Using intervention time series analyses to assess the effects of imperfectly identifiable natural events: a general method and example