Inventory of Glacial Lakes in the Koshi, Gandaki and Karnali River basins of Nepal and Tibet, China

Inventory of Glacial Lakes in the Koshi, Gandaki and Karnali River basins of Nepal and Tibet, China Identification of potentially dangerous glacial lakes and prioritization for GLOF risk reduction Submitted to UNITED NATIONS DEVELOPMENT PROGRAMME, (UNDP) NEPAL Submitted by INTERNATIONAL CENTRE FOR INTEGRATED MOUNTAIN DEVELOPMENT (ICIMOD) ii P a g e

March 2018 Content 1. Introduction... 1 2. Study area... 3 3. Approach and methodology... 5 3. 1 Data sources... 5 3. 2 Mapping method... 8 3. 3 Uncertainties and limitations... 9 3. 4 Glacial lake attributes... 11 3. 5 Identification of potentially dangerous glacial lakes... 13 3. 6 Ranking and prioritization of potentially dangerous glacial lakes... 17 4. Status of glacial lakes in 2015... 19 4. 1 Number and area of glacial lakes... 20 4. 2 Types of glacial lakes... 21 4. 3 Glacial lake size class... 26 4. 4 Altitudinal distribution of glacial lake... 29 4. 5 Distance to the source glacier... 31 5. Potentially dangerous glacial lakes and priority lakes for risk reduction... 33 5. 1 Characteristic parameters for lake stability... 33 5.1.1 Lake characteristics... 33 5.1.2 Dam Characteristics... 35 5.1.3 Mother (Source) glacier characteristics... 38 5.1.4 Physical condition of surroundings... 38 iii P a g e

5. 2 Identification and ranking of potentially dangerous (critical) glacial lakes... 41 5. 3 Prioritization of PDGL for GLOF risk reduction... 49 5.3.1 Socioeconomic value... 49 5.3.2 Priority of potentially dangerous glacial lakes for risk reduction... 56 6. Conclusions... 59 7. References... 60 8. Annexes... 66 Table A1: Number and area of glacial lakes by types (2015) in the Koshi, Gandaki and Karnali basins of the Nepal, TAR, China and India... 67 Table A2:Identification of potentially dangerous glacial lakes (PDGL) based on the characteristics of lake, dam and surrounding features including the source glacier.... 68 Table A3: Characteristics of lake, dam, source glacier and surroundings for the identification of potentially dangerous glacial lakes in Koshi, Gandaki and Karnali basins of Nepal and TAR, China. (yellow highlighted lake is removed from the list of PDGL)... 70 Figure A4: Identification of potentially dangerous glacial lakes (PDGL) based on the characteristics of lake, dam and surrounding features.... 73 Table A5: Information of parameters of all PDGL for dam breach model.... 74 Table A6: Distribution of types of lakes at different elevation zone in pdf.... 74 Table A7: Distribution of glacial lake sizes at different elevation zone in pdf.... 74 Table A8: Lake area classes vs type of lakes in pdf.... 74 Table A9: List of images used in the present study in pdf.... 74 Soft copy - Glacial Lake Inventory data including shp.file of 2000... 74 Soft copy - Glacial Lake Inventory data including shp.file of 2015... 74 iv P a g e

Key findings The glacial lakes ( 0.003 sq km) were mapped for 2015 based on Landsat images using remote sensing tools and techniques for Koshi, Gandaki and Karnali basins of Nepal and TAR, China. The study found 3,624 glacial lakes in three basins, of which 2,070 lakes in Nepal, 1,509 lakes in TAR, China and 45 lakes were in India. The lakes larger than 0.02 sq km were 1,410 in number, which are considered large enough to cause risk in the downstream if the lake breach. This potential of would be heightened Glacial Lake Outburst Flood (GLOF), if the lakes are associated with a large retreating glacier and steep sloping landforms at the surroundings. A total of the 47 glacial lakes were identified as potentially dangerous (critical) glacial lakes (PDGL) based on the criteria:1) characteristics of lakes and dam; 2) the activity of the source glacier and 3) morphology of the surroundings. Other factors such as the extreme climatic condition, seismic activity and malpractice or human interference on the lake and natural dam are not considered. Of the 47 lakes identified, 42 lakes are in the Koshi basin, 3 lakes in the Gandaki basin and 2 lakes are in the Karnali basin. With respect to the political boundary, 25 PDGL are in the territory of the TAR, China; 21 PDGL in Nepal; and one PDGL is in the Indian Territory. The number of PDGLs identified in 2011 and present study for Nepal equal in number same, although on 13 of them are common, while 8 PDGLs are different. The physical parameters were considered first to categories the PDGLs into three ranks depending on the hazard level. The socioeconomic parameters were then summated to categorise the PDGLs into three priorities (priority I, priority II and priority III) for a GLOF risk reduction. Of the total, 31 lakes are of priority I, 12 lakes are of priority II, and 4 lakes are of priority III. The priority I lakes are the critical ones, which are at the equilibrium of lake water and dam s strength. Slight change in lake water and dam s strength may breach out, which warrants immediate action for the potential GLOF mitigation measures. The priority II and III lakes have a potential of increase in the hazard level and hence need close and regular monitoring. The water level of four PDGL of priority I had already been lowered in the past to reduce the GLOF risk: two each from Nepal and the TAR, China. The water level of Tsho Rolpa and Imja Tsho Lakes of Nepal was lowered by more than 3m and 4m respectively. Similarly, the water level of GL088066E27933N and GL088075E27946N Lakes in China was also lowered to reduce the GLOF risk. Of the 21 PDGLs identified in Nepal, 6 lakes in priority I, 8 lakes in priority II and 9 lakes are in priority III. Similarly, 14 lakes are in priority I, two lakes are in priority II and 9 lakes are in priority III in TAR, China and one lake of priority II is in the Kali River of India. v P a g e

vi P a g e

vii P a g e

Acronyms and Abbreviations ALOS ASTER B (c) B (o) Co DEM Dl Dm DMS Ds E E (c) E (o) ETM+ GCF GIS GL GLIMS GLOBE GLOF GloVis GTOPO HKH I ICIMOD IPCC I(s) Advance Land Observing Satellite Advanced Spaceborne Thermal Emission and Reflection Radiometer Cirque Lake other bedrock dammed compressed and old dam Digital Elevation Model dam length distance to source glacier Degree minute second dam slope glacier erosional lake Cirque Lake other erosional lake Enhanced Thematic Mapper Plus Green Climate Fund Geographic Information Systems Glacial lakes Global Land Ice Measurements from Space Global Land One km- Base Elevation Project Glacial Lake Outburst Floods USGS Global Visualization Viewer Global Topography Hindu Kush Himalaya Ice dammed lake International Centre for Integrated Mountain Development Intergovernmental Panel for Climate Change supra-glacial lake viii P a g e

I(v) JAXA Km LIGG M M(e) M(l) M(o) NASA Nc NDWI NEA NIR NSIDC O OLI PDGL RS Sm Spot Sq. km SRTM TAR TM USGS UNDP WECS lakes dammed by tributary valley glaciers Japan Aerospace Exploration Agency kilometer Lanzhou Institute of Glaciology and Geocryology Moraine dammed lake end moraine dammed lateral moraine dammed other moraine dammed National Aeronautics and Space Administration no crest Normalized Difference Water Index Nepal Electricity Authority Near Infrared National Snow and Ice Data Center others glacial lakes Operational Land Imager potentially dangerous glacial lakes Remote sensing slope of source glacier Satellite Pour l'observation de la Terre square kilometer Shuttle Radar Topography Mission Tibet Autonomous Region Thematic Mapper United States Geological Survey United Nations Development Programme Water and Energy Commission Secretariat ix P a g e

ii P a g e

1. Introduction The cryosphere (glaciers, snow, river and lake ice and permafrost) is an integral part of the global climate system with important linkages with ecosystem and socio-economic benefits. The cryosphere has been changing rapidly in recent decades, and the changes vary depending on the spatial and temporal scale that they are examined on. Most Himalayan glaciers have been rapidly melting and shrinking since the 1980s (Bajracharya et al., 2014), concurrent with climate warming (Bhambri and Bolch, 2009; Bolch et al., 2012; Yao et al., 2012). Glacial loss and shrinkage not only affects water resources and hydrological processes, but also influences the formation and expansion of glacial lakes (Yao et al. 2010). The International Centre for Integrated Mountain Development (ICIMOD) has been involved in glacial lake inventories and the identification of potentially dangerous glacial lake since 1986 (Ives, 1986). ICIMOD, in collaboration with partners in different countries, embarked on the preparation of an inventory of glaciers and glacial lakes, and identification of potential sites for glacial lake outburst floods (GLOFs) in the Hindu Kush-Himalaya (HKH) region (ICIMOD 2010). A glacial lake inventory for Nepal and Bhutan was started in 1999 and for selected basins in China, India and Pakistan were started in 2002 (Mool et al., 2001; ICIMOD 2010). In 2011, a comprehensive study undertaken by the ICIMOD outlined the status of the glaciers of the HKH region (Bajracharya et al., 2011). To understand the changes in glacier area and extent, ICIMOD mapped glaciers from 1980s, 1990, 2000 and 2010 of Nepal, Bhutan, and some selected basins of HKH region based on satellite images. Analysis of the time series data revealed that the glaciers had lost almost a quarter of their initial area over the 30-year period (Bajracharya et al., 2014 b, c, 2016). Moreover, total ice reserves had decreased by 29 percent in Nepal between 1977 and 2010, whilst the number of glacial lakes had increased by 11 percent (Bajracharya et. al., 2007, 2008). The rapid melting and recession of many Himalayan glaciers due to of climate change is leading to the formation of new glacial lakes, whilst the enlargement of existing lakes is increasing the risk that the surrounding moraine dams will become destabilized(cruz et al., 2007; IPCC 2007; Rosenzweig et al., 2007). The moraine dams are mostly composed of loose debris and are susceptible to GLOFs (Randhawa et al., 2005). Himalayan GLOFs develop at high altitudes and can extend for long distances, damaging downstream infrastructure (Chen et al., 2007; Osti and Egashira, 2009; Liu et al., 2014). Examples of previous GLOFs in the Himalayas demonstrates that they constitute a 1 P a g e

serious threat to socio-economic and development endeavors (Rai, 2005; Wang et al., 2014; Worni et al., 2013). GLOFs from moraine-dammed glacial lakes have been assessed and modeled in several previous studies (Wang et al., 2012; Osti et al., 2013; Westoby et al., 2014). Several investigations of glacial lake changes have been conducted in the Himalayas, and large areal expansions with regional differences have been reported (Mool et al., 2001; Bolch et al., 2008; Gardelle et al., 2011; Li and Sheng, 2012). GLOFs are a crucial problem faced by regional countries and people in the Himalayas (Ageta et al., 2000). Some of them are associated with trans-boundary impacts (Xu Daoming et al.1989; Yamada and Sharma 1993; Reynolds 1998; Ives et al. 2010). Over 50 GLOF events that occurred in the region have been reported in the HKH region; however, records are available only for some areas of China, Nepal, Pakistan and Bhutan (Che et al, 2014; LIGG/WECS/NEA, 1987). Many more events may remain undocumented or unrecorded (Ives et al. 2010). An increase of GLOF events in the Himalayas has been reported over the period of 1940 to 2000, although the trend has been considered statistically insignificant (Richardson and Reynolds 2000; Bajracharya 2009). Until 2011, Nepal had experienced 24 GLOFs events (recorded only) where significant damage and loss of life was reported. The Dig Tsho GLOF of 1985 and the Tampokhari outburst in 1998 both led to considerable loss of life, property and infrastructure and severely affected the livelihoods of people living in downstream areas (Dwivedi, Acharya, & Simard, 2000; Vuichard & Zimmermann, 1987). Although GLOFs are not a recent phenomenon, they have started drawing considerable attention among scientists after the 1980s as the risk from potential GLOFs has increased (Xu Daoming 1988; Vuichard and Zimmermann 1986, 1987; Chen et al. 2013). About 1.6 million people living downstream within the territory of Nepal may be at risk from these natural hazards (Ghimire, 2004). However, GLOF risk can be reduced through the implementation of appropriate mitigation and adaptation measures. To achieve this, an updated and standardized glacial lake inventory should be conducted periodically in order to analyze the spatial distribution and temporal development of glacial lakes, produce a GLOF hazard assessment, and make plans for mitigation of identified potential GLOF risks. In addition, effective approaches should be available to the public to help in understanding GLOFs and their risk to the public. As a part of Green Climate Fund Readiness Programme for Nepal led by Ministry of Finance and to support the Green Climate Fund (GCF) Project proposal formulation by UNDP Nepal, 2 P a g e

a comprehensive mapping and assessment of glacial lakes of Nepal and Tibetan plateau (which are draining to Nepal) were conducted for 2000 and 2015 using remote sensing (RS) and a geographic information system (GIS).Different criteria are adapted to identify potentially dangerous (critical) glacial lakes, and to assess and estimate a qualitative or relative probability of GLOF and its impacts on downstream communities in Nepal. 2. Study area The present study area lies within the Koshi, Gandaki and Karnali basins, all of which are major tributaries of the Ganges River. The catchments of these river basins are transboundary to the Tibet Autonomous Region (TAR) China (upper section), Nepal (upper and middle section) and India (mostly lower section). Some of the tributaries are sourced in Tibet and China before flowing through Nepal to finally merge with the Ganges River in India (Figure 2.1). Figure 2.1: Study area showing the Koshi, Gandaki and Karnali basins in Nepal and Tibet, China. 3 P a g e

