Editing IPCC:AR6/SROCC/Chapter-2 (section)

== Box 2.1 Does Atmospheric Warming in the Mountains Depend on Elevation? ==

<div id="section-2-2-1-1surface-air-temperature-block-1"></div>

In mountain regions, surface air temperature generally tends to decrease with increasing elevation thus directly impacting how much of the precipitation falls as snow as opposed to rain. Therefore, changes in air temperature have different consequences for snow cover, permafrost and glaciers at different elevations. A number of studies have reported that trends in air temperature vary with elevation, a phenomenon referred to as elevation dependent warming (EDW; Pepin et al., 2015, and references therein), with potential consequences beyond those of uniform warming. EDW does not imply that warming is larger at higher elevation, and smaller at lower elevation, but it means that the warming rate (e.g., in ºC per decade) is not the same across all elevation bands. Although this concept has received wide attention in recent years, the manifestation of EDW varies by region, season and temperature indicator (e.g., daily mean, minimum or maximum temperature), meaning that a uniform pattern does not exist. The identification of the underlying driving mechanisms for EDW and how they combine is complex.

Several physical processes contribute to EDW, and quantifying their relative contributions has remained largely elusive (Minder et al., 2018 <sup>[[#fn:r17|17]]</sup> ; Palazzi et al., 2019 <sup>[[#fn:r18|18]]</sup> ). Some of the processes identified are similar to those explaining the amplified warming in the polar regions (Chapter 3). For example, the sensitivity of temperature to radiative forcing is increased at low temperatures common in both polar and mountain environments (Ohmura, 2012 <sup>[[#fn:r19|19]]</sup> ). Because the relationship between specific humidity and downwelling radiation is nonlinear, in a dry and cold atmosphere found at high elevation, any increase in atmospheric humidity due to temperature increase drives disproportionately large warming (Rangwala et al., 2013 <sup>[[#fn:r20|20]]</sup> ; Chen et al., 2014 <sup>[[#fn:r21|21]]</sup> ). Snow-albedo feedback plays an important role where the snow cover is in decline (Pepin and Lundquist, 2008 <sup>[[#fn:r22|22]]</sup> ; Scherrer et al., 2012 <sup>[[#fn:r23|23]]</sup> ), increasing the absorption of solar radiation which in turn leads to increased surface air temperature and further snowmelt. Other processes are specific to the mountain environment. Especially in the tropics, warming can be enhanced at higher elevation by a reduction of the vertical temperature gradient, due to increased latent heat release above the condensation level, favored in a warmer and moister atmosphere (Held and Soden, 2006 <sup>[[#fn:r24|24]]</sup> ). The cooling effect of aerosols, which also cause solar dimming, is more pronounced at low elevation and reduced at high elevation (Zeng et al., 2015 <sup>[[#fn:r25|25]]</sup> ). While many mechanisms suggest that warming should be enhanced at high elevation, observed and simulated EDW patterns are usually more complex (Pepin et al., 2015, and references therein). Numerical simulations by global and regional climate models, which show EDW, need to be considered carefully because of intrinsic limitations due to potentially incomplete understanding and implementation of relevant physical processes, in addition to coarse grid spacing with respect to mountainous topography (Ménégoz et al., 2014 <sup>[[#fn:r26|26]]</sup> ; Winter et al., 2017 <sup>[[#fn:r27|27]]</sup> ).

<div id="section-2-2-1-2rainfall-and-snowfall"></div>

<span id="rainfall-and-snowfall"></span>
==== 2.2.1.2 Rainfall and Snowfall ====

<div id="section-2-2-1-2rainfall-and-snowfall-block-1"></div>

Past precipitation changes are less well quantified than temperature changes and are often more heterogeneous, even within mountain regions (Hartmann and Andresky, 2013 <sup>[[#fn:r32|32]]</sup> ). Regional patterns are characterised by decadal variability (Mankin and Diffenbaugh, 2015 <sup>[[#fn:r33|33]]</sup> ) and influenced by shifts in large-scale atmospheric circulation (e.g., in Alaska; Winski et al., 2017 <sup>[[#fn:r34|34]]</sup> ). While mountain regions do not exhibit clear direction of trends in annual precipitation over the past decades ( ''medium confidence'' that there is no trend), snowfall has decreased, at least in part due to higher temperatures, especially at lower elevation (Table SM2.4, ''high confidence'' ).

