Guidelines for the management of glacial hazards and risks

Guidelines for the management of glacial hazards and risks R7816.142 1 INTRODUCTION Hazards related to glaciers and glacial lakes, known as glacial hazards, are an issue in many mountain ranges of the world, including the Himalaya (e.g. Reynolds, 1998a; Richardson and Reynolds, 2000a), the Andes (e.g. Lliboutry et al., 1977; Reynolds, 1992; Ames, 1998), the North American Rockies (e.g. Clague and Evans, 2000), the European Alps (e.g. Haeberli et al., 2001) and the Caucasus Mountains (Kääb et al., 2003). The purpose of this document is to set out a structured approach to the management of such hazards and their associated risks. Technical procedures for the assessment of hazard and vulnerability are outlined, followed by sections on subsequent management options and recommendations for management strategies. The guidance notes are aimed at increasing the objectivity and independence of the risk assessment process. Intended users of the notes include both technical personnel and non-specialists involved in the assessment or management of glacial hazards. This document is an output from a project funded by the UK Department for International Development (DFID) for the benefit of developing countries. It draws upon results from a three years research project to develop techniques of glacial hazard assessment and socio-economic vulnerability analysis aimed at assisting governments in the planning and implementation of risk reduction strategies. The views expressed are not necessarily those of the DFID. 2 DEFINITIONS Key terms related to risk management are defined here mainly based upon guidelines published by the International Strategy for Disaster Reduction (ISDR, 2002): Hazard: Vulnerability: Risk: A potentially damaging physical event, phenomenon or human activity, which may cause loss of life or injury, property damage, social and economic disruption or environmental degradation. A set of conditions and processes resulting from physical, social, economic and environmental factors, which increase the sensitivity of a community to the impact of hazards. The probability of harmful consequences, or expected loss (of lives, people injured, property, livelihoods, economic activity disrupted or environmental damage) resulting from interactions between natural or human induced hazards, vulnerable conditions and the ability to respond or cope with the consequences. Conventionally risk is expressed by the equation: Risk = Hazards x Vulnerability/Coping Capacity Risk management: The systematic management of administrative decisions, organisation, operational skills and responsibility to apply policies, strategies and practices for disaster risk reduction. or The process whereby decisions are made to accept a known or assessed risk and/or the implementation of actions to reduce the consequences or probabilities of occurrence. (Royal Society, 1992) August 2014 1

3 GLACIAL HAZARDS 3.1 Types of glacial hazard Figure 1 shows in schematic form some of the main hazards associated with glaciers in high mountain areas and their impacts downstream. These include: (a) Glacial lakes dammed by unstable ice-cored moraine complexes (D), which are prone to catastrophic drainage (Figure 2). (b) Low angle debris-covered glacier tongues on the threshold for formation of glacier-wide lakes (the largest and therefore potentially most hazardous type). Such lakes have been observed to form within a few decades. Perhaps the quickest reported development of a hazardous lake was that of Laguna 513 in the Cordillera Blanca of Peru, 12 km to the north-east of Carhuaz, which formed between 1988 and 1992 (Reynolds et al., 1998). (c) Ice avalanches impacting directly, or transforming into debris flows. Direct impacts from ice avalanches cause most concern in areas where potential targets lie within a few kilometres of glaciers. In the European Alps, events of this kind frequently lead to the loss of life and localised damage. Examples include the ice avalanche at Mattmark, Switzerland, in 1965 resulting in 88 deaths. (d) Failure of saturated glacial sediments, becoming debris flows. Examples include a debris flow from the glacier foreland of Nevado Salcantay, south-east Peru, in 1998, which blocked the main valley such that the backed-up water inundated the Machupicchu power station. (e) Catastrophic rock avalanches (sturzstroms), the most destructive type of landslide, triggered by earthquakes or melting permafrost in glacier headwalls. Notable examples include sturzstroms from Fig. 1: Schematic representation of the high mountain environment, indicating potentially hazardous glacier-related processes and their impacts downstream. Letters and numbers are referred to in the text. Huascarán Norte, Peru, in 1962 and 1970, the latter triggered by an earthquake, which resulted in 5,000 and 23,000 deaths respectively (Plafker and Ericksen, 1978). In 2002, a rock/ice avalanche from the headwall of August 2014 2

