Glaciers play an important role in the global hydrological cycle, through the storage of water for thousands of years (Figure 2.1). Water is stored in a series of reservoirs, including the ocean, lakes, groundwater, atmosphere, snowpack, and glaciers. Water movement is driven by energy: warmer air temperatures speed up the water cycle; colder air temperatures slow the water cycle down. Water movement from the atmosphere to the oceans and continents occurs as precipitation, including snow, sleet, and other forms of solid precipitation. Snow that accumulates for many years may turn into a glacier. This chapter reviews the current understanding of Hindu-Kush Himalayan (HKH) glaciers in the context of the modern climate setting, impacts of aerosols and black carbon1 on the energy budget affecting the glaciers, what paleoclimate records can tell us about current regional climate conditions, regional hydrology, and physical hazards in the Himalayas.
Glacial ice is characterized by (a) a density between about 830 and 920 kg m-3 (83 to 92 percent water content) and (b) air that is trapped in bubbles within the ice and no longer in contact with the atmosphere. When snow falls on a surface, it initially has a density of 50 to 70 kg m-3 and within a few days has a density on the order of 100 to 300 kg m-3 (10 to 30 percent water content). Over time, through compaction of overlying snow and through metamorphic processes, the density of snowpack gradually increases. Snow that does not melt is carried over to the next season, where it can be buried by subsequent snowfall. Snow that is older than a year but not yet glacial ice is called “firn” or “névé.” The density of firn gradually increases over time, and eventually the air trapped in pockets or bubbles is no longer in contact with the atmosphere. The firn has become glacial ice. Local climate determines the rate at which seasonal snow changes to glacial ice (cf. Cuffey and Paterson, 2010).
Glaciers move by gravitational processes, including internal deformation caused by shear stress imposed by overlying ice and snow, and potentially by basal sliding on a layer of liquid or quasi-liquid water. Ice masses can flow down slopes or across flat terrain because of the pressure produced by overlying snow and ice. Once a mass of ice flows as a solid, it is considered to be a glacier. Patches of ice and snow that do not flow are not glaciers.
The fundamentals of glacial behavior can be readily understood by recognizing that glaciers have both a zone of accumulation in which the volume of the glacier grows and a zone of ablation in which volume is lost. During the accumulation season (summer in the eastern HKH and winter in the western HKH), a glacier gains mass. During the melt season (summer in both the eastern and western HKH), some or all of that accumulation is lost. Thus, over the course of a year the size of a glacier may increase, decrease, or remain static. This is determined by whether accumulation or ablation predominates or whether they are equal. The accumulation area is the upper elevation zone where
1 Black carbon refers to particulate matter derived from the incomplete combustion of a hydrocarbon.
FIGURE 2.1 The global hydrological cycle, or water cycle, is the process by which water moves through a series of reservoirs, including the ocean, lakes, groundwater, atmosphere, snowpack, and glaciers. Water can be in any phase (solid, liquid, gas) in these reservoirs. Water moves from the terrestrial and oceanic reservoirs to the atmosphere through transpiration, evaporation, or sublimation. Water moves from the atmosphere to the terrestrial and oceanic reservoirs through precipitation. Precipitation can occur in liquid form (rain) or solid form (snow, sleet, other types). SOURCE: U.S. Geological Survey.
there is an annual net gain in mass, and the ablation area is the lower elevation zone where there is an annual net loss in mass. The equilibrium-line altitude (ELA) is the elevation where the accumulation and ablation zones meet and where the annual net mass balance is zero (Figure 2.2). The annual mass balance is the net difference between accumulation and ablation (cf. Cuffey and Paterson, 2010).
Accumulation includes all processes by which glaciers increase in snow and ice mass, such as snowfall, condensation, refreezing of rainfall, avalanche transport onto the glacier, and blowing snow transport onto the glacier. Ablation includes all of those processes by which glaciers lose snow and ice mass, such as snow-melt, icemelt, sublimation, blowing snow transport off the glacier, calving and avalanche removal (cf. Cuffey and Paterson, 2010).
When viewed as water supply systems, glaciers are analogous to lakes. Water storage in glaciers is analogous to the total quantity of water stored in a lake. Glacial accumulation is analogous to water input to a lake, which includes processes such as precipitation and water carried into the lake by streams, rivers, and groundwater channels. Glacial ablation is analogous to water removal from a lake, which includes processes such as evaporation, water carried out of the lake by streams, rivers, and groundwater channels, and extraction by humans. When water input sources equal water output sources, the lake is in steady state and the lake level does not change. With glaciers, when accumulation equals ablation, the volume of water stored in the glacier does not change and the ELA does not move. Glacial volumes decrease when ablation persistently exceeds accumulation, the ELA moves up, and the glacier in question ultimately disappears. This is analogous to a lake where persistent overdraft, in which extractions exceed water input, is always self-terminating.
Several important principles follow from this discussion. First, it is the change in the volume of the glacier, not the change in its downhill extent or areal extent that determines whether the net change is positive or negative. However, it is difficult to directly measure the volume of a glacier; thus measurements of glacial volumes are scarce throughout the world. Second, where the entirety of the glacier is below the equilibrium line, there will be no accumulation and with time the glacier will disappear. Third, glacial mass balance information will provide an important
FIGURE 2.2 Schematic of glacial mass balance indicating the accumulation area at higher elevation and the ablation area at lower elevation. Accumulation includes all processes by which solid ice (including snow) is added to a glacier, and ablation includes all processes by which ice and snow are lost from the glacier. The equilibrium-line altitude occurs at the elevation contour where the accumulation and ablation areas meet and the annual net mass balance is zero. SOURCE: Armstrong (2010).
link between variations in glacial volume and climate changes (Meier, 1962). A general understanding of glacial mass balance is essential to understanding what happens to glaciers over time.
Glaciers respond to climate to reach steady state, a state with no change in the mass balance or ELA over time. A glacier advances due to cooling temperatures or snowfall increase, resulting in a positive mass balance. Warming temperatures or a decrease in snowfall results in a negative mass balance and glacial retreat. A glacier that is in disequilibrium with a warming climate will retreat until equilibrium is reestablished or the glacier disappears.
Glaciers in mid-latitude mountain regions of the world, including those in the HKH region, experience melt in their ablation zones at some time in most years. Melting of glacial ice is a normal phenomenon. Most, if not all, mid-latitude glaciers contribute meltwater to streams and rivers. This contribution of glacial meltwater to the discharge of mountain streams and rivers occurs even in years with a positive glacial mass balance. A steady-state situation occurs when climate conditions are such that glacial melt equals accumulation and there is no change in the mass balance over some time period.
Glacier contribution to streamflow can be discussed in terms of the hydrological cycle (Figure 2.1) following the approach of Comeau et al. (2009). For their investigation of glacier hydrology in the Canadian Rockies, they defined “glacial melt” and “glacial wastage” in terms of the water equivalent. They simplified the annual glacial mass balance by treating sublimation as negligible, and assuming no snow inputs or outputs from avalanching, snowdrifting, or blowing snow, and no ice losses from calving. At the high elevations of the HKH, as well as in central and northern Tibet, where it is very cold and dry, sublimation is an important glaciological term in considering the mass balance of the glacier, but not in hydrological considerations. Therefore, the Committee has followed the approach of Comeau et al. (2009) in using the terms “glacial melt” and “glacial wastage” when discussing the relationship between glacial meltwater and streamflow. Because there are differences in meaning implied between glacial melt and glacial wastage for different disciplines, when reporting results from other sources, the Committee has been consistent with the language used in the original reference.
Comeau et al. (2009) defined the annual glacial mass balance as being equivalent to the annual snowfall minus the annual snowmelt from the glacier and minus the annual glacial icemelt. If a glacier is in equilibrium or has a positive mass balance, then the glacial icemelt term is defined as the icemelt volume that is equal to, or less than, the water equivalent of snow that accumulates into the glacier system in a hydrological year. If a glacial mass balance is negative, then glacial wastage is defined as the volume of icemelt that exceeds the water equivalent of the annual volume of snow accumulation into the glacier system, causing an annual net loss of glacier volume. In short, glacial melt does not by itself imply a negative mass balance or wastage. By these definitions, on an annual basis, the presence of a glacier in a basin affects total streamflow volume through wastage contributions only. Glacial melt is a storage term and does not contribute to increased total annual streamflow. Within the hydrological cycle, both glaciers and groundwater are storage reservoirs. Following the convention of Comeau et al. (2009), snowfall on glaciers is analogous to groundwater recharge, glacial melt is analogous to groundwater extraction (or outflow from artesian aquifers), and glacial wastage is analogous
to groundwater overdraft. Persistant glacial wastage and persistant overdraft are both self-terminating.
Glacial melt can affect total streamflow on a seasonal basis, and its significance is manifest in its timing, as water is stored as snow accumulation into the glacier system and the water equivalent runoff is delayed until icemelt in the late summer months of the otherwise low streamflow. Therefore, the importance of glacial melt in terms of percentage contribution to streamflow is primarily on a seasonal timescale.
An understanding of ice dynamics is required to understand the response of glaciers to climate change (Armstrong, 2010). If climate and ice dynamics result in a glacier extending farther downslope with time, the advance of the terminus2 will increase the total glacier area. A time lag on the order of decades or longer occurs between a change in climate and glacier advance or retreat, and year-to-year glacier terminus changes are likely a response to climatic events that occurred several decades or more in the past. The majority of glaciers in the HKH region have response times on the order of decades to a few centuries (Humphrey and Raymond, 1994; Johannesson et al., 1989). The response time is influenced by a glacier’s area and volume, precipitation regime, debris cover, and topographic shielding or shadowing (Kargel et al., 2011). All these factors vary widely over the HKH and High Mountain region of Asia.
The easiest glacial property to measure is the location of the terminus. Simply by walking uphill to the start of a glacier, one can locate the terminus of the glacier. The terminus position for that year can be marked in any number of ways, including a simple pile of rocks. Some glaciers have accurate records of their terminus position that go back a hundred years or more. However, this simple measurement may yield erroneous information about a glacier’s retreat and rate of retreat over short timescales of a decade or so. Prolonged retreat of the terminus of a glacier over time-scales of several decades does indicate that the glacier is retreating.
The “glaciological” method for determining glacier mass balance relies on a network of stakes and pits on the glacier surface and measuring the change in surface level between two fixed dates (an annual mass balance) or at the end of the ablation and accumulation seasons (a seasonal mass balance) (Racoviteanu et al., 2008). This method is considered the most accurate and provides the most information about spatial variation (Kaser et al., 2003). However, there are currently no long-term glaciological mass balance records for the HKH region, and few measurements of glacial mass balance at all (Kaser et al., 2006).
Mass balance can be estimated using the “geodetic method.” This indirect method consists of measuring elevation changes of the glacial surface over time from various digital elevation models constructed over the entire glacier surface (Racoviteanu et al., 2008). Because of large uncertainties, the geodetic method can only be used to estimate glacier changes at decadal or longer timescales (Kaser et al., 2003; Racoviteanu et al., 2008).
The logistical difficulties caused by the rugged topography and remote location of glaciers in the region make remote sensing techniques of particular interest. Remote sensing allows for regular monitoring of glacier area, length, surface elevation, surface flow fields, accumulation/ablation rates, albedo,3 ELA, accumulation area ratio, and mass balance gradient. A more detailed description of glaciological, hydrological, geodetic, and remote sensing glacier measurement methods is presented in Appendix C.
Glacier Extent and Retreat Rates
The HKH region is often referred to as the “third pole” because it contains the largest ice fields outside the polar regions. Some of the largest glaciers in the world are located here, including the Siachen glacier on the north slopes of the Karakoram Range, which stretches to a length of about 72 km and is the largest nonpolar glacier. Additionally, the mountains and glaciers of the Himalayas are culturally important to the region’s population (Box 2.1).
2 The glacier terminus, sometimes called the glacier snout, is the lower end of a glacier.
3 Albedo is the ratio of reflected solar radiation to incident solar radiation for a specific surface and has a value between 0 and 1. For example, fresh snow has an albedo of about 0.8 (AMS, 2000).
BOX 2.1 Cultural Importance of the Himalayas
People have traditionally revered mountains as places of sacred power and spiritual attainment, and the Hindu Kush-Himalayan (HKH) mountains play a central role in the spiritual, as well as practical, lives of millions of people (Bernbaum, 1998). It is from the Himalayas that the Ganges River, considered by Hindus to be the holiest of all rivers in India, rises and cuts its path through the valleys and gorges before it enters the plains. The Ganges River draining the southern area of the Himalaya is considered by Hindus to be both a goddess and a river, Ganga Mata (Mother Ganges; Eck, 1998, 2012), and is seen as sacred along its entire length. Many believe that bathing in the Ganges frees one from past sins and liberates the soul from the cycle of birth and death.
The glaciers have particular cultural importance as the perceived source of water for the Ganges and other rivers in the HKH. This is demonstrated by anecdotal evidence from pilgrims who bathe in rivers and lakes near the outlet of glaciers. For example, the Gangotri glacier is a traditional Hindu pilgrimage site. Devout Hindus consider bathing in the waters near Gangotri town a holy ritual.
The HKH region is also home to Mt. Kailash, in western Tibet (6,600 m in elevation), considered by many religions to be the holiest mountain in the world. This mountain is venerated by Hindus, Buddhists, Jains, Sikhs, and believers of Bonri, the ancient Tibetan religion (Peatty, 2011). For Hindus, Mt. Kailash is the heavenly abode of Lord Shiva and his consort Parvati. Tibetan Buddhists view Mt. Kailash as the pagoda palace of Demchog, the One of Supreme Bliss (Bernbaum, 2006). Mt. Kailash is considered sacred in these religions in part because it is the headwaters of four major rivers aligned in the cardinal directions: the Indus, the Brahmaputra, the Karnali (a major tributary of the Ganges), and the Sutlej (a major tributary of the Indus).
