IAP-24-062
Time-evolving hazard and risk from Glacial Lake Outburst Floods in Bhutan.
Himalayan glaciers are a vital water resource for the ~1 billion people living downstream [1], but these glaciers are melting rapidly as climate warms. Glacier shrinkage causes ice-marginal lakes to grow, which can burst, resulting in potentially highly destructive Glacial Lake Outburst Floods (GLOFs). Bhutan is the most vulnerable country globally to GLOFs because its population, infrastructure, cultural heritage and hydro-power generation capacity lie downstream of glacial lakes [2; Figs. 1 & 2]. Recently, its glacial lakes and downstream infrastructure have grown rapidly [Fig. 2], and have generated GLOFs requiring emergency action [3 & 4; Fig. 3]. Thus there is an urgent societal need to quantify the rising threat posed by GLOFs in Bhutan and the Himalya more broadly.
Numerical modelling can be used to assess potential GLOF characteristics, including flow routing, water depth and arrival time to infrastructure and/or population centres. However, there are many uncertainties in key physical parameters (e.g. dam breach style), so an ensemble modelling approach is often required, to determine the probability of inundation for a range of different physical scenarios. To assess potential impacts of GLOF inundation, information is also required on exposed downstream infrastructure and populations, and, in turn on their vulnerability, which can markedly alter impact. Physical parameters, exposure and vulnerability can all vary significantly over time, but the impact of this temporal variation on overall GLOF risk has yet to be comprehensively assessed and the relative importance of variability in each component is unclear. For example, we expect GLOF risk to lower seasonally, as local populations migrate to lower elevations in winter, but it is unclear how this interacts with seasonal changes in glacial lake volume. Furthermore, it is uncertain when physical and human factors combine to produce risk peaks and how this may vary spatially within a glacial-lake-fed catchment. Thus, this PhD project aims to assess the temporal variation of GLOF hazard and risk in Bhutan, via the following objectives:
Objective 1: Integrate remotely sensed and directly measured data to quantify temporal variability in GLOF hazard, exposure and vulnerability in Bhutan, for a range of timescales.
Objective 2: Utilise data from Objective 1 to conduct ensemble model runs for a range of physical and social scenarios and their variability over a range of time scales.
Objective 3: Quantify the relative importance of temporal variability in each parameter, identify any temporal risk peaks, and identify the timescale / parameter combinations with the greatest impact on GLOF risk.
Click on an image to expand
Image Captions
Figure 1. Location of the Punatsangchu catchment, including the Lunana complex, which contains Bhutan’s largest and most dangerous glacial lakes: Bechung Tsho (b), Raphstreng Tsho (c), Lugge Tsho (d), and Thorthormi Tsho (e). Downstream settlements are indicated with dashed arrows and key sites are numbered, including Punatsangchu hydroelectric project (PHP)-I and -II, which are Bhutan’s largest hydroelectric power plants. Rinzin et al., 2023.,Figure 2. Photo of Bhutan’s highest risk glacial lake, Thorthormi Tsho, which underwent an outburst in 2019.,Figure 3. Photo of settlements downstream of Thorthormi Tsho, which have been recently relocated due to rising GLOF threat.
Methodology
The PhD work will focus on the Punatsangchu catchment [Fig. 1] , as it has the 4th highest GLOF risk globally and the highest in Bhutan. Contingent on project direction, it may be expanded to other high-risk catchments in Bhutan.
Objective 1: A combination of remotely sensed and directly measured data will be used to quantify time-varying GLOF hazard, exposure and vulnerability, at time scales of days to years. This will include information on the hazard from high-resolution remote sensing data (e.g. glacial lake area from Planet labs imagery) and directly measured data (water level sensors installed as part of our broader, funded research project). A primary task will be to identify the datasets needed to quantify time-varying hazard and to identify any data gaps. Fieldwork will then be conducted by the PhD student to address these gaps where possible, e.g. via installation of additional water level sensors. Data on exposed infrastructure and buildings will be collected from Open Street Map and Google Earth imagery, which will be combined with Bhutan census data, to determine numbers of occupants and their demographics (e.g. age). Interviews and questionnaires will then be used to survey residents located in potential GLOF pathways, to determine temporal variability in their exposure, e.g. due to seasonal migration in winter.
Objective 2: The data on hazard and exposure collected for Objective 1 will be used as input for the GLOF modelling. The primary modelling tool will be HEC-RAS [5], which is capable of modelling sudden-onset floods and, crucially, of incorporating data on exposure and behaviours of at-risk populations, via the HEC-LifeSim extension. Depending on project direction, r.avaflow [6] may also be used to explore the entire GLOF process chain for specific, high risk sites from triggering mechanism, through to breaching / overtopping and downstream flood routing. r.avaflow has the advantage of providing greater insights into the possible spectrum of inundation scenarios, but requires more complex inputs than HEC-RAS and does not include human factors, so will be the secondary modelling tool. GLOF ensembles will be run in HEC-LifeSim with a range of time varying input parameters, based on the physical and human data collected in Objective 1, to determine the effect on GLOF inundation and impact. For example, the model will be run with the range of seasonal variations in glacial lake volume observed in the data and again with seasonal variations in population.
Objective 3: Together, the ensemble runs and physical / social data from Objectives 1 and 2 will be integrated to determine which time-varying factors have the greatest influence on GLOF impacts and to identify any temporal peaks in GLOF risk.
Project Timeline
Year 1
Familiarization with research area and thorough review of the existing literature; identification of key datasets and data gaps; initial data analysis; training on GLOF modelling.
Year 2
Complete data analysis to address Objective 1; first field season to address data gaps; initial modelling runs.
Year 3
Complete numerical modelling work; second field season to complete data gaps; finish data analysis.
Year 3.5
Complete any outstanding data analysis and finalise thesis write up. Thesis write up will be ongoing during the first three years, but will be the primary focus in the final 6 months.
Training
& Skills
Training will be provided in handling and integrating different remotely sensed data sources and in using GLOF modelling software. Fieldwork will be undertaken as part of a broader funded research project and training will be provided on both general fieldwork safety and operation and on the specifics of data collection. Broader-remit training will be provided on developing the project, scientific writing and data analysis.
References & further reading
[1] Immerzeel, WW, et al., Nature, 2020. 577(7790): p. 364-69. [2] Taylor, C, et al., Nature Communications., 2023. 14(1): p. 487. [3] Rinzin, S, et al., Journal of Hydrology, 2023. 619: p. 129311. [4] Rinzin, S, et al., Frontiers in Earth Science, 2021. 9. [5] CEIWR-HEC, 2021. Institute for Water Resources, Hydrologic Engineering Center, Davis, USA, 289 pp. [6] Mergili, M, 2017. Geoscientific Model Development, 10(2): 553-569.