Future projections of annual precipitation indicate increases of the order of 5 to 20% over the 21st century in many mountain regions, including the Hindu Kush and Himalaya, East Asia, eastern Africa, the European Alps and the Carpathian region, and decreases in the Mediterranean and the Southern Andes ( ''medium confidence,'' Table SM2.5). Changes in the frequency and intensity of extreme precipitation events vary according to season and region. For example, across the Himalayan-Tibetan Plateau mountains, the frequency and intensity of extreme rainfall events are projected to increase throughout the 21st century, particularly during the summer monsoon (Panday et al., 2015 <sup>[[#fn:r35|35]]</sup> ; Sanjay et al., 2017 <sup>[[#fn:r36|36]]</sup> ). This suggests a transition toward more episodic and intense monsoonal precipitation, especially in the easternmost part of the Himalayan chain (Palazzi et al., 2013). Increases in winter precipitation extremes are projected in the European Alps (Rajczak and Schär, 2017 <sup>[[#fn:r38|38]]</sup> ). At lower elevation, near term (2031 – 2050) and end of century (2081 – 2100) projections of snowfall all indicate a decrease, for all greenhouse gas emission scenarios ( ''very high confidence'' ). At higher elevation, where temperature increase is insufficient to affect rain/snow partitioning, total winter precipitation increases can lead to increased snowfall (e.g., Kapnick and Delworth, 2013; O’Gorman, 2014) ( ''medium confidence'' ).

<div id="section-2-2-1-3other-meteorological-variables"></div>

<span id="other-meteorological-variables"></span>
==== 2.2.1.3 Other Meteorological Variables ====

<div id="section-2-2-1-3other-meteorological-variables-block-1"></div>

Atmospheric humidity, incoming shortwave and longwave radiation, and near-surface wind speed and direction also influence the high mountain cryosphere. Detecting their changes and associated effects on the cryosphere is even more challenging than for surface air temperature and precipitation, both from an observation and modelling standpoint. Therefore, most simulation studies of cryosphere changes are mainly driven by temperature and precipitation (see, e.g., Beniston et al., 2018, and references therein).

Atmospheric moisture content, which is generally increasing in a warming atmosphere (Stocker et al., 2013 <sup>[[#fn:r40|40]]</sup> ), affects latent and longwave heat fluxes (Armstrong and Brun, 2008 <sup>[[#fn:r41|41]]</sup> ) with implications for the timing and rate of snow and ice ablation, and in some areas changes in atmospheric moisture content could be a significant driver of cryosphere change (Harpold and Brooks, 2018 <sup>[[#fn:r42|42]]</sup> ). Short-lived climate forcers, such as sulphur and black carbon aerosols (You et al., 2013 <sup>[[#fn:r43|43]]</sup> ), reduce the amount of solar radiation reaching the surface, with potential impacts on snow and ice ablation. Solar brightening caused by declining anthropogenic aerosols in Europe since the 1980s was shown to have only a minor effect on atmospheric warming at high elevation (Philipona, 2013 <sup>[[#fn:r48|48]]</sup> ), and effects on the cryosphere were not specifically discussed.

Wind controls preferential deposition of precipitation, post-depositional snow drift and affects ablation of snow and glaciers through turbulent fluxes. Near-surface wind speed has decreased on the Tibetan Plateau between the 1970s and early 2000s, and stabilised or increased slightly thereafter (Yang et al., 2014a <sup>[[#fn:r49|49]]</sup> ; Kuang and Jiao, 2016 <sup>[[#fn:r50|50]]</sup> ). This is consistent with existing evidence for a decrease in near-surface wind speed on mid-latitude continental areas since the mid-20th century (Hartmann et al., 2013 <sup>[[#fn:r51|51]]</sup> ). In general, the literature on past and future changes of near-surface wind patterns in mountain areas is very limited.

