A geospatial modelling for glacial lake outburst floods susceptibility analysis in the central Himalayas

In recent decades, anthropogenic factors have driven climate change, resulting in the rapid retreat of glaciers worldwide Rounce et al. (2023). In the Himalayan region, glacier retreat has led to the formation of precarious high-altitude glacial lakes (HAL), posing threats (Taloor et al. 2022) such as glacial lake outburst floods (GLOFs). The fragile Himalayan ecosystem faces additional risks from earthquakes, large infrastructure developments, and human activities in downstream regions (Ukita et al. 2011). The presence of glaciers and snow in the Himalayan Mountains is crucial for terrestrial geosystems, controlling water energy, wind circulation, and maintaining the complex interactions among climate and hydrology Zhong et al. (2021). The high mountains exhibit distinct spatiotemporal weather patterns, influenced by rising temperatures and affecting cryosphere dynamics (Gilany & Iqbal 2020; Abbas et al. 2021)). Glacier and snow formation, along with englacial hydrothermal processes, water flow dynamics, mass balance dynamics, and frontal ablation, significantly influence weather patterns in various environmental settings (Guan et al. 2015). Due to their unique characteristics and significant impact on local climates, glaciers receive global attention. Changing regional environmental patterns and extreme weather events, driven by human activities, indicate a rapid response to climate change (Abbas et al. 2022). The retreat of glaciers and variations in subsurface water levels due to climate change impact water availability for irrigation, agriculture, and hydropower production, affecting both economic and environmental aspects of life. Most glaciers worldwide are melting rapidly due to ongoing temperature increases (IPCC 2021), with sea levels predicted to rise significantly (Sakai et al. 2000; Allen et al. 2016; Emmer et al. 2022; Livingstone et al. 2022). The primary cause of glacial lake development is glacier melting and sediment movement (Hagg et al. 2021; Wangchuk et al. 2022). Unexpected glacial lake eruptions, caused by collapsing ridge barriers, can result in natural disasters like flash floods, large-scale water and sediment movements, and debris flows downstream (Stokes & Clark 2003). These events contribute to glacier mass loss and pose a significant risk of GLOFs, which can severely damage fragile alpine ecosystems. It is crucial to monitor these lakes using state-of-the-art intelligent energy systems (Kimothi et al. 2022) to mitigate risks. The increased variability of Himalayan water sources and potential flash floods from lakes necessitate prioritizing research into nonlinear dynamic climate models to achieve environmental sustainability in managing anthropogenic surface heat fluxes in the central Himalayas (Seinfeld & Pandis 2016; Kimothi et al. 2023a). The collective impact of regional human activities and rising temperatures frequently influences regional climate, leading to glacier melting in the Indian Himalayan sensitive high mountainous environments (Ashraf et al. 2012; Bosson et al. 2019; Anees et al. 2022; Kimothi et al. 2023b; Rongali et al. 2024).

Due to the Himalayas’ inaccessibility, hydrodynamic analysis is often limited to lower mountain ranges (Sattar et al. 2020). Devastating events like GLOFs and rock-slope failures have plagued the Himalayas, leading to the formation of new glacial lakes (Fischer et al. 2022); Thapliyal et al. 2023a). Abbas et al. (2023a, 2023b) used global climate models (GCMs) to study changes in winter precipitation in the western Pakistan river system and throughout South Asia in order to predict future changes in precipitation extremes. Maximum daily precipitation index increased along with an irregular declining trend of extreme precipitation over a wide geographic area. Significant challenges with water management and flood protection exist in the Himalayan region's vulnerable areas due to excessive precipitation. Erosion of bedrock slopes and glacier rock-falls increase the danger of lake collapses and catastrophic flooding downstream (Kirschbaum et al. 2019). Understanding the processes and mechanisms behind GLOFs is essential for predicting future hazards (Rounce et al. 2017). The recent flash-flood episodes across the Rishi-Ganga Basin underscore the need for geospatial technology to monitor HAL and develop early warning systems to mitigate natural disaster impacts in mountainous areas (Taloor et al. 2022). The Gya glacier inventories have been studied by Majeed et al. (2021), and GLOF events’ predictions are carried out in trans-Himalayan regions emphasizing the importance of glacial lake inventories to identify vulnerable sites (Arthur et al. 2020; Zheng et al. 2021; Livingstone et al. 2022). Similarly, Rawat et al. (2023) predicted GLOF scenarios employing hydrodynamic modelling with satellite data over the western part of the Indian Himalayas. As glaciers recede and melt, new lakes form near glacier and moraine borders, causing slope destabilization and increasing the risk of lake collapses and rock-falls (Quincey et al. 2007; Jones et al. 2022; Liu et al. 2022). Rapid advancements in glacier research using remote-sensing platforms and methods have been driven by the recognition of glaciers as climate-change indicators and their significant impact on global, regional, and local scales (Thapliyal et al. 2023b). High-resolution satellite synthetic aperture radar (SAR) data have been used to precisely map glacial lakes (Strozzi et al. 2012). Global warming has caused rapid retreat of basin glaciers, with satellite datasets revealing multidecadal changes in the Tibetan Plateau (Zhao et al. 2022; Zhou et al. 2022). Identifying lakes in glacial zones using high-resolution satellite SAR imagery is crucial for hazard assessment (Huang et al. 2011). Satellite imagery and Open-Street-Map data have been used to analyse potential flood risks, emphasizing the importance of remote sensing in studying glaciers and their changes.