One of Nepal’s most famous places of religious pilgrimage is Gosainkunda Lake (4,400 m in elevation). Every year during the Janai Purnima festival in August, thousands of Hindu and Buddhist pilgrims travel there by foot to bathe in the holy lake. Glacial meltwater is strongly associated with the major lakes and rivers in the HKH region. Rivers, glaciers, and mountains in the HKH are intertwined with the daily activities, spiritual lives, and the cultures of the local populations. Uncertainty surrounding the health of the glaciers and the rivers and lakes resonates deeply throughout these cultures.
The major concentrations of glaciers in the high mountain area of Asia cross more than 12 mountain ranges (Dyurgerov and Meier, 2005). There are currently no complete glacier inventories, but there is general agreement on the area of the glaciers in the region (Armstrong, 2010; Bolch et al., 2012). The total glacier coverage of the HKH and the Tibetan Plateau north to the Tien Shan4 is thought to exceed 110,000 km2, with about 50,000 identifiable glaciers (Dyurgerov and Meier, 2005). Table 2.1 summarizes glacial area estimates from different sources. However, comparisons of glacial area among different studies are difficult because spatial extents are often different or not well categorized.
Recent work by Jacob et al. (2012) shows that although previous estimates of mass loss in the region ranged from 47 to 55 Gt yr-1, the rate may be closer to 4 ± 20 Gt yr-1. The gaps and discrepancies in these various reports emphasize the need for a comprehensive glacial inventory of the region. In addition, more information about how glacier area is distributed with elevation would lead to a better understanding of how much glacial area is in vulnerable low-elevation areas (i.e., below the ELA). Glacial hypsometry plots the distribution of glacial area with elevation. Bajracharya et al. (2011) have developed a glacial hypsometry for Nepal (Figure 2.3). The hypsometry shows that glacial ice ranged in elevation from about 3,200 m to 8,500 m. The highest amount of glaciated area was in the 100-m-elevation band centered around 5,400 m. Glacial area decreases with both increasing and decreasing elevation.
Detailed glacier area measurements are not available for the full study area. However, the Committee calculated the proportion of glacier area in different elevation bands for the Indus and Ganges/Brahmaputra. In both basins, the majority of glacier area is in the 5,000- to 6,000-m band (Figure 2.4), with a significant amount in the 4,000- to 5,000-m band. The Indus Basin has a slightly greater proportion of its glacier area below 4,000 m than the Ganges/Brahmaputra Basin, whereas the Ganges/Brahmaputra has a slightly greater proportion of its glacier area above 6,000 m. Although these values should be considered qualitative, they are consistent with the more rigorous hypsometry data from Nepal (Figure 2.3). The differences are small, but they suggest that glacial retreat would be more sensitive to changes in climate in the Indus Basin than in the Ganges/Brahmaputra Basin; however, this qualifies
4 A large mountain system located in Central Asia and to the north of this report’s study area.
TABLE 2.1 Glacial Area Estimates from Different Studiesa
|Region||Glacier Area (km2)||Data Source|
|116,180||Xu, J., et al. (2009)|
|99,261||Yao et al. (2012)|
|Central HKH||33,050||WGMS (2008)|
|32,182||ICIMOD: Eriksson et al. (2009)|
|71,182||Indian Space Agency: ISRO (2011)|
|Himalayas||33,050||Dyurgerov and Meier (1997, 2005)|
|Karakoram||15,400||Dyurgerov and Meier (1997)|
|16,000||Dyurgerov and Meier (2005)|
|16,600||Yao et al. (2012)|
|Indus Basin||32,246||ISRO (2011)|
|Ganges Basin||18,392||ISRO (2011)|
|Brahmaputra Basin||20,542||ISRO (2011)|
|China||59,406||Chinese Academy of Sciences: Liu, et al. (2000)|
|India||37,959||Geological Survey of India in ICIMOD (2011b)|
|aComparisons of glacial area among different studies are difficult because spatial extents are often different or not well categorized.|
evidence that glaciers are more stable in the western Himalayas. This is because glacial retreat is sensitive to more factors than simply elevation, including precipitation regime, local temperatures, and debris cover.
Rates of glacial retreat in the HKH are not well understood because of a lack of field data (Kargel et al., 2011; Thompson, 2010), making it difficult to understand regional climate change impacts (Scherler et al., 2011b). One of the most studied glaciers in the region, AX010 in Nepal, has consistently been shown to have a negative mass balance. If the climate conditions remain consistent with the period 1992 to 1996, AX010 has been predicted to disappear by the year 2060 (Kadota, 1997). However, this glacier is relatively small, with an area of only 0.38 km2, and exists at a low altitude, extending to only 5,300 m, and thus only represents small, low-elevation glaciers (Fujita and Nuimura, 2011). However, approximately 50 percent of the area of Nepal glaciers is at altitudes above approximately 5,400 m (Alford et al., 2010; Bajracharya et al., 2011). Therefore, glacier AX010 is not a good indicator of general trends in the HKH region. In a study of glaciers in northern India, Kulkarni et al. (2011) found that glaciers smaller than 1 km2 lost an average of 28 percent of their area between 1962 and 2001, while glaciers greater than 10 km2 lost an average of 12 percent of their area in the same time period, further indicating that smaller glaciers cannot be used to determine regional trends.
Extrapolation of these few mass balance studies over the greater High Asian region has been used to estimate a rate of water loss from glacial retreat between 2002 and 2006 of -55 Gt yr-1 for this entire region, with -29 Gt yr-1 over the eastern Himalayas alone (Dyurgerov, 2010). In contrast, Jacob et al. (2012) used new information from the Gravity Recovery and
FIGURE 2.3 Glacial area in Nepal is shown as a function of elevation. Glacial ice ranges from about 3,200 m to 8,500 m in elevation. Total glacial area decreased between 2001 (red line) and 2010 (black line) although the highest amount of glacial ice remained at about 5,400 m in elevation. Glacial area decreases as the elevation increases or decreases from 5,400 m. SOURCE: Bajracharya et al. (2011).
Climate Experiment (GRACE)5 satellite mission to estimate a mass loss of only -4 Gt yr-1 for the region of High Asian mountains for the period 2003 to 2010. The much lower estimate of glacier loss from analysis of the GRACE data is at least in part because the GRACE satellite information integrates over the entire region, in contrast to the study by Dyurgerov (2010), which by necessity extrapolated the few glacial mass balance measurements collected at low elevations over the entire region.
Similarly, a recent time series using the geodetic approach based on recently released stereo Corona satellite imagery (years 1962 and 1970), aerial images, and recent high-resolution satellite data (Cartosat-1) to determine mass changes for 10 glaciers south and west of Mt. Everest, Nepal, show a specific mass loss between 1970 and 2007 of 0.32 ± 0.08 m of water equivalent per year. These results are consistent with the global average (Bolch et al., 2011). Terminus measurements of 466 glaciers in the Chenab, Parbati, and Baspa basins in the Indus catchment showed significant deglaciation (Kulkarni et al., 2007). Various studies have estimated the retreat of individual glaciers in the region: the Bhagirath Kharak glacier in Uttarakhand, India, retreated 7 m per year between 1962 and 2005 (Nainwal et al., 2008); the Dokriani glacier in Uttarakhand, India, retreated 550 m between 1962 and 1995 (Dobhal et al., 2004); the Parbati glacier in Himachal Pradesh, India, retreated 578 m between 1990 and 2001 (Kulkarni et al., 2005); the Satopanth glacier in
FIGURE 2.4 Glacial area is shown as a function of elevation for the Indus (black bars) and Ganges/Brahmaputra (gray bars) basins. In both basins, the majority of glacier area is in the 5,000- to 6,000-m-elevation band. Comparing the two basins, the Indus Basin has a greater proportion of its glacier area below 4,000 m in elevation than the Ganges/Brahmaputra, and the Ganges/Brahmaputra has a greater proportion than the Indus of its glacier area above 6,000 m in elevation. SOURCE: Based on data from the Natural Earth dataset of the Digital Chart of the World product and the Hydrosheds database.
5 The GRACE signal is heavily influenced by groundwater extraction and subcrustal mass and plate movement. The HKH region is a large, complex, and tectonically active area with substantial groundwater depletion. Therefore, use of GRACE satellite data for mass balance measurements in the HKH region leads to substantial uncertainties (e.g., Bolch et al., 2012). Moreover, the use of GRACE satellite data to understand groundwater is complicated by the fact that the coarse resolution of GRACE disallows understanding of groundwater overdraft patterns at the scale of individual or local community consumptive use.
Uttarakhand, India, retreated 23 m per year between 1962 and 2005 (Nainwal et al., 2008). Although there is remaining uncertainty about the retreat rates of specific glaciers (e.g., the Gangotri glacier in northern India; Ahmad and Hasnain, 2004; Bhambri et al., 2011; Kumar, K., et al., 2008; Kumar, R., et al., 2009; Naithani et al., 2001) and more mass balance measurements are needed, glaciers of the eastern HKH region, in general, have a negative mass balance and are retreating, but not at higher rates than other mid-latitude glaciers (Bolch et al., 2012; Racoviteanu, 2011). Uttarakhand, India, retreated 23 m per year between 1962 and 2005 (Nainwal et al., 2008). Although there is remaining uncertainty about the retreat rates of specific glaciers (e.g., the Gangotri glacier in northern India; Ahmad and Hasnain, 2004; Bhambri et al., 2011; Kumar, K., et al., 2008; Kumar, R., et al., 2009; Naithani et al., 2001) and more mass balance measurements are needed, glaciers of the eastern HKH region, in general, have a negative mass balance and are retreating, but not at higher rates than other mid-latitude glaciers (Bolch et al., 2012; Racoviteanu, 2011).
In contrast to the eastern HKH, Hewitt (2005) report that in the western HKH, there has been expansion of the larger glaciers in the Karakoram region6 since about 1990, particularly those at higher altitude. Similarly, Scherler et al. (2011b) report that for the Karakoram region, 58 percent of the studied glacier fronts were stable or slowly advancing with a mean rate of about +8 ± 12 m yr-1. Surging of glaciers has been observed in Karakoram glaciers, but more field observations are needed to confirm whether this indicates a positive mass balance. Data from the late 1980s indicated a possible trend of negative mass balance for the Siachen Glacier (Bhutiyani, 1999), but more recent evidence from remote sensing data shows that glaciers in the central Karakoram had a slightly positive mass balance between 1999 and 2008 (Gardelle et al., 2012). The western end of the HKH appears to show slower rates of retreat, less formation of pro-glacier lakes associated with flood hazard, and frequent observations of advancing glaciers, compared with the eastern region (Armstrong, 2010; Bolch et al., 2012; Hewitt, 2005). For the region as a whole, the loss of glacial ice over the last decade is much less than previously thought (e.g., Dyurgerov, 2010).
Possible Changes in Glacier Extent and Volume
There are few studies of the response of HKH glaciers to changes in climate (Cogley 2011). Glacial mass balances in the Karakoram area of the HKH appear to be stable, with some of the larger glaciers experiencing positive mass balances. These results suggest that there will be little change in glacier extent over the next several decades in this part of the HKH region. In the eastern HKH, glaciers are retreating, at rates similar to those in the rest of the world. Recent evidence shows that glaciers may be receding at smaller rates than previously estimated ( Jacob et al., 2012), although there is still uncertainty in estimates of glacial retreat. The evidence to date suggests little change in rates of glacial retreat and glacial extent in the eastern HKH over the next two to three decades. Even if this is the case, the rate of glacial retreat could increase in the future with appropriate changes in climate forcing.
Currently, retreat rates in the eastern HKH are highest at elevations below 5,000 m. This is particularly serious for glaciers with maximum elevation below 6,000 m. These small, low-elevation glaciers are expected to sustain high rates of retreat. High-elevation communities and activities that depend on glacial meltwater generated by these small glaciers are most likely to experience the impact of these retreating glaciers. The Committee cannot state with certainty whether major changes in either rates of glacial retreat or glacial extent in the HKH region will occur for the next several decades. However, below is a worst-case scenario over a timescale of several decades that could result in very high rates of glacial retreat.
A worst-case scenario of extensive glacial retreat is within the bounds of possibility. One scenario involves albedo feedback processes. Because of the large energy-albedo feedbacks associated with snow and ice, small changes in the amount and timing of snow, and in the overall energy balance can have large effects on a glacier’s mass balance. Fresh snow has an albedo range of about 0.75 to 0.95. In contrast, glacial ice has an albedo range of about 0.3 to 0.4. Removing snow from a glacier, holding other factors (air temperature, cloud cover, etc.) constant, results in a 200 to 300 percent increase in the delivery of energy to the surface of the glacier. As the exposed glacier ice heats up and then melts, more nearby snow also melts, resulting in more energy delivery to the glacier, and more glacial wastage. This process results in a runaway positive feedback signal that can accelerate the wastage of glacial ice. This albedo feedback process is currently occurring in Arctic sea ice.
Another scenario involves a regional change in monsoonal activity that reduces snowfall in the Himalayas, coupled with high amounts of black carbon depo-
6 The Karakoram region is a large mountain range spanning the border between Pakistan, India, and China.
sition, resulting in very high wastage rates. Imposition of increased air temperatures caused by black carbon heating of the atmosphere would accelerate the rates of glacial wastage. A change in monsoonal activity could result in less snowfall and more exposed glacial ice, which could itself lead to high wastage rates. As discussed later in this chapter, high amounts of black carbon are being entrained in the atmosphere and deposited in the HKH region, decreasing the albedo of glacier ice and snowpack. This decrease, even with no increase in air temperature, could lead to increased surface wastage. In the monsoonal region of the Himalayas, decreased albedo from deposition of black carbon is mitigated by repeated, almost daily, snowfall during the monsoon. Black carbon deposition is also mitigated by snow turnover processes. However, if new snow does not fall because of changes in monsoonal activity, then black carbon transported from the Indo-Gangetic Plain could accumulate on the snow surface, causing an acceleration of wastage rates. Furthermore, in this scenario with less monsoon precipitation and more glacial wastage, the contribution of glacial wastage to summer stream-flow would become more important. Such accelerated wastage rates could occur even without a change in air temperature. However, high atmospheric loads of black carbon heat the atmosphere, which would further accelerate melt rates. Another important effect of black carbon could be to change the phase of precipitation (e.g., from snowfall to rainfall). Changes in climate could also result in a shift in the precipitation phase and number of snow days in the region. For example, Shekhar et al. (2010) found significant variations in snowfall trends in the western Himalayas. More precipitation phase data are needed to fully understand whether snowfall events in the region are changing and how such changes will affect glacial mass balance. With the right conditions, accelerated rates of glacial retreat beyond present rates are a possibility. If such a situation does arise, most likely it would be local in origin and not global or even consistent throughout the entire HKH region.