The glaciers and glacial lakes are distributed only in the upper and middle sections of the river basins. Hence, the study area is confined to Nepal and the TAR, China. The major tributaries of the Koshi River are the Tamor River, the Arun River (Pumqu River in China), the Dudh Koshi River, the Tama Koshi (Rongxer River in China), the Likhu River, the Sun Koshi (Poiqu River in China), and the Indrawati River. The Gandaki River is fed by the Trishuli, Budhi Gandaki, Seti, Marsyangdi, and Kali Gandaki rivers. Some upper sections of the Budhi Gandaki and Trishuli rivers lie in the TAR. The Karnali River of western Nepal is also known as Ghaghara River in India. The main tributaries of the Karnali River are the Mahakali (Kali), Bheri, Humla Karnali, Mugu Karnali, Kawari, West Seti, and Tila rivers. Some tributaries of the upper Humla River are sourced in China and the catchment of the upper Kali River spans Nepal, India and China. The catchment area of the study area is considered in the TAR, China and Nepal only. Approximately 57.3%, 88.1% and 95.7% of the total catchment areas of the Koshi, Gandaki and Karnali basins lay in Nepal respectively. The remainder is located in the TAR, China (Table 2.1). Table 2.1: Catchment area of the Karnali, Gandaki and Koshi basins in Nepal and Tibet, China. Basin Area (sq. km) China Nepal Total Area (%) Area (sq. km) Area (%) Area (sq. km) Karnali 3016 4.3 67865 95.7 70881 Gandaki 4353 11.9 32116 88.1 36469 Koshi 29450 42.7 39456 57.3 68906 4 P a g e

3. Approach and methodology The glacial lake inventory consists of water bodies that are situated proximal to present glaciers as well as those located in lowland areas that were covered by glaciers in the past. The lakes are generally formed by glacial melt water. Glacial lakes may also exist beneath (subglacial) or within (englacial) glaciers, but are usually not visible in aerial/optical images, and their detection is challenging as it requires field-based methods to acquire the necessary information. Thus, subglacial and englacial lakes are not included in this inventory because they cannot be mapped from aerial/optical satellite images. To date, the latest glacial lake inventory of Nepal and adjacent region was conducted between 2003 and 2007 using Landsat ETM+ satellite images. In the context of climate change and global warming in the region, glaciers are shrinking and retreating rapidly resulting in large changes in the status of glacial lakes. Whilst glacial lakes are a source of fresh water, they are also a potential source of disaster if the dam containing the lake is breached. Knowledge of glacial lakes and related disaster risks is important to reduce the GLOF risk to the lower riparian community. To understand the latest status of the glacial lakes, high resolution satellite images are used to produce an inventory of glacial lakes of 2000 and 2015. A 5 m digital elevation model (DEM) is used to examine the geometric properties of the lakes. 3. 1 Data sources Landsat OLI Landsat satellite images have been widely used to map the extent of glaciers and glacial lakes globally due to their high spatial resolution and accessibility throughout the region (Bolch et al. 2010). Landsat data are used in this study due to their consistent spatial coverage in the region, high spatial resolution and free accessibility through the GLOVIS web portal (http://glovis.usgs.gov). We have used the Landsat Operational Land Imager (OLI) to prepare the current glacial lake inventory of the region. The OLI, built by the Ball Aerospace & Technologies Corporation, measures the visible, near infrared, and short wave infrared portions of the spectrum. Its images have 15-meter (49 ft.) panchromatic and 30-meter multi-spectral 5 P a g e

spatial resolutions along a 185 km (115 mile) wide swath. The images cover wide areas of the Earth whilst retaining sufficient resolution to distinguish features such as glaciers, glacial lakes, urban centers, farms, forests and other objects. OLI is more reliable and provides improved performance than ETM+. In addition, the use of satellite images of higher spatial resolution has been emphasized in this study. The OLI images acquired covering the region during 2014 to 2016 years were used for mapping glacial lakes. Images different years and derived from different sensors were also used to verify the existence of the mapped lakes. The images between September and December were used primarily because the likelihood of snow or cloud cover is lower during this period than other months of the year. Digital elevation model (DEM) Topographic information of the lakes and adjacent areas was used to identify and categorize the lakes and establish a ranking of dangerous glacial lakes. Digital elevation models (DEMs) can be used to extract the topographic parameters of the lakes, associated glaciers, moraines and the adjacent areas. A DEM can be defined as a regular gridded matrix representation of the continuous variation of relief over space and is a digital model of land surface form. The primary requirement of any DEM is that it should have the desired accuracy and resolution and be devoid of data errors. Their steady and widespread application can be further attributed to their easy integration within a GIS environment. Before the year 2000, the base elevation models depicting a global coverage were available in a 1 km resolution, e.g. GTOPO-30 (Global Topography in 30 arc-sec) and GLOBE (The Global Land 1 km- Base Elevation Project). However, in the last decade, more advanced global DEMs with higher spatial resolutions, such as the Shuttle Radar Topography Mission (SRTM) (version 4, C-Band DEM of 3 arc-second, 90 m resolution) and the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) (version 2, 30 m resolution), have become available. Apart from these freely available DEM datasets, stereo images from a number of satellites (e.g. Cartosat 1, Landsat 7 ETM+, QuickBird, IKONOS, SPOT, ASTER sensors, among others) have also been used to create DEMs using various software applications for examining landscapes. 6 P a g e

High resolution images are capable of obtaining more accurate surface information. For this study, we used a 5 m resolution ALOS DEM for Nepal and a 12 m resolution PALSAR DEM for the TAR, China. 7 P a g e

3. 2 Mapping method A number of remote sensing methods had been developed for generating glacial lake inventories (Kääb, 2000; Mool et al. 2001a, b; Huggel et al. 2002, 2006; Ives et al. 2010). We adopted the method used by Maharjan et al., 2018 for the inventory of glacial lakes. The method is summarized in Figure 3.1. The Normalized Difference Water Index (NDWI; Eq.(1)) method provides an automatic way to detect water bodies, including glacial lakes, on the basis of Landsat Thematic Mapper (TM) or Enhanced Thematic Mapper Plus (ETM+) images, was adopted in this study. Figure 3.1: The remote sensing based glacial lake inventory process. 8 P a g e

NDWI = NIR (or Band 4) Blue (or Band 1) NIR (or Band 4) + Blue (or Band 1).. (1) The ratio images of NDWI is created by arithmetic calculation of Band 4 near infrared (NIR) and Band 1 (Blue) of the Landsat images and the NDWI threshold value is applied to classify the glacial lakes in the images. A NDWI threshold value of -0.6 to -0.9, as adopted by Huggel et al. (2002), was used to prepare the inventory of HKH glacial lakes. Although this automatic classification method can speed up the detection of glacial lakes, this method is not applicable to a wide region due to some uncertainties created by atmospheric and physical processes. For example, if lakes are frozen or covered with snow, or cloud cover or shadow obstructs the image, they cannot be detected using this automatic classification method. In such cases, manual delineation method was used to map the lakes. The automated delineation of glacial lakes was validated and modified if necessary by overlaying the Landsat images over the previous inventory datasets whenever they were available (Mool et al. 2001a, b, 2003). Thus, any misclassified lakes were corrected, and missing lakes were added manually. Further, the mapped lakes were overlain with high resolution images if available in the Google Earth environment for validation. Generally, pixels in the raster images do not give the homogenous reflectance and represent only one object unless it is perfectly aligned in single object. Thus, at least four pixels are required to map the exact boundary of the object (lake) from the images. Therefore, the smallest glacial lake that can be mapped from the images should be covered by 4 pixels, which is 0.0036 sq. km area in the case of Landsat images. Hence, the glacial lake area of 0.003 sq. km is the minimum threshold for lake size that has been applied for mapping in the present glacial lake inventory. 3. 3 Uncertainties and limitations The uncertainties or accuracy of mapping of glacial lakes or glaciers from the satellite images used depends, typically, on the spatial resolution, seasonal/temporal snow cover, shadow, and contrast between the glacial lakes pixels and surroundings pixels (DeBeer and Sharp 2007; Bajracharya et al. 2014c). Landsat images with the least snow cover and cloud cover were 9 P a g e

selected for mapping to increase the quality of the automatic mapping approach and reduce manual correction of the boundary. The lake data was overlaid on the high resolution images in Google Earth and also cross-checked with the previous inventory data wherever it is available in order to validate and improve the accuracy of mapped lakes from the automatic approach. Also the glacial lake data were thoroughly checked by overlaying on the same Landsat images used for automatic mapping along with cross-checking in the high resolution images in Google Earth and any mismatches of the boundary of the lakes due to the seasonal/temporal snow cover and shadows were manually corrected using additional Landsat images. Although this cross-checking improved the quality of the data, the mapped lakes boundary were affected by various other types of obscurities, which are mostly dependent on image resolution. The uncertainty of the glacial lake boundary could not be greater than half of the image resolution (i.e., ±15 m in TM, ETM+ and OLI) (Bajracharya et al. 2014c). Hence the uncertainty of the glacial lake boundary was estimated by variation of area bounded by the lake polygon, which is calculated by number of image pixels bounded by each lake polygon and the total number of image pixels bounded by the 15 m buffer of each lake polygon. The equation used for calculating total uncertainty is given as: (a â)2 RMSE = n i=1 n Where, ai is the area of glacial lake from the total pixel bounded by glacial lake polygon and âi is the area of glacial lake from the total pixel bounded by the 15 m buffer of glacial lake boundary. The total uncertainty of glacial lake area is ±2% and this uncertainty were also observed in the glacial lake and glacier mapping of the HKH region (Maharjan et al., 2018; Bajracharya et al., 2014c). Depending on the mapping scale of the glacial lakes, the 30m resolution of Landsat images satisfies enough to map the lake boundary. The accuracy will be higher in the high spatial resolution images. Field verification is utmost necessary for the confirmation of the information before mitigation measures. 10 P a g e

3. 4 Glacial lake attributes Once the final glacial lake polygons were generated, the attributes of the glacial lake were generated in ArcGIS. Each lake polygon is given a unique ID containing the longitude and latitude of the centroid of the l polygon in the same way as the GLIMS ID is developed for glaciers by National Snow and Ice Data Center (NSIDC), University of Colorado, USA. GLIMS stands for Global Land Ice Measurement from Space. The GLIMS ID consists of 14 letters, e.g. GxxxxxxEyyyyyN for glaciers, and 15 letters (GLxxxxxxEyyyyyN) for glacial lakes, where G stands for Global, N for North, E for East and 5x stands for 3-digit degree decimal latitude and 6x stands for 3-digit degree decimal longitude. This is the first time that lake IDs are being used in a lake inventory study (Maharjan et al., 2018. The initial letter G in the GLIMS ID is replaced by GL representing Glacial Lake. Other parameters such as glacial lake area and elevation etc. were calculated automatically in ArcGIS using the DEM. The lakes were morphologically classified by manually overlaying high resolution lake images with terrain data in Google Earth. The lakes were classified as either moraine dammed, ice dammed or rock dammed (Table 3.1). Table 3.1: Classification of Glacial Lakes (modified after ICIMOD 2011) Moraine dammed (M) Ice dammed Glacial lake type Code Definition End-moraine dammed lake Lateral moraine dammed lake Other moraine dammed lake Ice-dammed lake M(e) M(l) M(o) I Lakes dammed by end (terminal) moraines. Water usually touches the walls of the side moraines. Water is usually held back by the end moraine (dam) but is not necessarily in contact with the glacier. Glacial ice may be present at the bottom of the lake (defined in some other classifications as an advanced form of supraglacial lake). Lakes dammed by lateral moraine(s) (in the tributary valley, trunk valley, or between the lateral moraine and the valley wall, or at the junction of two moraines). Lake is held back by the outside wall of a lateral moraine, i.e. away from the former glacial path. Lakes dammed by other moraines (includes kettle lakes and thermo-karst lakes). Lakes dammed by glacier ice, including lakes on the surface of a glacier or lake dammed by glaciers in the tributary/trunk valley, or between the glacier margin and valley wall, or at the junction of two glaciers. Supra-glacial lake I(s) Bodies of water (pond or lake) on the surface of a glacier. Dammed by tributary valley glacier I(v) Lakes dammed by glacier ice with no lateral moraines; can be at the side of a glacier between the glacier margin and valley wall 11 P a g e

