Adaptation strategies for high order discontinuous Galerkin methods based on Tau-estimation