Statistical significance of trends in monthly heavy precipitation over the US Academic Article uri icon


  • Trends in monthly heavy precipitation, defined by a return period of one year, are assessed for statistical significance in observations and Global Climate Model (GCM) simulations over the contiguous United States using Monte Carlo non-parametric and parametric bootstrapping techniques. The results from the two Monte Carlo approaches are found to be similar to each other, and also to the traditional non-parametric Kendall's test, implying the robustness of the approach. Two different observational data-sets are employed to test for trends in monthly heavy precipitation and are found to exhibit consistent results. Both data-sets demonstrate upward trends, one of which is found to be statistically significant at the 95% confidence level. Upward trends similar to observations are observed in some climate model simulations of the twentieth century, but their statistical significance is marginal. For projections of the twenty-first century, a statistically significant upwards trend is observed in most of the climate models analyzed. The change in the simulated precipitation variance appears to be more important in the twenty-first century projections than changes in the mean precipitation. Stochastic fluctuations of the climate-system are found to be dominate monthly heavy precipitation as some GCM simulations show a downwards trend even in the twenty-first century projections when the greenhouse gas forcings are strong. 2011 Springer-Verlag.

published proceedings


author list (cited authors)

  • Mahajan, S., North, G. R., Saravanan, R., & Genton, M. G.

citation count

  • 9

complete list of authors

  • Mahajan, Salil||North, Gerald R||Saravanan, R||Genton, Marc G

publication date

  • April 2012