Karakoram Anomaly Project
-Pakistan -
“The mountains are calling and I must go, and I will work on while I can, studying incessantly.”
— John Muir
The goal?
To strenghten the resilience of mountain communities to glacial floods
The Karakoram Anomaly Project (KAP) is a multidisciplinary and charitable initiative to address the risk of glacial hazards in Shimshal Valley of Northern Pakistan. Funded in part by the National Geographic Society and Royal Geographical Society among other, and advised by Utrecht University and Imperial College London across the natural and social sciences, policy development and science communication spectrums, the project aims is to reduce human and material losses deriving from glacial floods and strengthen community-based resilience and emergency preparedness capacity in Shimshal Valley, Pakistan.
This article describes in detail the scientific work we carried out in the valley from 2015 to 2018. If you are interested in the more adventurous elements of the project, then please read the ascent on Yukshin Gardan Sar in 2015 and the Exploration of Khurdopin Glacier in 2018. An overview of the activities carried out in Shimshal Village to improve the resilience of communities to flooding can be explored in this article.
Climate change and climate-induced disasters often greatly affect the poorest and most isolated communities. The Hindu Kush Himalayan (HKH) region is particularly prone to natural hazards such as floods, glacial lake outburst, droughts, landslides, avalanches and earthquakes. The unstable geological conditions and steep terrain, combined with climate change and frequent extreme weather conditions, pose myriad challenges for the communities[1]. The frequent occurrence of flash floods, one of the major natural disasters in the HKH, threatens lives, livelihoods and infrastructure, both in the mountains and downstream. Vulnerable groups such as the poor, women, children, the elderly and people with disabilities often suffer the worst impacts. These lack the financial resources to prepare and cope with disasters and have limited awareness of early warning. Furthermore, the lack of watershed management plans and community-based disaster risk reduction programme reduces their ability to address the impacts of climate change such as increased flood frequency, damage, and other extreme weather events. Several communities in the Karakoram region of Pakistan are becoming more vulnerable to climate- and glacier induced flooding and water stress (UNDP, 2019).
Figure 1. Shimshal. ©Ana Pavalache, 2018
Situated at 3100m, Shimshal (Figure 1) is the highest settlement in Pakistan’s northern Hunza region and the last village before the Chinese border. It is reachable only via one rocky road carved into the steep mountains, which is considered one of the world’s most dangerous. Shimshal Valley is known for a rather peculiar environmental phenomenon: surging glaciers. While for most alpine areas, scientists are concerned that global warming and retreating ice will swell glacial lakes and increase the risk of floods downstream, Shimshal has the same concern, but for a somewhat different reason. Several of the glaciers that flow into this valley surge, meaning they cycle through periods when they flow forward several times faster than usual (Bhambri et al., 2017)
To address the risk of glacial hazards in Shimshal, the KAP – a multidisciplinary and charitable initiative has been working across the natural and social sciences, international policy development and science communication spectrums, to improve understanding of the surging phenomenon and thus boost the resilience of vulnerable mountain communities to glacial lake outburst floods (GLOFs). This paper presents the preliminary findings and observations surrounding the subject matter following two field visits carried out by the team in 2015 and 2018.
Introduction
Figure 2. Typical Karakoram landscape. Here, a team of researchers performs a Real Time Kinematic GNSS Survey on Khurdopin Glacier, Shimshal Valley. ©Ana Pavalache, 2018
The Karakoram range, which spans parts of India, Pakistan and China, has the largest concentration of glaciers outside the Polar Regions (Figure 2) and is home to five of the highest peaks in the world, including K2. The perennial snow and ice cover exceeds 16,000 km2 and sustains nearly 200 million people living downstream, of which over 130 million people are in Pakistan (Rankl et al., 2014; Hewitt, 2014). For these reasons, the Karakoram (and the wider Hindukush-Himalaya range) is often called the Third Pole and the Great Asian Water Tower.
Glaciers in the Karakoram are also highly active and flow over extreme topography, which makes them among the steepest glaciers in the world (Sevestre, unpublished). What’s interesting about this region is that during the past decades' several glaciers have revealed a balanced condition or depicted slight mass gain along with frequent advancing and surging occurrences – a phenomenon later termed the ‘Karakoram Anomaly (Hewitt, 2005; Hewitt, 2011).
Surging glaciers switch between periods of slow flow, when they store mass at elevation, and periods of fast flow when they discharge this mass downstream (Figure 3) . During active phases, they can reach velocities 10 or even 100 times higher than normal, while also transporting large volumes of ice from an upper reservoir zone into a lower, receiving zone (Bhambri et al., 2017).
Figure 3. Sketch of the model system based on the Variegated Glacier. A glacier surge often propagates down-glacier from a triggering region accompanied by a traveling surface bulge. @Thøgersen et al., 2019. Surging glaciers are broadly defined as flowing at least 10 times and as much as 100 times faster than normal glaciers. ©Heidi Sevestre, undated.
Worldwide, only 1% of glaciers are known to surge (or roughly 2,300), however, this number is likely to increase as glaciers come under closer surveillance by remote sensing. They are concentrated in geographic hot spots including Svalbard, Canada's Yukon territory, Alaska, western Tibet, and the Karakoram and Pamir mountain ranges of Central Asia (Qiu, 2017). The behaviour of surging glaciers is complex, and their trigger mechanism and the potential link to a changing climate is still shrouded in uncertainty.
Some experts think that glaciers in the Karakoram are prone to surging because of their steepness. For example, as mass builds up from heavy snowfalls near the top of a glacier, gravity alone may be able to trigger a surge. However, most experts believe that there is a combination of factors that determines whether and how often glaciers can surge. This includes water from precipitation and melting, which can pool on the surface, then infiltrate into the glacier through crevasses where it reaches its base, warming and lubricating the ice/bedrock contact and, ultimately, releasing the ice (Qiu, 2017).