Bed rock dammed Blocked Bedrock dammed lake Cirque lake Other glacier erosion lake Other glacial lakes B Bodies of water that form as a result of earlier glacial erosion. The lakes accumulate in depressions after the glacier has retreated or melted away. B(c) A small pond occupying a cirque. B(o) O Bodies of water occupying depressions formed by glacial erosion. These are usually located on the mid-slope of hills, but not necessarily in a cirque. Lakes formed in a glaciated valley and fed by glacial melt. Damming material is not directly part of the glacial process e.g. debris flow, alluvial, or landslide blocked lakes. The Albers Equal Area Conic projection is used to calculate the area of glacial lake, the unit of the area adopted were in square kilometer (sq. km). A detailed list of glacial lake attributes is given in Table 3.2. 12 P a g e

Table 3.2: Fields and formats of glacial lake attributes S.N. Field Name Type Format Description 1 GLIMS_ID string GLxxxxxxEyyyyyN 2 Basin name string Text 3 Sub-basin name string Text Combination of longitude (X) and latitude (Y) of the centroid of the lake polygon. GL glacial lake, E East, N North. Drainage basin name based on maps and literature. Drainage sub-basin name based on maps and literature. 4 Longitude string DMS Longitude of center of glacial lake. 5 Latitude string DMS Latitude of center of glacial lake. 6 Altitude integer 7 Area float km 2 meter above mean sea level (m.a.s.l.) 8 Gl_Type string Text Type of glacial lake. Water level of glacial lake. Extracted from SRTM. Area of glacial lake. Calculated based on the Albers Equal Area Conic projection. 3. 5 Identification of potentially dangerous glacial lakes The step-by-step approach to identify critical or potentially dangerous glacial lakes (PDGL) developed by ICIMOD (Mool et al 2001, ICIMOD, 2011) has been previously applied to the Koshi basin (Shrestha et al. 2017) and is used in this study with some modifications. The stability of a lake depends on its characteristics and damming material (Figure 3.2). The dam should have sufficient strength to hold the lake water if it is to be considered stable, otherwise the lake may be breached in an outburst flood. Detailed lake and dam features were analyzed using remote sensing to assess the stability of each lake. However, stable lakes may still outburst due to the activity of the source glacier and failure of the surroundings, which may impact the lake and/or dam. Hence, the physical condition of the source glacier and the surroundings of the lake and dam is also considered in this study. Other triggering agents including earthquakes, extreme climatic events, unsafe anthropogenic intervention etc. may also breach the dam, but their potential impact could not be evaluated in this study. The criteria to identify PDGLs are discussed by Mool et al. (2001), Bajracharya (2007), and ICIMOD (2011) and are also considered in this study along with some additional criteria such as catchment of the lake and the elevation difference and length of the dam. 13 P a g e

Figure 3.2 Identification of potentially dangerous glacial lakes and prioritization for GLOF risk reduction. The following updated approach has been considered to identify PDGLs using remotely sensed data: Criteria 1. Lake characteristics The current mapping of glacial lakes includes all lakes with areas greater than 0.003 sq. km. If the lakes are larger than 0.02 sq. km, and on a steep slope, an outburst may occur, causing serious damage, especially in the highly populated areas with 14 P a g e

extensive infrastructure located downstream. The expansion rates of lakes larger than 0.02 sq. km was analyzed to estimate the possible volumetric increase in water. The following criteria were used to analyze the stability of the lakes. I. Lake size and rate of expansion II. III. IV. Increase in water level or volume of water Presence of cascading lakes Intermittent activity of supraglacial lakes Rapid changes in lake area and the lake s potential to grow in the near future due to the size of the catchment, presence of cascading lakes and chances of merging of supraglacial lakes are assessed. An expansion in lake area indicates an increase in water volume, which may increase the risk of outburst. 2. Dam characteristics The condition of the dam is an important parameter to consider in lake stability assessments. Dams constructed out of thin and loose moraine have a large potential to rupture. Narrow crested moraines will cause a lake to have a relatively high outburst potential compared to those with a wide crest. Thus, lakes with thinner moraine crests may be associated with a cause for GLOFs. Due to erosion and landslides, these moraine crests are usually angular and narrow at the top. This geometry is vulnerable to any surge wave generated by ice or snow avalanches, ultimately triggering a GLOF. The steepness of moraine wall slopes also determines the likelihood of dam outburst. The main dam characteristics that can be derived from remote sensing are as follows: I. Type of damming material II. III. IV. Crest width Slope of the dam wall Elevation difference of the moraine (height of the dam) V. Length of the dam VI. VII. VIII. Erosional activity or presence of landslides on the dam Presence or absence of drainage outflow Breached and closed in the past and the lake refilled again with water 15 P a g e

IX. Seepage through the damming walls X. Existence and stability of ice core and/or permafrost within dam 3. Source glacier characteristics The potential of a lake to burst and cause damage downstream can be heightened if the lakes are associated with a glacier. If the lake is in contact with a glacier retreating on a gentle slope, there is possibility for lake expansion. On the other hand, if the glaciers with crevasses on a steep slope, ice masses may detach from the glacier and fall into the lake. Falling ice masses can be disturbed the stability of the dam and/or the lake. Even stable lakes in such an environment can be considered as potentially dangerous and become a candidate that needs to be monitored. The steepness of the glacier tongue can also make the associated lake a potentially dangerous. The main characteristics that can be relevant to the condition of the glacier are as follows: I. Condition of associated glacier (source glacier) II. III. IV. Distance between the glacial lake and the source glacier(s) Steepness of glacier tongue Debris cover on the lower glacier tongue V. Presence of crevasses and ponds on the glacier surface VI. VII. Calving of ice from the glacier front Icebergs breaking off the glacier terminus and floating into the lake 4. Physical conditions of the surroundings The physical factors that can destabilize a lake and trigger its outburst can be identified using remote sensing of the lake surroundings. The factors that are considered in this study are: I. Hanging glaciers that are in contact with or very close to the lake II. Potential rock-fall/slide (mass movements) sites around the lake 16 P a g e

III. IV. Large snow avalanche sites immediately above the lake Sudden advance of a glacier towards a lower tributary or the main glacier which has a well-developed frontal lake. 5. Other factors Some other unseen factors that can destabilize a lake and/or dam to trigger lake outbursts are out of the scope of the remote sensing technique used in this study. Earthquakes and extreme climate events at the vicinity of the lake and dam are other major triggering factors which may result in a lake outburst. In addition, unscientific and inappropriate human intervention at the lake and dam may also lead to uncontrolled breaching. The first two factors are unpredictable. However, there is a need to increase stakeholder awareness about the danger associated with the lakes and for any further activity in the lake itself. The following triggering agents are not considered in this study: I. Earthquake generated waves in the lake, resulting in the deterioration and collapse of the dam. II. Extreme climate conditions resulting from excessive and continuous precipitation in the lake, dam and catchment area. III. Anthropogenic interference in the lake and dam which may destabilize the containing walls. Lake stability depends directly on the physical condition of the lake and the dam. Other factors which may significantly impact the stability of the lake include the activity of source glacier and the stability of the surroundings. This study considers data on lake characteristics, dam properties, source glaciers characteristics and the physical conditions of the area surrounding the lake to identify potentially dangerous glacial lakes in the study area. 3. 6 Ranking and prioritization of potentially dangerous glacial lakes Potentially dangerous glacial lakes were identified and ranked based on the physical characteristics of the lakes and morphology of the dam, source glacier and its surroundings following the criteria outlined in the section 3.4. Not all of the potentially dangerous glacial lakes do not constitute an equal risk to the community and infrastructure in the river basins. Based on their 17 P a g e

socioeconomic importance, the PDGL are prioritized in order to reduce the GLOF risk to the community. The socioeconomic parameters included in this study were number of household (Hh), population, road lengths (path, track and trial, motor-able and highway), the number and type of bridges (wooden, suspension, motor-able, and highway bridges), hydropower projects (number and capacity of hydropower projects in megawatts), in the path of 500m buffer zone and the catchment of a potential outburst. The information for preliminary assessment was derived from Google Earth images and the Nepal census data of 2011. 18 P a g e

4. Status of glacial lakes in 2015 The glacial lakes equal to and larger than 0.003 sq. km were mapped from Landsat satellite images between 2014 and 2016 for the Koshi, Gandaki and Karnali basins of Nepal and TAR (Figure 4.1). A unique identity (GLIMS ID) is assigned to each individual glacial lake present in the study area. The lake area, elevation and topographic features of individual glacial lakes were calculated from the 5 m ALOS DEM for Nepal and 12 m PALSAR DEM for the TAR using ArcGIS. Lake size was divided into 7 different classes according to their size. The type of lake was identified and classified into bedrock dammed, moraine dammed, ice dammed and other lake classes. The altitudinal distribution of lakes was analyzed using DEMs and lake boundaries in ArcGIS. The distance to the source glacier was measured manually only for moraine dammed lakes equal to or larger than 0.02 sq. km. Figure 4.1: Distribution of glacial lakes in the Koshi, Gandaki and Karnali basins of Nepal and the TAR, China. 19 P a g e

4. 1 Number and area of glacial lakes A total of 3,624 glacial lakes were mapped in the Koshi, Gandaki and Karnali basins of Nepal and the TAR. The Koshi basin had the largest number (2064) of glacial lakes, followed by the Karnali basin (1128) and the Gandaki basin (432). The distribution and area of glacial lakes in the Koshi, Gandaki and Karnali basins is shown in Figure 4.1 and given in Table 4.1. Table 4.1: Number and area of glacial lakes in the sub-basins of the three different basins (2015). Basin Koshi Gandaki Karnali Sub-basin Number Area Count % sq. km % Avg. (sq. km) Tamor 283 7.81 9.12 4.68 0.03 Arun 909 25.08 68.58 35.15 0.08 Dudh Koshi 355 9.80 16.92 8.67 0.05 Likhu 17 0.47 0.41 0.21 0.02 Tama Koshi 307 8.47 14.68 7.52 0.05 Sun Koshi 181 4.99 22.13 11.34 0.12 Indrawati 12 0.33 0.16 0.08 0.01 Sub-total 2064 56.95 132.01 67.65 0.06 Trishuli 242 6.68 8.19 4.20 0.03 Budhi Gandaki 49 1.35 1.58 0.81 0.03 Marsyangdi 59 1.63 6.22 3.19 0.11 Seti 4 0.11 0.15 0.08 0.04 Kali Gandaki 78 2.15 3.05 1.56 0.04 Sub-total 432 11.92 19.19 9.83 0.04 Bheri 164 4.53 9.20 4.72 0.06 Tila 82 2.26 4.11 2.11 0.05 Mugu 239 6.59 6.22 3.19 0.03 Kawari 28 0.77 1.04 0.53 0.04 West Seti 51 1.41 1.43 0.73 0.03 Humla 498 13.74 20.21 10.36 0.04 Kali 63 1.74 1.53 0.78 0.02 Karnali 3 0.08 0.19 0.10 0.06 Sub-total 1128 31.13 43.93 22.51 0.04 Total 3624 100.00 195.12 100.00 0.05 The Arun sub-basin of the Koshi basin had the highest number and total area of glacial lakes among the sub-basins and also a higher sum than the Gandaki and Karnali basins. The Arun sub basin contributes more than a quarter of the total number of lakes in the inventory and over a third of lakes by total lake area. The Humla Karnali sub-basin of the Karnali basin contributes the second highest number of glacial lakes (13.7%) and is the third highest in total lake area (10.36%). The Sun Koshi sub-basin of the Koshi basin consists of only 181 20 P a g e

