University of Twente Student Theses


Quantifying uncertainties in Landslide Runout Modelling

Zahra, Tuba (2010) Quantifying uncertainties in Landslide Runout Modelling.

[img] PDF
Abstract:The most essential part of landslide hazard assessment revolves around the prediction of the failure and the post failure movement or the runout of landslides. This approach requires the accurate prediction of the intensity or magnitude of the landslide with special reference to the runout behaviour. The runout behaviour maybe characterised by the quantitative spatially distributed runout parameters like, runout distance, runout width, depth of the moving mass & deposited material, velocity, pressure, volume of the material, scouring processes involved and saturation. In order to get a detailed understanding of the dynamic characteristics of debris flow often numerical methods are applied. A continuum method numerical solution includes conservation equations of mass, momentum, energy and describes the dynamic motion of debris along with the rheological models that determine material behaviour of the landslides (Dai et al. 2002). Attempts to streamline and create a framework for an acceptable method of uncertainty quantification for landslide runout modelling are largely lacking. In this research, an attempt is made to quantify the spatial uncertainty (including extent, depth and volume) that emerges from numerically modelling landslide runout. This research uses a numerical model called MassMov2D, which is a two dimensional numerical model of mass movement runout over complex topography to model the debris flow kinematics based on the depth-averaged form of the equation of motion for a fluid continuum applying the Voellmy rheology. The model has been implemented using the raster based environmental modelling language PCRaster. The observed areal extent, volume and deposit depth were used to back analyse the event and estimate the values for the leading rheological parameters, turbulence coefficient and basal frictional angle. An attempt was made to calibrate the model for souring rate, though a difficult process, to estimate the influence on runout deposit depth and volume. The results for calibration were used to back analyse the event and to adjust the parameters for the observed values of area, depth and volume. A sensitivity analyses of the model revealed that the acceptable range of parameters for which the uncertainty analysis was conducted was in the range of 100-1000 m.sec-2 for turbulence coefficient, a basal frictional angle of 9- 34 degrees and a synthetic range was chosen for the scouring rate. It was observed that the variability in the deposit depth was found to be nearer to the observed value of 6.4m with turbulence coefficient values between 200-300 m.sec-2, basal frictional angle values between 24-26 degrees and scouring rate values between 0.0053-0.0095m.sec. The Monte Carlo technique was used to derive random parameter values within these range. Owing to the time taken for each simulation the random number simulation was limited to 100 scenarios. Probabilities for runout reaching each pixel were calculated and also for reaching particular deposit depths (0-2m, 2-4m, 4-6m and above 6m). Maps were generated showing the probabilities of runout in each of the deposit depths. Accurate predictions were found for 62% of the area whereas an area of 23% was considered to be moderately uncertain because there were occurrences of runout observed in this area though they were very low falling in the probability interval of 0.2-0.8.
Item Type:Essay (Master)
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Programme:Geoinformation Science and Earth Observation MSc (75014)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page