Figure 4. Conceptual diagram showing how GLOFs form as a result of surging glaciers. GLOFs are sudden events which can release millions of cubic metres of water and debris leading to the loss of lives, property and livelihoods amongst remote and impoverished mountain communities. It is estimated that over 7.1 million people in Northern Pakistan are vulnerable to these hydrological disasters. ©UNDP Pakistan, 2019
The presence of surges has hazard-related implications such as the creation of glacial lake outburst floods (GLOFs) due to the formation of ice dams that can break catastrophically and suddenly release large quantities of meltwater, ice and sediment that devastate the valleys downstream (Figure 4). These ice dams form rapidly and rarely last more than one melting season. However, the size of ice-dammed lakes is considerably larger than their moraine-dammed counterparts, hence storing more potential energy. In the upper Indus and Yarkand basins of the Karakoram for example, historical records suggest that more than 146 GLOFs can be traced to ice-dammed lakes (Hewitt & Liu, 2010; Iturrizaga, 2005). Nearly all the large GLOFs in the Karakoram originate from the advance of the terminal lobe of a surge-type glacier into a main, ice-free river valley of which the glacier is a tributary (Bhambri et al., 2019).
Considering the implications for loss of life as well as the uncertainty regarding the contribution to sea level-rise, the surging phenomenon requires further investigation to advance our understanding of ice dynamic processes, glacially driven erosion, sediment transport and to derive a model that can predict when a glacier will surge. This in turn will help to provide early warning to vulnerable communities and limit the loss of life (Steiner, et al., 2018).
Case Study
Shimshal Valley and Khurdopin Glacier
Figure 5. Landsat 8 image, showing several glaciers flowing into the Shimshal Valley perpendicular to the flow of the river. In recent decades, the most prone to blocking the river has been Khurdopin Glacier. ©NASA Earth Observatory
Figure 6. Evidence from the surge event and subsequent lake damming visible at the tongue. ©Steiner et al., 2018.
The majority of known Karakoram ice dammed-lakes have been concentrated in a small number of valleys. One of them is Shimshal, a tributary of the Hunza on the north flank of the main Karakoram. Glacial floods are a huge problem here due to the cluster of five surging glaciers, which often form dangerous lakes. Historical records show that these glaciers have triggered at least 20 GLOFs over the past century, of which 11 have been major — making the valley one of the most vulnerable in the Karakoram Range (Figure 5).
Past floods have caused damage with such frequency that villagers in the river towns of Passu and Shimshal even established a fire beacon warning system (Nasa Earth Observatory, 2017). The greatest number of GLOFs has been associated with Khurdopin (36°20’18’’N, 75°28’3’’E), the largest surge-type glacier yet identified in the Karakoram, which blocks the ice-free lower Virjerab valley regularly.
Khurdopin Glacier is approximately 41 km in length, 1.5 km in width and has an elevation range between 3300 m above sea level (a.s.l.) in the Shimshal Valley and 7760 m a.s.l. The glacier is heavily debris covered on the lower 10 km of the tongue and distinct meandering debris bands typical for surge type glaciers are present up to 20 km from the terminus (Steiner, et al., 2018).
Khurdopin has surged most recently in 2017. After years of little movement, the glacier began a rapid advance in October 2016, accelerating to a rate of roughly 15 meters per day by the spring of 2017—one of the fastest rates observed for a mountain glacier globally. As ice and sediment pushed into the river, a sizable lake pooled up in March 2017 (Figure 6). By July 2017, the river had carved an outlet through the glacial debris before the lake could grow extremely large (Nasa Earth Observatory, 2017). Later in July, the size of the lake dropped from 100,000 m2 to about 40,000 m2 in a few days—a rapid enough change to produce a flood downstream that damaged bridges, farmland, and a key road (Pamir Times, 2017). The flooding and road damage left hundreds of people stranded in the Shimshal Valley awaiting evacuation by helicopter (High Asia Herald, 2017). By August 3, 2017, the lake had completely drained. Similar occurrences were registered again in 2018 and 2019.
Karakoram Anomaly Project
Figure 7. Karakoram Anomaly Project Phase II Team
Phase II, 2018
The Karakoram Anomaly Project (KAP) Phase II was developed from the hypothesis that the Khurdopin surge activity may continue and the dammed lake form again and grow bigger in the melt season of 2018. This poses a continuous threat to downstream settlements and infrastructure in case of a sudden breach. The KAP Phase II comprised an international team of researchers and creatives (Figure 7) such as environmental specialist Federica Chiappe (UK), photographer Ronald Patrick (CH), filmmaker Jeremy Janeczko (USA), cinematographer Ross Fairgrieve (UK), research assistant Ana Pavalache (CH) and science officer and team leader, Sergiu Jiduc (UK). The endeavour was funded in part by National Geographic Society (USA), Rab Equipment (UK), Transglobe Expedition Trust (UK), Gilchrist Educational Trust (UK), Mountain Fuel (UK) and Firepot by Outdoor Food (UK). This included cash and in-kind funding. Technical and scientific partners and consultants included Utrecht University (NL), Imperial College London (UK) and Geopraevent (CH).
Methodology
Figure 8. Conceptual model showing the embedder explorative case study research design.
Natural Science Component
The team combined desk- with field-based techniques to bring new insights about the trigger mechanism and behaviour of Khurdopin surge, as well as to assess the likelihood and impact of GLOFs in Shimshal Valley. Central to achieving this goal was to quantify the surface elevation and volume changes, and ice flow velocities at Khurdopin glacier using digital elevation models and high-resolution imagery from the Planet and Landsat satellites (Steiner, et al., 2018).
To derive spatial velocities we use cross-correlation feature tracking using the COSI-Corr software (Leprince et al., 2007) on selected Landsat imagery between 2000 and 2017 (30 m resolution), and on Planet high-resolution im-agery (3 m resolution) between 2016 and 2017 (Planet Team, 2017) (Supplement Table S1). Volume changes were com-puted using the SRTM from 2000, a TanDEM-X DEM from 2011 and a DEM generated from ASTER imagery from May 2017. The ASTER DEM was generated using the open-source Ames Stereo Pipeline software (Shean et al., 2016). We compared the DEMs in stable off-glacier terrain and cor-rected the products accordingly. Using the GlabTop2 model (Frey et al., 2014) and the SRTM, we com-puted ice thickness for the glacier and inferred the bed topog-raphy. Details on the specific COSI-Corr settings as well as the imagery used are provided in the Supplement. The poten-tial lake volume was calculated by intersecting the visually derived lake perimeter with the TanDEM-X DEM.
To investigate velocities on Khurdopin, we separated the tongue into 25 bins at 1 km equidistance along the centreline, and calculated the mean velocity within the bin. Using high-resolution imagery from the Planet satellites with sub-weekly overpasses (Planet Team, 2017), we were able to characterize the surge event and the surface dynamics on the lower tongue and near the glacier terminus.