The spatiotemporal analysis of HAL can be used to build an early warning system for government officials to notify the local population of an approaching dangerous situation and provide initial information on the appearance of GLOF. Modern technologies such as the Internet of Things (IoT), machine learning, and advanced satellite sensors can be useful in hazard reduction, strategic planning, and understanding the Himalayan region's most vulnerable areas (Kimothi et al. 2022). This study focuses on the influence of GLOF over the Purvi Kamet Mountain in the Alaknanda Basin in the central Himalayas, utilizing hydrodynamic modelling to make recommendations for reducing flood risk and analysing the impact of GLOFs using vulnerability maps and flood flow simulation. The assessment of glacial lakes using multispectral satellite data over the past three decades highlights the challenges of monitoring GLOF-related hazards in downstream areas.

The details of the remote-sensing datasets used in the study are provided in Table 1. The temporal datasets of Landsat-TM and Sentinel 1 and 2B datasets are used to map the lake extent. The Advanced Space-borne Thermal Emission and Reflectance Radiometer (ASTER) global digital elevation model (DEM) provided the terrain data for GLOF modelling. The DEM is open access (https://earthexplorer.usgs.gov/) and available in a spatial resolution of 30 m with 10–20 m vertical accuracy. Despite its lack of sensitivity and specificity, the ASTER GDEM was considered in the remote Himalayan region for a number of valid reasons. It is the most publicly accessible as well as including easy access to difficult terrain and areas where elevation data are sometimes hard to obtain and offers extensive global coverage. Second, alternative high-resolution DEMs might be too costly or difficult to obtain for initial evaluations and extensive research in the Himalayan region. In this case, ASTER GDEM's higher resolution and free availability make it a valuable tool. The normalized difference wetland index approach used the difference in the green and near-infrared bands and was applied for identifying glacial lakes.

The Hydrologic Engineering Centre's River Analysis System (HEC-RAS) program is open source, making it ideal for hydrological research work. Furthermore, the HEC-RAS 1-D turbulence hydrologic simulation was used to estimate the probability of GLOF phenomena and impact analysis in the downstream regions. The maxima of GLOF graphs calculated by the system are congruent with the peaks generated from established empirical correlations in the current investigation. In a hypothetical GLOF incident, the total amount of water discharged varies proportionally to the breach depth.

By assuming a constant velocity, it aimed to simulate both the duration and the geographical expanse of flash floods over the central Himalayan region. This includes various hydrological modelling components, such as flow and drainage basin depth analysis. For the constant flow, hydrological profile estimation, and 2D unstable flow simulation, the lake volume is the single most essential factor in determining the most hazardous zone. In the investigation of hazardous levels, the four primary characteristics are considered: the extent of the lake, outlet position of the basin, frequency of changing size, and lake gradient.

