic carbon (DOC), particulate organic carbon and mercury in six Arctic rivers were reported to increase with significant deepening of active layers caused by climate warming during 1999–2015 ( [[#Mu--2019|Mu et al., 2019]] ). Sustained warming in Lake Tanganyika in Zambia during the last ∼ 150 years reduced lake mixing, which has depressed algal production, shrunk the oxygenated benthic habitat by 38% and further reduced fish and mollusc yield ( [[#Cohen--2016|Cohen et al., 2016]] ). From 1994 to 2010, coastal benthos at King George Island in Antarctica have observed a remarkable shift primarily linked to ongoing climate warming and the increased sediment runoff triggered by glacier retreats ( [[#Sahade--2015|Sahade et al., 2015]] ). The recovery time of macroinvertebrates from floods was found longer in cases of pre-existing pollution problems ( [[#Smith--2019a|Smith et al., 2019a]] ).

In summary, although climate-induced water quality degradation due to increases in water and surface temperatures or melting of the cryosphere has been observed ( ''medium confidence'' ), evidence of global-scale changes in water quality is ''limited'' because many studies are isolated and have limited regional coverage.

<div id="4.2.8" class="h2-container"></div>

<span id="observed-changes-in-soil-erosion-and-sediment-load"></span>
=== 4.2.8 Observed Changes in Soil Erosion and Sediment Load ===

<div id="h2-10-siblings" class="h2-siblings"></div>

AR5 established potential impacts of climate change on soil erosion and sediment loads in mountain regions with glacier melt ( ''low to medium evidence'' ) ( [[#Jiménez%20Cisneros--2014|Jiménez Cisneros et al., 2014]] ). SRCCL (Olsson et al., 2020) reported with ''high confidence'' that rainfall changes attributed to human-induced climate change have already intensified drivers of land degradation. Nonetheless, attributing land degradation to climate change alone is challenging because of the role of land management practices ( ''medium evidence, high agreement'' ).

Climate change impacts soil erosion and sedimentation rates both directly from increasing rainfall or snowmelt intensity ( [[#Vanmaercke--2014|Vanmaercke et al., 2014]] ; [[#Polyakov--2017|Polyakov et al., 2017]] ; [[#Diodato--2018|Diodato et al., 2018]] ; [[#Golosov--2018|Golosov et al., 2018]] ; [[#Li--2020a|Li et al., 2020a]] ; [[#Li--2020b|Li et al., 2020b]] ) and indirectly from increasing wildfires ( [[#Gould--2016|Gould et al., 2016]] ; [[#Langhans--2016|Langhans et al., 2016]] ; [[#DeLong--2018|DeLong et al., 2018]] ), permafrost thawing ( [[#Schiefer--2018|Schiefer et al., 2018]] ; [[#Lafrenière--2019|Lafrenière and Lamoureux, 2019]] ; Ward [[#Jones--2019|Jones et al., 2019]] ) and vegetation cover changes ( [[#Micheletti--2015|Micheletti et al., 2015]] ; [[#Potemkina--2015|Potemkina and Potemkin, 2015]] ; [[#Carrivick--2017|Carrivick and Heckmann, 2017]] ; [[#Beel--2018|Beel et al., 2018]] ). In addition, accelerated soil erosion and sedimentation have severe societal impacts through land degradation, reduced soil productivity and water quality ( [[#4.2.7|Section 4.2.7]] ), increased eutrophication and disturbance to aquatic ecosystems ( [[#4.3.5|Section 4.3.5]] ), sedimentation of waterways and damage to infrastructure ( [[#Graves--2015|Graves et al., 2015]] ; [[#Issaka--2017|Issaka and Ashraf, 2017]] ; [[#Schellenberg--2017|Schellenberg et al., 2017]] ; [[#Hewett--2018|Hewett et al., 2018]] ; [[#Panagos--2018|Panagos et al., 2018]] ; [[#Sartori--2019|Sartori et al., 2019]] ) ( ''medium confidence'' ).