Kolka Glacier in the Caucasus Mountains, Russia, triggered a major slide of the glacier tongue and mudflow, killing 120 people (Figures 3 and 4; Kääb et al., 2003). Slope stability problems associated with degrading permafrost are also prevalent in the European Alps (Haeberli et al., 1997). (Note: there are many types of landslide, some of which are glacier-related, in addition to the examples given here). Fig. 2: Moraine breach and GLOF sediments deposited during an outburst from Sabai Tsho, Khumbu Himal, Nepal, in 1998. Note the houses (bottom right) for scale. Fig. 3: Avalanche source area, Kolka Glacier, Russia, for an event on 2002. Picture shows source area of headwall collapse (1), area of glacier slide (2), and lateral limit of avalanche (3). Igor Galushkin, Ministry of Natural Resources, North Ossetia. August 2014 3

Fig. 4: Debris flow deposits (centre) at Karmadon village, resulting from the headwall collapse and glacier slide of Kolka Glacier, Russia (Fig. 3). Igor Galushkin, Ministry of Natural Resources, North Ossetia. Other types of glacial hazard are related to a combination of floods, debris flows and avalanches, from various physical settings on, in or under the glacier and in association with glaciers in various stages of advance/retreat or surging (Table 1). Of increasing importance in recent years are the less dynamic hazards such as the disappearance of water resources through melting mountain glaciers. 3.2 Glacial hazards and climate change High mountains in temperate to tropical latitudes are now recognised as one of the most sensitive environments to climatic changes, and particularly the trend of warming that was observed through the 20th Century to the present day (Brecher and Thompson, 1993; IPCC, 2001; OECD, 2003). This warming trend is well illustrated by results from ice cores from the Quelccaya and Huascarán ice caps in Peru and the Sajama ice cap in Bolivia. These show an increase in NO3 - and 18 O values (positively correlated to warming) over the last 200 years. Indeed, the 18 O values for the twentieth century recorded in the Huascarán cores are the most enriched (i.e. warmest) of the last 6000 years (Thompson, 2000). In parallel, several studies have shown reductions in volume and extent, and negative mass balances of glaciers from Peru and neighbouring countries. For example, the Yanamarey glacier in the Cordillera Blanca reduced in volume at an increasing rate between 1948 and 1988 (Hastenrath and Ames, 1995). Similarly, Glacier Santa Rosa in the Cordillera Raura is experiencing 2 ma -1 (water equivalent) of surface lowering (Ames and Hastenrath, 1996). More than half the water discharge from this glacier is not being recharged by precipitation, but is supplied through glacier thinning, a trend reflected in the negative mass balances. Ames and Hastenrath (1996) predict that the glacier will only survive a further 40 years under present conditions. Similar conclusions are reached by Ramirez et al. (2001) for glaciers in the Bolivian eastern cordilleras. Other areas have experienced a comparable pattern of glacier wastage. The European Alps have lost about half their ice volume since the mid 19 th Century and have also experienced a less-well recorded change in permafrost area (Haeberli and Beniston, 1998). Field observations of debris-covered glaciers in the eastern Himalaya indicate they are responding to climatic changes by thinning substantially. A particular concern with August 2014 4

