Premium
Hurricane Lifespan Modeling through a Semi‐Markov Parametric Approach
Author(s) -
Masala Giovanni
Publication year - 2013
Publication title -
journal of forecasting
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.543
H-Index - 59
eISSN - 1099-131X
pISSN - 0277-6693
DOI - 10.1002/for.2245
Subject(s) - parametric statistics , markov chain , computer science , econometrics , markov chain monte carlo , markov process , event (particle physics) , homogeneous , parametric model , monte carlo method , meteorology , operations research , environmental science , statistical physics , statistics , mathematics , geography , machine learning , physics , quantum mechanics
The estimation of hurricane intensity evolution in some tropical and subtropical areas is a challenging problem. Indeed, the prevention and the quantification of possible damage provoked by destructive hurricanes are directly linked to this kind of prevision. For this purpose, hurricane derivatives have been recently issued by the Chicago Mercantile Exchange, based on the so‐called Carvill hurricane index. In our paper, we adopt a parametric homogeneous semi‐Markov approach. This model assumes that the lifespan of a hurricane can be described as a semi‐Markov process and also it allows the more realistic assumption of time event dependence to be taken into consideration. The elapsed time between two consecutive events (waiting time distributions) is modeled through a best‐fitting procedure on empirical data. We then determine the transition probabilities and so‐called crossing states probabilities. We conclude with a Monte Carlo simulation and the model is validated through a large database containing real data coming from HURDAT. Copyright © 2012 John Wiley & Sons, Ltd.