132.02 19.18 43.94 432 2064 1128 lakes but is the second highest by total glacial lake area (11.34%) and is the highest in terms of average lake area. The Sun Koshi sub-basin therefore has a higher proportion of large lakes than other basins in the study area. The Tamor, Dudh Koshi and Tama Koshi sub-basins each contribute to approximately 7 %to 10 % of the total number of glacial lakes, and 4 % to 8% of the total lake area. Other sub-basins in the study area account for less than 5 % of the total number of lakes and total lake area in the study region. Very few glacial lakes exist in the Likhu, Indrawati, Seti and Kawari sub-basins. Glacial lakes occupy a combined area of 195.12 sq. km in the study area (Table 4.1). Of this, ~132.02 sq. km (67.65%) is in the Koshi basin, 19.18 sq. km (9.83%) is in the Gandaki basin, and 43.94 sq. km (22.51%) is sourced from the Karnali basin (Figure 4.2). The average mean area of the glacial lakes within each sub basin ranges from 0.01 in the Indrawati sub-basin to 0.12 sq. km in the Sun Koshi sub-basin. Both of these are sub-basins of the Koshi basin. The average mean area of glacial lakes is 0.06 sq. km in the Koshi basin and 0.04 sq. km in the Karnali and Gandaki basins. The overall average mean area of the glacial lakes in the study area is 0.05 sq. km per lake. 10000 1000 Number Area (km2) 100 10 1 Koshi Gandaki Karnali Figure 4.2: Number and total area of glacial lakes in three different basins. 4. 2 Types of glacial lakes Glacial lake are often formed as the result of glacier retreat, a process which may leave behind large debris deposits. Lakes may be formed either within part of the eroded landform (e.g. bed rock dammed) or build behind a dam formed by moraines, ice and/or landslide debris. 21 P a g e

The lakes in this inventory were classified based on the damming material such as bedrock dammed (B), moraine dammed (M), ice dammed (I), and others (O) including a landslide dam, moraine or ice dam adjoining the glacier. Moraine dammed lakes are split into more detailed classes including end moraine (e), lateral moraine (l) and other moraine (o). Similarly, ice dammed lakes are classified into supraglacial (s) and valley (v) types. Bedrock dams are split into cirque (c) and other erosional landforms (o) (Table 3.1). The majority of the glacial lakes in all basins of the study area are moraine dammed (Table 4.2). Moraine dammed lakes comprise about 55% of the total lakes (2002) followed by bedrock dammed lakes (35%) (1256) and ice dammed lakes including others (10%) (339). Among the moraine dammed lakes, the end-moraine dammed lakes comprise about 15.8% (573) and nearly 2.3% (82) were lateral moraine dammed type. The other moraine dammed lakes contribute to more than 37.2 % (1347) of the total. Among the ice-dammed lakes, supraglacial lakes comprise 9.3 % (337) and other ice-dammed lakes contribute to only 0.1% (2) of the total. Cirque glacial lakes constitute about 7.8% (281) of the total, whereas other erosional glacial lakes comprise ~26.9% (975). In general, moraine-dammed lakes consist of loose, coarse material with little cementing content. This composition is easy to erode and thus lake moraines comprised of this material are vulnerable to GLOFs. The distribution of lake types within each basins is shown in Figure 4.3. Table 4.2: Number, and area of glacial lakes in three basins by lake type (2015). Major Basin Koshi Gandaki Karnali Total Type Moraine-dammed lake (M) Ice-dammed lake (I) Bedrock dammed /erosion lake (B) M(e ) No. Area (sq. km) No. Area (sq. km) No. Area (sq. km) No. Area (sq. km) 359 74.98 75 9.93 139 11.72 573 96.63 M(l) 37 7.48 20 1.44 25 1.92 82 10.84 M(o ) 668 12.61 143 3.47 536 12.54 1347 28.62 Sub-total 1064 95.07 238 14.84 700 26.18 2002 136.09 I(s) 234 3.36 46 0.36 57 0.46 337 4.18 I(v) 0 0 0 0 2 0.04 2 0.04 Sub-total 234 3.36 46 0.36 59 0.5 339 4.22 B(c ) B(o ) 132 7.27 31 1.33 118 5.42 281 14.02 625 24.06 116 2.55 234 5.7 975 32.31 22 P a g e

Number (%) Sub-total 757 31.33 147 3.88 352 11.12 1256 46.33 Others O 9 2.26 1 0.1 17 6.14 27 8.5 Total 2064 132.02 432 19.18 1128 43.94 3624 195.14 The distribution of different glacial lake types for each sub-basin is displayed in Table 4.3. The total number and area of moraine dammed lakes are comparatively higher than bedrock dammed, ice-dammed and other types of lakes across all of the basins. The Arun sub-basin has the highest number of moraine dammed lakes, followed by the Humla Karnali sub-basin of the Karnali basin. The Dudh Koshi and Tama Koshi sub-basins of the Koshi basin have more than 200 moraine dammed lakes and more than 100 moraine dammed lakes are located in the Tamor, Sun Koshi, Trishuli, and Mugu Karnali sub-basins. The sub-basins of the Koshi, Gandaki and Karnali basins had fewer moraine dammed lakes. The moraine dammed lakes in the Koshi basin are comparatively more vulnerable to GLOFs than those in the other basins. The ice dammed lakes present in this inventory are mostly supraglacial lakes, which are dynamic in nature, appearing and disappearing periodically depending on glacier melt rates. The growth and merging of supraglacial lakes are characteristic of this class of lakes, which ultimately convert into moraine dammed lakes. Imja Tsho and Tsho Rolpa are typical examples of moraine dammed lakes that have developed from supraglacial lakes (Bajracharya et al., 2007). This type of lake are more vulnerable to GLOFs. If the supraglacial lakes do not merge, there is high possibility of lake disappearance over time. The Dudh Koshi sub-basin contains the highest number of ice dammed (supraglacial) lakes. 40 30 20 10 0 Koshi Gandaki Karnali Moraine-dammed lake (M) Bedrock-dammed lake (E) Ice-dammed lake (I) Others (O) Figure 4.3: Number and percentage of the different types of glacial lakes in the three basins. 23 P a g e

Bedrock dammed lakes are the most stable type of lakes. These lakes have a lower probability of GLOF occurrence; however, flash floods may be generated by falling ice, snow or debris within the lake, causing the water to overflow the bedrock sill. There is little possibility of breaching of the bedrock dam by foreign masses landing on the dam or in the lake. 24 P a g e

Table 4.3: Number and area of glacial lakes by types (2015) in the sub-basins and basins of the Koshi, Gandaki and Karnali River of Nepal and the TAR Number Area (km 2 ) Icedammedammed Other dammed dammed Other Bedrock- Ice- Bedrock- Sub Basin Morainedammed lake (M) Total lake (M) Total Moraine-dammed lake (I) lake (E) lake (I) lake (E) M(e) M(l) M(o) I(s) I(v) B(c) B(o) O M(e) M(l) M(o) I(s) I(v) B(c) B(o) O Tamor 37 9 78 36 0 34 87 2 283 3.00 0.17 1.50 0.38 0.00 2.03 1.60 0.46 9.12 Arun 167 4 218 39 0 64 411 6 909 36.80 0.23 4.64 0.98 0.00 3.78 20.36 1.79 68.58 Dudh Koshi 64 17 134 103 0 4 32 1 355 9.77 3.04 2.22 1.40 0.00 0.17 0.31 0.01 16.92 Likhu 2 0 8 0 0 3 4 0 17 0.15 0.00 0.11 0.00 0.00 0.10 0.05 0.00 0.41 Tama Koshi 55 2 161 38 0 15 36 0 307 9.31 1.19 2.56 0.30 0.00 0.77 0.56 0.00 14.68 Sun Koshi 34 5 64 18 0 11 49 0 181 15.96 2.85 1.50 0.30 0.00 0.41 1.12 0.00 22.13 Indrawati 0 0 5 0 0 1 6 0 12 0.00 0.00 0.08 0.00 0.00 0.02 0.06 0.00 0.16 Sub-Total 359 37 668 234 0 132 625 9 2064 74.98 7.48 12.61 3.36 0.00 7.27 24.06 2.26 132.01 Trishuli 34 13 62 35 0 18 80 0 242 3.13 0.49 1.80 0.29 0.00 0.97 1.52 0.00 8.19 Budhi Gandaki 11 2 18 8 0 6 4 0 49 0.76 0.12 0.28 0.06 0.00 0.10 0.27 0.00 1.58 Marsyangdi 14 3 23 3 0 4 12 0 59 4.89 0.37 0.61 0.02 0.00 0.15 0.19 0.00 6.22 Seti 0 0 0 0 0 0 3 1 4 0.00 0.00 0.00 0.00 0.00 0.00 0.05 0.10 0.15 Kali Gandaki 16 2 40 0 0 3 17 0 78 1.15 0.47 0.77 0.00 0.00 0.13 0.53 0.00 3.05 Sub-Total 75 20 143 46 0 31 116 1 432 9.93 1.44 3.47 0.36 0.00 1.33 2.55 0.10 19.19 Bheri 21 1 56 3 0 16 65 2 164 1.44 0.01 1.07 0.02 0.00 0.73 0.98 4.96 9.20 Tila 2 0 20 2 0 24 32 2 82 0.52 0.00 1.32 0.01 0.00 1.50 0.54 0.23 4.11 Mugu 29 0 127 7 0 31 40 5 239 1.91 0.00 2.38 0.10 0.00 1.03 0.65 0.16 6.22 Kawari 5 0 10 0 0 6 7 0 28 0.50 0.00 0.21 0.00 0.00 0.27 0.06 0.00 1.04 West Seti 9 3 22 9 0 4 3 1 51 0.84 0.14 0.24 0.08 0.00 0.10 0.01 0.00 1.43 Humla 61 16 285 22 2 31 74 7 498 6.10 1.63 6.77 0.16 0.04 1.53 3.18 0.79 20.21 Kali 12 5 16 14 0 3 13 0 63 0.42 0.14 0.55 0.08 0.00 0.07 0.27 0.00 1.53 Karnali 0 0 0 0 0 3 0 0 3 0.00 0.00 0.00 0.00 0.00 0.19 0.00 0.00 0.19 Sub-Total 139 25 536 57 2 118 234 17 1128 11.72 1.92 12.54 0.46 0.04 5.42 5.70 6.14 43.93 Total 573 82 1347 337 2 281 975 27 3624 96.63 10.84 28.61 4.18 0.04 14.02 32.31 8.50 195.12 Basin Koshi Gandaki Karnali 25 P a g e

4. 3 Glacial lake size class The glacial lakes were classified into seven different size classes: Class 1 (<= 0.02 sq. km), Class 2 (>0.02 0.05 sq. km), Class 3 (>0.05 0.1 sq. km), Class 4 (>0.1 0.5 sq. km), Class 5 (>0.5 1 sq. km), Class 6 (>1 5 sq. km) and Class 7 (> 5 sq. km). Class 1 lakes are most frequent (2214) contribute 61 % of the total inventory (Table 4.4). However, the total area of these lakes is only 20.25 sq km and their contribution to the total lake area is only 10.38%. The average area of Class 1 glacial lakes is 0.01 sq km per lake. The Class 1 glacial lakes are not analyzed further for the identification of potentially dangerous glacial lakes. Class 2 lakes are the second highest in terms of total lake numbers and contribute about 21 % to the total lake area. The number of glacial lakes decrease with respect to the increasing class number. However, although class 4 lakes are ranked 4 th in terms of their absolute number, they are associated with the greatest total area of glacial lakes. Only one class 7 lake is present in the inventory (0.03%) with an area of 5.41 sq km. Although this is the largest glacial lake, it contributes to only 2.77% of the total lake area. Table 4.4 shows the distribution of glacial lake numbers and areas in different size classes within the basins. The Koshi basin exhibits the highest number of lakes in all classes. The largest glacial lake is also present in the Koshi basin. There is an inverse relationship between the number and area of lakes according to the size classes. Similar observations were made in the Pumqu (Arun) River basin (Che et al. 2014) and the Poiqu (Bhote Koshi) basin (Wang et al. 2014) with large numbers of smaller lakes of the same size as the class 1 and 2 lakes here. Table 4.4: Number and area of glacial lakes by size class (2015) in the Koshi, Gandaki and Karnali basins of Nepal and TAR, China. Koshi Gandaki Karnali Total Class Area (sq. km) Area (sq. km) Area (sq. km) No Area (sq. km) No No No Total Avg Total Avg Total Avg Total Avg 1 1239 11.04 0.01 261 2.46 0.01 714 6.75 0.01 2214 20.25 0.01 2 433 13.35 0.03 95 2.94 0.03 231 7.19 0.03 759 23.48 0.03 26 P a g e