these glaciers is that they are out of equilibrium with climate and on the threshold of lake formation (Benn et al., 2001). Table 1: Types of glacier-related hazards. Hazard types Description Examples References Mass movements Ice avalanche Rock avalanche Debris flow (Aluvión) Sudden mass movement of ice down slope. Starting areas can be classified as ramp- or cliff-type. High velocity (~90-350 kph) transport of fractured rock mass, often starting as a rock fall or slide on step slopes. Can be triggered by freeze-thaw cycles (permafrost degradation), seismicity, or stress release following glacier retreat. Also known as sturzstrom ( Hsü, 1978). Typically a slurry of mixed grade soil and water. Commonly triggered by ice/rock avalanches and/or glacier/glacial lake floods. Debris flows, irrespective of origin, are known in S. America as aluviones (sing. = aluvión). Mattmark, Switzerland, 1965. 88 dam construction workers killed. Nevado Huascarán, Peru, 1962 & 1970. 18-23,000 killed in the 1970 event. Kolka Glacier, Russia, 2002. Avalanche from glacier headwall, turning into debris flow; 120 deaths. Nevado Salcantay, Peru, 1998. Failure of saturated proglacial sediments. Hydropower plant destroyed (costs $160 million). Margreth and Funk, 1999 Browning, 1973 Kääb et al., 2002 Glacial floods Glacier outburst Glacial Lake Outburst Flood (GLOF) Catastrophic discharge of water from the subglacial and/or englacial system. Often triggered by geothermal heating (jökulhlaup), opening of englacial channels at the start of the ablation season (spring events) or intense rainfall. Catastrophic flood from the breaching of a moraine dam. Term used in Himalayan regions (synonymous with the French term dêbacle). Grímsvötn jökulhlaup, Iceland, 1996. Peak discharges of 34,000 m 3 s -1 reported. Damage to infrastructure. Mt. Rainier, USA, frequent. Damage to infrastructure/ property. Dig Tsho, Nepal, 1985. HEP plant destroyed. Luggye Tsho, Bhutan, 1994. 26 deaths, 200+ km runout. Guðmundsson et al., 1997; Russell et al., 1997 Walder & Dreidger, 1995 Vuichard and Zimmermann, 1987 Richardson & Reynolds, 2000a Ice-dammed lake outburst Catastrophic flood from the failure of an ice dam. Often occurs periodically from the same lake/glacier. Keyajir Lake, Xinjiang, China, every 1-4 years. Zhang, 1992 Glacier length & volume changes Glacier fluctuations Glacier surge Glacier advance or retreat, causing Inundation of land (advance) or water shortages through glacier wastage. Short-lived sudden increase in velocity, of an order of magnitude, often expressed as an advance. Glacier Chacaltaya, Bolivia, contemporary glacier wastage. Chiring Glacier, Karakoram, Pakistan. Ghiacciaio del Belvedere, Italian Alps, 2000-2002. Ramirez et al., 2001. Hewitt, 1998 Haeberli et al., 2002 A warming climate has a number of potential consequences for glacial hazards. The formation and disappearance of ice- and moraine-dammed lakes, some of which may be dangerous, are often associated with changes in glacial extent (Haeberli and Beniston, 1998). Debris flows are also a well-recorded consequence of glacier retreat (Ballantyne and Benn, 1994). Existing moraine-dams may experience accelerated degradation through melting of ice cores (Richardson and Reynolds, 2000b) and their lakes may increase in volume due to greater melt water input (Clague and Evans, 2000). Steep hanging glaciers that are partially or entirely frozen to their beds may become prone to ice avalanching as their snouts become warmbased (Haeberli et al., 1997). Continued glacier mass-balance losses will also result in water shortages after an initial period of increased discharge (Braun et al., 2000). The potential loss of glaciers as water resources is a greater source of anxiety to the people and industries that rely on glacial melt during dry seasons. Many tropical glaciers exist close to the melting point (Thompson, 1980) and even a modest increase in temperature may have drastic affects. A further widespread problem that may develop with global warming is a shift in hazard zones (Haeberli and Beniston, 1998). In other words, glacial and related hazards may begin to affect areas with no history of such events, where regulatory and social aspects of preparedness will not exist. August 2014 5

3.3 Economic consequences and the threat to development An accelerated use of cold mountain areas, over the last century, has brought human activities into conflict with glacial hazards. Such activities include settlement, hydropower production, forestry, mining and wilderness tourism (Haeberli, 1992; Clague and Evans, 1994, 2000; Reynolds, 1998a,b; Tweed and Russell, 1999; Richardson and Reynolds, 2000a, Taylor and Richardson, 2003). The Himalayan country of Nepal provides poignant examples of the economic impacts of GLOFs. The development of Nepal depends a great deal on the development of water resources and is described as the highest priorities in the socio-economic development of the country (Yamada, 1993, 2). For example, it is estimated that 43,000 MW of hydropower could be developed commercially, 1.24 % of which was developed by 2003 (Anon., 2003). However, as hydropower development has moved into the High Himalaya, glacial hazards have become an important issue. One of the more devastating and better described examples is the flood from Dig Tsho (tsho lake) in the Khumbu Himal of Nepal, in August 1985. Among the many consequences of the flood described by Vuichard and Zimmermann (1986, 1987) was the destruction of a small hydropower plant with the delay of promised electricity for the region (Figure 5). The projected costs associated with the destruction of a mature hydropower plant in Nepal have since been estimated at $500 million, and could jeopardise national economic development (Richardson and Reynolds, 2000a). Fig. 5: GLOF sediments in the Bhote Kosi valley, Nepal. Note lateral channel erosion and remains of concrete channel on the right bank of the river (left of picture, arrowed). Looking upstream. In Peru, the fear of a loss of hydropower was realised in 1998 with the destruction of the Machupicchu hydropower plant. Direct costs associated with this event (loss of production revenue over three years and rebuilding costs) amounted to $160 million. 3.4 The need for glacial risk management guidelines In the light of past economic loses from glacier-related events, donors, project financiers and/or insurers increasingly require a risk assessment for new development projects. However, little or no guidance is provided as to what actually comprises a glacial risk assessment. From the authors experience, the scopes defined within the tender documents for some of these projects have been shown to be inappropriate. Hopefully the guidelines presented here can begin to address this issue. August 2014 6

