Glacial lake outburst flood hazard under current and future conditions: first insights from a transboundary Himalayan basin

1Department of Geography, University of Zurich, Zurich, CH-8057, Switzerland 2Institute for Environmental Science, University of Geneva, CH-1205, Geneva 3School of Geography and Sustainable Development, University of St Andrews, St Andrews, KY16 9AL, UK 4Key Laboratory of Tibetan Environmental Changes and Land Surface Processes, Institute of Tibetan Plateau Research, 10 Chinese Academy of Sciences (CAS), Beijing, China 5CAS Center for Excellence in Tibetan Plateau Earth Sciences, Beijing, China 6Department of Remote Sensing & GIS, JIS University, Kolkata 700109, India

future deglaciated conditions. Meanwhile, Zheng et al. (2021a) have elaborated such analyses for the entire High Mountain Asia, revealing an almost 3-fold increase in GLOF risk and the emergence of new hotspots of risk over the course of the 21 st century. Significantly, the number of lakes posing a transboundary threat within border areas of China and Nepal could double in the future, particularly within the eastern Himalayan region (Zheng et al., 2021a). While such large-scale, first-order studies are important for raising general awareness of the future challenges that mountain regions will face (Hock et al., 2019), there 70 are limitations in the extent to which these studies can directly inform planning and response actions at the ground level.
The need for forward-looking, anticipatory approaches to hazard and risk modelling is clearly recognised within recent international guidelines on glacier and permafrost hazard assessment (GAPHAZ, 2017), yet practical examples on how to integrate future lake development for GLOF assessment and risk management are lacking. International best practice is framed 75 by both a first-order assessment undertaken at large scales (to identify potentially critical lakes), followed by a detailed assessment for these lakes using numerical models to simulate downstream flood intensities as a basis for hazard mapping (GAPHAZ, 2017). This is a common approach for existing threats, where the time, data, and expertise needed to invest in comprehensive hazard modelling and mapping can be well justified for a lake that is determined to be critical. However, for future lakes, where the formation of the lake and its eventual dam characteristics remain highly uncertain, there remains a 80 methodological gap in the hazard assessment process, as authorities are unlikely to undertake sophisticated hazard mapping for a threat that may not even eventuate. In this study we aim to address this gap, by providing an illustrative example of how the threat of a potential future lake can be feasibly assessed along-side that of current lakes, and how this information can feed practically into decision-making and response planning in a transboundary context.

85
Focusing on the transboundary Poiqu river basin in the central Himalaya, the specific objectives of the study are to 1) apply hydrodynamic modelling and systematic criteria to assess the magnitude and likelihood of worst-case outburst events from two potentially critical lakes in the Poiqu river basin, 2) compare the results with a potential outburst from a large lake that is anticipated to develop in the future, and 3) discuss the implications for early warning or other risk reduction strategies. This study is intended to provide timely input to the scoping and design phase of future GLOF risk reduction strategies in the Poiqu 90 basin, to ensure early warning systems and other measures remain suitable under possible future scenarios.

Study area
This analyses focuses on a ca. 40 km stretch of the lower Poiqu river basin originating from Galongco glacial lake, considering GLOF impacts in Nyalam town (capital of Nyalam county, Tibetan Autonomous Region), and downstream to the border with https://doi.org/10.5194/nhess-2021-167 Preprint. Discussion started: 28 June 2021 c Author(s) 2021. CC BY 4.0 License.
Nepal at Zhangmu (Fig. 1). The elevation range of the study area extends over 6000 metres, from the summit of Shishapangma 95 at 8,027 m a.s.l, whose glacierised slopes feed Galongco, to 2000 m a.s.l in the river valley at Zhangmu. According to Wang and Jiao (2015), mean annual temperature and mean annual precipitation in Nyalam (3810 m asl) are 3.8°C and 650.3 mm respectively, with sub-zero temperatures lasting from November -March each year. Temperatures peak in July (10.8°C), while highest precipitation rates are recorded in September (87.9 mm/month). In total, 60% of the annual rainfall falls during the monsoon months of July -September (Wang et al., 2015b) 100 The Poiqu basin is the Tibetan portion of the large transboundary Poiqu/Bhote Koshi/Sun Koshi River Basin, along which the economically important Friendship Highway links China to Nepal, and where significant hydropower resources are located (Khanal et al., 2015b). Based on a larger study across Tibet, the Poiqu basin has been identified as a clear hotpot of transboundary GLOF danger (Allen et al. 2019 -Fig. 1), where at least 6 major GLOF events reported over the past century, 105 including repetitive events from Jialongco in 2002 (Chen et al., 2013), and Cirenmaco in 1964, 1981and 1983. The 1981 event resulted in numerous fatalities, and estimated losses of up to US$4 million as a result of damage to houses, roads, hydropower, and disruption to trade and transportation services (Khanal et al., 2015a). Meanwhile an outburst of 1.1 × 10 5 m 3 from Gongbatongshacuo (adjacent to Cirenmaco) in July 2016, resulted in significant damage to hydropower and roads, exacerbating losses inflicted one year earlier by the Gorka earthquake (Cook et al., 2018). Whereas 110 Gongbatongshacuo has completely drained, Cirenmaco remains a large and persistent threat, considered as one of the most dangerous lakes in Tibet Wang et al., 2018).
In the current study, we focus not on Cirenmaco, which has already been the subject of comprehensive investigations , but rather on two other well documented threats of Jialongco and Galongco, owing to their potential to cause 115 damage to the Tibetan county capital of Nyalam, and downstream in Nepal. Both lakes have expanded rapidly over the past decades, with Galongco, the largest lake in the basin, increasing its area by 450% from 1.00 to 5.46 km 2 in the period 1964-2017 (Wang et al., 2015b;Zhang et al., 2019) . https://doi.org/10.5194/nhess-2021-167 Preprint. Discussion started: 28 June 2021 c Author(s) 2021. CC BY 4.0 License.