3 181 12.54 0.07 38 2.65 0.07 88 6.40 0.07 307 21.59 0.07 4 163 35.24 0.22 36 6.84 0.19 90 16.27 0.18 289 58.34 0.20 5 29 21.24 0.73 1 0.92 0.92 4 2.40 0.60 34 24.56 0.72 6 18 33.19 1.84 1 3.38 3.38 1 4.92 4.92 20 41.50 2.08 7 1 5.41 5.41 0 0.00 0 0 0.00 0 1 5.41 5.41 Total 2064 132.01 0.06 432 19.19 0.04 1128 43.93 0.04 3624 195.13 0.05 Table 4.5 displays the proportional distribution of lake size verses types of glacial lake. A considerable number of lakes consist of other moraine and end moraine dammed lake types. The end moraine dammed (M(e)) lakes are higher in all size classes of the lakes (Figure 4.4). Fifty-five moraine dammed lakes are larger than 0.5 sq km and out of this, 17 lakes are larger than 1 sq km. Most of the larger lakes are moraine dammed lakes, although some are blocked lakes (Table 4.6). The danger level of larger glacial lakes dammed by ice or moraine are high. Table 4.5: Number of glacial lakes by size class and lake type (2015). Class Size Types of lake M(e) M(l) M(o) I(s) I(v) B(c) B(o) O Total Class 1 <0.02 138 34 936 301 1 108 688 8 2214 Class 2 0.02-0.05 142 14 305 28 1 75 187 7 759 Class 3 0.05-0.1 106 11 68 4 0 62 54 2 307 Class 4 0.1-0.5 149 17 37 4 0 36 38 8 289 Class 5 0.5-1.0 23 5 0 0 0 0 6 0 34 Class 6 1.0-5.0 14 1 1 0 0 0 2 2 20 Class 7 >5.0 1 0 0 0 0 0 0 0 1 Total 573 82 1347 337 2 281 975 27 3624 27 P a g e

Figure 4.4: Number of glacial lakes by size class and lake type (2015) of Koshi, Gandaki and Karnali basins of Nepal and TAR, China. 28 P a g e

Elevation (masl) 4. 4 Altitudinal distribution of glacial lake The glacial lakes are found between 2400 and 6100 m.a.s.l. (Table 4.6). Around 6000 m.a.s.l. is the highest elevation of debris covered glaciers in the Nepal Himalaya (Bajracharya et al. 2014). Above this elevation, the landforms are steeper and mostly exist in a frozen state meaning that no (permanent/perennial) lakes can exist above this altitude. Figure 4.5 displays the distribution of glacial lakes in 100 m elevation zones. Only one glacial lake is located at 2400 m.a.s.l. in the Seti River subbasin of the Gandaki basin - the rest are above 3300 m.a.s.l. The majority of glacial lakes (58.3%) are located at elevations between 5000 and 6000 m.a.s.l. Similarly, 40.3% of the total lakes are located between 4000 and 5000 m.a.s.l. About 99% of the glacial lakes are M(e) M(l) M(o) I(s) I(v) B(c) B(o) O 6200 6000 5800 5600 5400 5200 5000 4800 4600 4400 4200 4000 3800 3600 3400 3200 3000 2800 present between 4000 and 6000 m.a.s.l. Similar altitudinal differences in glacial lake distribution have been reported in the Himalaya in other studies (Nie et al. 2013; 2600 2400 2200 2000 0 400 Number Wang et al. 2014). For example, in Poiqu basin, Wang et al. (2014) Figure 4.5: Altitudinal distribution of number and various types of lakes in the study area. showed that glacial lakes are distributed within the altitudinal range of 4420 to 5860, with majority of the lakes (76%) situated at elevations>5000 m.a.s.l. The altitudinal distribution of lakes shown in Figure 4.5 clearly demonstrates that a larger proportion of moraine dammed 29 P a g e

Elevation (masl) lakes occurs at higher elevations whilst erosional and others types of glacial lake exist at lower elevations. Figure 4.6 indicates that the percentage of small glacial lakes (<0.02 sq. km) is higher in all the altitudinal zones except in below 2500 m.a.s.l. However, the percentage of large glacial lakes (>0.1 sq. km) is comparatively higher in altitudinal zone between 3500 and 5500 m.a.s.l. Table 4.6: Distribution of different types of glacial lakes in 1000 m elevation zones Elevation Zone 3000-4000 4000-5000 5000-6000 6000-7000 Total Type No % No % No % No % No % M(e) 7 15.56 180 12.33 386 18.26 0 0 573 15.81 M M(l) 0 0 43 2.95 37 1.75 2 50 82 2.26 M(o) 6 13.33 387 26.51 953 45.08 1 25 1347 37.17 I(s) 7 15.56 178 12.19 152 7.19 0 0 337 9.3 I I(v) 0 0 0 0 2 0.09 0 0 2 0.06 I(o) 0 0 0 0 0 0 0 0 0 0 B B(c) 7 15.56 202 13.84 72 3.41 0 0 281 7.75 B(o) 12 26.67 452 30.96 510 24.12 1 25 975 26.9 Others O 6 13.33 18 1.23 2 0.09 0 0 27 0.75 Total Number 45 1460 2114 4 3624 % 1.24 40.29 58.33 0.11 100 Note: Only one glacial lake (0.03%) is below 3000masl 6000 4 0 5000 1323 417 181 158 23 11 4000 861 336 123 121 118 3000 26 6 3 9 01 2000 0 1 0 0% 20% 40% 60% 80% 100% Number (in percent) Class 1 (<0.02 km2) Class 2 (0.02 - <0.05 km2) Class 3 (0.05 - <0.1 km2) Class 4 (0.1 - <0.5 km2) Class 5 (0.5 - <1 km2) Class 6 (1 - <5 km2) Class 7( 5 km2) Figure 4.6: Number of glacial lakes in percent by size class in different elevation bands. 30 P a g e

4. 5 Distance to the source glacier The distance of glacial lakes to the source glaciers is another important factor in the stability of lakes. The distance from the lake to the source glacier is generated based on the 2005 lake data (Maharjan et al., 2018) for glacial lakes equal to and larger than 0.02 sq. km. Lakes that are more than 10 km away from the source glacier are not considered as glacial lakes as these are defined by their contact or close proximity to their source glacier. Changing glacier melt and retreat are closely associated with changes in the glacial lake environment. Lakes that are located within the glaciers are typically ice dammed lake and are about 13% in Koshi, 17% in Gandaki and 8% in Karnali basin (Table 4.7). The lakes in contact with the glacier are mostly moraine dammed lake and about 2% each of the lakes are in the Koshi and Karnali basins and 5 % in the Gandaki basin. The lakes at a distance less than 500m are 38%, 62% and 33% in Koshi, Gandaki and Karnali basins respectively. Rest the majority of lakes are more than 500m away from the source glaciers, which are not so much concerned with the stability of lakes and dam. Figure 4.7 show the distribution of number and area of lakes and distance to glaciers. The number and area of lakes had decreases with the increase in distance with the glaciers. Lakes closer to the glacier are higher in number and larger in lake area. Mostly the lakes far away from the glaciers are erosional or other types of lakes. Most of the moraine dammed lakes are within the 5 km distance from the glaciers. Highest number of moraine dam lakes are within 2 km distance from the glaciers. Table 4.7: Distribution of number and area of glacial lakes at different elevation zones Distance to source glaciers (m) No % Koshi Gandaki Karnali Area (sq km) % No % Area (sq km) % No % Area (sq km) Within 271 13 3.17 3 68 17 0.72 4 96 8 0.91 2 Contact with 50 2 11.31 9 20 5 1.80 10 26 2 2.05 5 >0 - <100 59 3 15.73 13 38 9 1.74 10 50 4 1.79 4 100 - <200 76 4 9.24 8 26 6 1.28 7 45 4 1.79 4 200 - <500 343 16 26.95 22 100 25 3.10 17 184 15 7.09 16 500 - <1,000 387 19 13.25 11 66 16 5.32 30 240 20 9.06 20 % 31 P a g e

Number Area (km 2 ) Number Area (km 2 ) 1,000 - <2,000 259 12 13.81 11 31 8 1.78 10 192 16 5.89 13 2,000 - <5,000 357 17 13.93 11 24 6 0.86 5 239 20 12.20 27 5,000 - <10,000 225 11 7.23 6 31 8 1.13 6 82 7 3.25 7 10,000 60 3 8.49 7 1 0 0.02 0 50 4 1.58 3 Total 2087 100 123.10 100 405 100 17.75 100 1204 100 45.60 100 300 250 Nunber Area 200 150 100 50 0-0.5 1.5 3.5 5.5 7.5 9.5 11.5 13.5 15.5 17.5 19.5 21.5 23.5 25.5 27.5 29.5 Distance from Glaciers (km) 20 18 16 14 12 10 8 6 4 2 0 a- Koshi 80 70 Number Area 60 50 40 30 20 10 0-0.5 1.5 3.5 5.5 7.5 9.5 11.5 13.5 15.5 17.5 19.5 21.5 23.5 25.5 27.5 29.5 Distance from Glaciers (km) 20 18 16 14 12 10 8 6 4 2 0 b- Gandaki 32 P a g e

Number Area (km 2 ) 140 120 Number Area 100 80 60 40 20 0-0.5 1.5 3.5 5.5 7.5 9.5 11.5 13.5 15.5 17.5 19.5 21.5 23.5 25.5 27.5 29.5 Distance from Glaciers (km) 20 18 16 14 12 10 8 6 4 2 0 c- Karnali Figure 4.7: Glacial lake number and area distribution to the distance from glacier (blue line: snout of glacier) in 200m zone in the study area: a- Koshi basin, b- Gandaki basin, c- Karnali basin. 5. Potentially dangerous glacial lakes and priority lakes for risk reduction 5. 1 Characteristic parameters for lake stability 5.1.1 Lake characteristics Changes in the glacial lakes mirror changes in their source glaciers. In the 33 year period from 1977 to 2010, the glaciers of Nepal have decreased by almost a quarter of their initial area (Bajracharya et al. 2014). Accordingly, the number of glacial lakes in the Koshi basin has increased from 1,160 in 1977 to 2,168 in 2010 and the total area of lakes has increased from 94.4 sq. km to 127.6 sq. km in 2010 (Shrestha et al., 2017). The number of lakes has increased by 86.9% and the total lake area has increased by 35.1%. We had mapped the glacial lakes for 2000 and 2015 and compared this inventory against existing data gathered in 2005 to produce a change assessment (Table 5.1). The number of glacial lakes in Koshi basin decreased from 2119 in 2000 to 2087 in 2005 and to 2064 in 2015. In contrast, the number of glacial lakes has increased in the Gandaki basin from 377 in 2000 to 405 in 2005 and 432 in 2015. Similarly, the glacial lakes in the Karnali basin have 33 P a g e

increased from 1105 in 2000 to 1204 in 2005 but decreased to 1128 in 2015. This decrease in number of glacial lakes is not an indication of the loss of glacial lakes as this pattern is produced by the merging of lakes with neighboring glacial lakes. As a result, the total area of the glacial lakes has increased from 179.56 sq km in 2000 to 186.44 sq km in 2005 and 195.39 sq km in 2015. The increase in number of glacial lakes is indicative of the rapid melting of glaciers and formation of new lakes particularly dammed by glacial ice and moraines. Table 5.1 shows the number and area of glacial lakes in 2000, 2005 and 2015. The threshold value of distance to the glacier is slightly different in 2005 than in the present study. Hence the number of glacial lakes in 2005 is slightly higher than 2015 in the Koshi and Karnali basins. However, the lake area had increased by 12% in the Koshi basin, 8% in the Gandaki basin and 1.27% in the Karnali basin between 2000 and 2015. The number of supraglacial lakes in the Koshi basin is higher and merging of supraglacial lakes reduces the number of lakes but overall lake area had increased. Table 5.1: Number and total area of glacial lakes in 2000, 2005 and 2015 Basin No 2000 2005 2015±1year Difference (2000 to 2015) Area (sq km) No Area (sq km) No Area (sq km) Koshi 2119 118.42 2087 123.10 2064 132.27 No % - 55-2.60 Area (sq km) % 13.9 11.70 Gandaki 377 17.76 405 17.75 432 19.19 55 14.59 1.4 8.07 Karnali 1105 43.38 1204 45.60 1128 43.93 23 2.08 0.6 1.27 Total 3601 179.56 3696 186.44 3624 195.39 23 0.64 15.8 8.82 This study has identified about 1410 glacial lakes that have an area equal to and larger than 0.02 sq. km. In comparison to the glacial lake inventory of 2000 and 2015, the number of lakes with areas equal to or larger than 0.02 sq km has increased in all basins (Table 5.2). Sixty-nine lakes in the Koshi basin, 31 lakes in the Gandaki basin and 38 lakes in the Karnali basin have increased since t the inventory of 2015. This increase in lake area is associated with a corresponding increase in GLOF risk. All lakes with an area equal to or larger than 0.02 sq. km are analyzed further to understand the stability of the lake and dam. 34 P a g e

