Open AccessSimulation of Storm Occurrences Using Simulated Annealing
Author(s) -
R. S. Lokupitiya,
Leon E. Borgman,
Richard Anderson-Sprecher
Publication year - 2005
Publication title -
journal of climate
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 3.315
H-Index - 287
eISSN - 1520-0442
pISSN - 0894-8755
DOI - 10.1175/jcli3546.1
Subject(s) - storm , univariate , climatology , tropical cyclone , simulated annealing , environmental science , meteorology , parametric statistics , poisson distribution , computer science , geology , statistics , geography , algorithm , mathematics , multivariate statistics , machine learning
Modeling storm occurrences has become a vital part of hurricane prediction. In this paper, a method for simulating event occurrences using a simulated annealing algorithm is described. The method is illustrated using annual counts of hurricanes and of tropical storms in the Atlantic Ocean and Gulf of Mexico. Simulations closely match distributional properties, including possible correlations, in the historical data. For hurricanes, traditionally used Poisson and negative binomial processes also predict univariate properties well, but for tropical storms parametric methods are less successful. The authors determined that simulated annealing replicates properties of both series. Simulated annealing can be designed so that simulations mimic historical distributional properties to whatever degree is desired, including occurrence of extreme events and temporal patterning.