Diagnosing United States hurricane landfall risk: An alternative to count‐based methodologies

AbstractAssessing hurricane landfall risk is of immense public utility, yet extant methods of diagnosing annual tropical cyclone (TC) activity demonstrate no skill in diagnosing U.S. hurricane landfalls. Atlantic TC count itself has limited skill, explaining less than 20% of interannual variance in landfall incidence. Using extended landfall activity and reanalysis data sets, we employed empirical Poisson modeling to produce a landfall diagnostic index (LDI), incorporating spatially and temporally averaged upper level divergence, relative sea surface temperature, meridional wind, and zonal shear vorticity. LDI captures 31% of interannual variability of U.S. hurricane landfalls and offers physical insight into why indices that successfully capture TC activity fail to diagnose landfalls: there is inherent tension between conditions likely to steer hurricanes toward the U.S. and conditions favorable for TC development. Given this tension, attempting to diagnose, predict, or understand TC count is inadequate for quantifying societal impacts due to landfalling hurricanes.