Methodological approach 125
In line with recent international guidance in GLOF hazard assessment (GAPHAZ 2017), in this study we consider lake susceptibility, which determines the likelihood of a given outburst scenario to occur, and use hydrodynamic modelling to determine downstream impacts. In order to compare the threat posed by the two current lakes with a future anticipated lake, we focus on worst-case scenario modelling -that is to say, the maximum outburst volume that could be produced from Jialongco, Galongco, and the anticipated future lake. 130

GLOF Modelling
The total volume of water potentially released during a GLOF event is of critical importance for hydrodynamic modeling of a GLOF scenario (Westoby et al., 2014). In this study, the volumes for Jialongco and Galongco were estimated by multiplying 135 https://doi.org/10.5194/nhess-2021-167 Preprint. Discussion started: 28 June 2021 c Author(s) 2021. CC BY 4.0 License. mapped lake area (A) by estimated mean depth (Dm), where Dm is calculated according to the empirical relationship of Fujita et al. (2013) which has been established based on lake data from the Himalayan region: where A and Dm are the lake area (km 2 ) and mean depth (m). Lake area was mapped using Google Earth imagery from 2019. 140 For GLOF modelling of the future lake, the location and maximum volume of the potential lake upstream from Jialongco is based on a modelled overdeepening in the glacier bed topography using GlabTop (Linsbauer et al., 2012). The model is now well established for providing a first-order indication of where lakes may develop in the future (e.g., Allen et al., 2016;Haeberli et al., 2016a;Linsbauer et al., 2016;Magnin et al., 2020). The ice thickness distribution from GlabTop is subtracted from a surface DEM to obtain the bed topography, i.e. a DEM without glaciers, from which overdeepenings in the glacier bed can be 145 detected and volumes estimated. Inputs to the model include manually edited glacier branch lines, and a DEM -in this case the NASA Shuttle Radar Topography Mission (SRTM) Version 3.0 (void filled) was used, at 30 m resolution. While the model predicts several possible locations in the Poiqu basin where large future lakes can develop, we focussed on the largest of these lakes that threaten the town of Nyalam. Beyond its potential size, this overdeepening was selected owing to its position in an area of the low surface gradient behind a pronounced terminal moraine, beneath a tongue where supraglacial ponds are already 150 developing, and at an elevation that is lower than other overdeepenings in the area. All factors provide favourable preconditioning for the formation of a large proglacial lake (Frey et al., 2010;Linsbauer et al., 2016) Based on the total estimated volume of the lakes, we then establish the potential flood volume (PFV) for each lake following the concept of Fujita et al. (2013), that assumes full incision and removal of the downstream slope of the dam (Fig 2a). Only 155 where the height of the potential breach (hb) is greater than the mean depth of the lake is the full release of the lake volume possible: For example, in the case of Jialongco, the breach height is estimated at 40 m, which is less than the mean depth of the lake 160 suggesting that even following full moraine incision, some water will remain in the lake (Fig 2b). The resulting PFV is therefore estimated at 24.8 m 3 10 6 (40 m x 0.62 km 2 ). In comparison, the well documented 1981 outburst from the smaller Cirenmaco was estimated to have involved a breach height of up to 60 m and an outburst volume of 19 m 3 10 6 (Xu, 1988). In principle, dam geometries can be measured directly in Google Earth, although there can be severe distortions in the imagery in some regions and the DEM accuracy is unknown. Therefore, to achieve a higher level of accuracy, we measured hb and other 165 topographic parameters using spot elevations extracted from a higher resolution (1 m grid cell) Digital Elevation Model, generated from 0.5 m resolution tri-stereo Pleiades imagery acquired in October 2018, covering the whole Poiqu basin (Bhattacharya et al., 2021). Subsequent breach parameters were calculated according to Froehlich (1995) for each outburst scenario: 170 Bw = 0.1803Ko (Vw) 0.32 (hb) 0.19 (3) where Bw is the breach width (in m), Ko is a constant which is considered to be 1.4 for overtopping failures, Vw is the volume above hb of the lake (in m 3 ), and Tf (in min) is the time taken for the breach to form (where distances Bw and hb are fully 175 obtained). https://doi.org/10.5194/nhess-2021-167 Preprint. Discussion started: 28 June 2021 c Author(s) 2021. CC BY 4.0 License.
The HEC-RAS (v 5.0.7) dam-break module was used to set up different breach scenarios for the three lakes (Table 1). Dambreak simulations were performed where the frontal moraine (dam) is defined to fail, given the calculated breach parameters (after Froehlich, 1995). Here, a progressive breach mechanism was assumed for all the scenarios where overtopping failure 180 initiated at the crest of the moraine spreading downwards and sidewise. The outputs in the form of outflow hydrographs (discharge vs. time) were then used as boundary conditions for downstream two-dimensional GLOF routing with HEC-RAS (v 5.0.7) as far as Zhangmu (Fig 3). This hydraulic model solves the Full Saint Venant equations two-dimensionally in an unsteady flow. Two-dimensional routing requires accurate terrain information as a primary input. While several freely available DEMs were tested (e.g., ALOS PALSAR at 12. 5 m or HMA at 8 m), topographic artefacts led to modelling errors.  The assessment follows a systematic approach that considers wide-ranging atmospheric, cryospheric and geotechnical factors that can influence lake susceptibility, and thereby the likelihood of a GLOF occurring (after GAPHAZ 2017). As a desk-based assessment, we draw on remotely sensed data to the extent possible, to enable a semi-qualitative comparison of susceptibility factors across the three lakes. Factors assessed, their primary attributes, and sources used are provided in Table 2. Topographic 205 characteristics (dam geometry, slope angles etc) were precisely measured using the high resolution 1m DEM generated from Pleiades imagery. To establish the potential for ice and/or rock avalanche triggering, additional GIS-based analyses were undertaken. The overall likelihood of rock (or debris) avalanches triggering an outburst was calculated based on the concept of topographic potential Romstad et al., 2009) which identifies within each lake watershed a) the potential for rock to detach (parameterized by slope angles >30°), and b) the potential for the resulting avalanche to reach the glacial 210 lake (parameterized by overall trajectory slopes >14° (tanα = 0.25). Potentially unstable zones of glacial ice were identified in Google Earth, based on orientation and density of crevassing, with a subsequent estimate of the ice thickness and volume provided from the GlabTop model output (Table 3). Furthermore, the time series of Google Earth imagery was examined to identify any evidence of historical mass movements, that could indicate an enhanced threat to the lakes below.

