LAUSR

The impact of anthropogenic forcings on tropical North Atlantic hurricane potential intensity (PI) is evaluated in Climate Model Intercomparison Project 5 models for the period 1958–2005. Eleven models are examined, but only seven models have a forced response that is distinguishable from internal variability. The use of discriminant analysis to optimize detectability does not yield a clear, common climate change signal. Of the seven models with a significant response, one has a negative linear trend while two have a positive linear trend. The trend in PI is not even consistent among reanalyses, although this difference is not statistically significant because of large uncertainties. Furthermore, estimates of PI internal variability have significantly different variances among different reanalysis products. These disagreements between models, reanalysis products, and between models and reanalyses, in conjunction with relatively large uncertainties, highlight the difficulty of detecting and attributing observed changes in North Atlantic hurricane potential intensity.