In order to understand the recent changes and movements and develop a three-dimensional spatial story of the surging glacier, the field team also assessed the glacier surface in detail to describe crevassing, meltwater systems and possible lakes occurring between the headwall and the ice. This was achieved via terrestrial photogrammetric techniques and repeat photography.
Figure 9. The concept of integrated vulnerability with the determining factors and indicators. While some of the factors in exposure and vulnerability are physical and environmental, other factors are socioeconomic, such as poverty, human settlement and habitat, lack of preparedness, susceptibility, and adaptive capacity. Poverty leaves many people in the Karakoram region with few resources when trying to rebuild their homes and livelihoods. ©Hegglin and Huggel, (2008). Disaster risks are a function of interplay between three key elements: hazard, exposure, and vulnerability. Natural hazards are increasing in magnitude and occurrence due to various drivers of change including climate change. The susceptibility of a community to the impact of a natural hazard increases due to both exposure of people and property to the hazard, and vulnerability. ©Vaidya R.A. et al. 2019.
Social Science and Community Development Component
The field research team also combined qualitative research methods embedded in an exploratory case study research design (Figure 8) to explore sustainable livelihood approaches that can inform the development, implementation and management of a community-based disaster risk management (DRM) action plan in Shimshal Valley. The underlying goal for this was to help improve the resilience of communities to catastrophic floods by sharing scientific knowledge and risk information, measuring vulnerability and identifying gaps in preparedness, response and recovery. Addressing DRM through a livelihoods approach, vulnerability and resilience were analysed according to the six different asset categories - natural, physical, social, human, financial and political. These assets together contribute to people's resilience to floods (Figure 9).
To achieve this, the field team carried out a series of focus group discussions and semi-structured interviews with local community members, DRM implementation agents and decision-makers, as well as emergency response actors who share direct or indirect experience with Shimshal glacial hazards.
The focus groups explored the socio-economic impact and vulnerability of GLOFs in terms of loss of individual and community assets including livelihood, agriculture, livestock, roads, bridges, irrigation canals as well as intangible assets such as education and culture. They also collected information egarding the response to such hazards, preparedness, risk assessment and traditional mitigation options. Key results and highlights of the project so far are presented in the section below.
Results
Highlights
• The 2017 Khurdopin surge confirms a return period of 20 years, with observations of floods caused by lake drainage after a surge in 1901-04, 1923, 1944, 1960, and observed surges in 1979, 1999 and 2017.
• The 2017 surge was characterized by peak ice flow velocities that reached 5000 m/year or roughly 15m/day – the fastest rate recorded for a mountain glacier in the region.
• There is a division point at 12 km up-glacier that separates two distinct reaches of the tongue: (a) the upper reach where velocities gradually increase during the build-up phase and mass continuously accumulates during the 19 years between surges, and (b) the lower reach where velocities peak during the surge.
• During the surge in May 2017, the glacier surface between km-3 and km-12 gained 50 to 160 m in height, which in turn suggests a net volume gain between 2015 and 2017 of around 900 megatons.
• At this stage, it is difficult to ascertain which are the main drivers for the regular surges on Khurdopin. In combination with a gradual accumulation of mass on the upper tongue during quiescence, and a resulting steepening surface gradient, the actual surge starts rapidly when a tipping point is reached. The increase in velocity at this specific location has possibly caused a switch from an otherwise cold to a temperate bed, initiating a rapid increase in basal sliding from 2015 onwards. The sudden absence of supraglacial ponds on the terminus during the surge and the formation of a supraglacial pond in May 2000 after the last surge exactly at the location of the clear line of change around km-12, could also point at a disturbed englacial network playing a role.
Velocities during surge events
Figure 10. Velocities measured from cross-correlating Landsat imagery of 1 year of the quiescent phase (a; 17 October 2010–18 September 2011), the last year of the build-up (b; 28 August 2015–10 May 2016) and the surge peak in May 2017 (c; 13 to 29 May 2017). Panel (d) shows mean values of the bins compared against bed elevation and all available velocity pairs for 2017, between December 2016 (dark red) and September (yellow), both for Landsat and Planet. Values for quiescence are shown at 1 order of magnitude larger, while values for the surge phase are shown at 1 order of magnitude smaller than measured. Note the difference in scales for the different phases. All coordinates are in UTM WGS84 Zone 43. ©Steiner et al., 2018.
Mean average surface velocities on the 25 km long main tongue of Khurdopin during a quiescent phase are below 5 m/year, with a small peak of 15 m/year at around 12 km along the tongue. The peak corresponds to a markedly steeper section of the profile. While lack of cloud-free imagery or poor image quality does not always allow accurate identification of the onset, peak and termination of the surge, the data suggest that a gradual increase of surface velocities over multiple years led to surging peaks with velocities up to 4000 m/year in 1979 and 1999 (Quincey and Luckman, 2014).
The most recent quiescent phase lasted from 2000 until at least 2011. By 2013 the glacier had reached surface ve-locities above 100 meters/year beyond the steep section (km-12), but still smaller than 10 m/year in the lower 5 km. The build-up phase between the quiescent phase and the actual surge peak between 2015 and 2016 was characterized by increasing surface velocities in the tongue’s upper reach (Figure 10). Between early 2017 and be-ginning of June velocities increased up to 5200 m/year and dropped again to below 200 m/year in most parts by September. While this extreme acceleration and deceleration happened within less than 9 months, the velocity peak along the longitudinal profile remained relatively stable. The gradual build-up and then relatively short surge peak support earlier findings based on less frequently available data (Quincey and Luckman, 2014).
Ice volume changes during surge events
Figure 11. Elevation change rates during the quiescent phase (a), and during the build-up and surge phase (b). Mean values per bin are shown in panel (c). Note that values for the surge phase are shown at 1 order of magnitude smaller than measured. All coordinates are in UTM WGS84 Zone 43. ©Steiner et al., 2018.
Apart from increased velocities, surges logically also result in large amounts of displaced ice volume. In many cases this results in a rapid extension of the position of the glacier’s snout. However, in the case of Khurdopin the apparent terminus does not advance and has not done so during at least the recent surges, since it has turned into a stable moraine, dynamically decoupled from the active part of the glacier. This makes detection of actual length changes of the active tongue visually difficult (Figure 9).