Future lake development
To establish the possibility of lake development and the likely future trajectory of lake area growth on the parent glacier (RGI60-15.09475), we examine the surface velocity, rate of thinning and the evolution of the geometry (surface slope) of the glacier in recent decades. Previous studies (Quincey et al., 2007) have identified glacier surface attributes which may precondition the surface of debris-covered glaciers for supraglacial lake development. Glaciers bounded by large lateral and 220 terminal moraines which have a flat or gently sloping (<~2˚), slowly flowing (<~10 m a -1 ) main tongue commonly host networks of supraglacial ponds as surface meltwater cannot drain from the glacier surface. Such pond networks expand when the mass balance of the glacier is negative and coalesce to eventually form a supraglacial lake at the hydrological base level of the glacier-the lowest point where the glacier surface intersects the terminal moraine (Benn et al., 2012).
We used the Pleiades DEM and glacier surface elevation change data generated by King et al. (2019) to examine the evolution 225 of the geometry of glacier RGI60-15.09475 since the 1970s. Glacier surface slope estimates were derived by the fitting of linear regression models through 'average' (mean of 5 evenly spaced) elevation profiles of the glacier surface split into 750 m long segments (King et al., 2018). We also assessed the current flow regime of the glacier using surface velocity data, which was generated through the tracking of glacier surface features visible in Sentinel 2 imagery over the period 2017-2019 (Pronk et al., 2021). Examination of these parameters established that the conditions at the surface of the glacier (Fig. 7) are well 230 suited to imminent glacial lake development considering the factors outlined by Quincey et al. (2007).
To investigate the likely size of such a lake in the coming decades we consider two different scenarios of glacier thinning between 2015 and 2100 and follow a similar method to that of Linsbauer et al. (2013) to simulate glacier thickness into the future, but employ different criteria to determine future lake area. Our first scenario is based on the assumption that the 235 acceleration in glacier thinning in the Poiqu basin measured by King et al. (2019) (Fig. 7) is replicated by the year 2100. Such an increase in thinning will be driven by a further 1˚C increase in temperature by 2100 (Kraaijenbrink et al., 2017), further to the ~1˚C increase in temperature which has occurred in the central Himalaya (Maurer et al., 2019) since the 1970s. The second scenario is based on the premise that the increase in thinning which has occurred between 1974 and 2015 ( Fig. 7) will be replicated over subsequent equivalent time periods (by 2056, 2097, etc). We extrapolated the thinning rates from King et al. 240 https://doi.org/10.5194/nhess-2021-167 Preprint. Discussion started: 28 June 2021 c Author(s) 2021. CC BY 4.0 License.
(2019) and integrated the resulting elevation changes between 2015 and 2100. We then assumed that once the glacier surface had lowered to a height below the hydrological base level of the glacier (4890 m a.s.l.) meltwater ponding would occur and that DEM pixels with an elevation of less than this threshold represented lake area at that point in time.

Results
Based on the three assessed lake outburst scenarios, we focus below on results relating to the core hazard dimensions of GLOF 245 magnitude and likelihood (or probability), and assess the exposure of buildings in the town of Nyalam. A full hazard and risk assessment, including a complete range of outburst scenarios and vulnerability mapping, is beyond the scope of this study.

GLOF impact
Worst-case outburst scenarios for the three lakes were simulated until the border between China and Nepal (Zhangmu). Of the 250 two current lakes assessed, the modeled peak discharge from Galongco is more than 14 times larger than that from Jialongo, leading to flow depths up to 5 m higher and velocities up to 2 m 3 s -1 faster impacting the town of Nyalam. At the border, 20 km downstream, inundation depths are up to 10 times larger for the Galongco event as the flow becomes constricted in the narrow topography of the valley (Table 1, Fig. 3). A worst-case outburst from the potential future lake, with a release volume of 70 x 10 6 m 3 , and peak discharge of 42,917 m 3 s -1 , would result in flow depths (20.1 m) and velocities (13.9 m 3 s -1 ) in Nyalam 255 that would exceed events from both Jialongco and Galongco, while downstream at the border, flow depths would be lower than that of the Galongco outburst (23.8 vs 27.9 m), but with significantly higher velocities (13.9 vs. 9.4 m 3 s -1 ). Differences later. An outburst from the potential future lake has the quickest arrival time of only 42 minutes in Nyalam, reaching the Nepalese border 30 minutes later (compared to 40 minutes later for the existing lakes).
Upstream of Nyalam a backwash effect is produced by the narrowing of the valley, extending for 600 m up the Poiqu river, 265 with maximum flow depths of 25 m. We note that model simulations undertaken using several coarser DEMs (e.g., ALOS PALSAR at 12. 5 m or HMA at 8 m) all resulted in significant modelling artefacts in this region immediately up-and downstream from Nyalam owing to voids in the DEMs in this area of complex topography. As a consequence, physically implausible flow depths exceeding 100 m were simulated due to artificial blockages along the river path, while the timing of the floodwave was effected by the stagnation of the flow occurring behind these blockages. 270 Potential processes that could significantly enhance and/or modify the GLOF magnitude include entrainment of large volumes of sediment along the flow path leading to bulking of the flow volume, blockages of a river by GLOF deposits leading to secondary outburst events, and a process chain involving more than one lake. Significant erosion of sediment and a catastrophic transformation into a debris flow event is considered unlikely for any of the three outburst scenarios, given that average 275 trajectory slope angles measured along the flow paths (Fig. 3) are well below those needed to entrain sediment from within a channel (Huggel et al., 2004). In the absence of significant entrainment of sediment, there is limited potential for large deposits to block adjacent waterways, although erosion and destabilisation of the river banks as a result of the flood waters means that such secondary hazards cannot be excluded, particularly in the steep sided gorge downstream of Nyalam.  Results indicate that an outburst event from the potential future lake could slam into, pool up, and eventually overtop the lateral moraine of Jialongco, producing a potential chain reaction where Jialongco also breaches (Fig. 4). Maximum flow heights measured at the surface of Jialongco reach 27 m, suggesting a significant volume of water could enter the lake via overtopping. 285 Simultaneously, the outburst from upstream would lead to erosion at the front distal slope of the Jialongco dam area, as the flow is constrained in this area leading to high energy levels. The combined high-impact low-probability chain reaction involving near-simultaneous breaching of the potential future lake and Jialongco requires more sophisticated modeling to fully analyse downstream impacts, but in a first approximation could lead to maximum combined flow depths >30 m in Nyalam.