The HKH region features one of the world’s steepest slopes of an extended mountain range, rising from its base in the alluvial Indo-Gangetic Plain near sea level to the great height of the Tibetan Plateau (~ 8,000 m) in a distance of 100 to 400 km across the width of the arc. Together with the Tibetan Plateau, the Himalayas exert great influence on the powerful Asian monsoon system (Figure 2.5). As such, there is a very high climatic gradient across the region.
The region’s climate ranges from tropical at the base of the foothills to permanent ice and snow at the highest elevations. During the late spring and early summer, the Plateau surface heats up quickly and serves as an elevated heat source, which draws warm and moist air from the Indian Ocean toward the Himalayas and Tibetan Plateau region. As the monsoon flow transports moisture from the Arabian Sea to the Indian subcontinent, it spurs heavy monsoon rain over the Indo-Gangetic Plain and the Bay of Bengal. During the winter, the low-level monsoon flow reverses to northeasterly, with prevailing large-scale subsidence and relative dry conditions over India.
Over the Tibetan Plateau, rainfall is scarce all year round with annual totals of 100 to 300 mm. Most of the precipitation falls in the form of snow in winter, with more than 50 percent of the land at elevation above 5,000 m covered by snow. In the summer, the snow
FIGURE 2.5 The moist air currents that drive the South Asian Monsoon are indicated by white arrows. Monsoon flow transports moisture from the Arabian Sea to the Indian subcontinent, resulting in heavy monsoon rain over the Indo-Gangetic Plain and the Bay of Bengal. SOURCE: Hodges (2006).
cover fraction drops to below 30 percent at the same elevation. The melted snow reveals an arid stepped landscape interspersed with scattered glaciers and large brackish lakes.
The climatic gradient is strong not only across, but also along the arc of the Himalayas (Figure 2.6). The Sutlej valley serves as a rough dividing line between the climate regimes of the western and eastern Himalayas (Bookhagen and Burbank, 2010). In the Karakoram in the west, about two-thirds of high-altitude snowfall is due to the mid-latitude westerlies. In the east, more than 80 percent of annual precipitation is from the summer monsoon. (Bolch et al., 2012).
The westernmost portion of the region includes the high mountains and glaciers of the Hindu Kush and the Karakoram, with a large number of rivers flowing into the upper Indus River Basin in Pakistan, eventually draining into the northern Arabian Sea. This region adjoins the arid, rugged regions of Afghanistan in the west, and the Thar Desert of northwestern India to the south. It has a relatively dry climate, with annual precipitation of 400 to 600 mm, primarily from wintertime storms associated with the mid-latitude westerlies. In the cold arid regions of Ladakh, India, the precipitation is somewhat higher in summer, but the mean annual precipitation is as low as 115 mm per year (Thayyen and Gergan, 2010).
FIGURE 2.6 The climate varies across the HKH region. In the west, indicated by purple, the climate is alpine and dominated by the mid-latitude westerlies. Most precipitation takes the form of winter snow. This area adjoins a cold arid climate regime, indicated by blue. In the east, indicated by yellow, the climate is dominated by the summer monsoon, with most of the precipitation coming during the summer months. The Indus, Ganges, and Brahmaputra watersheds are also shown. SOURCE: Thayyen and Gergan (2010).
Further east along the arc is the Greater Himalayan Range. This region includes snow-capped high mountains and foothills in northwestern India, Nepal, and Bhutan, forming the northern boundary of the fertile and populous Indo-Gangetic Plain, where the Ganges River flows. Rainfall is higher in the east, mostly from summer monsoon rain. The Bay of Bengal, in which the Ganges/Brahmaputra rivers flow out to sea is the wettest part of the Indian monsoon region. Bookhagen and Burbank (2010) reviewed precipitation data for the 10-year period from 1998 to 2007. They showed that mean annual rainfall ranges from ~1 to more than 4 m in the monsoon-precipitation-dominated portions of the region.
Role of Aerosols in Regional Climate
Aerosols are suspended fine particles in the atmosphere that have both natural and manmade sources. Aerosols from natural sources such as desert dusts have been known to coexist with the Indian monsoon in the eastern HKH region for a long time. During April and May, dusts are transported by the mid-latitude westerlies from the deserts in the Middle East and Afghanistan and the Thar Desert in northwestern India to the Indo-Gangetic Plain and Himalayas.
Since the Industrial Revolution, atmospheric loading of aerosols from manmade sources such as factories, power plants, cooking and heating, and slash-and-burn agricultural practices has greatly increased, making the Indo-Gangetic Plain one of the most polluted regions in the world. These aerosols often appear in the form of a brownish haze known as atmospheric brown clouds (Ramanathan et al., 2005). A key component of these brown clouds is black carbon, commonly known as soot. Black carbon sources include internal combustion engines, power plants, heat boilers, waste incinerators, slash-and-burn agricultural activities, and forest fires. Although some aerosol species have a cooling effect, airborne black carbon strongly absorbs solar radiation and heats up the atmosphere. Recent studies have shown that aerosols, in particular, black carbon because of its ability to heat the atmosphere, can affect the regional and global water cycles, including the Himalayan snowpack and glaciers, by altering the radiation balance of the Earth’s atmosphere and surface and modulating cloud and rain formation processes
(Lau et al., 2010; Ramanathan et al., 2005; Rosenfeld et al., 2008). Because aerosols have the capability to regulate atmospheric heat sources and sinks, modulate monsoon rainfall, surface evaporation, and river runoff, and possibly affect melting of high mountain snowpack and glaciers, they are an integral component of the monsoon climate system.
The atmospheric loading of aerosols is measured in terms of the aerosol optical thickness (AOT), which is quantitatively determined by the amount of solar radiation attenuation at Earth’s surface by the aerosol. During the late spring and early summer season (April to June), the AOT builds up dramatically over the Indo-Gangetic Plain and northwestern India (Figure 2.7, upper panels). The monsoon flow is blocked by the Tibetan Plateau and forced to rise over the Himalayas foothills and northern India. As a result, aerosols transported from remote deserts and from local emissions accumulate against the Himalayan foothills to a great height (>5 km) and spread over the entire Indo-Gangetic Plain and regions to the south (Figure 2.7, lower panel). Additionally, the southwest monsoon flow brings increasingly warm, moist, and unstable oceanic air from the Indian Ocean and the Arabian Sea to the Indo-Gangetic Plain and the Himalayan foothills. The mixture of dust and black carbon in the deep aerosol layer provides efficient heating of the atmosphere. It interacts with the warm moist monsoon air and maximizes the atmospheric water-cycle feedback, and may significantly modulate the summer monsoon rainfall (Lau et al., 2006, 2008).
In contrast, during winter, the prevailing monsoon flow is cold, dry northeasterly with large-scale subsidence. Local emissions of aerosols from the Indo-Gangetic Plain are transported by the northeasterly flow in the form of atmospheric brown cloud plumes emanating from the source region over the Indo-Gangetic Plain to the adjacent ocean (Figure 2.8, upper panels), and are trapped within the stable and low boundary layer (<1 km; Figure 2.8, lower panel). In winter, the atmospheric brown clouds have a higher contribution from local black carbon emissions, but little contribution from dust, because of lack of deserts upwind. The black carbon aerosol heats the boundary-layer air, but cools the land surface, thus further increasing atmospheric stability, suppressing convection and the already-scarce wintertime rainfall. The wintertime high aerosol loading within the boundary layer in the
FIGURE 2.7 Climatological monthly distribution of aerosol optical thickness (AOT) during April, May, June from MODIS (upper panels) and vertical distributions across the Tibetan Plateau from CALIPSO (lower panel, horizontal scale shows latitude/longitude coordinates) show the deep and extended layer of aerosols over vast regions of the Indo-Gangetic Plain and Himalayan foothills. Some aerosols can be detected over the top of the Tibetan Plateau. SOURCE: upper panel, Gautam et al. (2010); lower panel, Gautam et al. (2009b).
FIGURE 2.8 Monthly distribution of aerosol optical thickness (AOT) during November, December, January from MODIS (upper panels) and vertical distributions across the Himalayan Tibetan Plateau from CALIPSO (lower panel, horizontal scale shows latitude/ longitude coordinates) show high concentration of aerosols confined within the shallow boundary layer over the Indo-Gangetic Plain and Himalayan foothills. SOURCE: Ritesh Gautam, personal communication.
Indo-Gangetic Plain is well known for its adverse impacts on local climate, especially on visibility, aviation safety, and human health. However, the aerosols will have minimal large-scale atmospheric water-cycle feedback because the large, stable boundary layer and the free atmosphere above are effectively isolated from the aerosol forcing.
Possible Effects of Aerosols on
Glacier and Snowpack
The growth and decay of Himalayan snowpack and glaciers are strongly dependent on changes in the fluctuations of South Asian monsoon climate, and aerosols can affect monsoon temperature and rainfall (Ramanathan and Carmichael, 2008). Ramanathan et al. (2005) suggested that through the reduction of surface solar radiation—the so-called dimming effect—aerosols cool the Earth’s surface beneath the aerosol layer. The cooling reduces the land-sea thermal contrast and north-south sea surface temperature gradient, which in turn may lead to weakening of the Asian monsoon.
Other studies have focused on the effect of solar heating by absorbing aerosols (desert dust and black carbon) and induced atmosphere-land feedbacks on the Asian monsoon (Bollasina et al., 2008; Lau and Kim, 2006; Lau et al., 2006, 2008; Meehl et al., 2008; Menon et al., 2002; Wang et al., 2009). Observation studies have shown that desert dust aerosols found over the Indo-Gangetic Plain are more absorbing compared with natural dusts from other parts of the world (Eck et al., 2005; Ramana et al., 2004). These studies suggest that dust particles transported from the adjacent deserts may have been coated with fine soot particles during their passage over the highly industrial regions of the Indo-Gangetic Plain, thus becoming stronger absorbers of solar radiation. It is possible that these highly absorbing aerosols could be instrumental in providing anomalous heating of the atmosphere over northern India and the Tibetan Plateau, through the atmospheric water-cycle feedback—the so-called elevated heat pump (EHP) effect—causing monsoon rainfall to shift northward toward the Himalayan foothills and northern India during the late spring to early summer monsoon season.
The EHP effect is thought to be most effective in May and June because of the great depth and vast horizontal extent of the aerosol layer over the Indo-Gangetic Plain, and its interaction with the warm, moist unstable environment. In contrast, the EHP effect is minimized in the winter because of the cold, dry stable air mass and the confinement of aerosols within the shallow atmospheric boundary layer and narrow regions over the Indo-Gangetic Plain. On the basis of experiments with a global climate model, Lau et al. (2010) showed that the EHP effect caused by dust and soot over the Indo-Gangetic Plain during late spring and early summer could lead to early and accelerated melting of the Himalayan seasonal snowpack via aerosol-induced atmosphere-land surface feedback.
Other studies, however, have questioned and debated the importance of the EHP based on analyses of observations (e.g., Lau and Kim, 2011b; Nigam and Bollasina, 2010). Several global modeling studies have suggested that much of the aerosol forcing of monsoon changes may be caused by nonlocal processes. These include changes in radiative fluxes reaching the ocean surface that help power the monsoon (Wang et al., 2009), consistent with theoretical work on the importance of moist processes rather than land-sea contrasts in creating the monsoon (Boos and Kuang, 2010), remote sea surface temperature forcing (Meehl et al., 2008), and large-scale atmospheric circulation changes (Bollasina et al., 2011). Various studies have also shown evidence that the summer monsoon has weakened over the past 50 years. Analysis of rainfall data for the second half of the 20th century showed a decreasing trend in the length of the monsoon (Dash et al., 2009; Ramesh and Goswami, 2007). Model results also indicate reduced rainfall over India, with small increases over the Tibetan Plateau (Meehl et al., 2008). Although current studies are consistent in finding that aerosols can substantially perturb the timing and intensity of the South Asian monsoon, further work is needed to clarify the magnitude of these effects and to determine the relative importance of the various proposed mechanisms.
An additional important mechanism whereby absorbing aerosols may cause early melting of snow-pack and accelerated glacial retreat is the so-called snow-darkening effect. Dust and soot transported from emission sources in the Indo-Gangetic Plain and from remote regions have been found as impurities deposited in Himalayan snowpack and glaciers. These impurities caused a characteristic reduction in the reflectivity of the snow and ice surface in the visible range (Warren and Wiscombe, 1980), lowering the albedo of snow and ice and leading to more solar radiation absorption and possible accelerated rate of snowmelt and glacial retreat.
