Reconstructing glacial lake outburst floods in the Poiqu River basin, central Himalaya

The Poiqu River basin in central Himalaya is a transboundary basin between China and Nepal that has experienced rapid glacier ice loss and glacial lake growth, some of which have drained in catastrophic glacial lake outburst floods (GLOFs). Here, we focus on the outbursts of two glacial lakes in this basin, Gongbatongsha Co in 2016, and Poiqu No.1 in 2002, and integrate multiple data sources and approaches to improve our understanding of past GLOFs and project the magnitude of potential future GLOFs. We identify their potential triggers and assess downstream damage by combining remote sensing observations and field surveys. We find that after the outbursts, Gongbatongsha Co shrank by 0.14 ± 0.003 km² (∼91 %) in 2016 and almost disappeared, and Poiqu No.1 shrank by 0.05 ± 0.001 km² (∼80 %) in 2002 but then expanded by 0.10 ± 0.002 km² (∼670 %) by 2022. Our study identifies ice avalanches from the parent glaciers as the dominant trigger of the two GLOF events, and clarifies that the 2002 GLOF event stemmed from Poiqu No.1 instead of the reported Jialong Co. We also reconstructed the flow depths, velocities, and inundation areas using the hydrodynamic models HEC-RAS and FLO-2D. For Gongbatongsha Co, the models suggest that the discharge peaked at ∼550 m³s⁻¹ about 15 min after the outburst, and the flood reached the settlement of Zhangmu with a peak flow of ∼1600 m³s⁻¹ about 8.5 h later. The peak discharge during the GLOF from Poiqu No.1 was ∼2700 m³s⁻¹. The flood reached the town of Nyalam 1.3 h later, with a peak discharge of ∼1100 m³s⁻¹. The simulation of a future GLOF from Poiqu No.1 shows significant potential impacts to Nyalam town and other downstream communities due to the high flow depth. A large GLOF scenario for Poiqu No.1 would result in a peak discharge of more than 2500 m³s⁻¹, with the flood wave reaching Nyalam in less than 40 min. We show that very high-resolution digital elevation models such as obtained from uncrewed aerial vehicles perform better in flood simulation. Our approach can be applied to potentially dangerous glacial lakes in other high mountain areas.