GLOF likelihood 300
The second component of GLOF hazard concerns the likelihood or probability of an event occurring considering the wideranging factors that can condition or trigger an outburst. Taking a systematic approach (after GAPHAZ 2017), we compare the relative susceptibility of the three lakes considered in this study, considering also how this susceptibility might evolve in the future (Table 3). The table distinguishes those factors that condition and/or trigger an outburst event, while also linking to those factors that can influence outburst magnitude (see 4.1). Located in a transitional zone to the north of the main Himalayan 305 divide, the upper Poiqu basin is subject to heavy rainfall during the Asian summer monsoon. With a significantly larger watershed area, Galongco is considered more susceptible to heavy rain and/or snow melt leading to high lake water levels, and under future deglaciated conditions the lake may become fed by a well-developed paraglacial stream network. However, even under these conditions, the relatively favourable dam geometry (low width to height ratio and 15 m dam freeboard) suggests that the likelihood of a catastrophic outburst via this triggering mechanism is low. Similarly, self-destruction via warm 310 temperatures and melting of ground ice within the moraine dam can be effectively discounted. Creeping permafrost features visible in the vicinity of Galongco, and a partially hummocky appearance of the lake dam, suggests some presence of an icecored moraine, but the width and gentle downstream slope of the dam would make a catastrophic failure extremely unlikely. https://doi.org/10.5194/nhess-2021-167 Preprint. Discussion started: 28 June 2021 c Author(s) 2021. CC BY 4.0 License.
As with the majority of large glacial lakes across the Himalaya (Liu et al., 2013;Richardson and Reynolds, 2000;Sattar et al., 315 2021), the main triggering threat is considered to come from large slope instabilities, impacting into the lake. Under current conditions, Jialongco is considered to be most susceptible to ice avalanches, given the presence of a steep, highly crevassed tongue positioned directly behind the lake (Fig. 5). With an average slope of 36°, and large crevasses marking a sharp break in topography, full collapse of the glacier tongue (~20 x 10 6 m 3 ) is feasible ( Table 4). The mass would impact the lake in a direction parallel to the longitudinal axis of the lake, leading to maximum overtopping wave heights and swashing effect, 320 meaning even a partial collapse of the unstable ice mass could be sufficient to displace the full potential flood volume of the lake, irrespective of whether or not the dam is deeply incised. However, further retreat of the tongue will see a reduction in the potential avalanche volume, and eventually this threat will be eliminated completely as the ice retreats to a point behind the topographic break.

