Precursory motion and deformation mechanism of the 2018 Xe Pian-Xe Namnoy dam Collapse, Laos : insights from satellite radar interferometry

Abstract

The sudden failure of hydraulic dams poses catastrophic hydrological risks to local communities and environments. The 2018 Xe Pian-Xe Namnoy Dam collapse in Southern Laos unleashed a flood wall that resulted in more than 70 deaths and left thousands homeless. While the causes of this event were clearly known as geological and engineering issues, the detailed process of dam degradation and potential hazards of other dams in the area remain unclear. In this study, we use radar satellites to investigate the precursory deformation of major dams in the Xe Pian-Xe Namnoy hydropower project. By combining independent component analysis and an Earth loading model to decompose the observed deformation into various time-dependent components, we study the possible deformation mechanism. We identify a precursory vertical deformation acceleration on the collapsed Saddle Dam D reaching up to 13 cm/yr two months before the failure. Two earth-fill saddle dams (E and F) also experienced accelerating subsidence of ∼ 9 cm/yr two months before the dam failure. The deformation pattern of these saddle dams resembles that of the collapsed Saddle Dam D, indicating a critical need for reinforcement interventions before being returned to service. In comparison, the rock-fill Xe Pian and Xe Namnoy Dams were relatively stable before the 2018 failure event. We show that water pressure, consolidation, thermal expansion, and precursory acceleration are the major mechanisms controlling dam deformation. Our results provide additional details for understanding the cause of the dam-break, and also to warn that other saddle dams associated with the project may also break if no further treatment is done. In addition, this study demonstrates the importance of space-borne radar interferometry and hybrid data analysis for early warning and hydrological safety management of small-to-medium-size dams, where in situ monitoring systems are often unavailable.