In the largest river basin of the Colombian Andes, regional climate change and land use activities (ploughing, grazing and deforestation) caused a 34% erosion rate increase over 10 years, with the anthropogenic soil erosion rate exceeding the climate-driven erosion rate ( [[#Restrepo--2018|Restrepo and Escobar, 2018]] ). Sedimentation increases due to soil erosion in mountainous regions burned by wildfires, as a result of warming and altered precipitation, is documented with ''high confidence'' in the USA ( [[#Gould--2016|Gould et al., 2016]] ; [[#DeLong--2018|DeLong et al., 2018]] ), Australia ( [[#Nyman--2015|Nyman et al., 2015]] ; [[#Langhans--2016|Langhans et al., 2016]] ), China ( [[#Cui--2014|Cui et al., 2014]] ) and Greece ( [[#Karamesouti--2016|Karamesouti et al., 2016]] ) and can potentially damage downstream aquatic ecosystems ( [[#4.3.5|Section 4.3.5]] ) and water quality ( [[#4.2.7|Section 4.2.7]] ) ( [[#Cui--2014|Cui et al., 2014]] ; [[#Murphy--2015|Murphy et al., 2015]] ; [[#Langhans--2016|Langhans et al., 2016]] ) ( ''medium confidence'' ). In Australia, for instance, sediment yields from post-fire debris flows (113–294 t ha –1 ) are 2–3 orders of magnitude higher than annual background erosion rates from undisturbed forests ( [[#Nyman--2015|Nyman et al., 2015]] ). The positive trend in sediment yield in small ponds in the semiarid southwestern USA over the last 90 years was not entirely related to the rainfall or runoff trends, but was a result of complex interaction between long-term changes in vegetation, soil and channel networks ( [[#Polyakov--2017|Polyakov et al., 2017]] ).

Regional climate changes (precipitation decrease) and human activities (landscape engineering, terracing, large-scale vegetation restoration, soil conservation) over the Loess Plateau (China) caused a distinct stepwise reduction in sediment loads from the upper-middle reach of the Yellow River, with 30% of the change related to climate change ( [[#Tian--2019|Tian et al., 2019]] ). Substantial increases in sediment flux were identified on the Tibetan Plateau ( [[#Li--2020a|Li et al., 2020a]] ; [[#Li--2021a|Li et al., 2021a]] ), for example, the sediment load from the Tuotuohe headwater increased by 135% from 1985–1997 to 1998–2016, mainly due to climate change ( [[#Li--2020a|Li et al., 2020a]] ). In 1986–2015, the sedimentation rate in dry valley bottoms of the Southern Russian Plain was two-2–5 times lower than in 1963–1986 due to the warming-induced surface runoff reduction during spring snowmelt ( [[#Golosov--2018|Golosov et al., 2018]] ). Declining erosion trends are primarily associated with soil conservation management in northern Germany ( [[#Steinhoff-Knopp--2018|Steinhoff-Knopp and Burkhard, 2018]] ) and reforestation in southwestern China ( [[#Zhou--2020|Zhou et al., 2020]] ).

The climate change impact on erosion and sediment load varies significantly over the world ( [[#Li--2020b|Li et al., 2020b]] ) ( ''high confidence'' ). There was a statistically significant correlation between sediment yield and air temperature for the non-Mediterranean region of western and central Europe ( [[#Vanmaercke--2014|Vanmaercke et al., 2014]] ) and northern Africa ( [[#Achite--2016|Achite and Ouillon, 2016]] ). Still, such correlation is yet to be found for the other European rivers ( [[#Vanmaercke--2015|Vanmaercke et al., 2015]] ). Increased sediment and particulate organic carbon fluxes in the Arctic regions are caused by permafrost warming ( [[#Schiefer--2018|Schiefer et al., 2018]] ; [[#Lafrenière--2019|Lafrenière and Lamoureux, 2019]] ; Ward [[#Jones--2019|Jones et al., 2019]] ). [[#Potemkina--2015|Potemkina and Potemkin (2015)]] demonstrate that regional warming and permafrost degradation have contributed to an increased forested area over the last 40–70 years, reducing soil erosion in eastern Siberia. The sediment dynamics of small rivers in the eastern Italian Alps, depending on extreme floods, is sensitive to climate change ( [[#Rainato--2017|Rainato et al., 2017]] ). In the northeastern Italian Alps, precipitation change during 1986–2010 affected soil wetness conditions, influencing sediment load ( [[#Diodato--2018|Diodato et al., 2018]] ). Regional warming in northern Africa (Algeria) dramatically changed river streamflow and increased sediment load over four decades (84% more every decade compared to the previous) ( [[#Achite--2016|Achite and Ouillon, 2016]] ).

A long-term global soil erosion monitoring network based on the unified methodological approach is needed to correctly evaluate erosion rates, detect their changes and attribute them to climate or other drivers.

In summary, in the areas with high human activity, factors other than climate have a more significant impact on soil erosion and sediment flux ( ''high confidence'' ). On the other hand, in natural conditions, for example, in high latitudes and high mountains, the influence of climate change on the acceleration of the erosion rate is observed ( ''limited evidence, medium agreement'' ).

<div id="4.3" class="h1-container"></div>

<span id="observed-sectoral-impacts-of-current-hydrological-changes"></span>