Statistical modelling of tropical cyclones’ longevity after landfall in Australia

Abstract

Most of the devastation caused by a tropical cyclone occurs on land, and therefore, its longevity after landfall is of critical importance. Published literature identifies many factors including inland environmental characteristics that influence this longevity as well as power dissipation rate. These have been studied in this research in the context of tropical cyclones that hit Australian coasts during the period 1984-2010. For obvious reasons, tropical cyclones which manifested recurrence or multiple landfalls have been excluded. After applying several statistical tools to the observed data, it has been found, from the list of variables identified in literature related to tropical cyclones, that storm intensity at landfall, translation speed, relative humidity, surface temperature, upper level divergence, and surface roughness exhibited statistical significance. However, stepwise regression retained only surface roughness and central pressure which yielded a coefficient of determination of 76 percent during calibration and 59 percent during validation with 60%/40% split in data. The influence of surface roughness is well understood, but as yet, no consistent metric for the purpose of tropical cyclones’ propagation exists, and therefore, this paper introduces a method of assigning surface roughness based on terrain characteristics.