Revisiting the problem of sediment motion threshold