<span id="snow-cover"></span>
=== 2.2.2 Snow Cover ===

<div id="section-2-2-2snow-cover-block-1"></div>

Snow on the ground is an essential and widespread component of the mountain cryosphere. It affects mountain ecosystems and plays a major role for mass movement and floods in the mountains. It plays a key role in nourishing glaciers and provides an insulating and reflective cover at their surface. It influences the thermal regime of the underlying ground, including permafrost, with implications for ecosystems. Climate change modifies key variables driving the onset and development of the snow cover (e.g., solid precipitation), and those responsible for its ablation (e.g., air temperature, radiation). The snow cover, especially in low-lying and mid-elevation areas of mountain regions, has long been identified to be particularly sensitive to climate change.

The mountain snow cover is characterised by a very strong interannual and decadal variability, similar to its main driving force solid precipitation (Lafaysse et al., 2014 <sup>[[#fn:r52|52]]</sup> ; Mankin and Diffenbaugh, 2015 <sup>[[#fn:r53|53]]</sup> ). Observations spanning several decades are required to quantify trends. Long-term ''in situ'' records are scarce in some regions of the world, particularly in High Mountain Asia, Northern Asia and South America (Rohrer et al., 2013 <sup>[[#fn:r54|54]]</sup> ). Satellite remote sensing provides new capabilities for monitoring mountain snow cover on regional scales. The satellite record length is often insufficient to assess trends (Bormann et al., 2018 <sup>[[#fn:r55|55]]</sup> ). Evidence of past changes from regional studies is provided in Table SM2.6. At lower elevation, there is ''high confidence'' that the mountain snow cover has generally declined in duration (on average by 5 snow cover days per decade, with a ''likely'' range from 0 to 10 days per decade), mean snow depth and accumulated mass (snow water equivalent) since the middle of the 20th century, with regional variations. At higher elevation, snow cover trends are generally insignificant ( ''medium confidence'' ) or unknown ''.''

Most of the snow cover changes can be attributed, at lower elevation, to more precipitation falling as liquid precipitation (rain) and to increases in melt at all elevations, mostly due to changes in atmospheric forcings, especially increased air temperature (Kapnick and Hall, 2012 <sup>[[#fn:r56|56]]</sup> ; Marty et al., 2017 <sup>[[#fn:r57|57]]</sup> ) which in turn are attributed to anthropogenic forcings at a larger scale (Section 2.2.1). Formal anthropogenic attribution studies provide similar conclusions in Western North America (Pierce et al., 2008 <sup>[[#fn:r58|58]]</sup> ; Najafi et al., 2017 <sup>[[#fn:r59|59]]</sup> ).

Assessing the impact of the deposition of short-lived climate forcers on snow cover changes is an emerging issue (Skiles et al., 2018 <sup>[[#fn:r60|60]]</sup> and references therein). This concerns light absorbing particles, in particular, which include deposited aerosols such as black carbon, organic carbon and mineral dust, or microbial growth (Qian et al., 2015 <sup>[[#fn:r61|61]]</sup> ), although the role of the latter has not been specifically quantified. Due to their seasonally variable deposition flux and impact, and mostly episodic nature in case of dust deposition (Kaspari et al., 2014 <sup>[[#fn:r62|62]]</sup> ; Di Mauro et al., 2015 <sup>[[#fn:r63|63]]</sup> ), light absorbing particles contribute to interannual fluctuations of seasonal snowmelt rate (Painter et al., 2018) ( ''medium evidence, high agreement'' ). There is ''limited evidence (medium agreement'' ) that increases in black carbon deposition from anthropogenic and biomass burning sources have contributed to snow cover decline in High Mountain Asia (Li et al., 2016 <sup>[[#fn:r64|64]]</sup> ; Zhang et al., 2018 <sup>[[#fn:r65|65]]</sup> ) and South America (Molina et al., 2015 <sup>[[#fn:r66|66]]</sup> ).