Recent studies in the HKH region implicate snow darkening by black carbon and dust as a possible cause of glacial retreat. Xu, B. et al. (2009) deduced from observed ice-core samples in Tibetan glaciers that black carbon deposition may have contributed to their rapid retreat. Large, shallow glaciers such as those in the western HKH will be affected more by black carbon deposition than small glaciers with steep surrounding mountains such as those in the eastern HKH, because their snow-turnover rate7 is higher. This indicates some uncertainty about the effect of black carbon on glacial retreat, but also that there are many factors that influence glacial retreat, and these factors interact in ways that are difficult to predict. Menon et al. (2010) found from model experiments a 0.9 percent reduction in snow/ice cover over the Himalayan region between 1990 and 2000 due to increased aerosol loading over the Indian subcontinent, with a large contribution from black carbon caused by emissions from coal and biofuel. Shindell et al. (2012) found that reductions of black carbon and coemitted pollutants lead to large decreases in Himalayan snow/ice albedo forcing, and that the net mitigation of surface warming by the emissions reductions was strongly enhanced in at least part of that area. However, because of the use of low-resolution models in these studies, it is unlikely that details of aerosol transport across the Himalayan ranges are captured. Further observations are needed to confirm the results of these model studies. For example, Shrestha et al. (2010) found that black carbon was about 10 percent of the total aerosol composition, based on measurements at two low-latitude sites in central Nepal. The effect of different types of snow cover, including black carbon, on albedo have been investigated using remote sensing (Negi and Kokhanovsky, 2011a,b; Negi et al., 2009). Surface observations of the amount of black carbon
7 Snow turnover is due to avalanches, gravitational processes, and event-driven snowfall.
deposited on glacier surfaces and how much it affects the albedo are needed.
Kaspari et al. (2011) found increased black carbon concentrations in century-old deep ice core consistent with black carbon from manmade sources being transported to high elevations of the Himalayas (although the reported total black carbon concentrations were surprisingly large, raising some concerns about data quality). Using atmospheric black carbon loading at the Ev-K2 Nepal Climate Observation-Pyramid site, Yasunari et al. (2010) estimated a plausible reduction of surface albedo due to black carbon deposition of 2 to 6 percent, and a corresponding 10 to 30 percent increase in the annual runoff for a typical Tibetan glacier. Using MODIS satellite data and comparing snow albedo before and after major dust storms, Gautam et al. (2011) estimated a 6 to 8 percent reduction in surface albedo. However, observations of dust and black carbon and their effects of surface albedo are still scarce, with large uncertainties due to a lack of instrumental records and limited sampling. Likewise, representation of snow-darkening processes in climate models is still in early development (Flanner et al., 2007, 2009; Yasunari et al., 2011).
Current Trends and Projections
of Regional Climate
Reports of surface air temperature and precipitation trends vary greatly across the Himalayas. An increase in maximum and minimum daily temperature of 1.0 to 3.4 °C was found across the Himalayas between 1988 and 2008 (Bhutiyani et al., 2007; Shekhar et al., 2010; Shrestha et al., 1999). In Nepal, the maximum daily surface temperature was found to be increasing at a rate of 0.5 to 1.0 °C per decade from the late 1960s to the mid 2000s. The temperature increase has a tight gradient across the Himalayas with the faster rates of warming at higher elevations, and slower warming rates or even a slight cooling trend found in the lower elevations. This rate of surface warming is more than five times that of global warming by greenhouse gases, suggesting the importance of local heating processes. Note, however, that many of the high-elevation stations are located in deep valleys, and are not actually high elevation. They are not evenly distributed and often are located in villages or urban centers, increasing the potential for urban heat island effects on the data.
In contrast, in the northwestern Himalayas and Karakoram, the trends are less clear. A decreasing trend of both maximum and minimum temperature of 1.6 and 3.0 °C, respectively, has been found (Shekhar et al., 2010). Other studies show a general warming trend, with an increase of 0.06°C per decade during the monsoon season and 0.14°C per decade during the winter (Bhutiyani et al., 2010). Over the same region, an increasing trend in winter precipitation, most likely associated with changes in the mid-latitude westerlies, has also been reported (Fowler and Archer, 2006). However, a recent review of instrumental records in the northwestern Himalayas and Karakoram show no trend in winter precipitation and a decrease in monsoon precipitation between 1866 and 2006 (Bhutiyani et al., 2010). Note that many of these findings are based on in situ historical observations from a very sparse network of stations, especially at higher elevations. Although they provide an extremely important picture of climate trends in the HKH region, they are subject to uncertainties due to inadequate spatial and temporal resolution, possible data inhomogeneity, and local effects.
Satellite remote sensing data have revealed critical spatial and temporal information regarding temperature and precipitation changes in recent decades in the HKH region. Microwave satellite measurements have revealed a widespread warming trend in the troposphere column above the Himalayan-Gangetic region (Figure 2.9).
The warming trend is most pronounced in the premonsoon season (April to May) over the western Himalayas and northern Pakistan region, with a maximum warming rate of approximately 0.8 °C per decade between 1979 and 2007 (Figure 2.9, right panel). This widespread warming also gives rise to an increase in tropospheric land-sea thermal gradient (Figure 2.9, left panel), which favors a stronger monsoon. Prasad et al. (2009) reported large seasonal variations in the tropospheric warming trend, with the statistically significant enhanced warming during the months December to May, and more warming over the western than the eastern Himalayas. Interestingly, the premonsoon tropospheric warming coupled with the cooling at the surface inferred from tree-ring records in Nepal in recent decades is consistent with the direct effect
FIGURE 2.9 (a) Zonal mean (40° to 100° E) latitudinal profile of mid-tropospheric temperature trend for the premonsoon season (March-April-May) from 1979 to 2007; (b) spatial distribution of the mid-tropospheric temperature trend over the Indian Monsoon region in May. Open circles denote significance of linear trends at 95 percent. The warming trend in the premonsoon period gives rise to an increase in the tropospheric land-sea thermal gradient. SOURCE: Gautam et al. (2009b).
of absorbing aerosols in warming the atmosphere and cooling the surface (Ramanathan et al., 2005; Satheesh and Ramanathan, 2000).
Figure 2.10 shows that, from satellite measurements, there is high coherence between interannual variations of tropospheric temperature and aerosol index for absorbing aerosols over the Himalayan foothills and the Indo-Gangetic Plain, with both temperatures and aerosols showing a steady rising trend between 1979 and 2007. From field measurements of aerosol optical properties and surface solar flux, the aerosols over the Himalayas-Gangetic Plain are found to be very absorbing with single scattering albedo of 0.89 ± 0.01 (at 550 nm). Scaled by the vertical distri-
FIGURE 2.10 Red and blue solid lines show interannual variations of temperatures in the middle (4 to 7 km) and lower (surface to 4 km) troposphere, respectively, over northern India (25 ° to 35 °N, 69 ° to 82 °E) for March-April-May (MAM) period between 1979 and 2007. Dashed lines indicate the linear trends within the middle (red) and lower (blue) troposphere, respectively. Aerosol index variations during MAM over northern India since 1979 are shown by the black dashed line. There is high coherence between interannual variations of tropospheric temperature and aerosol index for absorbing aerosols over the Himalayan foothills and the Indo-Gangetic Plain, with both temperature and aerosol showing a steady rising trend between 1979 and 2007. SOURCE: Gautam et al. (2009b)
bution of aerosols from the spaceborne Cloud Aerosol Lidar with Orthogonal Polarization (CALIOP) observations, this value of single scattering albedo yields a model simulated aerosol-induced solar heating profile with maximum heating in the middle troposphere (4 to 5 km) on the order of 2 to 5 °C per day.
In conjunction with the tropospheric temperature increases, Gautam et al. (2009a) also found that the early summer monsoon rainfall over northern India has been on the rise since the 1950s, with over 20 percent increase for the period 1950 to 2004. These results are in agreement with recent observations of increased premonsoon rainfall over northern India (Lau and Kim, 2010), and suggest a future possibility of an emerging rainfall pattern of a wetter monsoon over northern India in late spring and early summer followed by a drier period in central and southern India. Other studies have indicated a weakening of the monsoon, with a trend of fewer rain days over the latter half of the 20th century (Dash et al., 2009; Meehl et al., 2008; Ramesh and Goswami, 2007).
A growing number of recent modeling studies have suggested a causal relationship between Himalayan snow cover changes and the increased loading of absorbing aerosols in the Indo-Gangetic Plain. Menon et al. (2010) reported that the snow cover in the Himalayas estimated from satellites declined by about 5 to 10 percent during the period 1990 to 2001, and that about 30 percent of the decline is due to increased black carbon aerosols in India. Flanner et al. (2009) found that the effects of carbonaceous aerosols are comparable to those due to greenhouse gases in causing the decline in Eurasian springtime snow/ice cover. Lau et al. (2010) showed that the effect of black carbon and dust heating of the atmosphere can lead to an early thawing of the snowpack starting in April, and accelerated melting in May to June in the western Himalayas by 8 to 10 percent. Qian et al. (2011) showed deposition of black carbon in snow cover over the Himalayas and Tibetan Plateau can contribute to a significant change in the anomalous surface radiative fluxes, and rainfall redistribution in the Asian monsoon from preindustrial to present-day climate forcing.
In addition to greenhouse gas warming, regional effects of atmospheric heating and the darkening of the snow and ice surface by dust and black carbon may play an important role in causing the rainfall and temperature trends, as well as accelerated retreat of the Himalayan glaciers and snowpack (Flanner et al., 2009; Gautam et al., 2009a,b, 2010; Lau and Kim, 2006; Lau et al., 2008, 2010; Menon et al., 2010; Qian et al., 2011; Ramanathan et al., 2007; Yasunari et al., 2010). Moreover, black carbon on snowpack and glaciers could have a time-delayed effect. A recent study showed that in the Tien Shan Glacier in Tibet, black carbon concentration at the bottom of unmelted snowpack is much higher than at the surface, because of flushing of black carbon by snowmelt that refreezes at lower layers (Xu et al., 2012). If the ELA rises in response to warmer temperatures, the black carbon-rich snow may be reexposed, increasing solar absorption, and further accelerating the retreat of the snowpack or glacier.
The questions of how aerosols may have affected and will further affect the Asian monsoon and Himalayan glacial retreat in the future remain subjects of ongoing research. One scenario is that unless effective emission control on black carbon is achieved, the continued increased atmospheric loading of black carbon over Asia will exacerbate the solar dimming effect, causing further reduction in land-sea thermal contrast and meridional temperature gradient. Such reductions will in turn reduce evaporation and available moisture in the atmosphere, leading to a spindown of the monsoon large-scale circulation and thus a weakening of the monsoon with reduced rainfall over the Indian subcontinent (Meehl et al., 2008; Ramanathan et al., 2005). Another scenario (Gautam et al., 2009b, 2010; Lau et al., 2009; Wang et al., 2009) is that the initial heating of the atmosphere by increasing dust and black carbon in the premonsoon period may actually increase the meridional tropospheric temperature gradient and induce atmospheric feedback, bringing more moisture and rainfall into the Himalayan region in the late spring and early summer season. This effect will lead to early and accelerated seasonal wasting of the Himalayan snowpack. The increased cloudiness and rainfall in the early monsoon could lead to lower land surface temperature and, subsequently, an early termination or a weakened peak monsoon through atmosphere-land surface and cloud feedback. How much these two scenarios have been in effect in explaining historical trends in monsoon rainfall is still a matter of debate and active ongoing research. The fate of the Himalayan glaciers is strongly dependent on precipitation changes in the
region, associated with the vagaries of the Asian summer monsoon and the mid-latitude westerlies winter precipitation regime. Better data from both ground-based and space-based observations as well as improved modeling capability are required to understand and predict the trends of precipitation, temperature, and glacial retreat and their impacts on river runoff and freshwater supply affecting downstream populations.
Evaluation of model projections is largely based on comparison of their historical simulations with observations. As described previously, temperature shows a general warming trend, over the last three decades in particular (Christensen et al., 2007). Evidence shows that temperature is increasing more rapidly with altitude, although there are fewer observation stations at higher elevations, leading to a potential data bias. The eastern and central Himalayan regions (dominated by the monsoon) show a decline in precipitation over the past ~50 years, whereas the western Himalayas show no precipitation trend in the monsoon season.
Winter precipitation in the western Himalayas shows an increasing trend. The relative importance of various forcing agents and of unforced, internal variability in the climate system in driving these trends is not clear. Nonetheless, the ability of models to reproduce these trends remains one of the most important tests of the credibility of model projections.
Climate model projections show warming throughout the region and in all seasons, with greater magnitude over the Tibetan Plateau than over the South Asian continent south of the Himalayas (Figure 2.11). Temperature changes over the next two to three decades show similar spatial patterns but a smaller magnitude. These trends are consistent across models in the sense that the ensemble mean change is greater than the standard deviation across the individual models, and are consistent across scenarios although the magnitude varies somewhat (about 25 percent greater or less than the A1B results depending on the emission scenario). Models are generally fairly successful in reproducing the historical temperature trends, inspiring confidence in their projections.
Precipitation projections, however, vary substantially across models, and models often do not capture historical trends well. The most robust feature is a long-term projected increase in boreal summer precipitation in the eastern Himalayas and Bangladesh, which is seen in most models over the 21st century. This leads to a consistent projection of increased annual mean river runoff in the Ganges and Brahmaputra in the models run in support of the Intergovernmental Panel on Climate Change (IPCC) AR4. A majority of models show a drying trend in winter precipitation in the central Himalayan foothills, but a substantial fraction do not. Over the western Himalayas, models are about as likely to show increases as decreases in precipitation in either season. Many models show trends that vary over time, with, for example, negative rainfall trends in the next decade for the Himalayan foothills but enhanced precipitation in the latter half of the century. Results of the next-generation models run in support of the IPCC AR5 were being analyzed as this report was being written, and may provide more robust indications of likely precipitation changes in this region. Further research may also provide better understanding of the role of black carbon in driving precipitation changes in this region. This appears to be substantial, and thus could contribute to some of the divergence across AR4 models because many did not include black carbon in their projections.
There have been fewer analyses of projected changes in extremes even though there is increasing observational evidence of increased extreme heavy rain in monsoon regions (Gao et al., 2002; Goswami et al., 2006; Rajeevan et al., 2008). It appears that both global and regional models are somewhat more consistent in projecting an increase in the frequency of extreme precipitation events as the climate warms (Christensen et al., 2007). Such a change in variability may have more severe consequences than a change in the mean precipitation. However, confidence in these projections is again tempered by the limited ability of models to accurately represent observed variability, especially in the monsoon systems.