325
In comparison, the largest potential unstable ice masses surrounding Galongco would strike the lake perpendicular to the longitudinal axis of the lake (from the west) meaning most of the energy from a displacement wave would be dissipated on the opposing side of the lake. Steep ice cliffs located higher on the mountain slopes, including those found currently above where the future lake is expected to form, are estimated to have maximum volumes ranging from 0.1 -1 x 10 6 m 3 (Table 4), and therefore are considered insufficient to generate the worst-case outburst flood volumes simulated here. Hence, a large rock or 330 combined ice-rock avalanche is considered to be the most feasible mechanism capable of triggering the maximum potential outburst flood volume from either Galongco or the potential future lake. A greater likelihood of such an event is identified for Galongco, given the sheer size of the catchment meaning greater topographic potential for large rock failures, including from the slopes of Shishapangma rising nearly 3000 m above the lake. Similarly, the potential future lake is positioned directly beneath the ~ 2000 m high eastern face of Ramthang Karpo Ri. Given that Poiqu basin is located within a high seismic hazard 335 zone (Shedlock et al., 2000), large ice-rock avalanches of the magnitude needed to trigger a worst-case scenario from these lakes are possible, but remain extremely rare events. While displacement wave processes depend ultimately on the orientation of the incoming mass, and its interaction with lake bathymetry (Schaub et al., 2015), we estimate an avalanche volume in the order of 50 million m 3 would be needed to initiate a worst-case outburst from Galongco. This estimate accounts for the relatively stable dam geometry, requiring a significant amount of the flood volume to be released in the initial overtopping 340 wave, which, based on empirical evidence, can be estimated as being up to 10 times the incoming mass (Huggel et al., 2004).
Even on a global scale, avalanche volumes of this magnitude are extremely rare (Kääb et al., 2021;Schneider et al., 2011), making this a high magnitude, but very low likelihood process chain. Finally, all three lakes are susceptible to instantaneous or progressive landslides occurring from the adjacent lateral moraines, most notably for Jialongco where active instabilities are clearly evident. Recent studies have shown that large lateral failures, either instantaneous or progressive, can be sufficient 345 to initiate catastrophic process chains where dam geometries are sufficiently prone to erosion (Klimeš et al., 2016;Zheng et al., 2021b).
Based on the assessment results, a large outburst scenario involving the maximum potential flood volume is considered most likely under current conditions to originate from Jialongco, triggered by an ice avalanche or large failure of the lateral moraine 350 slopes. Large rock or combined ice-rock avalanches are a less likely, but potentially high magnitude trigger of an outburst from all 3 lakes considered. Given the large volume of water that would need to be displaced and breach depth that would need to occur, the probability of a worst-case scenario originating from Galongco is considered very low. The susceptibility of the potential future lake to avalanches, moraine instabilities, or rain and snowmelt, will ultimately depend on the its final dam geometry, and particularly its freeboard, which is highly uncertain from model results alone.   Landsat-based lake area mapping ; Area/depth scaling (Fujita et al., 2013), GlabTop  (Shedlock et al., 2000) https://doi.org/10.5194/nhess-2021-167 Preprint. Discussion started: 28 June 2021 c Author(s) 2021. CC BY 4.0 License.    We identify from Open Street Map and Google Earth imagery, the buildings in Nyalam exposed to different GLOF intensity levels according to simulated flood flow heights (after Pozzi et al., 2005). While classification schemes vary across countries, land areas potentially affected by high flood or debris flow intensities (calculated on the basis of flow heights and/or flow velocities), are typically considered as high hazard zones even for low probability events (GAPHAZ, 2017). In Nyalam, lower 375 flow heights associated with an outburst from Jialongco result in lower levels of exposure compared to simulated events from Galongco or the potential future lake (Fig. 6). The majority of buildings in Nyalam are located high above the river channel, where they are safe even in in the event of a worst-case outburst scenario. However, it is clear that the rapid expansion of infrastructure along the river banks north of the main settlement over the past several years has significantly increased the built area exposed to potential GLOF events, with many new buildings located in the high intensity flood zone. Overall, levels of 380 exposure are comparable for simulated outbursts from both Galongco and the potential future lake, with both events likely to disrupt the main national road and bridges linking to the town.

c) Geotechnical and Geomorphic
Downstream from Nyalam in the reach to the border with Nepal there are few buildings located along the river bank, and the main threat is to a 7.5 km stretch of the transnational highway (Fig. 4), of which the proportion affected by high intensity flood 385 levels is 74% and 96%, for modelled outbursts from Jialongco and Galongco respectively (up to 98% for the potential future lake scenario). While we did not simulate beyond the border owing to the limited coverage of the required high resolution Pleiades DEM, previous events (e.g., Cook et al., 2018;Wang et al., 2018), and assessment studies (Khanal et al., 2015a;Shrestha et al., 2010) have highlighted the significant risk to Nepalese communities, hydropower stations, and other infrastructure located along the banks of the Bhotekoshi river.

Trajectory of future lake development
The thinning of glacier RGI60-15.09475 over at least the last four decades has caused the development of a glacier surface that is well suited for supraglacial lake development (Fig. 8). The central 2.5 km of the glacier's ablation zone, where 400 supraglacial ponds are already forming, is effectively stagnant, very gently sloping and has become heavily pitted due to differential ablation in response to spatially variable debris thickness. These conditions will enable the further expansion of the supraglacial pond network, which is unlikely to drain quickly.

410
The extrapolation of thinning measured over the last four decades over glacier  suggests that a large portion of the glaciers surface will soon sit below an elevation where supraglacial meltwater would normally drain from the glacier surface, allowing for the development of a supraglacial lake. Under scenario 1 (1974-2015 thinning replicated by 2100), 0.6 km 2 of the glaciers surface will be below the hydrological base level of the glacier by 2100 (Fig. 8). The majority of this area 415 will be located within 1 km of the glacier's terminal moraine, although some small areas further up-glacier will also sit below the hydrological base level by 2100 due to the glacier's inverse ablation gradient (Fig. 7). Under scenario 2 (1974-2015 thinning replicated by 2056, 2097), up to 1.33 km 2 of the surface of glacier RGI60-15.09475 will sit below the hydrological base level of the glacier by 2100. In addition to the large area proximal to the terminus of the glacier which will sit below the hydrological base level, a large portion of the glacier surface above the overdeepening identified by GlabTop will also have become 420 susceptible to supraglacial lake expansion (Fig. 8c).