Lakes upstream and downstream of these lakes are also examined as these may impact lake stability. Table 5.2: Number and area of glacial lakes equal to and larger than 0.02 sq km in 2000 and 2015. Basin No 2000 2015 Difference Area (sq km) No Area (sq km) No Area (sq km) Koshi 771 106.64 840 121.49 69 14.85 Gandaki 140 13.97 171 16.73 31 2.76 Karnali 381 32.67 419 37.47 38 4.8 An enlargement in lake area increases the potential energy of the reservoir, whilst a decrease in the dam area reduces the dam s strength leading to the breaching of the dam. Types and the sizes of the glacial lakes are given in the Table 4.5. About 2214 lakes smaller than 0.02 sq. km are excluded from the analysis of potentially dangerous glacial lakes. In addition, a further 480 lakes including valley lakes (1), blocked lakes (bedrock dammed lake) (281-108=73) (Table 4.5), and other blocked lakes (19) are also excluded from the potentially dangerous glacial lake analysis. A total of 2694 lakes are excluded and 896 lakes remain to identify the potentially dangerous glacial lakes for level 2. Analysis for PDGL (Level 1) = Total lakes Class 1 I(s+v) B(c+o) - O = 896 (1) The number of glacial lakes derived from the Level 1 (result of equation 1) is further analyzed incorporating the dam characteristics of respective glacial lakes. 5.1.2 Dam Characteristics The flow of glacier meltwater can be obstructed by rocky terrain, glacier ice, loose moraine and landslides to form a lake. The obstructing feature functions like a natural dam. Rapid glacial melt results in either an increase in surface runoff or lake expansion. The expanded lake will either increase the potential energy to weaken the dam or flow over the dam. The condition of the dam is important in determining the stability of the lake. Most of the lakes close to the glacier snout are dammed by loose moraine. When lake waters flow over thin and loose moraine dams, these structures may be easily eroded resulting in a GLOF. Lakes 35 P a g e

dammed by narrow crest moraines have a relatively high potential for outburst compared to lakes with wide crest dams. Thus, dams with thinner crests are associated with a higher GLOF risk. Due to erosion and mass movements, dam crests are usually angular and narrow at the top. Any surge wave generated by ice or snow avalanches may ultimately trigger a GLOF. The steepness of the slope of the moraine wall also determines the likelihood of the outburst from a lake. Ice dammed and moraine dammed lakes are particularly susceptible to instability. Bedrock dammed lakes are more stable than lakes with other types of natural dams as the stability of other dammed lakes depends on the characteristics of dammed features. The main characteristics of the lake dams were analyzed for glacial lakes equal to and larger than 0.02 sq. km. The characteristics of these dams includes: 1) no dam crest (nc) inflow and outflow of lake is equivalent; 2) compressed and old dam material (co) more stable than loose debris; 3) dam length greater than 200m (dl) reduces the erosional capacity of overflow; 4) the outer slope of dam is less than 20 degrees (ds) a lower gradient will have less erosional capacity. These features define the main characteristics of a dam stability. Based on these parameters, eight lakes identified as PDGL in 2011 (listed in Table 5.3) were removed from the PDGL list in this study. Processing through four levels of analysis, the remaining 47 lakes are identified as potentially dangerous glacial lakes (Table 5.4 and Figure 5.1). 36 P a g e

Table 5.3: Glacial lakes identified as potentially dangerous (PDGL) in 2011 that have been removed from the PDGL list in this study. S.N. Lake ID/Name valley Description 1. 2. kotam_gl_0135/ Nagma Pokhari kodud_gl_0193/ Tam Pokhari Tamor Dudh Koshi 3. gakal_gl_0004 Kaligandaki 4. 5. koaru_gl_0012/ Barun Pokhari gabud_gl_0009/ Birendratal Arun 6. koaru_gl_0016 Arun Budhigandaki 7. gakal_gl_0008 Kaligandaki 8. gakal_gl_0022 Kaligandaki Outburst in 1980. Confined lake water outflow through the moraine and debris almost for 1700 m then only the drop at steeper slope. No further GLOF is expected. Outburst in 1998, presence of confined channel wider than 45 m and flow through the moraine and debris for almost 1450 m then only flow on the river valley. No further GLOF is expected. Blocked by possibly old landslide debris, existing wide and confined lake out let at the side of the dam, no further GLOF is expected. GLOF evidence in the past and the river valley is full of alluvial fan with gentle slope for about 300m,, no damming, compact debris at downstream less chances of huge ice avalanches to create GLOF. No damming, erosional land feature, compact debris at downstream, in contact with retreating glacier, in case of glacier topple only the possibility of overflow of splash water. No GLOF is expected. No damming, erosional land feature, compact debris at downstream, fed by lake at the glacier snout, dam length of about 1500m then landslide and steep slope. No GLOF is expected. On medial moraine, old and compact moraine. confined surface flow, lake to glacier distance is about 1km, no chances of ice fall, landslide or by any means to break the moraine. Valley and blocked lake with outlet, no chances of ice fall, landslide or by any means to GLOF. The lakes with the dam characteristics of no crest (nc), compressed and old dam material (co), dam length greater than 200m (dl) and outer slope of dam less than 20 degrees (ds) is assumed to be a stable lake and dam and hence will be subtracted from the level1 for the further analysis in the identification of potentially dangerous glacial lakes. A total of 295 lakes remained for the analysis of level 3. Analysis for PDGL (Level 2) = Level 1 nc- co dl - ds = 295 (2) 37 P a g e

5.1.3 Mother (Source) glacier characteristics Whilst lake and dam characteristics determine whether a lake is considered as stable, lake stability may be disturbed by its source glacier. This may occur either by rapid melting of glacier ice, resulting rapid growth of glacial lake, or by the toppling/collapse of a large mass of glacier ice to disturb the lake flow or damage the dam, causing it to breach. For lakes in direct contact with their source glacier, steep slopes and large lake-glacier elevation differences may allow large ice masses to fall into the lake, creating a large wave which may result in the rupture of the dam. This process is considered potentially dangerous, and glacial lakes that exhibit these properties become candidates for potentially dangerous lakes that need to be monitored. The information on the condition of hanging glaciers, distance to lakes, steepness of glacier tongues, debris cover on lower glacier tongues, presence of crevasses and ponds on glacier surfaces, and the possibility of toppling/collapsing ice from glacier fronts or icebergs breaking off the glacier terminus and floating into the lake are also examined in this study to identify potentially dangerous glacial lakes. Lakes situated further than 200m from the glacier terminus (dm) and those with source glaciers with surface slopes less than 60 degrees (sm) are assumed to be stable and are removed from further detailed analysis. 5.1.4 Physical condition of surroundings The stability of the surroundings of the glacial lake and dam are additional important factors which may destabilize the lake and dam. Much of the study area is located at high altitude where snow/ice avalanches and landslides commonly occur. These phenomena can disturb the dam and the lake to trigger a GLOF. A total of 243 lakes were safe both from the source glacier and surroundings, the remaining 52 lakes were analyzed for level 4. Analysis for PDGL (Level 3) = Level 2 - dm - sm - (S) = 52 (4) 38 P a g e

The remaining 52 glacial lakes were thoroughly checked in the high resolution satellite images available in Google Earth and identified 47 glacial lakes were potentially dangerous and selected for the potential GLOF risk reduction. Number and total area of the lakes at different level are given in the Annex (Table A). Table 5.4: Identification of potentially dangerous glacial lakes (PDGL) based on the characteristics of lake, dam and surrounding features including the source glacier. Country Nepal China India Basin Lake inventory Lake size and type Dam characteristics Surrounding features PDGL Zero level First level Second level Third level Fourth level Nr Nr Nr Nr Nr Koshi 834 199 91 19 18 Gandaki 255 65 18 2 2 Karnali 981 241 39 1 1 Sub-total 2070 505 148 22 21 Koshi 1230 309 123 28 24 Gandaki 177 52 17 1 1 Karnali 102 22 3 0 0 Sub-total 1509 383 143 29 25 Koshi 0 0 0 0 0 Gandaki 0 0 0 0 0 Karnali 45 8 4 1 1 Sub-total 45 8 4 1 1 Total 3624 896 295 52 47 39 P a g e

1 1 4 8 22 21 45 29 25 148 143 505 383 2070 1509 Number 10000 Nepal China India 1000 100 10 1 Lake Inventory Lake size and type Dam characteristics Surrounding features PDGL Figure 5.1: Identification of potentially dangerous glacial lakes (PDGL) based on the characteristics of lake, dam and surrounding features including the source glacier. 40 P a g e

5. 2 Identification and ranking of potentially dangerous (critical) glacial lakes In this current study, a remote sensing approach combined with a geographic information system was used to identify the potentially dangerous glacial lakes within the large population of lakes in the study area. First, the small lakes ( 0.02 sq km) were excluded. The remaining lakes were then evaluated using a range of geomorphological and physical criteria which assessed factors related to the lake, dam, source (mother) glacier and surrounding area to identify the potentially dangerous glacial lakes. Out of 3624 mapped lakes, 1410 lakes are equal to and larger than 0.02 sq.km. This area of water is considered large enough to cause damage downstream if the lake was to rupture. This potential is heightened if the lakes are associated with a large and retreating glacier. Out of 1410 lakes, 1230 lakes were removed based on the damming condition, the activity of the source glaciers and their surroundings. The remaining 180 lakes were analyzed further to identify potentially dangerous glacial lakes. Of these, 47 glacial lakes were identified as potentially dangerous including 42 lakes in the Koshi, 3 in the Gandaki and 2 in the Karnali basins (Figure 5.2). Out of these, 25 PDGL lakes are in the TAR with transboundary to Nepal, 21 PDGLs are situated in Nepal and one is located in India. Table 5.5 show the PDGLs by river basin and country, and Table 5.6 displays the characteristics of the lakes, dams, source glaciers and the surroundings of the potentially dangerous glacial lakes. Table 5.5: Summary of potentially dangerous glacial lakes in the Koshi, Gandaki and Karnali basins of Nepal and the TAR, China. Basin Sub-basin Nepal China India Total Tamor 4 x x Arun 4 13 x Koshi Dudh Koshi 9 x x 42 Tama Koshi 1 7 x Sun Koshi x 4 x Gandaki Trishuli 1 1 x Marsyangdi 1 x x 3 Karnali Kali x x 1 Humla 1 x x 2 41 P a g e

Total 21 25 1 47 The PDGLs were ranked to determine the priority of lakes for potential GLOF risk reduction. Precise ranking was not possible. However, outbursts can occur that have no historic precedence, especially in view of the current atmospheric warming. It is important to remember this limitation. Also, many of the GLOFs that have occurred have effectively disrupted the retaining end moraine dams to the extent where the likelihood of subsequent outbursts in the same locality is minimal. Another consideration is that where potential outbursts are tentatively identified in areas remote from human activity, they should not be prioritized. Reliable determination of the degree of glacial lake instability, at least in most cases, will require detailed glaciological and geotechnical field investigation (Ives et al. 2010). S.N. Lake ID S.N. Lake ID S.N. Lake ID S.N. Lake ID S.N. Lake ID 1 GL087945E27781N 11 GL087591E28229N 21 GL087092E27798N 31 GL086476E27861N 41 GL085870E28360N 2 GL087934E27790N 12 GL087930E27949N 22 GL086977E27711N 32 GL086447E27946N 42 GL085838E28322N 3 GL087893E27694N 13 GL088002E27928N 23 GL086958E27755N 33 GL086500E28033N 43 GL085630E28162N 4 GL087749E27816N 14 GL088019E27928N 24 GL086957E27783N 34 GL086520E28073N 44 GL085494E28508N 5 GL087596E27705N 15 GL088066E27933N 25 GL086935E27838N 35 GL086530E28135N 45 GL084485E28488N 6 GL087632E27729N 16 GL088075E27946N 26 GL086928E27850N 36 GL086532E28185N 46 GL082673E29802N 7 GL087771E27926N 17 GL088288E28017N 27 GL086917E27832N 37 GL086371E28238N 47 GL080387E30445N 42 P a g e

