Estimation of Maximum Scour Depth Downstream of Spillways

The problem of accurate prediction of the maximum depth of scour hole downstream of spillways has been based on the experimental studies and the equations developed are mainly empirical in nature. In this paper, prediction of maximum depth of scour downstream of a spillway using ski jump bucket type arrangement with empirical equations, linear regression, M5 tree and with polynomial & RBF kernel based SVM techniques have been attempted. The study shows that relatively less used RBF kernel based SVM scheme has emerged as the most satisfactory on the present data set as compared to empirical equations, linear regression and M5 tree suggested by the previous investigators. The results are also comparing well with regression based study^{[1, 2, 3]}.