Using three DEMs (SRTM in 2000, TanDEM-X in 2011 and ASTER in 2017) the elevation change rates for the quiescent and surge phases are quantified (Figure 11). To match the ice volume changes with the actual surge phase, we extrapolated the annual rate of surface elevation change observed between 2000 and 2011 until 28 August 2015 as a proxy for the qui-escent phase surface elevation change. Given the steep rise in velocity it is assumed that from 28 August 2015 onwards the surge phase started and the annual rate in surface elevation change was estimated from the remaining volume difference and the 2017 DEM. The transition from positive to negative elevation change during the quiescent phase is clearly no-table and coincides with the steep section of bedrock around km-12 (Figure 11a), an observation made earlier by Gardelle et al. (2012), who identified this distinct behaviour for other glaciers in the region as well. This distinction is again visible exactly at the same location for the surge, when elevation change is positive in the lower reach where mass is accumulating.
During the surge in May 2017 the glacier surface between km-3 and km-12 has gained height by 50 to 160 m. Based on the elevation changes we find a net volume gain be-tween 2015 and 2017 of 2485×106 m3 (±×106 m3 based on the DEM accuracy) between the steep section and the part of the terminus where no more surface change is visible. Aver-aged over the entire glacier we estimate that the overall vol-ume loss is slightly negative (see surface elevation change in Figure 10), similar to what is reported by Bolch et al. (2017).
Field Observations
Khurdopin’s ablation zone is characterised by irregular thickening with wave-like zones of higher ice moving down the glacier, particularly in the first 15km from the snout. Continuous sections of ice cliffs rise steeply to the glacier surface and over-ride some glacier margins. A relatively well-defined line of shear occurs between the active ice and passive margin. The glacier also shows rapidly changing supraglacial meltwater systems with the closing of ice margin conduits to subglacial drainage and ponding of water at the ice edge. The Virjerab River is still dammed by the glacier (i.e. 45m high dam) and is draining subglacially. Unless floating ice blocks these meltwater channels, the chance for new dangerous lakes to form this melting season are low. However, when compared with 2017, the incidence and magnitude of these geomorphological, glaciological and hydrological features, suggest that surging activity is slowing down, and ice mass is rapidly being wasted due to high temperatures. Further field observations are presented in the images below.
Figure 12. The view over the moraine allows us to see the added mass of the surge – a ‘normal’ glacier surface would not be visible from this vantage point as the moraine crest is generally the maximum height of the last glacial maximum. However, in this case the up to 50m added ice shows on top. The top right inset shows the shear that folds from the glacier surface. Glacier flow is happening in the shape of an onion. Glacier ice emerges in the ablation zone from below. This is why those faults come out at exactly the angle the ice emerges and the ice is covered in debris as it is transported from the base. @Jiduc and Steiner, unpublished.
Figure 14. IIn the background the black line marks the former height of the moraine from a previous – stronger – surge. Here the ice was another 50m higher than in 2018. In the foreground we see the emerged debris (red) and as a result of the ruptured melt channels water ponding on the surface that is heavy with sediments (rather than clean blue/green if it was just surface melt). ©Jiduc and Steiner, unpublished.
Figure 13. The view up-glacier allows us to grasp the added mass (the blue dashed line would be the surface before the surge). You can also clearly see the separation between the upper part – clean ice, where it depletes – and the lower part where ice emerges from the base heavy with debris. The second pane shows again the emerging ice through the shear planes. ©Jiduc and Steiner, unpublished.
Figure 15. Evidence of surges at the snout. Here the ice from the last surge is visible to be behind sediment deposits of earlier advances. ©Jiduc and Steiner, unpublished.
Another interesting observation is the stable mass balance of Khurdopin Glacier depicted over the past 90 years. Figure 16 below, shows a repeat photo pair of Khurdopin glacier in post-surge phase (after the 1923 and 2017 surge events respectively). By comparing the 1925 and 2018 images, we can observe that both the glacier tongue and the icy head walls are at the same level suggesting in turn that the glacier experienced little to no glacier thinning. This irregularity is often called the Karakoram Anomaly and was first noticed by Hewitt, (2005) before being confirmed by subsequent geodetic studies. Thus, glacier behaviour in the Karakoram is highly heterogeneous, both spatially and temporally, and its drivers are not yet fully understood.
Figure 16. Upper section of Khurdopin glacier as seen in 2018 (left) and 1925 (right). Comparison of the two images suggests that little ice mass loss has occurred over the past 98 years: a good, real example of the so-called Karakoram Anomaly. (Left) Capt. Visser, 2015 ©Royal Geographical Society with IBG, London and (Right) ©Ronald Patrick, 2018 for KAP.
GLOF Hazard
Figure 17. The Khurdopin-Virjerab ice dam as seen on 05.07.2018. ©Ronald Patrick.
Figure 18. The Khurdopin-Virjerab ice dam as seen from our UAV on 05.07.2018. ©Jiduc, unpublished.
• In April 2017, a dammed lake formed in lower Virjerab Basin, which grew quickly from 72 000m3 at the beginning of May to 1 million m3 a month later and peaking at 2 million m3 on June 28. The lake finally drained starting July 21 and had disappeared by 5 August. As a consequence, the river washed away the main Shimshal road at multiple locations, destroyed at least one main bridge and eroded local agricultural land, making the valley inaccessible for a week.
• In 2018, the field team observed that the Virjerab River remained dammed by a 45m-high ice wall and was draining subglacially. Unless floating ice blocks these meltwater channels, or the surging reactivates, the chance for new dangerous lakes to form next melting season is low (Figiure 17 and 18).
• In 2019 a similar lake formed which drained over a period of a couple of days.
The tongue of the Khurdopin Glacier reaches across the main valley floor. As a consequence, the glacier has blocked the local Vijerab River multiple times in the last century, caused by the tongue pushing towards the opposite headwall of the main valley (Figure 18). Most of the reported lake drainages were not catastrophic and they have rarely caused damages down-stream beyond eroded fields and damaged bridges (Hewitt and Liu, 2010; Iturrizaga, 2005). From historic Landsat im-agery it is obvious that a lake formed during the melt sea-son in two consecutive years after the surge in 1999, likely because the added mass required considerable time to be eroded. In late April 2017, the lake formed at exactly the same location, growing quickly from 72 000 m3 at the be-ginning of May to 1 × 106 m3 1 month later and peaking at 2 × 106 m3 on 28 June. The lake finally drained starting around 21 July and had disappeared by 5 August. As a con-sequence the river washed away the road at multiple loca-tions, destroyed at least one main bridge and eroded local agricultural land, making the valley inaccessible for a week.