Meteorological records from the HKH region are scarce and many of those that do exist extend only back to the mid-1950s to early 1960s. Thus, assessing climate change there, including its true long-term significance and likely future impact, is hampered by a paucity of long-term meteorological and glaciological observations from the higher and more remote elevations of
FIGURE 2.11 Mean projected changes in the AR4 models for surface temperature (top) and precipitation (center row) for the annual average (left), December to February (center) and June to August (right). Changes are shown for the mid-range SRES A1B scenario for 2080 to 2099 relative to 1980 to 1999. The lower row shows the number of models out of the 21 analyzed that projected increases in precipitation (increases in temperature were consistent across models). These projections show warming throughout the region and in all seasons, with greater magnitude over the Tibetan Plateau than over the South Asian continent south of the Himalayas. SOURCE: Christensen et al. (2007).
Asia. Regardless, evidence suggests that the eastern Himalayas and the Tibetan Plateau are warming overall (Diaz and Bradley, 1997; Liu and Chen, 2000; Shrestha et al., 1999) and this warming appears to be increasing with elevation, at least up to 4,800 m (Liu and Chen, 2000; Qin et al., 2009). In the western Himalayas, the situation is more complex with all but the Karakoram having warmed since at least the 1980s (Bhutiyani et al., 2007; Fowler and Archer, 2006; Shekhar et al., 2010). The Karakoram remains the principal regional climate anomaly as it likewise is regarding current glacier activity.
Unfortunately, these modern climate studies provide no meaningful insights into how significant the recent observed changes in climate have been compared with the natural variability of the region’s climate, because the observation records are too short. This provides the motivation for using centuries- to millennia-long annual tree-ring chronologies as well as ice cores to reconstruct climate over the HKH. Moraine dating can also provide useful information despite not having the same temporal resolution as tree-ring chronologies. Specifically, it can provide
evidence of past glacial behavior. Moraine-dating studies have indicated spatial and temporal variation in the advance and retreat of glaciers in the region. Studies suggest that HKH glaciers have not responded uniformly to climate forcing in the past. For example, Benn and Owen (1998) used moraine dating in the Kunlun Mountains along the northern edge of the Tibetan Plateau to determine that during the last glacial cycle, glacial advance was not synchronous. Other studies have shown that there may be a relationship between the spatial and temporal patterns of glacial advance and retreat in the region and the spatial and temporal patterns of the monsoon (Owen et al., 2002, 2005). A recent study of glaciers in northern India confirmed that on the millennial scale, patterns of glacial advance and retreat are related to patterns of monsoon variability (Scherler et al., 2010). More studies of this type would increase understanding of glacier behavior and climate response in the region.
Tree-ring chronologies can be used to reconstruct climate over the HKH, but to date have not been used to provide a direct history of glacial advance and retreat in the Himalayas using dendrochronological dating methods to reconstruct glacier fluctuations from dead trees overridden by advancing glaciers (e.g., Holzhauser et al., 2005; Luckman, 1993). This is because the HKH glaciers are almost all above tree line, which severely restricts the availability of tree-ring material for dating past glacial advances and associated retreats. With this in mind, three papers have been published on tree growth and related glacier activity in the Himalayas: Singh and Yadav (2000), Bhattacharyya et al. (2006), and Borgaonkar et al. (2009). None have directly dated the timing and extent of past glacier fluctuations in the Himalayas using the dendrochronological dating methods used elsewhere in Europe and North America. Rather, the Himalayan studies have inferred likely changes in past and present glacier activity through changes in tree growth thought to be related to climate (primarily temperature). Thus, we are limited to using tree rings for reconstructing climate over the HKH, using that information to determine how anomalous recent changes in climate are compared with the past, and inferring the degree to which those anomalous changes may be responsible for the observed retreats of the HKH glaciers.
Annual tree-ring chronologies are well suited for reconstructing past climate variability and change in the HKH region because they are annually resolved and exactly dated, making them useful for direct calibration and reconstruction of instrumental climate data back in time. When this is done, some form of statistical validation of the tree-ring estimates of past climate is commonly conducted by comparing climate data withheld from calibration exercises with the tree-ring estimates. This process of verification (Fritts, 1976) has a long history of use in dendroclimatology and is a direct measure of reconstruction accuracy. Cook and Kairiukstis (1990) and Cook and Pederson (2010) provide discussions of statistical validation in dendroclimatology. Estimates of reconstruction precision can also be obtained from error (or uncertainty) estimates associated with the calibration exercise. In regression analysis, this typically involves the use of the root mean square error of the fitted model to determine error estimates around each reconstructed value back in time. Doing so is a relatively recent development in statistical dendroclimatology, one that has not been done in the papers cited here on tree-ring research in the HKH region.
When it is not possible to directly calibrate tree-ring records against climate data, as can be the case in the data-poor regions of the HKH, it is often still possible to infer past changes in climate from them at certain highly stressed locations, like temperature change from tree growth at upper timberline locations. However, such inferences are far weaker for interpreting past changes of climate from tree growth compared with direct calibration and validation using instrumental climate data.
Using tree rings to study past climate in the Himalayas began over 20 years ago (e.g., Ahmed, 1989; Bhattacharyya et al., 1988; Hughes and Davies, 1987; Pant, 1979) and accelerated quite dramatically in the 1990s (e.g., Bhattacharyya et al., 1992; Borgaonkar et al., 1994; Bräuning, 1994; Esper et al., 1995; Yadav and Bhattacharyya, 1992). Thus, the literature on this topic is quite extensive and somewhat variable in its methods of analysis, findings, and interpretations, making a comprehensive literature review for this report impractical. However, the findings on current and past climate
change from tree rings in the HKH are nonetheless reasonably consistent for large regions of the HKH and for some seasons. This is most apparent from the many series that extend back into the Little Ice Age (LIA) roughly 300 to 500 years ago. It is far more difficult to make a useful assessment of what climate was like during the Medieval Warm Period in the Himalayas, some 1,000 years ago, because such long tree-ring records are relatively rare. Therefore, no further reference to the Medieval Warm Period will be made in this review. The following summary is divided into two climate change topics over the HKH: temperature and precipitation. For each of these topics, the review of tree-ring evidence for climate change then proceeds from the western to eastern ends of the HKH and up onto the eastern end of the Tibetan Plateau.
There is evidence for recent climate warming over large parts of the HKH since the LIA, with the latter decades of the 20th century often being the warmest overall. This is apparent from inferred or calibrated tree-ring reconstructions of past summer temperature in the western Tien Shan in Kirghizia (Esper et al., 2003) and the Karakoram in northern Pakistan (Esper et al., 2002). In the western and central Himalayas of India and Nepal the temperature histories from tree rings are somewhat less seasonally consistent with respect to climate warming since the LIA. In the western Indian Himalayas, Borgaonkar et al. (2009, 2011) described tree-ring evidence for anomalous winter warming that is especially evident since the mid-20th century. Cook et al. (2003) found similar evidence for long-term warming in an all-Nepal cold-season temperature reconstruction. In contrast, Yadav et al. (1997, 1999) and Yadav and Singh (2002) produced spring temperature reconstructions with very subdued warming trends since the LIA and even a small degree of cooling indicated during the latter half of 20th century. Independently, Cook et al. (2003) and Sano et al. (2005) produced similar results in reconstructions of February to June and March to September temperatures, respectively, from Nepal.
The appearance of a cooling trend in predominantly spring temperatures during the late 20th century prompted Yadav et al. (2004) to suggest that the western Indian Himalayas were defying global warming. This was apparently due to daily minimum temperatures in the western Himalayas decreasing about three times faster than daily maximum temperatures, thus resulting in an overall cooling trend in mean daily temperatures for the spring months. In contrast, Borgaonkar et al. (2009, 2011) produced several tree-ring records from the same region that showed accelerated growth due to warmer winter temperatures since the mid-20th century. Coupled with the long-term winter warming trend produced by Cook et al. (2003) for Nepal, this implies that winter temperatures in the western Himalayas have been trending opposite to predominantly spring temperatures in recent decades. It is unclear why this appears to be happening, although the cooling trend in surface temperature could be related to increasing atmospheric loading of aerosols in this region.
Moving eastward to Bhutan, recently developed unpublished results indicate a substantial summer season warming trend there since 1700 based on a large-scale composite of Himalayan spruce (Picea spinulosa) tree-ring data that begin in 1400 (Figure 2.12). There is a pronounced below-average growth period from 1590 to 1820 in the data, interpreted here as the LIA in Bhutan, followed by a nearly monotonic growth increase up to recent times that appears to emerge from the background after about 1880. That this tree-ring record can be interpreted as an index of summer temperature change over Bhutan is demonstrated by its correlation with gridded summer temperatures shown in Figure 2.13. This map shows the locations of significant correlations between the Bhutan spruce tree rings and mean monthly temperature data for the period 1901 to 2003. The greatest concentration of significant correlations is exactly over the Himalayas of Bhutan where the tree-ring data are located. The peak correlations over Bhutan (~0.50) are somewhat modest from the perspective of explained variance (~25 percent) between tree rings and summer temperatures, but the gridded temperature data used here is based solely on interpolated data from surrounding Indian meteorological stations. The climate records from Bhutan are extremely short (~10 to 20 years long) and insufficient for this kind of analysis. Therefore, the results presented here are likely to be a minimum estimate of how strongly the tree-ring record in Figure 2.13 is reflecting changes in current and past summer
FIGURE 2.12 The “All Bhutan” Picea spinulosa tree-ring chronology (1400-2005). The sample size (cores) per year is also shown, along with the chronology range above the dashed line where the expressed population signal (EPS) and sample size (N) are sufficient to produce a statistically valid tree-ring chronology (1450-2003 here) for climate modeling and interpretation (Wigley et al., 1984). The data indicate a pronounced below-average growth period from 1590 to 1820, interpreted here as the Little Ice Age in Bhutan, followed by a steady growth increase that indicates a substantial summer season warming trend since 1700. SOURCE: E. R. Cook and P. J. Krusic (unpublished data, 2011).
temperatures over Bhutan. Thus, the glacial retreat now under way in Bhutan (Bajracharya et al., 2007) appears to be associated with unprecedented summer warming over this part of the Himalayas.
Finally, over the Tibetan Plateau on the drier side of the Himalayas, there is less clear and somewhat conflicting evidence for climate warming since the LIA. In southwest Tibet, Yang et al. (2009) produced evidence for a modest summer warming that appeared to increase in strength toward the end of the 20th century. In contrast, a January to June temperature reconstruction from southeast Tibet produced by Yang et al. (2010) showed modest cooling. Comparisons made to other tree-ring reconstructions (e.g., Bräuning and Mantwill, 2004; Fan et al., 2008a; Li et al., 2011; Liang et al., 2008, 2009) from the Tibetan Plateau also yield conflicting evidence for widespread climate warming in the 20th century. This may be due to variations in the seasonal response of the trees to temperature, as previously suggested in the western Indian Himalayas, and also due to variations in site characteristics that could possibly affect moisture availability in this dry region.
FIGURE 2.13 Mean summer (June-July-August) temperature correlations in the Bhutan region (p < 0.10) with the “All Bhutan” Picea spinulosa tree-ring chronology shown in Figure 2.12. Correlations are calculated over the 1901-2003 period. The positive correlations indicate that the tree-ring data shown in Figure 2.12 can be interpreted as an index of summer temperature change over Bhutan. SOURCE: Produced with KNMI Climate Explorer using CRU TS 3.1 gridded mean temperature data.
Precipitation—Monsoon and Winter Snowfall
Reconstructions of past precipitation (and related hydrological variability) from tree rings in the HKH are less common than for temperature. However, there have been some notable successes that generally show a trend toward anomalously wetter conditions in the 20th century in the western portion of the HKH compared with earlier times. In the Karakoram of northern Pakistan, Treydte et al. (2006) measured stable oxygen isotope ratios in the annual rings of ancient junipers and showed that the 20th century has been the wet-
test period of the past millennium there. A bit farther eastward in the western Indian Himalayas, Singh and Yadav (2005), Singh et al. (2006, 2009), and Yadav (2011) all show that the latter half of the 20th century has been the wettest period since at least the LIA. In comparison, precipitation reconstructions for the eastern Tibetan Plateau (e.g., Fan et al., 2008b; Liu et al., 2011) show a more mixed record of interdecadal wetness and dryness, with a tendency for drier conditions in the last half of the 20th century followed by somewhat wetter conditions since that time.
Overall, the tree-ring climate reconstructions indicate that climate has been getting warmer and wetter since the LIA over significant portions of the HKH, but the story is regionally and seasonally complex with not all locations or seasons agreeing in the direction or size of the change. In the drier western part of the HKH, there is reasonably consistent evidence from tree rings for a warming trend in winter temperatures since the LIA, with the 20th century often being the warmest period overall. However, this warming trend does not appear to extend into the spring season. Instead, spring temperatures have generally declined there since the mid-20th century based on tree-ring evidence. The western Himalayas and Karakoram also appear to be getting wetter now compared with the past. This is consistent with the current stable or surging state of glaciers in the Karakoram (Bolch et al., 2012). In the eastern Himalayas of Bhutan, dramatic summer warming has been under way there since the LIA, which does not show any signs of slowing down. This is consistent with the glacial retreat now under way there (Bajracharya et al., 2007). On the Tibetan Plateau, changes in temperature and precipitation indicated by tree rings are mixed, with a tendency for drier conditions in the last half of the 20th century followed by somewhat wetter conditions since that time.
The general tendency for warmer and wetter conditions over large parts of the HKH can affect the glaciers in different ways, depending on whether temperature or precipitation is more important to the mass balance. In the western HKH where moisture delivered by the mid-latitude westerlies is most important, glacial mass balance appears to be reasonably stable because winter snowfall accumulation is presently sufficient to offset any climate warming during the summer ablation season. Whether this stability continues will depend at least in part on the continued delivery of sufficient snowfall to offset any future increases in ablation due to summer warming, a great unknown at this time. However, if the past is any indication, the present period of unusual above-average wetness may be hard to increase enough to offset the effects of future warming. In the summer monsoon-dominated eastern Himalayas, it appears that ablation is already outpacing accumulation based on glacial retreat in areas such as Bhutan, where 20th century summer temperatures appear to be warmer than at any time since 1450. It is hard to imagine how this trend might be reversed in the future because of the way temperatures are expected to continue to rise due to greenhouse warming.
