[Back]

S. Hartl:
"Analyse der Felslawinen Frank Slide und Vals mit Hilfe des Computercodes r.avaflow.";
Supervisor: A. Preh; Institut für Geotechnik, 2019; final examination: 2019-10-25.

The master thesis deals with the numerical modeling of rock falls and rock avalanches by means of the software "r.avaflow". The most accurate possible modelling of the spreading of rapid flow-like landslides (mass flows) such as rock avalanches using numerical methods is of great importance in determining the hazard and the risk to infrastructure and human life.
At present, there are two different approaches for the calculation of such flow-like landslides: on the one hand, continuum mechanics or equivalent fluid methods (e.g. DAN, RAMMS, r.avaflow) and on the other hand discontinnum mechanics methods (for example, the DEM method) are available. Most case studies as well as back calculations are based on continuum mechanics approaches and the Voellmy rheology to describe the runout processes of landslides. Prediction parameters for estimating the ronout of landslides exist for this combination. Developed as part of a research project funded by the German Research Foundation (DFG)and the Austrian Research Fund (FWF), r.avaflow is a new comprehensive continuum-mechanical approach that includes an automated methodology for back-calculating events and evaluating computational results against oberservations. r.avaflow can use either the Voellmy-Salm rheology or the Pudasaini two-phase mass flow model as part of the calculation process. Indices such as True Postive, True Negative, CSI (Critical Success Index) and AUROC (Area Under ROC Curve), which are mainly used in the analysis of statistical distributions, are used to evaluate the achieved model accuracy. The aim of the work is first to evaluate to what extent it is possible to reproduce rockfalls and rock avalanches with r.avaflow using the implemented Voellmy-Salm rheology and secondly, to what extent the results of the computation agree with the results of the continuum mechanics code DAN and the Voellmy rheology. For this purpose, two well-documented events, the rock avalanche "Frank Slide" (Alberta, Canada, 1903) and the rockfall "Vals" (Valsertal, Tyrol, 2017), are back-calculated and compared with DAN.
In the course of the case study Frank Slide, a very close match could be achieved between simulation and the observations of the deposits of the event. A recalculation of the landslide with the parameters of the Voellmy-Salm rheology φb = 6,5° und ξ = 580 m/s2 gives the best
results. In the course of the analysis of the modeled deposits, it can be stated that the runout distance was somewhat underestimated compared to observations and the lateral spread was slightly overestimated. The high accuracy of the r.avaflow model with Voellmy-Salm rheology is reflected in the characteristics CSI and AUROC.
The rockfall Vals has a much smaller dimension in terms of the impacted area, the detached rock volume and the differences in elevation than the rock avalanche Frank Slide. Therefore, the
influence of the resolution of the digital elevation model in the Z-direction on the calculation results was investigated. For this purpose, two different elevation models of the basal topography
were used for the back-calculation: in a first model variant, the elevation model provided was used with a resolution of 1 m in the Z-direction, in a second model variant a elevation model smoothed by the method of interpolation was used. The results of the case study Vals showed in both model variants only limited agreement of simulation and observation regarding the modeled deposits. In addition, the two model variants each differ in the parameters of the Voellmy-Salm rheology.
The results of the investigations carried out suggest that r.avaflow is well suited for the back-calculation of large rock avalanches such as the Frank Slide. For smaller volumes, as in the case Vals, the results with regard to the back-calculated Voellmy parameters and the modeled deposits are not clear. Prediction parameters of the same rheology determined with DAN can only be used to a limited extent as input parameters for the calibration of the model in r.avaflow.

Felssturz, Felslawine, numerische Modellierung, r.avaflow