Satellite observation and modeling of operating reservoir-dam systems

Abstract

Reservoir and dam systems are infrastructures for water management and power generation. Various natural processes (e.g., extreme rainfall and flooding) and anthropogenic activities (e.g., maintenance and operation strategies) may trigger different natural hazards (e.g., earthquakes and dam failures) in a catchment area. To date, there have been more than 40 earthquakes with M > 4, triggered by impoundment and subsequent reservoir operation. Over 1600 dam failure events have occurred since the 1850s, and more than 10 catastrophic events have occurred worldwide in the last five years (2016-2021). Therefore, effective monitoring techniques for the dam body and nearby banks, and knowledge of regional tectonics and rheology are critical for disaster prevention. Earth observations from space, especially space geodetic data (Interferometric Synthetic Aperture Radar (InSAR)), can track the surface deformation of a dam and the surrounding lakeshore. This indicates the surface stability in the reservoir-dam system and provides constraints and insights into geophysical and geotechnical processes. However, very few previous studies have focused on seasonal or transient deformations and their interaction with the hydrological loading in reservoir-dam systems; and embankment deformation in relatively small shapes (e.g., hundreds of meters along the axis). Motivated by these limitations, multi-remote sensing data and numerical models were integrated in this study to monitor four dam-reservoir systems and understand the interaction between deformation and hydrology, structural engineering, and seismology. First, we investigated the coseismic deformation near Mangla Reservoir, Pakistan, to explore the causality between reservoir impoundment and the 2019 Mw 5.7 New Mirpur earthquake. The results indicated that the abnormal 2019 summer monsoon applied external loading stress to a blind thrust, promoting fault failure. Second, we studied the seasonal loading-induced deformation near Tehri Reservoir, India. The good agreement between two independent space geodetic datasets (i.e., Global Navigation Satellite System (GNSS) and InSAR) and the Earth loading model suggested that the anthroponomic filling and emptying cycles modulate the elastic response near the lakeshore. Third, we proposed the RSDHI framework, which is a remote sensing-based solution for dam-related hazard investigations. We applied and validated the RSDHI framework to the 2020 dam failure event in Sardoba Dam (Uzbekistan). We found that even without in situ data, the InSAR and other remote sensing techniques can facilitate rapid accident investigations. The results suggested that secondary consolidation controlled dam deformation before the failure event, and a significant differential settlement may result in the final collapse. Finally, to extend the remote sensing workflow used for dam monitoring, we studied the dam failure in a complex hydrological project, the Xe Pian-Xe Namnoy (XPXN) project in Laos, which has multiple earth-fill and rock-fill embankments. The mixing method including Independent Component Analysis (ICA) and numerical models was used to separate different components of dam deformation before the 2018 dam failure. While the collapsed saddle dam shows an accelerating displacement, the similar deformation pattern on the other two remaining saddle dams reveals ongoing risks. In summary, this dissertation holds the promise of contributing to elucidation of different deformation patterns and mechanisms in reservoir-dam systems, using multi-sensor satellite data and numerical modeling. This study also contributes to a better understanding of Earth's rheology and seismic activity, and can be extended to a broader range of reservoir-dam monitoring for disaster mitigation.