Ice floes on the water surface indicate ice calving from the advancing tongue and could pose an additional threat as they could block a drainage channel temporarily and create a sud-den spill upon disintegration. Considering the height of the advanced glacier tongue – between 15 m at the fringe and up to 160 m on the surging tongue – and the fact that in 2000 the lake reached lake levels ca. 10 m higher than in 2017, we show potential lake extents that could reach beyond 1 km2 or 10 × 106 m3, possibly during the melt season of 2018 or 2019. Repeat floods in the 1 or 2 years after a possible surge event have been reported multiple times in the recent cen-tury as well (Hewitt and Liu, 2010). The volumes calculated could be decreased by sediments visibly deposited either by the surging glacier or the dammed Vijerab River.
Other Hazards and Frequency
• Shimshal faces multidimensional hazards. For example, GLOFs and rain derived flash floods (especially in the summer) have increased in frequency and magnitude. Furthermore, changes in the climate have been observed with regard to a shift in the snow and rain season, from the lower precipitation amounts in November-January to heavy snowfalls in mid-late March. Avalanches and landslides are also a constant threat due to the extreme gradient of the Karakoram mountains.
• Shimshal lacks the institutional mechanisms and capacity to deal with such a multi-hazard environment and cascading disasters.
• Shimshal is not the only valley in the Central Karakoram experiencing glacial and hydrological hazards. In 2018 and 2019, Ishkoman and Shishper, glaciers both caused medium to heavy damage due to flooding and rapid sliding and demonstrated the vulnerability of infrastructure from the National scale (Karakoram Highway also known as KKH) to the local (Ishkoman village). Hassanabad glacier is also increasing its flow rate, which could indicate another “surge in the making”. All of these suggest that surges (and their related impacts) are not isolated incidents but widespread phenomena that require urgent attention.
Discussion
Figure 19. Evidence from the surge event visible at the tongue. The left panel is based on a Planet image from the 28th of June 2017, the centrelines and the lake extent in 2017 are also mapped from Planet imagery. The lake extent in 2000 is mapped from the panchromatic band of Landsat-7. The projected lake extents and depths are computed based on the Shuttle Radar Topography Mission Digital Elevation Model. The yellow rectangles show areas of heavy crevassing. Arrows at (a) and (b) 5 denote angle of view for images on the right. Panel (a) shows an overview of the front of the tongue and panel (b) shows the front of the advance. ©Steiner et al., 2018
The data collected and analysed support earlier studies on Khurdopin in the observation of a relatively constant return period of a glacier surge of 20 years since the end of the 19th century, irrespective of a changing climate and surges of nearby glaciers (Hewitt and Liu, 2010; Quincey et al., 2011; Quincey and Luckman, 2014). Using distributed velocity and elevation change data we furthermore show that a division point exists at 12 km up-glacier that separates two distinct reaches of the tongue:
(a) the upper reach where velocities gradually increase during the build-up phase and mass con-tinuously accumulates during the 19 years between surges, and
(b) the lower reach where velocities peak during the surge and the ice mass previously accumulated in the upper reach is relocated within only a number of weeks. This line likely coincides with a steep bedrock section and is located just below a tributary that possibly supplies a lot of additional mass via avalanche deposits.
The surge of 2017 showed a similar 4-year build-up time as the surge in 1979 over which the glacier surface in the upper reach increased by approximately 3 m/year and decreased by up to 7 m/year in the lower reach. This period is defined by constantly increasing velocities in the upper reaches. It is difficult to ascertain which are the main drivers for the regular surges on Khurdopin Glacier (Quincey and Luckman, 2014). In combination with a gradual accumulation of mass on the upper tongue during quiescence and a resulting steepening surface gradient, the actual surge starts rapidly when a tipping point is reached.
Ice deformation (Greve and Blatter, 2009; Round et al., 2017), with a modelled ice thickness between 120 and 350 m, results in a velocity of 1 to 60 m/year for the 1–4 degrees steep surface gradient over the whole tongue and 3 m/year at the steep section. This is of the same order of magnitude as the measured velocities during quiescence and the early build-up phase. While these values stay relatively stable for most of the tongue and can account for the overall glacier flow velocities they increase by an order of magnitude to more than 50 m/year in the steep section due to the increase in surface gradient, thus making up more than 50 % of the observed surface velocity. The increase in velocity at this specific location has possibly caused a switch from an otherwise cold to a temperate bed, initiating a rapid increase in basal sliding from 2015 onwards.
The sudden absence of supraglacial ponds on the terminus during the surge (Figure 19) and the formation of a supraglacial pond in May 2000 after the last surge exactly at the location of the clear line of change around km-12, could also point at a disturbed englacial network playing a role (Kamb, 1987; Mayer et al., 2011). At least the last two surges occurred at the beginning of the melt season, which could further catalyse the surge if melt water reaches the ice-bedrock interface. Basal sliding is also most likely the dominant flow process as the cross-profiles of surface velocity indicate plug flow, characterized by flat rather than parabolic velocity profiles as was observed during the quiescent phase (Kamb et al., 1985).
As previously suggested, the surge on Khurdopin is hence likely triggered by the thermal switch, but the actual surge is dominated by basal sliding (Quincey and Luckman, 2014), similar to Kyagar Glacier (Round et al., 2017). Future field observations should focus on finding possible evidence for these processes and possible feedback processes, especially related to the deformation of water-saturated granular base material that could explain these extreme acceleration rates and peak velocities (Damsgaard et al., 2015).
The surface velocities observed during the peak surge in May 2017 on Khurdopin Glacier are, together with the recently observed surge on the neighbouring Hispar Glacier (Paul et al., 2017), the fastest so far reported for the region. In their magnitude and rapid acceleration and deceleration they are comparable to similar bursts at the closely investigated Variegated Glacier (Kamb et al., 1985), where observations with even higher temporal resolution were available. The increased velocity and asso-ciated ice volume redistribution resulted in increased strain rates, evidenced by crevasses appearing at the glacier surface since early May with a marked increase in size and num-ber since mid-June (Fig. 3d). The high peak rates of basal slip can result in erosion rates up to 0.5 m a−1 for a brief pe-riod (Humphrey and Raymond, 1994), a value several orders of magnitude higher than typical erosion rates in mountain ranges. Large amounts of additional sediments were visible at the glacier snout during the surge.
