Prediction of breach parameters caused by failure of landslide dams

Dams provides great advantages for living creatures but they also pose a threat because breaching of both landslide dams and constructed dams generate potentially devastating floods. Prediction of breach parameters (breach width, breaching time and peak discharge) is very important to understand and to avoid such threats of dam failures. In this study, 1044 landslide dam failure cases are examined. After missing information are discarded, new empirical models are developed by the help of gene expression programming (GEP) technique for breach width, breaching time, peak discharge and a different parameter from the literature which is time to failure i.e age of landslide dams at the time of failure. 75% of data set are used for training and remaining are used for testing of GEP models. Results obtained by models are compared with the existing models in literature in graphical form. Statistical results are also given for comparison. Since obtained R2 values are quite high and RMSE values are lower for all breach parameters, GEP model in this study gives more powerful results for landslide dams with respect to the existing models in the literature. Key Words: Landslide Dams, Breach Parameters, Dam Failure, Dam Safety, Genetic Programming