One of the critical functions that mountain glaciers serve is the preservation of detailed information about past climate and the ability of glaciers to respond to different climate variables. Currently, there are five long ice-core records (greater than 2,000 years) for the HKH region. Three are from the Tibetan Plateau, and only two are from the Himalayas. As discussed in the previous section, the climate of the region varies a great deal from east to west, and to some extent from north to south. Therefore, more ice-core data is needed to form a complete picture of the region’s paleoclimate. However, information from the five long ice-core records that do exist, much like tree-ring data, demonstrate the climatic complexity and diversity of the HKH region.
Proxy climate records spanning more than 500,000 years have been recovered from the Guliya ice cap in the far northwestern Kunlun Mountain region of the Tibetan Plateau, dominated by westerly air flow over the Eurasian land mass, while shorter records (< 10,000 years) have been recovered from ice fields in the central Himalayans to the south, where a monsoonal climate regime dominates and the annual accumulation is high. These records show that the Himalayan ice fields are sensitive to fluctuations of the South Asian monsoon, and are affected by rising temperature in the region.
Over the past 25 years the Ice Core Paleo climate Research Group at the Byrd Polar Research Center has collected ice cores from five sites across the Tibetan Plateau. Figure 2.14 shows the location of these five drill sites. Naimona’nyi (6,050 m a.s.l.) is included
FIGURE 2.14 Location of the five ice-core study sites in the Himalayas and across the Tibetan Plateau.
even though recent melting at the glacier surface has removed the upper 40 to 50 years of the record (Kehrwald et al., 2008). The decadal δ18O and dust data along with a 3-year moving average are illustrated in Figure 2.15. The data illustrate the effect of the recent warming across the Plateau on the mean δ18O values since 1950. There is a strong enrichment of 18O that is enhanced with altitude. The average values of δ18O from 1000 to 1950 are -10.850/00 for Dunde (5,325 m a.s.l.), -15.010/00 for Purguogangri (6,070 m a.s.l.), -14.280/00 for Guliya (6,200 m a.s.l.) and -20.480/00 for Dasuopu (7,200 m a.s.l.). From 1950 to the top of these records (except for Naimona’nyi where the most recent part of the record is missing) the δ18O averages have increased by 0.710/00, 1.350/00, 1.080/00, and 2.630/00 for Dunde, Puruogangri, Guliya, and Dasuopu, respectively. These trends are consistent with instrumental temperature records since the 1950s across the plateau (Liu and Chen, 2000) as well as with the model predictions of the vertical amplification of temperature across the region.
The mineral dust record for the same periods in these five cores shows a great deal of site-to-site variability because dust is often more of a local to regional signal. The only record that shows an increase in concentrations in the 20th century is the 7,200-m a.s.l. Dasuopu record. There is a very large difference in dust concentrations from region to region. The Dunde and Guliya ice cores located on the most northeastern and northwestern areas of the Tibetan Plateau contain 20 times higher dust concentration than that of the most southern ice-core study site, Dasuopu, at the top
FIGURE 2.15 One thousand year records of δ18O and mineral dust concentrations, shown as decadal averages, from five Tibetan Plateau ice-core study sites along with a three-decade moving average.
of the Himalayas. The larger concentration of dust in cores in the north is due in part to the proximity to Takla Makan and Gobi deserts and the fact that they are located in regions of the plateau that are dominated by the westerlies. Puruogangri, Naimona’nyi and Dasuopu located in the central Tibetan Plateau and the Himalayas to the south are under greater influence of the monsoons and contain far lower concentrations of mineral dust. The lowest mineral dust concentrations are found in the 7,200-m a.s.l Dasuopu ice-core study site. The ice-core data files along with the metadata files for Dunde, Guliya, Dasuopu, and Puruogangri ice cores are archived at the National Climatic Data Center (NCDC).8 Naimona’nyi ice core data files and metadata will be archived pending publication.
A high-resolution ice-core record from Dasuopu at 7,200 m in the central Himalayas (located just north of the China-Nepal border) shows that this location responds to fluctuations in the intensity of the South Asian monsoon. Measurements of dust and chloride concentrations yield information about changes in monsoonal intensity. Older sections of the ice cores reveal periods of drought in the region, with the drought of greatest intensity occurring from 1790 to 1796. Recent increases in anthropogenic activity in India and Nepal are shown by a doubling of chloride concentrations and a fourfold increase in dust in the upper sections of these cores. The Dasuopu ice core also suggests a 20th-century warming trend that appears to be amplified at higher elevations (Thompson et al., 2000).
The ice fields located in more arid regions of the Tibetan Plateau (i.e., Guliya, Puruogangri, Dunde) have rather similar annual net mass balances, averaging 220 mm water equivalent (liquid water obtained from melting snow or ice) per year for Guliya in the far northwestern Tibetan Plateau, 350 mm water equivalent per year for Puruogangri in central Tibetan Plateau, and 390 mm water equivalent for Dunde in northeastern Tibetan Plateau. Ice-core records from these three ice fields have rather similar histories with net balance higher in the 17th and 18th centuries, consistently lower values in the 19th century, and a general increase in the 20th century. In contrast, the net balance history on Dasuopu in southern Tibetan Plateau in the Himalayas shows a different pattern, with current net balances averaging 1,000 mm of water equivalent per year. The net balance history shows consistently high
values over the 19th century. Although the 600-year net balance history for the Himalayas (e.g., Dasuopu) is quite different from that in the Tibetan Plateau to the north, their oxygen isotope histories, or proxy temperature records, are remarkably similar at lower frequencies. During the 17th century, temperatures were warmer over Guliya and Puruogangri than over Dunde and Dasuopu, but there has been persistent, gradual warming from the 18th through the early 20th century and accelerated warming over the second half of the 20th century. The recent isotopic enrichment is consistent among the Tibetan Plateau sites and independent of the net balance (Duan et al., 2006; Thompson et al., 2006).
A sulfate record, which indicates deposition of sulfate aerosol, for 1000 to 1997 from the Dasuopu ice core shows that this site is sensitive to anthropogenic activity originating in southern Asia. Before 1870, sulfate concentrations in the atmosphere were relatively low and constant, but after 1870, concentrations increased and the rate of increase has accelerated since 1930. This trend in sulfate deposition is accompanied by growing SO2 emissions in South Asia. This is in contrast to sulfate concentrations derived from Greenland ice cores, which have declined since the 1970s. This is a result of regional differences between Europe and Asia in emission and transport of sulfate, as well as different levels of environmental regulation (Duan et al., 2007). As discussed earlier in this chapter, a number of recent studies have concentrated on the impact of black carbon and aerosols on atmospheric heating and glacier melting (Lau et al., 2010; Menon et al., 2010; Ramanathan and Carmichael, 2008).
Regional composites for the Tibetan Plateau have been constructed using decadal averages of oxygen isotopes over the last 2,000 years to reveal larger temporal-scale changes. The 2,000-year perspective from these Tibetan Plateau ice cores shows large and unusual warming at high elevations. The oxygen isotope record clearly shows that large-scale dynamics have changed over the past century, regardless of whether the record is used as a proxy for temperature, precipitation, or atmospheric circulation (Thompson and Davis, 2005; Thompson et al., 2006; Vuille et al., 2005; Yao et al., 1996). Similar to tree-ring chronologies, ice-core records collected across the Tibetan Plateau demonstrate that it is a climatically diverse and complex region. Ice-core datasets from exposed mountain summits away from the effects of urbanization and topographic sheltering provide relatively unbiased records of the planet’s climate.
Mountains are the water towers of the world, characterized by high precipitation and little evaporation because of lower air temperatures and longer snow coverage, resulting in large contributions of snowmelt and icemelt to the runoff of lowland areas (Viviroli et al., 2007). This is especially true for the HKH region, where the snow and ice stored in high-altitude glaciers in the Greater Himalayas are a source of water for almost every major river system in the region. However, a complete understanding of the regional hydrology— including the actual contribution of snow and glacial meltwater to surface waters and groundwater of the region—is lacking because of the same incomplete science and unresolved uncertainties discussed earlier in the report.
The lack of understanding of the regional hydrology is intimately tied to water security concerns in the region. Some reports and peer-reviewed publications suggest that glacial meltwater provides a large share of the water feeding discharge into major rivers such as the Ganges. This apparent contribution of glacial meltwater to these large rivers, combined with the misconception that the region’s glaciers are experiencing the highest rates of glacial retreat in the world, have combined to create a sense of water scarcity in the region (e.g., Kehrwald et al., 2008).
Uncertainty About the Contribution of Glacial
Melt to the Hydrology of the Region
Barnett et al. (2005) report that “there is little doubt that melting glaciers provide a key source of water for the region in the summer months: as much as 70 percent of the summer flow in the Ganges and 50 to 60 percent of the flow in other major rivers.” This statement is based on estimates of glacial melt contributions to river flow derived from models that relied on many assumptions because of the lack of field data (Singh and Bengtsson, 2004; Singh and Jain, 2002; Singh et al., 1997). Rees and Collins (2006) used a theoretical
modeling approach to conclude that for large distances downstream, the contribution of discharge from glacier icemelt often dominates flow, particularly when other sources of runoff are limited.
Attempts to improve the understanding of the contribution of glacial wastage to the regional hydrology confound these interpretations by identifying scientific gaps, important nuances in geography, and contrasting results when using different scientific methods. For example, modeling9 showed that in Nepal the glacial meltwater contribution to tributaries to the Ganges near the base of the Himalayan sub-basin streamflow varies from approximately 20 percent in the Budhi Gandaki Basin to approximately 2 percent in the Likhu Khola Basin. The average across nine basins in Nepal was 10 percent (Alford et al., 2010), a far lower percent than that discussed above. Using remote sensing approaches,10 Racoviteanu (2011) corroborated these lower values by showing that for the Langtang basin in Nepal, glacial meltwater contributes about 10 percent of discharge at an elevation of 900 m. This work also showed that the contribution of glacial meltwater to discharge increases with increasing elevation, reaching about 50 to 70 percent at an elevation of 3,800 m (Racoviteanu, 2011). Using ice-core records and measuring radioactivity, Kehrwald et al. (2008) suggested that reports of the relationship between glacial retreat and downstream water resources have not accounted for mass loss through thinning of high-elevation, low-latitude glaciers. For example, this is apparently occurring in the Naimona’nyi glacier in Tibet. Thinning of high-elevation glaciers could result in a decrease in water availability in regions where the water supply is dominated by high-elevation glacial melt. Armstrong (2010) reported that previous assessments of the relationship between glacial meltwater and surface water supply have been highly qualitative or local in scale. Direct evidence is lacking to support the higher-end values reported for the contribution of glacial icemelt to total river flow volume (i.e., 50, 60, 70 percent). However, it is generally accepted that the percent contribution increases from east to west across the region (Immerzeel et al., 2010).
Finally, contributing to the confusion about the relative importance of glacial melt to the discharge of rivers is the often-overlooked differentiation between relative contributions of snowmelt versus glacial melt. For example, any water source from high elevation is sometimes assumed to be glacial melt, when in fact snowmelt may be a major contributor. Snowmelt is a renewable resource that is replenished every year, in contrast to the fossil water11 contributed by glacial wastage (Barnett et al., 2005). In addition, some reported values include runoff from rain and other forms of precipitation that would occur with or without the presence of glacial ice. In most basins, the contribution of rain far outweighs the combined contributions of snowmelt and glacial wastage to discharge. For example, Andermann et al. (2012) found that snowmelt and glacial melt contributed roughly 10 percent to the discharge of the three main Nepal rivers. Therefore, in these catchments, rainfall could contribute as much as 90 percent to the total discharge.
A Complex Hydroclimate System
Glaciers are only one part of the complex HKH hydroclimate system,12 where the relative importance of the contribution of glacial meltwater to runoff depends on the magnitude of other components of the hydrological cycle. The different climate regimes of the region are characterized by differences in the spatial and temporal distribution of precipitation (type and amount) and runoff. As mentioned previously, summer monsoon rains dominate the annual precipitation cycle of the eastern Himalayas while the west is dominated by winter snowfall with low amounts of summer precipitation. A similar variation in the relative contributions of rain, snowmelt, and melt of glacier ice to the discharge of different rivers throughout the HKH is expected.
9 The model was based on limited mass balance measurements from glaciers in the region, remote sensing measurements of glacier area, and a variety of assumptions.
10 Datasets were derived using remote sensing resources discussed further in Appendix A, such as the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) sensor, declassified Corona imagery, and Landsat ETM+.
11 Fossil water is water that has been stored in a glacier or an aquifer for a long (years or more) period of time.
12 The hydroclimatic system includes the processes by which the climate system causes global and local variations in the hydrological cycle.
Methods comparing measurements or models of glacial meltwater production with measurements of downstream discharge volume are problematic. Glacial meltwater can be interpreted as raw volume input into the system, but downstream of the glaciers discharge has been modified by a number of factors, including precipitation, evaporation, irrigation, damming, and groundwater exchange. With increasing distance from the glaciers, these modifications increase in relative importance, while the relative contribution of glacial meltwater decreases. In a direct comparison between glacial meltwater and runoff downriver, the volume contribution from glaciers can be overestimated with increasing distance from the glaciers (Kaser et al., 2010).
Furthermore, limited data on the water cycle of the region, due to a variety of reasons ranging from difficult terrain to political instability, is a chronic problem. Scientists compensate by using many assumptions in models or remote sensing imagery. The overall result, however, is more overall uncertainty in the body of scientific literature than in other parts of the world that are not limited by these challenges. Despite this uncertainty, many studies have contributed to reducing this uncertainty associated with the complexity of the HKH hydroclimate system.