Read our academic paper in the Journal of Cryosphere.
Shimshal Village GLOF Vulnerability
Figure 20. GLOF risk map for Shimshal Valley as developed by Utrecht University Mountain Hydrology Group. This was complemented by a participatory mapping exercise in Shimshal Village which categorized assets according to their corresponding vulnerability to GLOFs. The table below provides further information. ©Steiner, unpublished.
Our research shows that Shimshal village has a flood-prone built environment with most physical assets (including the school and the new hospital), built on flood channels and plains, and lacking flood defence systems (Figure 20 to 21).
• The level of flood resilience of households is varied and depends strongly on the position and the materials of the houses. The most important animals, such as yaks and sheep, are kept in pastures. Some smaller ones are kept for domestic purposes. Agriculture is mostly carried out in flood-prone areas. There was no mention of the use of flood-resistant crops. There was however the mention of planting trees as protection against floods, and to recreate watersheds that might be affected. All the roads are close to the river and are built with local, simple materials. Electricity generation facilities and distribution lines are also close to the river, and so are bridges, which have some cement but are usually wooden. Irrigation channels are varied, some are less exposed than others. Despite the fact that the main school and hospital were built in the flood plain, they are new, so there is the hope that they would withstand glacial hazards.
• Whilst cultural heritage, education, and life-choices are impacted by floods, community social-cohesion, strengthened by religious practices, is high. Similarly, people’s strong awareness of mountain ecosystem services strengthens adaptive decision-making and in turn, coping mechanisms during times of adversity necessary to rebuild physical assets.
Figure 21. Flood vulnerability zones in Shimshal Village. @Jiduc, unpublished.
• Recovery capital however is limited: the government tends to operate reactively, and the gap between disaster preparedness and response is filled by resource-constrained NGOs. The lack of resources hinders the institutionalisation of village authorities to manage issues effectively. Partnerships are necessary to develop robust disaster-risk-management programmes.
• The Shimshal community has a wealth of indigenous knowledge and extremely useful coping mechanisms and practices which can be explored further in order to develop better strategies for mitigating flood hazard in the future.
• Thanks to the call to prayer, the community can effectively come together, when a flash flood occurs to rebuild impacted irrigation channels. This indicates that the community has a strong bonding social capital.
Shimshal’s response to GLOFs
The Shimshal community’s response to GLOFs is primarily based on self-help. The population has a self-sustaining resistance within their system for self-sustaining resistance within their system for people participate actively in providing evacuation and relief services to the victims using their self-and relief services to the victims using their self-to reduce the hazard risks. Due to poor livelihood conditions and lack of resources, they face challenges in taking effective, longer term response measures for risk reduction or mitigation. There is also very limited medical help during emergency situations. Though local committees are there for undertaking different tasks, gaps exist in their coordination and capacity building. Outside help from any government agency is rarely available or not available in time. In examining the roles of various departments and organizations involved in disaster planning and preparedness in the region, there seems to be some deficiencies in actual planning for preparedness, capacity and skill development before a disaster. The situation demands all-out efforts for integration among different elements and stakeholders, changing priorities for risk reduction and strengthening local capability for better preparedness and hazard mitigation.
The road to flood resilience: building local capacity
Through the KAP, the Shimshal community acquired an improved scientific understanding of glacial dynamics and learned what tools and methods are used by the scientific community to study glacial change. This includes learning about specific remote sensing methods to quantify ice flow velocity, volume and elevation changes (e.g. cross-correlation feature tracking with Landsat and Planet imagery, and topographic analysis via SRTM, TanDEM-X and ASTER imagery). Furthermore, the community learned that many of these tools and datasets are available free of charge for research over a limited area, at high resolution, and sub weekly acquisition. Anyone in the village with a connection to the Internet can download and process satellite imagery for hazard identification and risk assessment. However, there are limits to this approach as Internet availability is very poor in Shimshal.
A further important output is the improved awareness of early warning systems. By using videos and focus group discussions, the community learned what a radar-based system comprises, including level measurement sensors for rivers and lakes. This improved knowledge is extremely valuable for influencing stakeholder consultations with decision makers. In particular it will help the community to further the engagement with local institutions such as United Nations Development Programme (UNDP) Pakistan, working on holistic approaches to disaster risk management, to ensure that community-based adaptation is prioritised, and that the voice of the community is heard.
Strong and lasting relationships have been created with the community members: from elders who treasure the valley’s traditions and found space in the project to share them more widely, to the youth, who are looking for opportunities to improve the livelihoods of their families. In particular, the team developed a strong bond with a very talented young man who is currently building a small resort on his own and is interested in learning new skills from the team to build low-cost early warning systems. This and other approaches to improve the valley’s resilience are currently being explored by the Karakoram Anomaly Project.
Driving Disaster Risk Reduction (DRR) in Shimshal
When seeking ways to reduce flood risks and increase community resilience to hydrological disasters in the Karakoram region, a holistic approach to risk management is needed, where risks are transferred to the stakeholders that are best placed to manage them. This has been clearly emphasised in The International Centre for Integrated Mountain Developments (ICIMOD) GLOF disaster risk management framework (ICIMOD, 2019, Figure 22).
Figure 22. Issues and activities in GLOF risk management. ©ICIMOD (2015).
The focus on post-disaster response is in fact no longer adequate or effective in dealing with disasters. Also, more data and technology will not be sufficient on their own to improve people’s lives. What Shimshal needs is an integrated approach to managing the risks of disasters:
• Regional institutions are best placed to deal with transboundary GLOF issues, to facilitate information sharing and collaboration modalities
Figure 23. Cross-level partnerships can facilitate the coordinated implementation of four principles that integrate community and ecosystem objectives in disaster resilience (coloured rings) and help address paradoxes that pose challenges for DRR in mountains (grey ring). ©Klein et al., 2019
• National and sub-national government to develop and enforce policies and regulations, such as on water conservation and use, in addition to environmental and social safeguards. They can also coordinate emergency preparedness and response to the affected communities.
• Extension workers and NGOs to design and deliver training, capacity building, and awareness campaigns.
• Academic institutions to provide increased evidence on hydrological and glacial risks and solutions.
The Shimshal community, as the first to be impacted, needs to be the most ‘ready’ and fast responding. As such, community members need to be able to voice their needs and challenges to policymakers and practitioners15, be engaged in developing culturally appropriate and feasible solutions, and be empowered with resources and knowledge to respond and recover. Since GLOFs are a regular feature in Shimshal, flood risk reduction strategies should be based on adopting suitable community-based structural and non-structural measures.