8 GL087636E28093N 18 GL086304E28374N 28 GL086858E27687N 38 GL086314E28194N 9 GL087626E28052N 19 GL087134E28069N 29 GL086925E27898N 39 GL086157E28303N 10 GL087563E28178N 20 GL087095E27829N 30 GL086612E27779N 40 GL086225E28346N Figure 5.2 Location of potentially dangerous glacial lakes in the Koshi, Gandaki and Karnali basins of Nepal and TAR, China. The key physical parameters applied in the ranking were the distance between the lake outlet and dam crest and lakes enlargement over time, changes in the boundary conditions of the associated glaciers (frontal retreat), and the distance between the lake and the glacier: whether the two were in contact, close, or less than 500m apart. Lakes farther than 500m from their associated glaciers were not considered as potentially dangerous. The rating of the moraine dams included height, width, and steepness. Surroundings of the lake area included factors such as possible rock or debris slides, hanging glaciers, and potential avalanche paths. The danger level of the PDGL are categorized in to 3 levels: Rank I Large lake and possibility of expansion due to calving of glaciers. Lake close to loose moraine end, no overflow through the moraine dam, steep outlet slope, hanging source glacier, chances of snow and/or ice avalanches and landslide at the surroundings may impact to the lake and dam. Rank II Confined lake-outlet, lake-outlet close to compact and old end moraine, hanging lake, distinct seepage at the bottom of end moraine dam, gentle out moraine slope. Rank III Confined lake-outlet, gentle outward dam slope, large lake but shallow depth, moraine dam more than 200m, old and compact moraine. Of the 47 lakes reviewed, 31 lakes were classed as Rank I, 12 lakes as Rank II, and 4 lakes as Rank III (Table 5.6). The water level was already lowered from four of the lakes under Rank I, by more than 3m in Tsho Rolpa and 4m in Imja Tsho of Nepal and two lakes GL088066E27933N and GL088075E27946N in China. 43 P a g e

Table 5.6: List of potentially dangerous glacial lakes in Koshi, Gandaki and Karnali basins of Nepal and TAR, China. S.N. Lake ID/Name Rank Description River basin Country 1 GL087945E27781N I 2 GL087934E27790N III 3 GL087893E27694N III 4 GL087749E27816N I 5 GL087596E27705N I 6 GL087632E27729N III 7 GL087771E27926N I 8 GL087636E28093N I 9 GL087626E28052N I 10 GL087563E28178N III 11 GL087591E28229N II 12 GL087930E27949N I Less dam length, steep side slope and landslide, source glacier hanging but far from the lake. Lake close to dam end but confined outlet, hanging glacier, steep side wall slope. Shallow lake at steep slope and short dam length, hanging mother glacier Glacier in contact, less dam length, possibilities of avalanches, landslide on outer slope of dam. Lake expanding, cascading lake overflow may trigger the outburst, less dam width, erosion at end moraine. Lake outlet close to moraine dam end (2m), high gradient dam, Possibilities of lake expansion, calving source glacier, lake outlet near to dam end, steep side wall. Possibilities of lake expansion, lake outlet near to dam end, marking of outlet drainage, seepage at the bottom of the dam, steep outward dam slope and steep side wall. Possibilities of lake expansion, lake outlet near to dam end, no clear outlet drainage, seepage at the bottom of the dam, steep and eroded side wall with landslide and rock blocks, hanging source glacier. Large lake and expanding, no distinct lake outlet and surface runoff is appeared after 500m down the slope, one side steep wall, Large lake at the extreme end of moraine, dry outlet channel for almost 370m and then appeared the river flow, old and compact moraine, hanging source glacier. Possibilities of merging of small lake, less dam width, calving Tamor Tamor Tamor Tamor Arun Arun Arun Arun Arun Arun Arun Arun Nepal Nepal Nepal Nepal Nepal Nepal China China China China China China 44 P a g e

S.N. Lake ID/Name Rank Description River basin Country 13 GL088002E27928N I 14 GL088019E27928N I 15 GL088066E27933N I 16 GL088075E27946N I 17 GL088288E28017N II 18 GL086304E28374N II 19 GL087134E28069N I 20 GL087095E27829N II 21 GL087092E27798N Lower Barun I source glacier, steep side slope with possibilities of snow avalanches and landslide, Large lake and fed by many tributary glaciers, old and dry lake outlet channel of about 40m and then major seepage appeared, calving source glacier, PDGL GL088019E27928N is at the tributary valley. Possibilities of lake expansion, lake outlet at the dam end, couple of overflow channels, moraine dam seems old and compact, calving source glacier, Lake water lowered and done compaction at end moraine dam, calving glacier, possibilities of expansion, short dam length. Lake water lowered, lake expanding on both ends, dead ice on the end moraine, large source glaciers, large PDGL GL088066E27933N at left valley may damage the lake Lake at close to dam end but confined wide outlet channel, compact and old moraine, calving and steep slope source glacier, high gradient of moraine dam. Ice fall from source glacier, large lake, old moraine Possibilities of expansion of lake, Lake outlet almost at the dam end, minor overflow but many seepage, possibilities of landslide on the side wall. Hanging lake connected with the retreating glacier, landslide at the side wall. Possibilities of expansion of lake, calving source glacier, chances of landslide and ice avalanches at the right side wall of the lake, one lake and couple of small lakes at the at the upper catchment. Arun Arun Arun Arun Arun Arun Arun Arun Arun China China China China China China China Nepal Nepal 45 P a g e

S.N. Lake ID/Name Rank Description River basin Country 22 GL086977E27711N I 23 24 25 GL086958E27755N Chamlang GL086957E27783N Hongu 2 GL086935E27838N Hongu1 26 GL086928E27850N I 27 GL086917E27832N I 28 GL086858E27687N I 29 30 GL086925E27898N Imja Tsho GL086612E27779N Lumding II I I I I Lake expanding towards glacier, short dam length and steep, calving source glacier, high chances of ice toppling and avalanches Hanging glaciers, high chances of ice avalanches, lake formation near the end moraine, Ice underneath the dam but the dam length is more than 500m Hanging glacier, chances of avalanches, short dam length and steep slope with many erosional features Lake expanding towards the retreating glacier snout, hanging lakes on both side of the valley, lake outlet is close to dam end. Many cascading lakes in the lower old moraine., PDGL GL086928E27850N is at the hanging valley Lake out let near to the dam end, dam outer slope is steep, cascading lake in upstream, chances of landslide and ice avalanches at upstream, may also trigger to the lake GL086935E27838N Close to source glacier, short dam length and steep side slope with erosional features, Ice underneath the dam, Few meters of freeboard, outer dam slope steeper, hanging source glacier, chances of landslide, ice avalanches from the head water and from right valley. Lake water lowered by 4m in 2016, lake expanding level reduced, ice underneath end moraine, merging of supraglacial pond Lake expanding rapidly, in contact with calving glacier, 3 hanging lakes at the side valley, continues dam slope Dudh Dudh Dudh Dudh Dudh Dudh Dudh Dudh Dudh Nepal Nepal Nepal Nepal Nepal Nepal Nepal Nepal Nepal 46 P a g e

S.N. Lake ID/Name Rank Description River basin Country 31 GL086476E27861N Tsho Rolpa 32 GL086447E27946N I 33 GL086500E28033N II 34 GL086520E28073N I 35 GL086530E28135N II 36 GL086532E28185N I 37 GL086371E28238N I 38 GL086314E28194N I 39 GL086157E28303N I 40 GL086225E28346N II 41 GL085870E28360N Ganxico I II Lake water lowered in 2000, expanding rapidly with the retreat of calving source glacier, steep moraine dam and side moraine is thin and danger, hanging lake in tributary glacier. Lake expanding and possibility of merging with supraglacial ponds, evidences of small supra lake outburst at end moraine, dead ice at the end moraine, fed by three glaciers and calving at source glacier. Lake expanding, calving source glacier, additional small lake at end moraine, steep slope end moraine, one large hanging glacial lake in the side valley, lake extension is close to end moraine, hanging source glacier, erosional features at left valley wall Pond and ice in end moraine, confined outlet with gentle slope, hanging lake in the tributary valley, steep and calving source glacier, Lake snout at end of dam, possibilities of expansion, calving source glacier and crevasses near the lake Lake expanding in contact with long cascading glacier, no free board, short inward dam length, one supraglacial lake at the side valley Lake extension is close to end moraine, hanging source glacier with steep slope, hanging lake at the side valley, Lake extension up to end moraine, crevasse on the hanging source glacier, possibilities of ice avalanches Lake expanding on calving source glacier, short dam length but clear outlet in the lake Larger expanding lake, small outlet, gentle dam, possibility of Tama Tama Tama Tama Tama Tama Tama Tama Sun Sun Sun Nepal China China China China China China China China China China 47 P a g e

S.N. Lake ID/Name Rank Description River basin Country 42 GL085838E28322N Lumichimi 43 GL085630E28162N I 44 GL085494E28508N II 45 GL084485E28488N Thulagi 46 GL082673E29802N II 47 GL080387E30445N I II I ice in end moraine dam, possibilities of landslide and ice avalanches from side slope. Fast growing lake, steep source glacier, short end moraine dam, narrow lake outlet Lake at extreme dam end, chances of landslide from the right side wall and ice avalanches from the hanging source glacier. Lake is expanding and at debris covered glacier and hanging moraine, steep outer dam slope, chances of snow and ice avalanches, lake end and end moraine is about 100m. Large lake and expanding on the debris covered source glacier, possibility of landslide and snow avalanches from the side walls, evidences of subsidence of old and compact end moraine. Shallow lake but close to crest, overhanging boulder protecting the erosion of the dam, hanging source glacier with many crevasses. Lake close to top dam end, chances of expansion of lake due to debris cover source glacier, steep outer dam slope, chances of landslide at upstream, Sun Trishuli Trishuli Marsyangdi Mugu Kali China Nepal China Nepal Nepal India 48 P a g e

5. 3 Prioritization of PDGL for GLOF risk reduction 5.3.1 Socioeconomic value The socioeconomic assessment of the regions downstream of the identified potentially dangerous glacial lakes was carried out only for Nepal. The socioeconomic parameters included the number of household, population, the extent of motor-able road and highway sections, the number and type of bridges (wooden, suspension, motor-able, and highway bridges), and hydropower projects (number and capacity of hydropower projects in megawatts) in the 500m buffer zone in the path of a potential outburst and in the catchment. Population and households GLOF events have downstream impacts at different levels: individual household, ward/vdc, district, and national. At a household level, impacts are either direct from inundation or indirect by secondary erosion or landslides. At the ward/vdc level, people are affected by a loss of natural resources and service infrastructure. At the district level, damage to physical infrastructure disrupts the flow of goods and services, and at national level power supplies are disrupted because of damage to hydroelectricity projects, affecting populations living far beyond the GLOF area. The downstream population and infrastructure were estimated for the basins Tama Koshi, Dudh Koshi, Arun and Tamor down to the confluence with the Sun Koshi River. Similarly the information for Marsyangdi River and Trishuli River were calculated down to the confluence with the major Narayani River (Figure 5.3). The downstream population along the river valley of PDGLs were estimated based on the population census data of 2011 and the maximum number of house hold and population who could be directly affected along the river valley of about 500m buffer zone are counted on the high resolution satellite images available on Google Earth. Of these, at the household level, the number of people likely to be affected by a potential GLOF varied from 10,000 in Sun Koshi and 19,000 number in Dudh Koshi (Table 5.7). The remaining population 49 P a g e

could suffer a loss of environmental resources and service infrastructure. The maximum number who could be indirectly affected through infrastructural damage and loss of goods and services ranges from 51,000 for Tama Koshi to 62,500 for in Sun Koshi (Table 5.8). Figure 5.3: Household and population estimated from the catchment area for sub-basins of Koshi and Gandaki River. Table 5.7 Household distribution and estimation of population in 500m buffer zone along the river based on the Google images. Basin Sub-basin Household (HH) Population estimated Nepal China Nepal China Tamor 2,966 0 13,894 0 Arun 3,008 5,414 13,156 na Koshi Dudh Koshi 4,175 0 18,553 0 Tama Koshi 2,689 275 10,801 na Sun Koshi 2,332 1,377 9,821 na Gandaki Trishuli 3,066 154 12,057 na Marsyangdi 4,435 0 17,433 0 50 P a g e