There is strong seasonality in annual precipitation. Annual hydrographs for the Ganges and Indus rivers clearly show strong seasonality in the amount of discharge, leading to seasonal differences in water availability. The Ganges River exhibits a significant discharge during the summer months, resulting in a water surplus that can maintain in streamflows and in some areas recharge groundwater storage. However, water consumption exceeds natural runoff in the winter months of February and March when there is some reliance on groundwater and/or storage (Hoekstra and Mekonnen, 2011). The Indus River has a lower peak discharge and lower annual discharge than the Ganges River. The Indus River discharge also varies seasonally and interannually. (The relationship between natural runoff and water use in the major river basins in the study region is discussed in more detail in Chapter 3.)
Using a modeling approach, Immerzeel et al. (2010) concluded that glacial melt and snowmelt are “extremely important in the Indus Basin and important for the Brahmaputra basin, but plays only a modest role for the Ganges, Yangtze, and Yellow Rivers.” Preliminary results show that in the Indus, snow and glacial melt contribute one and a half times as much discharge as that generated naturally downstream below 2,000 m. In the Brahmaputra, snow and ice discharge is about a quarter of that generated downstream. The model found snow and glacial melt to be less important in the Ganges, with discharge being about one-tenth of that generated downstream. For the Indus, this indicates that the source of much of the streamflow at low elevations is snowmelt and/or icemelt from elevations greater than 2,000 m. The values for the Ganges are remarkably similar to those of Alford et al. (2010) and suggest that glacial melt is not a major contributor for river systems to the east but is much more important for river systems to the west.
Recently, Wulf et al. (2011) quantified the water resources and discharge components for the Sutlej River from 2004 to 2009, which is a major tributary of the Indus River flowing through northern India and Pakistan. As discussed earlier in this chapter, the Sutlej Basin is located at the interface between the monsoon-dominated precipitation regime to the east and the winter-snow-dominated regime to the west. Results indicate that the discharge of the Sutlej River at Bhakra located at low elevation and situated at the base of the mountains is sourced predominately by snowmelt (48 percent) followed by effective rainfall (rainfall-evapotranspiration,13 39 percent) and glacial melt (13 percent). Average runoff per square meter is less than 0.2 m yr-1 in the high-elevated, low-relief Transhimalayan part of the Sutlej Valley, peaks at about 1.5 m yr-1 in the snowmelt-dominated High Himalaya, and is about 0.9 m yr-1 at the rainfall dominated- mountain front. Snowmelt is thus a much more important component of discharge than glacial melt for the Sutlej Basin, where monsoon rains are less important than farther to the east in Nepal and Bhutan.
The hydrograph of the upper Indus Basin is highly seasonal, with about 85 percent of annual discharge occurring between May and September. Figure 2.16 shows the time series of monthly and yearly variability in May to September discharge at the historic Partab
13 Evapotranspiration is the combined processes through which water is transferred to the atmosphere from open water and ice surfaces, bare soil, and vegetation (AMS, 2000).
FIGURE 2.16 The monthly hydrograph of upper Indus River discharge at Partab Bridge, Pakistan, for the period 1962 to 2008 (top) and the cumulative discharge series for the May to September months (bottom) that collectively account for about 85 percent of the annual discharge on average. In both the top and bottom, the discharge on the y-axis is in units of cubic meters per second (m3 s-1), or CMS. The red curve is a 50 percent LOWESS robust smoothing of the yearly discharge data. The annual discharge has the appearance of two flow regimes: about 3,500 m3 s-1 from 1962 to 1987 and about 3,900 m3 s-1 from 1988 to 2008, an approximate 11 percent increase. This is reasonably consistent with periods of observed glacial recession from 1985 to 1995 and expansion from 1997 to 2002 in the Karakoram. Plateau ice-core study sites along with a three-decade moving average.
Bridge, Pakistan, for the period 1962 to 2008. Median discharge is 3,658 m3 s-1, but has the appearance of two flow regimes: 3,519 m3 s-1 from 1962 to 1987 and 3,902 m3 s-1 from 1988 to 2008, an approximate 11 percent increase. These two periods are reasonably consistent with periods of observed glacial retreat in 1985 to 1995 and expansion in 1997 to 2002 in the Karakoram (Hewitt, 2005). The upper Indus Basin record at Partab Bridge is too short to determine whether these are indicative of long-term trends or shorter random changes at the decadal scale, hence the value of producing a much longer May to September river flow reconstruction from tree rings (Ahmed and Cook, 2011). However, the results do indicate that recent measurements of increased snowfall in the area and positive glacial mass balances are correlated with recent increases in discharge. Furthermore, the sub-basins and tributaries of the Upper Indus Basin are not always in phase with respect to their relative contributions to the overall discharge. Fowler and Archer (2006) report conflicting signals of temperature and discharge in tributary inflows that ultimately discharge into the Tarbela Reservoir, farther downstream on the Indus Basin. In summary, if the observed +11 percent change is a short-term phenomenon, it falls within the interannual variability of flows and is of little import for water supply; if it indicates a long-term step increase in discharge, +11 percent over 5 months represents approximately 4 million acre-feet, which would be significant for downstream mountain communities and also for downstream reservoir operation, storage, and deliveries. But again, the period of record is too short to speculate about the trend, driving forces, or water resources significant of this record at present.
The Role of Groundwater
Groundwater is an important part of the hydrological system in any part of the world. It is the primary source of freshwater in many areas around the world and responds more slowly to meteorological conditions when compared with the surface components of the water cycle. The full role of groundwater in the HKH region is not clearly understood. For example, the amount of groundwater storage in the HKH region is largely unknown because storage is notoriously difficult to measure, especially in mountainous terrain. Furthermore, the relationship between glacial melt and snowmelt and groundwater recharge is not fully understood. Understanding the average spatial and temporal characteristics of groundwater fluxes, recharge, and discharge is a research frontier facing the hydrological science community (NRC, 2012a) and the HKH region is no exception; the mechanisms and pathways of groundwater recharge and discharge remain unclear.
There is no evidence that glacial melt or wastage contributes to groundwater recharge outside local regions in any significant way. Research from other mountainous regions has shown that groundwater recharge outside local impacts is negligible and there are no circumstances that hint that this would not be so in the HKH. The effect of glacial melt on ground-water recharge on the plains is likely small because the contribution of glacial meltwater to flows downstream is generally small, and therefore any groundwater recharge from major rivers downstream would be little affected by changes in flows of glacial meltwater. This is important because there is substantial evidence that groundwater withdrawals are increasing and it appears unlikely that increases in glacial wastage would have any significant impact on the supplies of groundwater available to meet the increased demands.
Groundwater extraction across northern India and the surrounding area in response to the growing demand for water14 is exceeding groundwater recharge, causing a lowering of the water table (Qureshi, 2003; Shah, 2009; Sikka and Gichuki, 2006). The GRACE15 satellite mission revealed a steady mass loss that has been proposed to be due to this excessive groundwater extraction (Rodell et al., 2009; Tiwari et al., 2009). Tiwari et al. (2009) estimate the region lost groundwater at a rate of 54 ± 9 km3 yr-1 between 2002 and 2008, which is likely the largest rate of groundwater loss for comparably sized regions. A more recent analysis by Moiwo (2011) of the loss of groundwater storage in this area using GRACE data shows a similar depletion. If this trend is sustained, there could be water shortages in the region when the aquifers become economically exhausted (Tiwari et al., 2009). More specifically, it is possible that groundwater withdrawals during the low-flow period of these river basins may cause these rivers to become seasonally dry. Finally, advances from satellite gravimetry in the GRACE mission have helped to quantify groundwater depletion in the plains of northwest India, but the spatiotemporal resolution of the current satellite system is too coarse to fully distinguish between all mass changing processes (groundwater, ice-melt, sediment, and tectonic forces (Bookhagen, 2012).
Although there is still a great deal of uncertainty, non-glacial mountain runoff (recharge) from the Himalayas toward low-elevation areas, in other words, the mountain water tower effect, could be a significant source of recharge to northern India. Preliminary results from a simple but effective hydrological model combined with daily rainfall and discharge measurements found that groundwater flow through bedrock is approximately six times the annual contribution from glacial icemelt and snowmelt to central Himalayan rivers (Andermann et al., 2012). However, this hydrological model used by Andermann and colleagues accounts only for snowmelt waters and not for the melting of glacial ice, which is likely to be an important discharge contribution during the late summer. Furthermore, with a model based on mean monthly values, it is difficult to establish the infiltration behavior of relatively short, but strong, monsoon storms (Bookhagen, 2012).
Groundwater storage in the middle and upper Indus River plains of Pakistan is also significant, and it has been increasingly pumped to fulfill agricultural and urban water demands. This is consistent with the decrease in discharge (discussed above) below the Partab Bridge over the last 10 years. However, ground-water in this region is recharged by the distribution of upper Indus basin runoff through Pakistan’s extensive canal irrigation system across the Indus plains, mitigating groundwater depletion. But in some areas the canal system contributes to waterlogging, a high water table resulting in the saturation of soil to the degree of hindering or preventing agriculture (Briscoe and Qamar, 2006). Salinization of waterlogged lands soon follows.
These groundwater balance processes are difficult to estimate, however, because they involve water-level fluctuations over the year, as well as high levels of spatial variation in depths to groundwater and salinity levels (Van Steenbergen and Gohar, 2005). Waterlogging and salinity issues increase with the successive reuses of water downstream in the canals of the lower Indus Basin (Bhutta and Smedema, 2005). Again, the signal of the contribution of glacial meltwater in the downstream hydrological cycle can be lost in the noise of other processes, such as irrigation and the resulting hydrological impacts.
In contrast to northern India and Pakistan, ground-water resources in the Kabul Basin of Afghanistan appear to be adequate for current needs and for at least the next several decades. A quantitative study of
14 Groundwater accounts for 45 to 50 percent of irrigation and 50 to 80 percent of domestic water use (Rodell et al., 2009).
groundwater resources in the Kabul Basin (a very arid region), Afghanistan by the U.S. Geological Survey and Afghanistan authorities (Mack et al., 2010),16 has shed light on the role of groundwater in this country. This study integrated a variety of hydrological datasets—for example, streamflow data, water quality data, and satellite imagery—into a groundwater flow model to assess current and future water availability in the region.
The study gleaned that groundwater from the upper aquifers has been the primary source of water for agriculture and municipalities (Mack et al., 2010). The availability of groundwater in the Kabul Basin primarily depends on (a) surface-water infiltration from rivers and streams, (b) water leakage from irrigated areas, (c) subsurface groundwater inflows from mountain fronts and, (d) groundwater storage in thick sediments. However, most recharge is derived from leakage of streamflow. Snowmelt in the mountains surrounding Kabul Basin, particularly the Paghman Mountains, contributes an unknown but important amount to the water resources of the basin (Mack et al., 2010).
Groundwater resources in the upper aquifer during years of normal precipitation and in the northern Kabul Basin are considerable. Existing community water-supply wells that are shallow, or screened near the water table, likely would be affected by increased groundwater withdrawals, however, and could be rendered inoperable or dry during summer months with groundwater-level declines as small as about 1 m. Simulations of the effects of increasing water use on groundwater levels indicate that a large percentage of existing shallow water-supply wells in urban areas may contain little or no water by 2057 (Mack et al., 2010).
Possible Changes in Regional Hydrology
The analysis of future climate change impacts on the hydrology of the HKH region is complex because of climate variability, sparse data, and uncertainties in climate projections and the response of glaciers to current and future changes in climate, as discussed earlier in this chapter.
Immerzeel et al. (2012) evaluated the hydrological response to future changes in climate for the Langtang Basin in Nepal using a high-resolution combined cryospheric-hydrological model that explicitly simulates glacier evolution and all major hydrological processes. The analysis showed that both temperature and precipitation are projected to increase over the next century. These increases will lead to greater evapotranspiration and greater snowmelt and icemelt. This, combined with more snow falling as rain, results in a steady decline of the glacier area in the model. Furthermore, the analysis shows that increased precipitation and icemelt will lead to increased streamflow. The seasonal peak in meltwater coincides with the monsoon peak; therefore no shifts in the hydrograph are expected.
If these results are representative of the region, the Committee expects little change in the hydro-graph of large rivers in the HKH region in response to changes in climate and potential glacial retreat over the next several decades. If anything, there may be an increase in discharge. Potential changes in climate that result in drier and/or warmer conditions will result in negative glacial mass balances, providing an additional water source for these large rivers over the next several decades. Decreases in available water from changes in climate (less precipitation, more evapotranspiration, etc.) will be compensated by the release of water from storage in glaciers. However, higher elevation areas can receive more than 50 percent of their annual water flow from glacial meltwater. Populations that live in high-elevation areas—or use them for activities such as seasonal grazing—may face water shortages in the near future (years to decades) if their basins have glaciers where the upper end of the glacier is at or near the elevation of the local ELA.
There is the possibility of reduced availability of groundwater below the front of the HKH, for example, the Gangetic plain in northern India. Andermann et al. (2012) have recently shown that groundwater storage in the fractured bedrock significantly influences the Himalayan river discharge cycle for the Ganges River Basin. They show that water from rainfall and snowmelt is stored temporarily in a groundwater reservoir with a characteristic response time of about 45 days. Further, they suggest that water traveling through groundwater reservoirs in the eastern HKH represents about two-thirds of annual discharge. Groundwater is the primary source of water during low-flow times
16 The study was conducted by the U.S. Geological Survey (USGS), under an agreement funded by the U.S. Agency for International Development (USAID). It was conducted with cooperation from the Afghanistan Geological Survey (AGS) and the Afghanistan Ministry of Energy and Water (MEW).
such as winter months, where consumptive use of water in the Ganges Basin is already greater than flow. The short response time and large amount of water flowing through the groundwater systems, in combination with the GRACE measurements that suggest a rapid reduction in the amount of groundwater in northern India, are consistent with a system that is experiencing overdrafting of groundwater. Current recharge rates do not appear to be able to replenish present rates of groundwater removal. Continued or accelerated rates of groundwater removal can easily lead to water shortages on the scale of years.
