Project Information

Project Description

Modelling of Cascading Hazards in Naturally-dammed Lakes
Abstract:Naturally-dammed lake outburst floods are sudden, catastrophic events releasing high volume of water and sediments with high peak discharges in the mountainous region. The lake outburst results from the interaction of the underlying series of natural events affecting one upon another, also known as cascading hazards. Insights of the cascading hazards can be understood by modelling them in an integrative way. Among several processes involved in the lake outburst flood, this study focusses on integrated modelling of three major cascading processes: slope stability, impulse wave and flood. The main objective of this research is to evaluate the outburst flood hazard modelling of naturally-dammed lakes by integrated process chain modelling under different scenarios which aim to support the disaster risk mitigation planning. For this, Lake Sarez and Lake Rivakkul were chosen as two case study areas from the mountainous region of Tajikistan. Three scenarios were designed for Lake Sarez and two scenarios were designed for Lake Rivakkul. In Scenario 1 of Lake Sarez, the volume and depth of potential rockslide were predicted from a physically-based slope stability model, Scoops3D using the geotechnical parameters assigned from the past literature. The outputs were then coupled with the Heller-Hager model which estimated the impulse wave characteristics. The optimistic and pessimistic range of parameter values from past studies were chosen in Scenario 2 and Scenario 3 respectively. These parameter values were assigned to Heller-Hager model to estimate the wave run-up heights, overtopping volume and duration. Simple hydrographs were constructed from the outputs of the Heller-Hager model which could be coupled with the flood model. For the Lake Rivakkul, an ideal rotational slip was assumed for the potential rockslide in Scenario 1 from which the volume and depth were predicted. Heller-Hager was used to estimate the wave run-up heights. In Scenario 2, the dam impounding Lake Rivakkul was assumed to breach and completely empty the lake. The peak discharge and time to peak were estimated from the empirical equations to construct the input hydrograph for the two-dimensional flood modelling in the HEC-RAS. Terrain data was extracted from SRTM DEM (30 m) with the computational mesh size of 50 m. A full momentum scheme with the timestep of 5 seconds and a uniform roughness coefficient of 0.06 was used. The resulting flood was routed at 24 km and 60 km downstream of the lake outburst till Rivak village and Khorugh respectively. Scenario 1 and Scenario 3 generated the run-up heights that overtopped the Usoi dam impounding Lake Sarez in the expense of several limitations. However, the Usoi dam was considered stable by numerous studies so downstream flood modelling was not done but the hydrograph was constructed from the overtopping volume and duration demonstrating its plausibility of coupling with the flood model. Scenario 1 of Lake Rivakkul did not generate enough run-up height to overtop the dam therefore, it was not able to couple with the flood model. In Scenario 2 of Lake Rivakkul, the peak flood reached at the Rivak village and Khorugh at 2.25 hours and 4.45 hours respectively. The average inundation depths at these places reached about 3.6 m and 4 m accordingly. The results of the chain modelling were represented only as the first-pass hazard assessment. Uncertainties and limitations of each process models along with the challenges of modelling the lake outburst hazard chain were highlighted. Modelling of the lake outburst hazard cascade will inherit some uncertainties and quantification of uncertainties are recommended for a range of acceptable results. To overcome these challenges in the data-scarce mountainous environment, recommendations were made regarding the selection of advanced numerical models complemented by field investigations and the future use of physically-based multi-hazard models which can analyse the entire event within a single run.