Statistical modelling of tropical cyclone tracks: a comparison of models for the variance of trajectories

TL;DR

This paper develops and compares statistical models for the variance of hurricane trajectories, finding that an anisotropic correlated variance model best predicts future hurricane positions.

Contribution

It introduces a semi-parametric approach to model the variance of hurricane track deviations, including isotropic, anisotropic uncorrelated, and anisotropic correlated models.

Findings

Anisotropic correlated model outperforms others in prediction accuracy.

Variance modeling improves understanding of hurricane trajectory fluctuations.

The approach uses a jack-knife method for optimal parameter estimation.

Abstract

We describe results from the second stage of a project to build a statistical model for hurricane tracks. In the first stage we modelled the unconditional mean track. We now attempt to model the unconditional variance of fluctuations around the mean. The variance models we describe use a semi-parametric nearest neighbours approach in which the optimal averaging length-scale is estimated using a jack-knife out-of-sample fitting procedure. We test three different models. These models consider the variance structure of the deviations from the unconditional mean track to be isotropic, anisotropic but uncorrelated, and anisotropic and correlated, respectively. The results show that, of these models, the anisotropic correlated model gives the best predictions of the distribution of future positions of hurricanes.