Effective risk management is needed to address the issues of changing patterns in vulnerability and the nature/distribution of hazards, and the severe economic and social consequences of glacier-related disasters. In particular, increasingly objective schemes are necessary to enable the effective prioritisation of sites for remediation. Only then can scarce resources in developing countries be allocated effectively. Professionals new to this increasingly important subject area require technical guidance, whilst less technically orientated information is needed for the non-specialists involved with management and communication. These requirements were illustrated recently through an unfortunate situation in the Cordillera Blanca, Peru. In April 2003 a major overseas science organisation issued a press release stating that a particular lake was dangerous. Their interpretation of the evidence was flawed but unfortunately the press release reached a global audience. NewScientist.com, for example, declared Glacier crack places Peruvian city in peril! stating that 60,000 people were in danger. The subsequent effects on the tourism industry were disastrous. Tourist numbers were reportedly down by 75%, and by July 2003 the economic losses associated with this incident had been estimated by Huaraz Municipality as exceeding $20 million and climbing. Clearly guidelines are needed not only for technical aspects, but also to highlight the responsibilities of risk assessment specialists. 4 A FRAMEWORK FOR GLACIAL RISK MANAGEMENT Procedures for the management of glacial risks tend to be lacking generally, yet there are other industry sectors in which formal risk management approaches have a proven track record. Dai et al. (2002) demonstrated how social and economic losses due to natural hazards can be reduced by effective planning and management. Whilst their work focuses on landslides, their proposed management framework could easily have been written for glacial hazards. They highlight the need to address: (a) the probability of the hazard occurring, (b1) its runout, (b2) the vulnerability of property and people, (c) the risk to those people and the property, and (d) management strategies and decision-making. Harris and Herbert (1994) with reference to contaminated land, suggest broadly similar elements of (a) hazard identification and assessment, (b) risk estimation, (c) risk evaluation and (d) risk control (letters a-d show the broad equivalents between these two schemes). A risk management approach suitable for application to glacial hazards is presented here based on the frameworks mentioned above. Four main elements may be considered (Figure 6): (i) Hazard identification and assessment. This considers the magnitude and the potential, or likelihood, of an adverse event. The aim is to detect and define the specific hazards at a site and to quantify the degree of hazard if possible. Often this is achieved with reference to empirical and generic reference data from past hazard events. (ii) Vulnerability assessment and risk estimation. This considers the potential consequences of a particular event by examining the exposure of the flood route and vulnerable targets. This stage may require the use of models to predict the variations in vulnerability under differing levels of hazard. By combining observed or predicted levels of hazard and vulnerability it is then possible to estimate levels of risk under different scenarios. (iii) Risk evaluation. This stage involves making judgements about the acceptability of the observed or predicted risks and is likely to involve several parties (e.g. scientific advisors; local, regional and/or national authorities; the site owner and/or manager; local communities). Non-scientific parties will need to base decisions on the information communicated to them by the scientific investigation team. Special attention is thus needed to ensure that scientific information is communicated accurately, in a transparent manner using terminology appropriate for non-specialists. (iv) Risk control. This involves specifying a course of action to contain, reduce, or remove risks. Approaches can include the reduction of the hazard (e.g. partially draining a glacial lake), the reduction/removal of vulnerability (e.g. building protective dams, relocating vulnerable people), or a combination of the two. Risk control measures can be designed according to the level of acceptable risk, as determined during the risk evaluation phase. The activities undertaken during hazard assessment, vulnerability assessment and risk evaluation combine to form the risk assessment phase (Figure 6). Risk reduction involves decisions made during the risk evaluation and the actual process of risk control, either by reducing hazard, vulnerability or a combination of the two. Interaction between the elements is essential to allow for better targeting of effort, a better definition of the problem and possible solutions, and improved technical and financial control over the work stages. August 2014 7

Fig. 6: Prototype risk management framework highlighting the relationships between the elements of risk management and stages in a typical work programme (after Harris & Herbert, 1994). With reference to landslides, Dai et al. (2002) say that risk assessment and management demands a pluralistic approach, with science acting in a socio-political framework, to allow value-focused decision-making. They call for a better understanding of the physical processes and greater sophistication, transparency and rigour in the application of science, but within a collaborative and consensual decision-making framework. This demand for reproducibility and factual correctness is echoed by the Swiss Government's instructions for natural hazard assessment (Heinimann et al., 1988). Employing a structured risk management approach allows for the systematic and objective assessment of risk, the specific assessment of uncertainty, and the provision of a rational, consistent, transparent, and defensible basis for discussion about proposed response strategies (Harris and Herbert, 1994). August 2014 8