Global Tropical Cyclone Landfall Statistics

Cyclones are of most interest to human beings when they make landfall. Yet most studies of tropical cyclones and climate focus on basin-wide statistics, presumably because landfalls are rarer and spatially localized along coasts. Here the spatial and temporal variability of annual tropical cyclone landfall frequency are studied globally, using principal component analysis (PCA) of anomalies in a landfall density function defined using kernel smoothing. The goal is to improve statistical significance by considering all basins simultaneously, as well as to look for the imprint of coherent climate modes in TC landfall statistics. Historical data from 1981 to 2014 from the IBTrACS are used.

The first empirical orthogonal function (EOF) mode explains close to 20% of the variance of the landfall density function anomaly, and has strong negative values over the western Pacific around Taiwan, Philippines, South China, and Vietnam, and weak positive values everywhere else. The second and third EOF modes explain ~10% of the variance each, and both have north to south tripole patterns over the western Pacific and Indian Ocean. The higher order the modes are, the more regional the spatial structures become. Physical explanations for the EOFs will be assessed by relating their spatial structures, as well as the temporal structures of the associated principal components, to various climate phenomena (e.g., ENSO and AMO) and to environmental conditions (e.g., large-scale circulation and sea surface temperature) which are known to affect regional TC track and landfall patterns.