In the context of climate-change impact on the glacial regions, it is highly recommended to monitor the health of glaciers as well as associated hazards. As the GLOFs are abrupt events and are responsible for disastrous floods or sediment flow in downstream regions, they should be analysed for GLOF risk and mitigation purposes. In this context, the HEC-RAS model was applied for the estimation of GLOF hazard risk extent for the low-lying areas of the glacial valley. LANDSAT-1994, LISS-III-2000, LISS-IV-2019, and Sentinel-2-2020 imageries were analysed for the periods of 1994, 2000, 2019, and 2020, and it was observed that the VL was the most potentially dangerous lake among the 11 lakes observed in the study region. The lake was chosen for further analysis in GLOF investigation as it has an extent of 0.29 km². The lake cross-section was generated using DEM, which depicted the dam-break structure and hence was used to model glacial lake outbursts. In addition, cross-sections of the river from the VL to the villages in the lower-lying region were used in the model to the highest elevation of the river cross-section. The river reach is approximately 23 km long with a steep slope at various points making it tedious to model the GLOF hydrograph, especially in this condition. To remove the barriers in creating the GLOF hydrograph, we have the following steps. First, VL has been breached in the model, and the hydrograph is routed downstream.

Global warming has resulted in the creation of multiple high glacial lakes in the Himalayan region. Among the major glacial lakes of the region, the VL has an elevation of 4,676 m above sea level and is located in the central Himalayas.

The GLOF hazard depends on various sets of circumstances such as several rupture levels, rupture sizes, and time of glacial collapse and will be subjected to multiple dynamic simulations. Considering the flat shape of the prefrontal moraine of the lake, hazard evaluation was carried out for the breach depths. Three scenarios are based on varied breach depths of 60, 30, and 15 m.

The basic data were used for providing a conceptual model of VL in the Alaknanda Basin of the Himalayan region. In this study, ASTER DEM data were used to extract river cross-sections to prepare the HEC-RAS simulation and for GLOF routing. Fresh water was considered in the simulation model; however, this is invariably associated with sediments accumulating near lake streams and along the channel. The study also assessed a number of probable GLOF scenarios for VL and its associated influence on the nearby downstream community. The preliminary GLOF hydrograph's vulnerability to the rupture process requires a time of failure (T_f) assessed by evaluating various hypothetical rupture situations with varying times of the collision.

The long-term effects of climate change could be adversely affected by biodiversity changes brought about by glacier retreat IPCC (2021); Thapliyal et al. (2024)). Monitoring the amount of snow cover extent (SCE) on a regular basis highlights the importance of the cryosphere in maintaining river flows brought on by glacier melt. Because of the SCE's diminishing mineral concentration, glaciers are receding, getting smaller, and losing volume, which has an adverse effect on livelihoods, vegetation, and water quality. The specifics of snow conditions and permafrost's function in alpine hydrology are not well understood. This diverse region depends on the sustainable management of its natural resources, which are essential for feeding millions of people and are susceptible to the effects of climate change Chauhan et al. (2023).

Figure A2(a) and A2(b) (Supplementary Information) demonstrates the potentially hazardous lakes derived through Sentinel 1 and 2B datasets, which clearly indicate that the glacial lakes have a larger spatial extent. Sentinel-1 SAR data product has shown the formation of large glacier lakes (GLs); however, in Sentinel-2B data, small emerging water bodies are also observed. As there is a scarcity of field data over these regions, which have complex terrains and accessibility is difficult, in this study, we confirmed the volume and thickness of the largest observed GL using empirical relationships (Equations (3)–(5)). The breach parameters for the model were determined using Froehlich's method, and the results are reported in Table 2.

The first scenario, S1, was created in the model with breach depth (h_b = 60 m), breach width (B_w = 80.83 m), and time of failure (T_f = 0.24 h), and the 9,283m³/s discharge obtained is observed as the worst-case scenario in GLOF. The model parameters are as follows: h_b = 30, B_w = 64.40 m, T_f = 0.32 h, with the discharge being half the volume of the lake, generating a peak of 5,129 m³/s attained in 11 min after the initial breach event in Scenario S2. Similarly, using a set of breach parameters h_b = 15, B_w = 51.59 m, T_f = 0.41 h, it is observed that 1/4 of the total lake volume will release a maximum discharge peak of 1,306 m³/s in 18 min after the initial glacial rupture in Scenario S3.