Projected changes of mountain snow cover are studied based on climate model experiments, either directly from GCM or RCM output, or following downscaling and the use of snowpack models. These projections generally do not specifically account for future changes in the deposition rate of light absorbing particles on snow (or, if so, simple approaches have been used hitherto; e.g., Deems et al., 2013), so that future changes in snow conditions are mostly driven by changes in meteorological drivers assessed in Section 2.2.1. Evidence from regional studies is provided in Table SM2.7. Although existing studies in mountain regions do not use homogenous reference periods and model configurations, common future trends can be summarised as follows. At lower elevation in many regions such as the European Alps, Western North America, Himalaya and subtropical Andes, the snow depth or mass is projected to decline by 25% ( ''likely'' range between 10 and 40%), between the recent past period (1986 – 2005) and the near future (2031 – 2050), regardless of the greenhouse gas emission scenario (Cross-Chapter Box 1 in Chapter 1). This corresponds to a continuation of the ongoing decrease in annual snow cover duration (on average 5 days per decade, with a ''likely'' range from 0 to 10). By the end of the century (2081 – 2100), reductions of up to 80% ( ''likely'' range from 50 to 90%) are expected under RCP8.5, 50% ( ''likely'' range from 30 to 70%) under RCP4.5 and 30% ( ''likely'' range from 10 to 40%) under RCP2.6. At higher elevations, projected reductions are smaller ( ''high confidence'' ), as temperature increases at higher elevations affect the ablation component of snow mass evolution, rather than both the onset and accumulation components. The projected increase in winter snow accumulation may result in a net increase in winter snow mass ( ''medium confidence'' ). All elevation levels and mountain regions are projected to exhibit sustained interannual variability of snow conditions throughout the 21st century ( ''high confidence'' ). Figure 2.3 provides projections of temperature and snow cover in mountain areas in Europe, High Mountain Asia (Hindu Kush, Karakoram and Himalaya), North America (Rocky Mountains) and South America (sub-tropical Central Andes), illustrating how changes vary with elevation, season, region, future time period and climate scenario.

<div id="section-2-2-2snow-cover-block-2"></div>

<span id="figure-2.3"></span>
<!-- START IMG -->
<!-- IMG TITLE -->
'''Figure 2.3'''

<span id="figure-2.3-projected-change-19862005-to-20312050-and-20802099-of-mean-winter-december-to-may-june-to-august-in-subtropical-central-andes-snow-water-equivalent-winter-air-temperature-and-summer-air-temperature-june-to-august-december-to-february-in-subtropical-central-andes-in-five-high-mountain-regions-for-rcp8.5-all-regions-and-rcp2.6-european"></span>
<!-- IMG CAPTION -->
'''Figure 2.3 | Projected change (1986–2005 to 2031–2050 and 2080–2099) of mean winter (December to May; June to August in Subtropical Central Andes) snow water equivalent, winter air temperature and summer air temperature (June to August; December to February in Subtropical Central Andes) in five high mountain regions for RCP8.5 (all regions) and RCP2.6 (European […]'''
<!-- IMG FILE -->
[[File:fa917e5e7d37593bc83c825a05e40b5f IPCC-SROCC-CH_2_3.jpg]]

Figure 2.3 | Projected change (1986–2005 to 2031–2050 and 2080–2099) of mean winter (December to May; June to August in Subtropical Central Andes) snow water equivalent, winter air temperature and summer air temperature (June to August; December to February in Subtropical Central Andes) in five high mountain regions for RCP8.5 (all regions) and RCP2.6 (European Alps and Subtropical Central Andes). Changes are averaged over 500 m (a,b,c) and 1,000 m (d,e) elevation bands. The numbers in the lower right of each panel reflect the number of simulations (note that not all models provide snow water equivalent). For the Rocky Mountains, data from NA-CORDEX RCMs (25 km grid spacing) driven by Coupled Model Intercomparison Project Phase 5 (CMIP5) General Circulation Models (GCMs) were used (Mearns et al., 2017 <sup>[[#fn:r44|44]]</sup> ). For the European Alps, data from EURO-CORDEX RCMs (12 km grid spacing) driven by CMIP5 GCMs were used (Jacob et al., 2014 <sup>[[#fn:r45|45]]</sup> ). For the other regions, CMIP5 GCMs were used: Zazulie (2016) <sup>[[#fn:r56|56]]</sup> and Zazulie et al. (2018) for the Subtropical Central Andes, and Terzago et al. (2014) and Palazzi et al. (2017) <sup>[[#fn:r48|48]]</sup> for the Hindu Kush and Karakoram and Himalaya. The list of models used is provided in Table SM2.8.