Discussion
The results from this study demonstrate how, on the primary basis of remotely sensed datasets and GIS tools, GLOF risk management planning at the basin-scale can be expanded to consider new threats that may develop in the future. In doing so, this study has taken established approaches for lake susceptibility assessment (GAPHAZ 2017) and outburst modelling (Westoby et al., 2014) and applied these approaches for the first time to consider also an outburst scenario from a potential 435 future lake. To the extent possible, the assessment was based on freely available data and imagery. However, in steep, complex topography such data can have limitations, and a high-resolution DEM derived from Pleiades imagery was required to achieve accurate GLOF modelling results for Poiqu River basin. While not intended to substitute the type of comprehensive model and field-based hazard mapping that needs to support decision-making (e.g., Frey et al., 2018), the results from this study provide an intermediary step for risk management planning. Using the tools and approaches demonstrated here, authorities can 440 effectively bridge the knowledge gap between the known existing threats to which they must immediately respond, and those potential, yet poorly constrained threats that are anticipated emerge in the future.
For the Poiqu basin, these results come at an opportune time, given that local authorities over the past year appear to have initiated major engineering work at Jialongco (Fig. 9). In principle, the focus of authorities on Jialongco is supported by the 445 results of this study, which indicate that the lake poses the greatest immediate threat to the village of Nyalam, and, under a worst-case scenario, will lead to significant flood heights and velocities downstream in Nepal. While less likely, a large https://doi.org/10.5194/nhess-2021-167 Preprint. Discussion started: 28 June 2021 c Author(s) 2021. CC BY 4.0 License.
outburst from Galongco would result in a higher intensity flood event, although full drainage of the lake volume is not considered feasible. In fact, despite its rapid expansion over recent years (Wang et al., 2015b;Zhang et al., 2019), the maximum potential flood volume of Galongco, as limited by the dam geometry and potential height of the moraine breach, would likely 450 not have changed. Nonetheless, our simulations reveal a potential peak discharge under a worst-case scenario that is more than 10 times larger than indicated by previous modeling (Shrestha et al., 2010), suggesting that previously estimated potential property losses of up to US$197 million in downstream communities of Nepal are a lower limit to what could feasibly occur.
Despite the threat the lake poses, the focus at Jialongco on hard engineering strategies to reduce GLOF risk could prove both 455 costly and inefficient, if not complimented by a more comprehensive and forward looking strategy. Although the overall strategy of authorities is not clear, the recent removal of much of the frontal moraine and apparent enhancement of the outlet channel has had only a minimal effect on the overall lake size. Conversely, the resulting removal of the dam freeboard now leaves the lake more susceptible to an overtopping wave, caused by a potential ice avalanche or instability of the lateral moraine wall. In general, increasing exposure of people and assets is seen as a main driver of disaster risk in mountain regions (Hock 460 et al., 2019), and this is clearly evidenced through the rapid increase in built infrastructure upstream of Nyalam over a twoyear period, directly within the high intensity zone of potential GLOF paths (Fig. 6). Significant and permanent lowering of the water level in Jialongco would reduce the threat to these buildings from an outburst from this lake, but similar action would need to be repeated at Galongco and as new lakes emerge in the future, in order to minimise potentially larger, albeit, lower probability threats. We would therefore argue that a focus on early warning systems coupled with effective land use zoning 465 and capacity building programs (e.g., Huggel et al., 2020), would provide a more effective, ecologically responsible, and forward looking response strategy, reducing the risk not only from an outburst from Jialongco, but also future-proofing against large outburst scenarios from Galongco or potential new lakes that may develop in the future.