The lake surface was analysed for years 1994, 2000, and 2020 by visual interpretation and recorded; the total lake surface areas were measured to be 0.15, 0.17, and 0.29 km², respectively. Geospatial (satellite data) monitoring of the study area reveals the growth of VL in the last three decades. Since the lake is in its pro-glacial phase of evolution, future growth is expected as the glacier continues to retreat.

The simulation model approaches complex processes, and modelling methods can be used to assess the risk of potential GLOFs. Here, we attempted to simulate the GLOF start, propagation, and potential impact of VL in the prospect of the GLOF scenario. A numerical model for GLOF simulation provides the current status, dam-failure situations, and flow propagations in the central Himalayan region.

From the position of the lake and the closest settlement downstream, the channel's inundation and stream mean velocity are found to be inconsistent and varied according to the width of the stream. To measure the sensitivity of the 2D routing of the prospective moraine, the breach was achieved by a series of validated coefficients (N). The inundation depth and velocity maps are divided into five groups. Analysis of GLOF model sensitivity depends on various factors such as monitoring glacier/snow mass flow rate (maximum and mean) for different depths and lake boundary conditions. Additionally, the inundation depth and top width were computed for various cross-sections in relation to the mean velocity and the mean speed. It was determined across cross-sections that the modelled GLOF flooding depends on the flow rate of the particular stream during scenarios S2 and S3, while Scenario S1 shows cumulative impact.

The vulnerability analysis of GLOF in glacial lakes especially in the Himalayan region is a critical input for ecological sustainability. The modelling of GLOF events over the Himalayan high ridges is a highly required parameter for early warning systems to mitigate natural disasters. In this context, the present research work deals with the GLOF vulnerability of VL in Uttarakhand, which has not yet been explored or is even neglected. Here, we integrate the most relevant and advanced datasets to evaluate the analysis of the possibilities of flash floods downstream. However, ground information related to the area of a glacial lake, the water storage volume, and so forth is needed to improve the accuracy of the predictions. Especially, the use of high-resolution datasets such as RapidEye, Pleiades, PlanetScope, and Google Earth can improve the output of the model.

For the various scenarios, the flow velocities along the channel going from VL to the adjacent downsteam settlement are assessed and demonstrated in Table 2. Spatially disseminated stream inundation depth and mean depth (m) along the VL channel in most hazardous scenarios (S1, S2, and S3) have been explored in this study. In S1, S2, and S3, 100%, 50%, and 25% volume of discharge will be released and the time of breach will be 0.24, 0.32, and 0.41 h, respectively. These results demonstrate how the geographical patterns of depth range and velocity distribution from VL to the furthest habitation downwards are influenced by various roughness coefficients (N) throughout the channel. A vulnerability map indicates that the sites closest to the river are among the most susceptible. Due to dam breaches, GLOF can move in potentially devastating masses (mixed water and silt) at high discharge velocities for distances of a few to hundreds of kilometres. Numerical modelling and simulation models for complex systems may be used to assess the danger of anticipated GLOFs. The variances in flow depth, as well as the maximum and mean flow speeds for varied downstream boundary conditions, are used to assess the model's response. In connection to mean velocity and mean speed, the inundation depth and top breadth were also calculated for various scenarios. This may enhance the possibilities of reducing causalities during flash-flood events as well, as settlements along the riverside are under high floodplain risk zones. It is highly recommended to use real-time monitoring systems for glacial health as well as meteorological datasets with the early warning system to reduce the impact of flash-flood disaster on regional livelihood.

Because of the existence and growth of glacial lakes from the Sikkim Himalaya, the Himalayan region is particularly vulnerable to GLOF, which can destroy infrastructure and result in fatalities in the downstream zone. Numerous geographic factors, including the glacial lake's extension, affect how vulnerable it is to GLOF. By utilizing a variety of topographic parameters, Hazra & Krishna (2022) tried to assess Lake Shako Cho's GLOF risk in terms of real growth.