<!-- END IMG -->
<span id="glaciers">
</span>
=== 2.2.3 Glaciers ===

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

The high mountain areas considered in this chapter (Figure 2.1), including all glacier regions in the world except those in Antarctica, Greenland, the Canadian and Russian Arctic, and Svalbard (which are covered in Chapter 3) include ~170,000 glaciers covering an area of ~250,000 km 2 (RGI Consortium, 2017) with a total ice volume of 87 ± 15 mm sea level equivalent (Farinotti et al., 2019 <sup>[[#fn:r70|70]]</sup> ). These glaciers span an elevation range from sea level, for example in south-east Alaska, to &gt;8,000 m a.s.l. in the Himalaya and Karakoram, and occupy diverse climatic regions. Their mass budget is determined largely by the balance between snow accumulation and melt at the glacier surface, driven primarily by atmospheric conditions. Rapid changes in mountain glaciers have multiple impacts for social-ecological systems, affecting not only biophysical properties such as runoff volume and sediment fluxes in glacier-fed rivers, glacier related hazards, and global sea level (Chapter 4) but also ecosystems and human livelihoods, socioeconomic activities and sectors such as agriculture and tourism, as well as other intrinsic assets such as cultural values. While glaciers worldwide have experienced considerable fluctuations throughout the Holocene driven by multidecadal variations of solar and volcanic activity, and changes in atmospheric circulation (Solomina et al., 2016 <sup>[[#fn:r71|71]]</sup> ), this section focuses on observed glacier changes during recent decades and changes projected for the 21st century (Cross-Chapter Box 6 in Chapter 2).

Satellite and ''in situ'' observations of changes in glacier area, length and mass show a globally largely coherent picture of mountain glacier recession in the last decades (Zemp et al., 2015 <sup>[[#fn:r72|72]]</sup> ), although annual variability and regional differences are large (Figure 2.4; ''very high confidence'' ). The global trend is statistically significant despite considerable interannual and regional variations (Medwedeff and Roe, 2017 <sup>[[#fn:r73|73]]</sup> ). Since AR5’s global 2003 – 2009 estimate based on Gardner et al. (2013) <sup>[[#fn:r74|74]]</sup> , several new estimates of global-scale glacier mass budgets have emerged using largely improved data coverage and methods (Bamber et al., 2018 <sup>[[#fn:r75|75]]</sup> ; Wouters et al., 2019 <sup>[[#fn:r76|76]]</sup> ; Zemp et al., 2019 <sup>[[#fn:r77|77]]</sup> ).

These estimates combined with available regional estimates (Table 2.A.1) that the glacier mass budget of all mountain regions (excluding Antarctica, Greenland, the Canadian and Russian Arctic, and Svalbard) was ''very'' ''likely'' -490 ± 100 kg m -2 yr -1 (-123 ± 24 Gt yr -1 ) during the period 2006 – 2015 with most negative averages (less than -850 kg m -2 yr -1 ) in the Southern Andes, Caucasus/Middle East, European Alps and Pyrenees. High Mountain Asia shows the least negative mass budget (-150 ± 110 kg m -2 yr -1 , Figure 2.4), but variations within the region are large with most negative regional balance estimates in Nyainqentanglha, Tibet (-620 ± 230 kg m -2 yr -1 ) and slightly positive balances in the Kunlun Mountains for the period 2000 – 2016 (Brun et al., 2017 <sup>[[#fn:r78|78]]</sup> ). Due to large ice extent, the total mass loss and corresponding contribution to sea level 2006 – 2015 is largest in Alaska, followed by the Southern Andes and High Mountain Asia (Table 2.A.1). Zemp et al. (2019) estimated an increase in mean global-scale glacier mass loss by ~30% between 1986–2005 and 2006–2015.

<div id="section-2-2-3glaciers-block-2"></div>

<span id="figure-2.4"></span>
<!-- START IMG -->
<!-- IMG TITLE -->
'''Figure 2.4'''

<span id="figure-2.4-glacier-mass-budgets-for-the-eleven-mountain-regions-assessed-in-this-chapter-figure-2.1-and-these-regions-combined.-mass-budgets-for-the-remaining-polar-regions-are-shown-in-chapter-3-figure-3.8.-regional-time-series-of-annual-mass-change-are-based-on-glaciological-and-geodetic-balances-zemp-et-al.-2019.-superimposed-are"></span>
<!-- IMG CAPTION -->
'''Figure 2.4 | Glacier mass budgets for the eleven mountain regions assessed in this Chapter (Figure 2.1) and these regions combined. Mass budgets for the remaining polar regions are shown in Chapter 3, Figure 3.8. Regional time series of annual mass change are based on glaciological and geodetic balances (Zemp et al., 2019). Superimposed are […]'''
<!-- IMG FILE -->
[[File:a33193757f586f23ffad4aba67192b7b IPCC-SROCC-CH_2_4.jpg]]