While GlabTop and other similar modelling approaches (see Farinotti et al., 2019a) have been widely used to anticipate future 470 glacial lake locations and assess related risks and opportunities (e.g., Farinotti et al., 2019b;Haeberli et al., 2016a;Magnin et al., 2015), large uncertainties remain as to if and when specific overdeepenings will transition into lakes. In this study, we have focussed on a very large overdeepening positioned beneath a flat, heavily debris covered glacier tongue -a classic geomorphological setting in which large proglacial lakes typically develop (Benn et al., 2012;Haritashya et al., 2018), and analogous to the setting of Galongco. Coupled with the fact that conditions at the surface of the glacier have already allowed 475 supraglacial lakes to form in the ablation zone of the glacier, there can be a high degree of confidence that a future proglacial lake will develop in this location, trapped behind the prominent terminal moraine. The extrapolation of measured thinning rates over the glacier (Fig. 7) allowed for the estimation of when a glacial lake may begin to develop within the boundary of the overdeepening beneath the glacier (Fig. 8). If the acceleration in thinning of the glacier which has occurred over the last four decades is replicated by 2100 (Scenario 1), or over an equivalent time period to that examined by King et al. (2019King et al. ( ) (1974 2015-Scenario 2), 0.6-1.3 km 2 of the glaciers surface will sit below the hydrological base level of the glacier and therefore will likely host supraglacial meltwater. Under scenario 1, supraglacial lake area equivalent to the current area of Jialongco will be replicated on glacier RGI60-15.09475 by 2100, and by ~2067 under scenario 2 (Fig. 8). These two scenarios of thinning may still represent a conservatively slower trajectory of lake development on this glacier. Both the development of extensive supraglacial ponds and ice cliff networks and the transition of a supraglacial lake to a full depth proglacial lake can increase 485 the overall thinning rate in the ablation zone of debris-covered glaciers (King et al., 2020;Mölg et al., 2020;Thompson et al., 2016). Our simple extrapolation of current thinning rates and patterns does not account for the initiation or expansion of these https://doi.org/10.5194/nhess-2021-167 Preprint. Discussion started: 28 June 2021 c Author(s) 2021. CC BY 4.0 License. ablative processes. Therefore, we would rather expect greater thinning than our results predict in the lowermost ~1.5 km of the glacier over coming decades once a substantial amount of meltwater has ponded at the glaciers surface. In general, the results from this study suggest that hazard mapping and land use planning that accounts for worst-case outburst 495 threats from Jialongco, and particularly Galongco, would largely remain valid for the future lake scenario, given only small differences in the potential built area affected (Fig. 6). In other words, despite uncertainties in the potential speed of future lake development, the opportunity cost of extending hazard zones and related planning to include areas potentially affected under future scenarios is minimal, particularly when considering the protection of critical infrastructure and services. Likewise for early warning, simulations show that warning times could be reduced by up to 20 minutes for downstream communities in 500 Nepal, under a future outburst scenario. Hence, in order to ensure warning systems and response strategies remain robust over the longer-term, it is recommended that authorities consider such future scenarios in the design phase, under the philosophy of preparing for the worst, while hoping for the best. Particularly in complex transboundary regions requiring communication and collaboration between countries, minutes lost or gained can be critical for effective early warning and evacuation.