Community based disaster risk reduction (CB-DRR) has become increasingly popular over the last 20 years (Allen, 2006) as a way to reduce vulnerabilities and build local-level capacity for disaster response and recovery. There is growing awareness in CB-DRR of the value of local knowledge for securing livelihoods and building community resilience to address extreme events (Mercer et al., 2010). In CB-DRR, solutions come from the community in a bottom-up process, which empowers them to develop and manage locally appropriate strategies that are tailored to the particular locations and people in that area (Newnham et al., 2015). While policymakers recognize the significant economic, human, and environmental costs of disasters, practical guidelines for implementing DRR are limited for vulnerable populations like those in the mountains (Wymann von Dach et al., 2017).
Ecosystem-based disaster risk reduction (Eco-DRR) on the other hand, emphasizes ecosystem conservation, restoration, and sustainable management as key elements for DRR. We believe that by integrating the emerging field of Eco-DRR with community-based DRR (CB-DRR), it will help to address the increasing vulnerabilities of mountain people and ecosystems in Shimshal (Figure 23) (Klein et al., 2019). This integrated CB- and Eco- DRR framework comprises four principles:
• governance and institutional arrangements that fit local needs;
• empowerment and capacity-building to strengthen community resilience;
• discovery and sharing of constructive practices that combine local and scientific knowledge; and
• approaches focused on well-being and equity.
Recommended actions for Shimshal
Governance
• Develop partnerships among voluntary agencies,governments, donors, multi-lateral agencies, businesses, and local community for DRR (e.g. Shimshal Trust with UNDP, Agha Khan Agency for Habitat, and Pakistan Meteorological Department);
• Examine local adaptation strategies and their linkages with planned adaptation in the current policy environment;
• Lobby to allow local community leaders to have a seat on the local government’s decision-making board for flood control.
Capacity
• Provide training for and implement flood resistant construction and retrofitting;
• Develop disaster preparedness using games, role playing, and simulation exercises including during monthly women’s clubs’ meetings;
• Provide disaster-based insurance to poor smallholder farmers. If possible, also paid through labour on community based DRR activities;
• Facilitation of farmer access to credit, also through innovative decentralised banking systems;
• Implement a community-based flood early warning system using sensors, solar transmitters, mobile phones to warn downstream villages (Box 3).
Knowledge
• Provide training to farmers about risks, climate smart agriculture, weather index-based insurance, and financial literacy through participatory techniques such as games, theatres, storytelling; weather index-based insurance may combine international and local knowledge;
• Carry out field-based 'mobile workshops' where local participants can share their experiences in monitoring glacial lakes with communities and scientists;
• Develop civil society-private sector partnership with climate adaptation strategies shared across players and sectors through community social networks.
Equity
• Create gender-sensitive disaster preparedness and response plans, addressing specific needs of diverse groups within the community;
• Identify impacts on and adaptive capacities of women and men;
• Define strategies for equitable access to resources for minority, indigenous people, and women;
• Empower grassroot women networks to gain support of local and national governments and donors to bring their priorities and practices to the forefront of policy and programming.
A suitable early warning system
Figure 24. Example of an early warning system as developed by Geopraevent in the Shaksgam Valley of the Chinese Karakoram. ©Geopraevent, 2019
Early warning systems comprising a simple instrument installed upstream to detect either floods or debris flows can reduce the threat of GLOFs by automatically generating flood signals that are relayed to downstream communities ( ICIMOD, Indus Basin CBFEWS). A radar-based lake and river level measurement system coupled with webcams at the tongue of Khurdopin Glacier would provide continuous surveillance of the situation and several radar level meters would provide flow height data further downstream. All data, as well as meteorological parameters, would be transmitted to a data portal several times a day via satellite link and the system would automatically generate alerting SMS that are sent to local authorities if dangerous levels are reached (Figure 24) (Geopraevent, 2019). To be effective, community based early warning systems should be based on four elements: risk knowledge and scoping, community-based monitoring and early warning, dissemination and communication and response capability and resilience.
Policy Recommendations
• NGOs, GOs and community institutions need to urgently adopt a multi-hazard risk assessment approach. Such an approach should address primary, secondary, and cascading hazards.
• Action in the four pillars of information, infrastructure, institutions, and insurance is urgently needed. Different instruments within command and control, monetary incentives, persuasion, and nudging can be developed and made available. Sensitization to gender differential vulnerabilities within society must increase, in order to reduce vulnerability.
• Countries of the Karakoram region need to cooperate more extensively and effectively by sharing data, information, and scientific and indigenous knowledge, and by fostering transboundary disaster risk management practices. Institutional arrangements for collective action should be enhanced and capacity building programmes organized for strengthening of such regional cooperation.
• Awareness creation of flood hazard preparedness and risk reduction among target communities and key stakeholders is required. The role of media and education institutions would be beneficial in mass communication and capacity building for this purpose.
• Particular attention needs to be paid to designing and delivering training programmes targeted at women, who work in the fields and so are at high risk of flood hazard. They can be provided basic disaster management capacity building. Gender emerges as one of the most significant socioeconomic factors affecting vulnerability. While both men and women in the Karakoram have valuable knowledge, skills, experience, and coping capacities, these strengths tend to differ by gender. And unlike men’s capacities and knowledge, those of women are often ignored in policies and formal arrangements related to development, mitigation and recovery.
Action Plan: What can the Karakoram Anomaly Project do?
• Deepen environmental research to help develop a model that can predict glacier surges;
• Connect and partner with other programmes and entities nationally, regionally, and globally, for joint evidence building for policy making;
• Continue the social research started, and link the research in the Shimshal Valley with case studies in other regions;
• Develop innovative communication material (starting with a documentary) combining environmental and social research with expedition ‘with purpose’;
• Fundraise / crowdfund / seek co-finance to develop and deploy cost-effective early warning systems (EWSs) in the most vulnerable valleys in the Karakoram;
• Disseminate findings through participation in conferences and other events.
References
Allen K.M. (2006). Community based disaster preparedness and climate change adaptation: local capacity building in the Philippines, Disaster, 30.1, pp 81-101.
