f A Neural network picker for VSP data

Artificial neural networks (ANN) are presented as a way of automatically picking shear-wave arrival onset times in VSP data. Picking is achieved by utilizing the relative vector modulus of two horizontal component recordings as the neural network input. A discriminant function, F(t) , determined by the output of the trained ANN, is then employed to pick the arrival onset. The results demonstrate this neural network architecture is successful after training by a few selected datasets.