Conclusions 505
The Poiqu basin in the central Himalaya has been well established as a hotspot from which transboundary GLOF threats can originate. In the current study, we have focused on two lakes that directly threaten the Tibetan town of Nyalam and areas downstream, comparing the likelihood, potential magnitude, and impacts of large outburst events from these lakes. In addition a future scenario has been modelled, whereby an outburst was simulated for a potential new lake, anticipated to form upstream https://doi.org/10.5194/nhess-2021-167 Preprint. Discussion started: 28 June 2021 c Author(s) 2021. CC BY 4.0 License.
of Jialongco. For all lakes, worst-case scenarios were simulated, assuming release of the full potential flood volume of the lake 510 as defined by the maximum breach height of the moraines. The study has recognised that:  Jialongco, although smaller in size, poses the greatest current threat to Nyalam and downstream communities, owing to the high potential for an ice avalanche to trigger an outburst, feasibly leading to release of the full potential flood volume. Even though engineering work has started the threat persists as the lake volume remains large and the reduced 515 dam freeboard now leaves the lake more susceptible to an overtopping wave.
 An ice avalanche on its own is considered unlikely to initiate a large outburst from Galongco, although a lowprobability/high impact event involving a catastrophic rock/ice avalanche into the lake should be considered as a realistic scenario, particularly given the seismic activity in the region.
 A future scenario, involving the anticipated new lake would lead to flow depths and velocities in Nyalam that exceed 520 either of the current lakes, and the peak wave would reach the border with Nepal up to 20 minutes faster than for the current lakes.
 While previous studies have focused on rapid lake expansion in the region, for the town of Nyalam, it is rather the expansion of infrastructure directly within the high intensity flood zone of both current and future lakes that has significantly increased GLOF risk levels. 525 On the basis of these findings, a comprehensive and forward-looking approach to disaster risk reduction is called for, combining early warning systems with effective land use zoning and capacity building programs. Hard engineering strategies that address only the hazard source are a socially and environmentally less desirable option, as such strategies do nothing to address underlying risk drivers of exposure and vulnerability, and are likely unsustainable in the face of ongoing environmental 530 changes.