The household and population calculated down to the confluence with the Sun Koshi River. The Tama Koshi and Dudh Koshi river confluences with the almost west-east flowing Sun Koshi River, whereas for the Arun River and Tamor River are considered up to the confluence of Tamor and Sun Koshi. Table 5.8: Household and population downstream within the Koshi River valley that could be affected (based on Census, 2011). S.N. Lake ID / Name Basin HH POP Male Female Basin HH POP Male Female 1 GL087945E27781N 12466 58462 27370 31092 2 GL087934E27790N 12466 58462 27370 31092 Tamor 3 GL087893E27694N 12396 58261 27286 30975 12572 58883 27591 31292 4 GL087749E27816N 12413 58320 27308 31012 5 GL087596E27705N 12343 56108 26571 29537 6 GL087632E27729N 12343 56108 26571 29537 7 GL087771E27926N 12479 56741 26884 29857 8 GL087636E28093N 12479 56741 26884 29857 9 GL087626E28052N 12479 56741 26884 29857 10 GL087563E28178N 12519 56925 26978 29947 11 GL087591E28229N 12519 56925 26978 29947 12 GL087930E27949N 12519 56925 26978 29947 13 GL088002E27928N 12519 56925 26978 29947 Arun 14 GL088019E27928N 12519 56925 26978 29947 12961 59275 28150 31125 15 GL088066E27933N 12519 56925 26978 29947 16 GL088075E27946N 12519 56925 26978 29947 17 GL088288E28017N 12519 56925 26978 29947 18 GL086304E28374N 12519 56925 26978 29947 19 GL087134E28069N 12519 56925 26978 29947 20 GL087095E27829N 11782 53859 25551 28308 21 GL087092E27798N / Lower Barun 11782 53859 25551 28308 22 GL086977E27711N 6549 31481 15076 16405 23 GL086958E27755N / Chamlang 6549 31481 15076 16405 24 GL086957E27783N / Hongu 2 6549 31481 15076 16405 GL086935E27838N / 25 6549 31481 15076 16405 Hongu 1 Dudh 11205 51993 25221 26772 26 GL086928E27850N Koshi 6549 31481 15076 16405 27 GL086917E27832N 6549 31481 15076 16405 28 GL086858E27687N 5948 28692 13769 14923 29 GL086925E27898N / Imja Tsho 7241 33127 15873 17254 30 GL086612E27779N / Lumding 6096 29154 13873 15281 51 P a g e

S.N. Lake ID / Name Basin HH POP Male Female 31 GL086476E27861N / Tsho Rolpa 12010 50643 23867 26776 32 GL086447E27946N 12071 50898 24010 26888 33 GL086500E28033N 12071 50898 24010 26888 34 GL086520E28073N Tama 12071 50898 24010 26888 Koshi 35 GL086530E28135N 12071 50898 24010 26888 36 GL086532E28185N 12071 50898 24010 26888 37 GL086371E28238N 12071 50898 24010 26888 38 GL086314E28194N 12071 50898 24010 26888 39 GL086157E28303N 14744 62493 29969 32524 40 GL086225E28346N 14744 62493 29969 32524 41 GL085870E28360N / Sun Ganxico Koshi 14744 62493 29969 32524 42 GL085838E28322N / Lumichimi 14744 62493 29969 32524 Note: HH House hold; Pop Population Basin HH POP Male Female 12104 50972 24036 26936 14744 62493 29969 32524 Exposure of infrastructure Several existing and proposed hydropower projects are commissioned along or near the river valley that they are threatened from the GLOF risk (Table 5.9 and Figure 5.4). The roads, tracks & trails, and bridges built along the river valley are at higher risk (Table 5.10). The hydro-powers are already on operational and some are under construction and plan in the river basins like Tamor, Tama Koshi, Sun Koshi and Trishuli (Table 5.9). The level of exposure to a potential GLOF in basins such as the Bhotekoshi/Sunkoshi and Marsyangdi is comparatively higher because of big hydropower projects - Marsyangdi and Middle Marsyangdi at Marsyangdi and Bhotekoshi, Sanima and Sunkoshi hydropower at Bhotekoshi/Sunkoshi. The total capacity of hydropower in Marsyandi basin is 922 MW but only two plants are under operation generating 74 MW. Similarly, a capacity of 208 MW is planned from seven hydropowers in Bhotekoshi/Sunkoshi while only 61 MW is operational from four plant. The highway and main trails in these basins are well developed and are likely to be affected to varying extent depending on a GLOF event. For example, the Lumuchimi Lake (GL085838E28322N) and Ganxi Co Lake (GL085870E28360N) in the Poiqu basin in China can damage a huge loss in the downstream settlements such as Liping, Tatopani, Larcha, Barhabise, Lamosangu and Khandichaur along the Kodari Highway 52 P a g e

in the Bhotekoshi/Sunkoshi basin in Nepal. It can destruct the settlements, markets, bridge, trials, roads, highway and other infrastructure on its path. Other lakes like Imja Tsho (GL086925E27898N) at Dudh Koshi, Thulagi (GL084485E28488N) at Marsyangdi and Tsho Rolpa (GL086476E27861N) at Tama Koshi also have high risk in the downstream area as the flow of people, goods, settlements, markets and infrastructure have increased along these valleys due to the growth of tourism and trade. At Tama Koshi basin, the seven plants (implemented/proposed) can produce 1,053 MW of energy and the lake Tsho Rolpa (GL086476E27861N) can affect two hydropower producing 70 MW and 160 km of main trail in a GLOF event (Table 5.9 and 5.10). A GLOF from Imja Tsho (GL086925E27898N) and Lumding (GL086612E27779N) Lakes at Dudh Koshi can damage the main trail, 39 bridges trials and tracts, settlements and markets but not a hydropower plant (683 MW) as three plants are under implementation phase. Several hydropower are in implementation phase in Tamor (five plants producing 811 MW) and Trishuli (eleven plants producing 633 MW) basins so the damages can occur only in the trails, settlements and other goods that exists along the river valley. Two hydropower are in operation generating 38 MW of energy in the downstream of Trishuli. In Arun basin, two hydropower are in implanting phase that can produce 1000 MW of energy. A GLOF from Lower Barun (GL087092E27798N) in Arun basin can pose risk to 1 bridge road, 30 bridge trails and tracks, 260 main trails and 37 other roads (Table 5.10). However, this risk can be reduced if proper attention is given to GLOF risks and implement a mitigation and adaptation measures in the areas directly susceptible to the GLOF risk. Table 5.9: Location of hydropower projects in the river valley susceptible to a potential GLOF risk. SN River Hydro Power Project Latitude Longitude MW Status 1 Tamor Upper Tamor 27.4531 87.7083 415 GLA 2 Tamor Middle Tamor 27.3992 87.6717 57 GLA 3 Tamor Tamor Storage 27.0344 87.5242 200 SLI 4 Tamor Tamor Mewa 27.3333 87.5833 101 SLI 5 Kabeli Kabeli-A 27.2281 87.6819 38 GLI 6 Arun Arun 3 27.5000 87.2000 900 SLI 7 Barun Khola Lower Barun Khola 27.6833 87.3444 10 SLI 8 Dudh Koshi Dudhkoshi-6 27.6667 86.6917 93 SLA 9 Dudh Koshi Dudhkoshi-2 27.3603 86.6667 240 SLI 10 Dudh Koshi Dudhkoshi-4 27.5583 86.6667 350 SLI 53 P a g e

SN River Hydro Power Project Latitude Longitude MW Status 11 Dudh koshi Dudhkoshi Storage 27.1139 86.5833 300 SLI 12 TamaKoshi Tamakoshi-3 TA-3 27.4928 86.0761 650 GLA 13 Tamakoshi Upper Tamakoshi-A 27.9167 86.1967 85 SLA 14 TamaKoshi Tamakoshi-V 27.7583 86.1833 87 SLI 15 Sipring Sipring Khola 27.8106 86.2297 10 OP 16 Jum Jum Khola 27.9378 86.2131 62 SLA 17 Khimti Khimti -I 27.4744 86.1011 60 OP 18 Lapche Upper Lapche Khola 28.0500 86.1456 99 SLI 19 Bhote Koshi Middle Bhotekoshi -1 27.8833 85.9103 40 GLA 20 Bhote koshi Madhya Bhotekoshi 27.8189 85.8636 102 GLI 21 Bhote Koshi Upper Bhotekoshi 27.9111 85.9203 45 OP 22 Sun Koshi Sun Koshi 27.7500 85.8361 10 OP 23 Sun Koshi Sunkoshi Small 27.7667 85.8833 3 OP 24 Sun Koshi Lower Balephi-3 27.7194 85.7625 5 SLI 25 BhairabKund Bhairab Kund Khola 27.9311 85.9314 3 OP 26 Bhote koshi Rasuwagadhi 28.2347 85.3561 111 GLI 27 Bhote koshi Rasuwa Bhotekoshi 28.1750 85.3333 120 SLI 28 Bhote koshi Chilime Bhotekosi 28.1556 85.3222 53 SLA 29 Trishuli Devighat Cascade 27.8522 85.0967 10 GLA 30 Trishuli Trishuli Galchhi 27.7978 84.9722 75 GLA 31 Trishuli Upper Trishuli 3A 28.0225 85.1878 60 GLI 32 Trishuli Third Trishuli Nadi 27.8019 84.7867 20 GLI 33 Trishuli Upper Trishuli 3B 27.9867 85.1697 37 GLI 34 Trishuli Devighat 27.8853 85.1319 14 OP 35 Trishuli Trishuli 27.9192 85.1458 24 OP 36 Trishuli Middle Trishuli 27.9633 85.1750 65 SLI 37 Trishuli Upper Trishuli-1 28.0742 85.2111 216 SLI 38 Trishuli Upper Trishui-2 28.1286 85.3011 102 SLI 39 Trishuli Langtang Khola 28.1514 85.3428 10 GLA 40 Marsyangdi Upper Marsyangdi -2 28.3678 84.3583 600 GLA 41 Marsyangdi Upper Marsyangdi 1 28.3139 84.3917 138 GLA 42 Marsyangdi Marsyangdi Besi 28.2000 84.3542 50 GLA 43 Marsyangdi Upper Marsyangdi A 28.2853 84.3653 50 GLI 44 Marsyangdi Madhya Marsyangdi 28.1389 84.4050 70 OP 45 Nyadi Lower Nyadi Hp 28.3083 84.4014 10 SLA 46 Radhi Radhi Small 28.3967 84.4094 4 OP 54 P a g e

Note: MW Mega Watt; GLI - Generation license issued; GLA - Generation license applied; SLA Survey license applied; OP Operational project Figure 5.4: Location of hydropower projects in the river valley to a potential GLOF risk. Table 5.10: Road, bridge, track and trails within 500m zone along the river valley to a potential GLOF risk. Sub-Basin Bridge Road No Length (km) Bridge Trails & tracks No Length (km) District Road Highway Main Trail Other Road No Length (km) No Length (km) No Length (km) No Length (km) Tamor 2 0.26 29 3.01 6 8.18 300 253.93 20 24.1 Arun 1 0.14 30 3.10 260 256.40 37 40.4 Dudh Koshi 39 4.23 159 135.70 165 237.2 Tama Koshi 5 0.36 34 3.06 10 8.30 195 159.72 98 136.9 Sun Koshi 13 1.43 19 2.41 77 60.02 177 102.97 109 156.1 Trishuli 41 3.84 32 3.49 31 19.21 170 95.89 341 218.16 53 62.9 Marshyangdi 33 3.61 21 12.21 44 34.86 199 147.55 46 67.5 Mugu Karnali 1 0.07 37 4.35 1 0.00 392 382.59 43 49.4 Mahakali 17 1.46 168 135.63 259 336.0 Grand Total 62 5.91 166 17.73 31 19.21 257 164.21 1468 1133.97 830 1110.4 55 P a g e

5.3.2 Priority of potentially dangerous glacial lakes for risk reduction The PDGL is ranked and prioritized considering both of the socioeconomic and physical parameters. The physical parameters include the boundary conditions of the associated glaciers and lake enlargement over time; the distance between lake and glacier; the height, width and steepness of the moraine dams; physical condition of the surroundings (potential for rock and debris slides and avalanches); and the presence of hanging glaciers. The socioeconomic parameters include the size of downstream settlements, the number and type of bridges, hydropower projects, the agricultural land, and any other important infrastructure or activities of economic value. Based on these parameters, the PDGL were categorized into: I) high priority lakes requires extensive field investigation and GLOF risk reduction activities; II) medium priority lakes requires close monitoring and reconnaissance field surveys; and III) low priority lakes warrants periodic observation. The key physical parameters were considered first for the selection of PDGL and later categorized into three ranks depending on the level of hazard. The socioeconomic parameters were then added to categories the priorities for a GLOF risk reduction. Eighteen lakes falls under the Priority 1, 11 lakes in Priority II and again 18 lakes in priority III to reduce the potential GLOF risk (Table 5.11). Among them 6 lakes of Priority I, 8 lakes of Priority II and 7 lakes of Priority III are in Nepal. Similarly, 12 lakes of Priority I, 2 lakes of Priority II and 11 lakes of Priority III are in TAR, China and one lake of priority II is in India. Depending upon the Rank level and priority level, PDGLs were further classified into four level of risks: very high, high, medium or moderate and low. Of the total, 17 PDGLs are in very high (11 in TAR, China and 6 in Nepal (Figure 5.5 and Table 5.11). Similarly, 56 P a g e

11 PDGLs are in high risk (3 in TAR, China and 7 in Nepal). These high and very high risk lakes requires extensive field investigation and mitigation measures to reduce GLOF risks. Figure 5.5: Priority level of potentially dangerous glacial lakes in TAR, China and Nepal. 57 P a g e