The largest changes to the hydrological system over the next decade or two will most likely be because of changes in the timing, location, and intensity of monsoonal activity Interannual variability of the monsoon strongly affects spatial patterns. At some locations in the central Himalayas, one monsoon depression alone can account for 10 to 20 percent of all monsoon rainfall. A small shift in storm path from one year to another can cause large differences in water availability. An example is the flooding in Pakistan in July and August 2010 that resulted from an unusual combination of severe weather events (Lau and Kim, 2011a). The 2010 Pakistan flood caused historic social and economic losses for the country. The flooding was primarily due to heavy rain that fell during late July and early August 2010, from a shift of monsoon activities from the Bay of Bengal to northern Pakistan. Glacial wastage from the Himalayas likely did not play a role in this case.
While flow in the Indus and the Ganges/ Brahmaputra basins will be highly affected by changes in precipitation, climate change will also have other impacts on the hydrological cycle. For instance, evapotranspiration rates may increase over large parts of the irrigated area of both basins. To compensate, farmers may apply more irrigation water, further drawing down surface water and groundwater. This trend may be most important during the dry months of the year, and may be most significant on the Indus River, which already has dry periods when available flow does not meet demand for irrigation water.
As discussed in Chapter 1, lack of observational data in the HKH region has led to misunderstandings about the effects of climate change on glacial retreat rates. One such misunderstanding is that retreating glaciers will lead to widespread flooding. While this is not likely, the region does face other physical hazards, including flash flooding due to extreme precipitation, flooding due to monsoon rainfall, lake outbursts, landslides, and avalanches. Monsoon flooding and lake outbursts are covered in detail here.
Monsoon flooding can cause loss of life and property, and potentially economic and social calamities. During the historic flooding in Bangladesh in 1998, two-thirds of the country was submerged; about 1,000 people perished from flooding or succumbed to water-borne diseases such as typhoid and cholera. Nearly 16,000 km of roads and more than 700,000 hectares of cropland were damaged or destroyed, and over one million people were displaced (BBC News, 1998; del Ninno et al., 2001).
Monsoon flooding can have important political impacts. Because of the enormous cost involved in rebuilding businesses, agriculture, and infrastructure, a major monsoon flood can have profound and long-term impacts on the policy and politics of the local and national governments in the Himalayan region. The way governments respond, and how they interact with international relief groups in managing the relief efforts, may also contribute to public perception of inefficiency, favoritism, political discord, and unrest, as in the case of the 2010 Pakistan flood. Another example is the 2008 flooding of the Kosi River, which flows from the Himalayan foothills of Nepal to a confluence with the Ganges in the Bihar region of northern India. The flood event led to a dispute between Nepal and India regarding mismanagement of the Kosi River (Malhotra, 2010).
The main cause of monsoon floods is heavy rain. Monsoon floods occur most frequently in the foothill regions along the arc of the Himalayas, and low-lying areas in the head of the Bay of Bengal during the summer monsoon season from June through September. The heavy monsoon rains range from Bangladesh, Bhutan, Nepal, and northeastern India, to north and northwestern India and Pakistan. With few exceptions, every year somewhere in the region, some degree of monsoon flooding will occur, due to the uneven distribution of monsoon rain.
A number of additional factors have the potential to contribute to the severity of monsoon floods by compounding the impact of heavy rains (NRC, 2012a). For example, if the heavy rains are coupled with unusually high volumes of runoff from melting snowpack in the Himalayas (especially rain-on-snow events), the result might be devastating. Land use changes and deforestation could also affect the severity of monsoon floods through an increase in surface-water flow leading to more severe flooding. Changes in the distribution of monsoon rainfall in response to climate and other factors may bring heavy rain to relatively dry regions (e.g., the 2010 Pakistan flood), where the local population may be less prepared or equipped to carry out prevention, evacuation, and mitigation measures. Although these scenarios are speculative, it is clear that how hydrological extremes (in this case, monsoon flooding) are intertwined with anthropogenic effects is poorly understood (NRC, 2012a).
The monsoon varies with many factors, including air-sea interactions and land processes. Increased aerosol concentrations may also influence the monsoon through local heating. The severity of monsoon flooding also depends on the local topography and infrastructure. Although the magnitude of the heavy monsoon rain during the Pakistan flood of 2010 was small compared with that of the 1988 Bangladesh flood, the impacts of the flooding were equally devastating. One-fifth of the country was underwater; nearly 2,000 people perished; 20 million people were affected. The direct damage caused by the floods was estimated to be US$6.5 billion, with an additional US$3.6 billion in indirect losses (Asian Development Bank and World Bank, 2010).
A glacial lake outburst flood (GLOF) is a type of flood that occurs when water dammed by a glacier or a moraine is rapidly released by failure of the dam (e.g., Bajracharya et al., 2007; Hewitt, 1982; Xin et al., 2008). There are two distinctly different forms of glacial lake outbursts: those that result from the collapse or overtopping of ice dams formed by the glacier itself, and those that occur when water drains rapidly from lakes formed either on the lower surface of glaciers (supraglacial) or between the end moraine and the terminus of a retreating glacier (moraine-dammed; Figure 2.17). The phenomenon of ice-dammed lakes is more prevalent in the Karakoram Mountains in northern Pakistan and the Pamirs. In the eastern Himalayas (e.g., Bhutan, Nepal) GLOFs are generally caused by water draining rapidly from supraglacial lakes or the collapse of moraines. Failure of the confining dam can have a variety of causes, including earthquakes, catastrophic failure of slopes into the lake (avalanches, rock slides, ice fall from a glacier into the lake), a buildup of water pressure, or even simple erosion of the confining dam over time.
An example of a potential threat from ice-dammed lakes is the Medvezhi Glacier in the Pamir Mountains. In 1963 and 1973, the surge of the glacier was large enough that the ice dam exceeded 100 m in height, creating a lake of over 20 million m3 of water and debris. A series of large floods resulted from the outburst of that lake, but there were no victims because of monitoring and early warning systems. Infrastructural damage, however, was significant (UNEP, 2007).
New glacial lake formation and the enlargement of existing lakes have resulted from thinning and retreat of glaciers in the HKH region. Many glacial lakes in Nepal are growing at a considerable rate, increasing the risk to local populations. Twenty-four GLOF events have occurred in Nepal in the recent past, causing considerable loss of life and property. For example, the 1981 Sun Koshi GLOF damaged the only road link to China and disrupted transportation for several months, and the 1985 Dig Tsho GLOF destroyed the nearly completed Namche Small Hydroelectric Project, in addition to causing other damage farther downstream (Bajracharya et al., 2007).
FIGURE 2.17 Schematic diagram of a moraine-dammed glacial lake formed by glacial meltwater. Failure of the confining moraine dam leads to an outburst flood.
There is no doubt that people and property for considerable distances downstream from the unstable lakes are facing a serious threat; the problem, however, is how to determine the degree of probability of such an event. Analysis of the rapidly growing worldwide literature, including field and theoretical knowledge, on the outburst of glacial lakes, led a recent ICIMOD commission on GLOFs in Nepal to conclude that it is not feasible to make a reliable prediction of a specific occurrence on the basis of existing knowledge ( ICIMOD, 2011a). Because direct predictions cannot be made, a careful selection of prioritized lakes needs to be monitored on a regular basis.
GLOFs are not the only outburst lake hazards in the HKH region. Another is the landslide lake outburst flood (LLOF), which is a catastrophic release of impounded water from behind a natural dam formed by a landslide. In the steep mountainous Himalayas, landslides are a common event, whether they are triggered by normal weathering and erosion processes, extreme rainfall events, or earthquakes. The release potential of water by LLOFs can exceed that for GLOFs (Hewitt, 1982). This is because landslide dams can be very large. Dunning et al. (2006) describe a landslide dam that formed in Bhutan in 2003 and subsequently impounded 4 × 106 to 7 × 106 m3 of water before its failure in 2004. Landslide lakes can also occur at much lower elevations than glacial meltwater lakes, where they can impound runoff from larger upstream catchment areas compared with the areas contributing to glacial meltwater lakes. Hewitt (1982) provides a detailed history of outburst floods in the Karakoram, including some massive LLOFs, and Gupta and Sah (2008) present an example of a LLOF in the Satluj catchment in Himachal Pradesh, India, well below the termini of any glaciers above it.
Like GLOFs, LLOFs pose a serious hazard to people, property, and infrastructure downstream from the landslide dams. However, compared with the large number of glacial meltwater lakes forming today because of climate warming and glacial retreat, landslide lakes are less commonly formed in the high Himalayas (e.g., Hewitt, 1982). Therefore, the risk posed by future LLOFs is likely to be significantly less than that posed by GLOFs in the HKH region. In the lower trans-Himalayan regions where landslide lakes replace meltwater lakes as hazards (e.g., Gupta and Sah, 2008), LLOFs will be a greater future risk. That being said, the development of landslide lakes is almost certainly less predictable than meltwater lakes because landslides are more random and where they will occur is harder to predict than glacial retreats. Thus, LLOFs are even less predictable than GLOFs.
Key features of the physical geography of the HKH region were identified at the workshop by the breakout groups on Climate and Meteorology and on Hydrology, Water Supply, Use, and Management. Starting from those concepts, the Committee used its expert judgment, reviews of the literature, and deliberation to develop the following conclusions:
• The climate of the Himalayas is not uniform and is strongly influenced by the South Asian monsoon and the mid-latitude westerlies. Projecting impacts of climate change in the Himalayas is challenging because of complex interactions between global, regional, and local forcing and responses.
• Evidence suggests that the eastern Himalayas and the Tibetan Plateau are warming, and this trend is more pronounced at higher elevations. However, a lack of sufficient paleoclimate data makes assessing the long-term significance of this warming trend a challenge.
• There are sparse historical climate data in the region, but scientists are fairly confident about projections of future temperature increases. There is more uncertainty in projections of amounts and timing of precipitation.
• Aerosols from the combustion of fossil fuels, wood, and other sources are increasing in the Indo-Gangetic Plain and the foothills of the Himalayas. Absorbing aerosols such as desert dust and black carbon may contribute to the rapid warming of the atmosphere, and model results indicate that this may in turn contribute to accelerated melting of snowpack and retreat of glaciers. Black carbon deposited directly on non-debris-covered glaciers and snowpack could increase the rate of retreat by reduction of surface albedo.
• Over the next few decades, atmospheric concentrations of greenhouse gases are projected to continue to increase globally, while black carbon aerosols are
likely to continue to increase in the Indo-Gangetic Plain and Himalayas. Unless these trends are stabilized or reversed, the impacts of greenhouse gases and black carbon on the rate of Himalayan glacial retreat will increase. That is, both the rate and volume of the glacial retreat will be relatively greater than it would be otherwise.
• The rate of retreat and growth of individual glaciers is highly dependent on glacier characteristics and location. The most vulnerable glaciers are small glaciers at low elevation and with little debris cover. These characteristics also make glaciers more susceptible to black carbon deposition, and model results indicate black carbon deposition may make them more vulnerable to retreat.
• In the eastern and central Himalayas there is evidence of glacial retreat with rates accelerating over the past century. Retreat rates are comparable to other areas of the world. Glaciers in the western Himalayas appear to be more stable overall, with evidence that some may even be advancing.
• In the short term, climate change is likely to increase glacial wastage. In the longer term the impact of continued retreat of glaciers is not clear. The rate of glacial retreat depends not only on temperature, but also on precipitation changes associated with the summer monsoon in the central and eastern HKH and the winter westerlies in the western HKH. Black carbon aerosols, via atmospheric heating and deposition on snowpack and glaciers, may increase the rate of glacial wastage.
• Surface-water flow is highly seasonal and varies across the region, as does the relative importance of glacial meltwater. In most instances, the annual contribution of snowmelt and rainfall to streamflow exceeds that of glacier wastage. Recent literature indicates that the importance of the glacial contribution to runoff has previously been overestimated.
• The contribution of glacial wastage can be more important when the glacial wastage acts as a buffer against hydrological impacts brought about by a changing climate. For example, in the late summer when all snow has melted and the monsoon-rainfall contribution is declining or in the eastern HKH during times of drought.
• Although retreating glaciers will subsidize surface flow and mitigate immediate losses to discharge by retreating glaciers, the loss of glacier “insurance” becomes more problematic for flows in the upper reaches of the eastern HKH over the long term.
• In the western HKH where more of the surface-water flow is from higher elevations, the contribution of glacial wastage could be particularly important in affecting the timing and volume of surface-water discharge.
• Overall, retreating glaciers over the next several decades are unlikely to cause significant change in flows at lower elevations, which depend primarily on monsoon rain. However, for high-elevation areas, current glacial retreat rates, if they continue, appear to be sufficient to alter the seasonal and temporal streamflow in some basins. Removing water stored as glacial ice does not imply any a priori effect on average annual discharge in the long term, assuming annual precipitation remains the same. In the short term with constant annual precipitation, glacial wastage will augment the quantity of streamflow.
• Limited streamflow data in upper basin regions, along with government constraints on scientific access to international streamflow data, increase the uncertainties surrounding hydrological trends, variability, and extreme events in the region. Limited streamflow data also limit the understanding of the relative contributions of rain, snowmelt, and glacial meltwater, as well as groundwater recharge mechanisms in the region.
• Uncertainties in the role of groundwater in the overall hydrology of the region are even greater than those of surface water. Current understanding of groundwater in the region is confounded by a variety of limitations including knowledge gaps about the interaction between surface water and groundwater; difficult terrain; the fractured and variable nature of the underlying geological substrate; and the inability to easily distinguish the contributions of snowmelt, glacial meltwater, monsoonal precipitation, and human actions such as groundwater overdraft to flows. Evidence suggests that sizable and extensive overdraft in the central Ganges Basin is likely to have an earlier and larger impact on water supplies than foreseeable changes in glacial wastage.
• For upstream populations, GLOFs and LLOFs are the dominant physical hazard risk. For downstream populations in the central and eastern Himalayas, floods from changes in monsoon rainfall and cyclones are more likely to be important, along with changes in the timing of extreme events.