• maximum likelihood estimation;
  • model selection;
  • point process data;
  • tornado incidence risk

We propose a point process model with multiplicative risk for the study of tornado reports in the United States. In particular, we implement a rigorous statistical procedure to evaluate whether tornado report counts are significantly related to topographic variability. The model we propose also includes flexible nonparametric components for spatial and seasonality effects. We apply the proposed model and methodology to the analysis of tornado report data from 1953 to 2010 in the United States. Our analysis shows that in addition to the spatial and seasonal effects, the topographic variability is an important component of tornado risk. Copyright © 2013 John Wiley & Sons Ltd