Bhambri, R., Hewitt, K., Kawishwar, P., B. Pratap, B. (2017). Surge-type and surge-modified glaciers in the Karakoram. Sci. Rep., 7, p. 15391.
Bolch, T., Kulkarni, A., Kääb, A., Huggel, C., Paul, F., Cogley, J. G., et al. (2012). The state and fate of himalayan glaciers. Science 336, 310–314. doi: 10.1126/science.1215828.
Bolch, T., Pieczonka, T., Mukherjee, K., and Shea, J. (2017). Brief communication: Glaciers in the Hunza catchment (Karakoram) have been nearly in balance since the 1970s, The Cryosphere, 11, 531–539, https://doi.org/10.5194/tc-11-531-2017.
Bonekamp, P. N. J., Kok, R. J. De, Collier, E., Immerzeel, W. W., & Grogan, D. S. (2019). Contrasting Meteorological Drivers of the Glacier Mass Balance Between the Karakoram and Central Himalaya, 7(June), 1–14. https://doi.org/10.3389/feart.2019.00107.
Brun, F., Berthier, E., Wagnon, P., Kääb, A., and Treichler, D. (2017). A spatially resolved estimate of High Mountain Asia glacier mass balances from 2000 to 2016. Nat. Geosci. 10:668. doi: 10.1038/ngeo2999.
Damsgaard, A., Egholm, D. L., Piotrowski, J. A., Tulaczyk, S.,Larsen, N. K., and Brædstrup, C. F. (2015). A new methodology to sim-ulate subglacial deformation of water-saturated granular mate-rial, The Cryosphere, 9, 2183–2200, https://doi.org/10.5194/tc-9-2183-2015.
Gardelle, J., Berthier, E., and Arnaud, Y. (2012). Slight mass gain of karakoram glaciers in the early twenty-first century. Nat. Geosci. 5, 322–325. doi: 10.1038/ngeo145.
GeoPraevent. (2019). Level Measurements. Available from: https://www.geopraevent.ch/technologies/level-measurements/?lang=en.
GeoPraevent. (2019). Glacial lake monitoring in western China (Shaksgam-valley). Available from: https://www.geopraevent.ch/project/glacial-lake-monitoring-western-china-shaksgam-valley/?lang=en.
Greve, R. and Blatter, H. (2009). Dynamics of Ice Sheets and Glaciers, Springer Berlin Heidelberg, Berlin, Heidelberg.
Frey, H., Machguth, H., Huss, M., Huggel, C., Bajracharya, S.,Bolch, T., Kulkarni, A., Linsbauer, A., Salzmann, N., and Stof-fel, M. (2014). Estimating the volume of glaciers in the Himalayan-Karakoram region using different methods, The Cryosphere, 8, 2313–2333, https://doi.org/10.5194/tc-8-2313-2014.
Hegglin E, Huggel C. (2008). An integrated assessment of vulnerability to glacial hazards—a case study in the Cordillera Blanca, Peru. Mt Res Dev 28:299 309.
Hewitt, K. (2005). The Karakoram anomaly? Glacier expansion and the “elevation effect”, Karakoram Himalaya. Mt. Res. Dev. 25, 332–340.
Hewitt, K. and Liu, J. (2010). Ice-dammed lakes and outburst floods, Karakoram Himalaya: Historical perspectives and emerging threats. Phys. Geogr. 31, 528–55.
Hewitt, K. (2011). Glacier change, concentration, and elevation effects in the Karakoram Himalaya, upper Indus basin Mt. Res. Dev., 31, pp. 188-200.
Hewitt, K. (2014). Glaciers of the Karakoram Himalaya: Glacial Environments, Processes, Hazards and Resources. Springer Science & Business Media.
High Asia Herald. (2017). Floods cause massive destruction in GB; hundreds stranded in valleys, on roads Available from: http://thehighasia.com/floods-cause-massive-destruction-in-gb-hundreds-stranded-in-valleys-on-roads/.
Humphrey, N. F. and Raymond, C. F. (1994). Hydrology, erosion and sediment production in a surging glacier: Varie-gated Glacier, Alaska, 1982–83, J. Glaciol., 40, 539–552, https://doi.org/10.1017/S0022143000012429.
ICIMOD. (2018). Pakistan Government Credits CBFEWS for Zero Loss of Human and Animal Lives in Sherqilla Floods. Available from: https://www.icimod.org/?q=30038.
Iturrizaga, L. (2005). New observations on present and prehistorical glacier-dammed lakes in the Shimshal valley.
Kamb, B. (1987). Glacier surge mechanism based on linked cavity configuration of the basal water conduit system, J. Geophys. Res., 92,9083–9100, https://doi.org/10.1029/JB092iB09p09083.
Kamb, B., Raymond, C. F., Harrison, W. D., Engelhardt, H.,Echelmeyer, K. A., Humphrey, N., Brugman, M. M., and Pfeffer, T. (1985). Glacier surge mechanism: 1982–1983 surge of Variegated glacier, Alaska, Science, 227, 469–479, https://doi.org/10.1126/science.227.4686.469.
Leprince, S., Ayoub, F., and Avouac, J.-P. (2007). Automatic and precise orthorectification, coregistration, and subpixel correlation of satellite images, application to ground deforma-tion measurements, IEEE T. Geosci. Remote, 45, 1529–1558, https://doi.org/10.1109/TGRS.2006.888937.
Mayer, C., Fowler, A. C., Lambrecht, A., and Scharrer, K (2011.: A surge of North Gasherbrum Glacier, Karakoram, China, J. Glaciol., 57, 904–916, https://doi.org/10.3189/002214311798043834.
Paul, F., Strozzi, T., Schellenberger, T., and Kääb, A. (2017). The 2015 Surge of Hispar Glacier in the Karakoram, Remote Sens., 9, 888, https://doi.org/10.3390/rs9090888.
Quincey, D. J. and Luckman, A. (2014). Brief Communication: On the magnitude and frequency of Khurdopin glacier surge events, The Cryosphere, 8, 571–574, https://doi.org/10.5194/tc-8-571-2014.
Round, V., Leinss, S., Huss, M., Haemmig, C., and Hajnsek, I. (2017). Surge dynamics and lake outbursts of Kyagar Glacier, Karako-ram, The Cryosphere, 11, 723–739, https://doi.org/10.5194/tc- 11-723-2017.
Royal Geograohical Society with IBG Photographic Archives (Undated). Various 19th and 20th centuries expeditions to the Karakoram Mountains Archives