Using object-based categorization (OBC) and analytical hierarchy process (AHP), Mohanty & Maiti (2021) investigated the probabilities of GLOF across the vast Himalayan region. Throughout the Indian Himalayan zones, 60, 164, and 3,974 lakes were determined and marked to be at high, medium, and lower risk of GLOF. For precision, OBC and AHP were integrated with the top 15 predictors. The AHP method prioritized avalanches, lake-size change rate, moraine-dam steepness, lake freeboard, and earthquakes. Independently, 107 and 224 notable lakes were identified using OBC and AHP approaches. Combining the GIS overlay analysis techniques results in 60 important lakes, followed by 164 and 3,974 glacial lakes with a moderate to lesser response to GLOF. In the Himalayan ranges, almost 95% GLOF-probable lakes are found in the north-eastern part, i.e. Kosi, Pelkhu, and Tista river basins, in Sikkim, Everest, Bhutan, and the Langtang areas, respectively. Similarly, in the western Himalayan region, it is found that the glacial lakes are prominently increasing in numbers as well as width.

A comprehensive analysis of the GLOF is required to study sediment transportation using empirical models and geological approximations. However, modelling is quite difficult as there are limitations in obtaining crucial input parameters conventionally. To overcome these limitations, use of satellite dataset inputs with the hydrodynamic model is highly recommended to simulate and analyse GLOF for future hazard assessments. Real-time glacial area monitoring is advised, particularly for GLOF modelling, as this could help to protect natural resources, limit the effects of natural disasters, and save the nation's economy.

The investigation of satellite-based monitoring of glacial lakes in remote mountain locations aims to identify and prioritize disaster mitigation strategies. This study suggests localization and enhancements when reproducing in other places. The outcomes will be helpful for raising public awareness of the potential impacts of climate change in the Himalayas as well as for researchers, planners, and global decision-makers.

In the Indian Himalayas, glacial flash-floods are frequent with devastating magnitude and have a substantial impact on the fragile alpine environment. GLOF hazard was evaluated in the Vasundhara region for this study utilizing the dam-breach and hydrodynamic models. GLOF in VL is investigated and potentially dangerous situations are identified to predict the landslide-related hazard in the downstream regions. GLOF sensitivity-rating maps illustrate that this location is vulnerable to moderate to extreme levels of flash floods. The study has proven the pathway of getting preliminary information and effects of probable GLOF events that may be used to design early warning systems and management plans. Glacier-dominated areas present unique challenges to downstream communities for increased threats of GLOFs, which have significant impacts on regional, social, environmental, and economic systems. Global-warming-derived climate change and extreme events, such as increasing instability of high mountain slopes, which is a matter of concern, should be addressed. Based on the rupture process, which necessitates a time of failure (T_f) determined by analysing several rupture scenarios with different collision timings, the uncertainty in the GLOF vulnerability analysis is determined. There are various factors impacting the uncertainties arising in outburst scenarios, such as climate dynamics, remote-sensing data limitation, and hydrological and geomorphic factors. The complex processes that govern the formation of the Himalayan glacial lakes, such as glacier dynamics, ice-melt rates, and moraine stability, are shrouded in an abundance of unexplored territory. Apart from this, there are uncertainties in environmental parameters such as temperature fluctuations and precipitation patterns; it is challenging to anticipate the quantity and the size of glacial lakes throughout time. While using remote-sensing data has its benefits, it can also have drawbacks, particularly when dealing with diverse terrain. For example, it can be challenging to locate and characterize small or hidden lakes when there is cloud cover or in hard-to-reach areas. Errors regarding image resolution, temporal coverage, and data-processing techniques might affect the identification and measurement of glacial lakes. These errors may also affect GLOF predictions. The GLOF model needs feedback systems that ensure accuracy in the most crucial model variables. Regional climate models, extreme weather events, precipitation variability, and greenhouse-gas emissions are some of the uncertainties affecting the comprehension of long-term outburst hazards by affecting the rate and extent of glacier retreat and glacial-lake expansion. By addressing these uncertainties, we might enhance risk-management plans and offer a more thorough comprehension of the possible effects of GLOF events in the complex terrain of the Indian Himalayas.

Currently, there is no study available for the highly sensitive VL; therefore, GLOF estimation using the geospatial technique is highly required for early warning systems to mitigate natural disasters, especially in the hazardous potential zones over the trans-Himalayan region. It is concluded that GLOF vulnerability analysis using HEC-RAS is a suitable flash-flood modelling tool and must be utilized in vulnerable zones.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.