Figure 2.4 | Glacier mass budgets for the eleven mountain regions assessed in this Chapter (Figure 2.1) and these regions combined. Mass budgets for the remaining polar regions are shown in Chapter 3, Figure 3.8. Regional time series of annual mass change are based on glaciological and geodetic balances (Zemp et al., 2019). Superimposed are multi-year averages by Wouters et al. (2019) based on the Gravity Recovery and Climate Experiment (GRACE), only shown for the regions with glacier area &gt;3,000 km2. Estimates by Gardner et al. (2013) were used in the IPCC 5th Assessment Report (AR5). Additional regional estimates available in some regions and shown here are listed in Table 2.A.1. Annual and time-averaged mass-budget estimates include the errors reported in each study. Glacier areas (A) and volumes (V) are based on RGI Consortium (2017) and Farinotti et al. (2019), respectively. Red and blue bars on map refer to regional budgets averaged over the period 2006–2015 in units of kg m-2 yr-1 and mm sea level equivalent (SLE) yr-1, respectively, and are derived from each region’s available mass-balance estimates (Appendix 2.A).

<!-- END IMG -->
<div id="section-2-2-3glaciers-block-3">
</div>

It is ''very likely'' that atmospheric warming is the primary driver for the global glacier recession (Marzeion et al., 2014 <sup>[[#fn:r79|79]]</sup> ; Vuille et al., 2018 <sup>[[#fn:r80|80]]</sup> ). There is ''limited evidence'' ( ''high agreement'' ) that human-induced increases in greenhouse gases have contributed to the observed mass changes (Hirabayashi et al., 2016 <sup>[[#fn:r81|81]]</sup> ). It was estimated that the anthropogenic fraction of mass loss of all glaciers outside Greenland and Antarctica increased from 25 ± 35% during 1851–2010 to 69 ± 24% during 1991–2010 (Marzeion et al., 2014 <sup>[[#fn:r82|82]]</sup> ).

Other factors, such as changes in meteorological variables other than air temperature or internal glacier dynamics, have modified the temperature-induced glacier response in some regions ( ''high confidence'' ). For example, glacier mass loss over the last seven decades on a glacier in the European Alps was intensified by higher air moisture leading to increased longwave irradiance and reduced sublimation (Thibert et al., 2018 <sup>[[#fn:r83|83]]</sup> ). Changes in air moisture have also been found to play a significant role in past glacier mass changes in eastern Africa (Prinz et al., 2016 <sup>[[#fn:r84|84]]</sup> ), while an increase in shortwave radiation due to reduced cloud cover contributed to an acceleration in glacier recession in the Caucasus (Toropov et al., 2019 <sup>[[#fn:r85|85]]</sup> ). In the Tien Shan mountains changes in atmospheric circulation in the North Atlantic and North Pacific in the 1970s resulted in an abrupt reduction in precipitation and thus snow accumulation, amplifying temperature-induced glacier mass loss (Duethmann et al., 2015 <sup>[[#fn:r86|86]]</sup> ). Deposition of light absorbing particles, growth of algae and bacteria and local amplification phenomena such as the enhancement of particles concentration due to surface snow and ice melt, and cryoconite holes, have been shown to enhance ice melt (e.g., Ginot et al., 2014; Zhang et al., 2017 <sup>[[#fn:r87|87]]</sup> ; Williamson et al., 2019 <sup>[[#fn:r88|88]]</sup> ) but there is ''limited evidence'' and ''low agreement'' that long-term changes in glacier mass are linked to light absorbing particles (Painter et al., 2013 <sup>[[#fn:r89|89]]</sup> ; Sigl et al., 2018 <sup>[[#fn:r90|90]]</sup> ). Debris cover can  modulate glacier melt but there is ''limited evidence'' on its role in recent glacier changes (Gardelle et al., 2012 <sup>[[#fn:r91|91]]</sup> ; Pellicciotti et al., 2015 <sup>[[#fn:r92|92]]</sup> ). Rapid retreat of calving outlet glaciers in Patagonia was attributed to changes in glacier dynamics (Sakakibara and Sugiyama, 2014 <sup>[[#fn:r93|93]]</sup> ).

Departing from this global trend of glacier recession, a small fraction of glaciers have gained mass or advanced in some regions mostly due to internal glacier dynamics or, in some cases, locally restricted climatic causes. For example, in Alaska 36 marine-terminating glaciers exhibited a complex pattern of periods of significant retreat and advance during 1948–2012, highly variable in time and lacking coherent regional behaviour (McNabb and Hock, 2014 <sup>[[#fn:r94|94]]</sup> ). These fluctuations can be explained by internal retreat-advance cycles typical of tidewater glaciers that are largely independent of climate (Brinkerhoff et al., 2017 <sup>[[#fn:r95|95]]</sup> ). Irregular and spatially inconsistent glacier advances, for example, in Alaska, Iceland and Karakoram, have been associated with surge-type flow instabilities largely independent of changes in climate (Sevestre and Benn, 2015 <sup>[[#fn:r96|96]]</sup> ; Bhambri et al., 2017 <sup>[[#fn:r97|97]]</sup> ; Section 2.3.2). Regional scale glacier mass gain and advances in Norway in the 1990s and in New Zealand between 1983–2008 have been linked to local increases in snow precipitation (Andreassen et al., 2005 <sup>[[#fn:r98|98]]</sup> ) and lower air temperatures (Mackintosh et al., 2017 <sup>[[#fn:r99|99]]</sup> ), respectively, caused by changes in atmospheric circulation. Advances of some glaciers in Alaska, the Andes, Kamchatka and the Caucasus were attributed to volcanic activity causing flow acceleration through enhanced melt water at the ice-bed interface (Barr et al., 2018 <sup>[[#fn:r100|100]]</sup> ).

Region averaged glacier mass budgets have been nearly balanced in the Karakoram since at least the 1970s (Bolch et al., 2017 <sup>[[#fn:r101|101]]</sup> ; Zhou et al., 2017 <sup>[[#fn:r102|102]]</sup> ; Azam et al., 2018 <sup>[[#fn:r103|103]]</sup> ), while slightly positive balances since 2000 have been reported in the western Kunlun Shan, eastern Pamir, and the central and northern Karakoram mountains (Gardelle et al., 2013 <sup>[[#fn:r104|104]]</sup> ; Brun et al., 2017 <sup>[[#fn:r105|105]]</sup> ; Lin et al., 2017 <sup>[[#fn:r106|106]]</sup> ; Berthier and Brun, 2019 <sup>[[#fn:r107|107]]</sup> ). This anomalous behavior has been related to specific mechanisms countering the effects of atmospheric warming, for example, an increase in cloudiness (Bashir et al., 2017 <sup>[[#fn:r108|108]]</sup> ) and snowfall (Kapnick et al., 2014 <sup>[[#fn:r109|109]]</sup> ) spatially heterogeneous glacier mass balance sensitivity (Sakai and Fujita, 2017 <sup>[[#fn:r110|110]]</sup> ), feedbacks due to intensified lowland irrigation (de Kok et al., 2018), and changes in summer atmospheric circulation (Forsythe et al., 2017 <sup>[[#fn:r111|111]]</sup> ).

There is ''medium evidence (high agreement'' ) that recent glacier mass changes have modified glacier flow. A study covering all glaciers in High Mountain Asia showed glacier slowdown for regions with negative mass budgets since the 1970s and slightly accelerated glacier flow for Karakoram and West Kunlun regions where mass budgets were close to balance (Dehecq et al., 2019 <sup>[[#fn:r112|112]]</sup> ). Waechter et al. (2015) <sup>[[#fn:r113|113]]</sup> report reduced flow velocities in the St. Elias Mountains in North America, especially in areas of rapid ice thinning near glacier termini. In contrast Mouginot and Rignot (2015) found complex ice flow patterns with simultaneous acceleration and deceleration for glaciers of the Patagonian Icefield as well as large interannual variability during the last three decades concurrent with general thinning of the ice field.

<div id="section-2-2-3glaciers-block-4" class="box"></div>

<span id="ccb.6-glacier-projections-in-polar-and-high-mountain-regions"></span>