Model for Seasonal North Atlantic Hurricane Counts

To a first approximation on the annual time scale, high ocean heat content, low values of wind shear, and westerly steering currents increase the risk of hurricanes (Gray 1968; DeMaria et al. 2001; Elsner 2003). Indexes that track variations in these factors are used to construct skillful statistical models of coastal hurricane activity and potential financial losses (Saunders and Lea 2005; Jagger et al. 2008). Table 1 lists the coefficient estimates of a generalized linear regression model for Atlantic tropical cyclone counts (tropical storms and hurricanes) using data that starts at different years. The model covariates include sea-surface temperature

Table 1 Coefficients of a generalized linear model (Poisson) of tropical cyclone counts. The model uses the logarithm of the rate as the link function to a linear regression of the covariates. Model coefficients are determined from a maximum likelihood procedure. The covariates include the May through June averaged North Atlantic Oscillation (NAO) index in units of standard deviation, the August through October averaged Southern Oscillation Index in units of standard deviation, and the August through October averaged SSTs in the main development area of the central North Atlantic Ocean (see Fig. 2). For a one unit change in the covariate, the difference in the logarithms of expected tropical cyclone counts changes by the respective model coefficient given that the other covariates are held constant. The reduction in deviance from a model with no covariates is between 40 and 48% depending on start year. The intercept term is not included in the table




z value


TS + H 1900-2006
















TS + H 1914-2006
















TS + H 1944-2006
















(SST) as an indicator of ocean heat content averaged over the main development area of the Atlantic Ocean (80W to 20W by 5N to 25N), the Southern Oscillation Index (SOI) as a remote indicator of shear, and the North Atlantic Oscillation index (NAO) as an indicator of steering currents.

Monthly values of the NAO and SOI are obtained from the Climatic Research Unit of the University of East Anglia. The May and June values of the NAO are averaged to produce the NAO covariate. The August through October values of the SOI are averaged to produce the SOI covariate. Both covariates have units of standard deviations. The monthly SST values were obtained from the U.S. National Oceanic and Atmospheric Administration (NOAA) in Boulder, Colorado, USA ( and are from the Kaplan SST V2 data (Kaplan et al. 1998). The data are averaged over the main development area of the Atlantic Ocean (80W to 20W by 5N to 25N) from 2 degree by 2 degree latitude-longitude grids and have units of degrees Celsius.

The generalized linear model for tropical cyclone count data is the Poisson regression (Elsner and Schmertmann 1994; McDonnell and Holbrook 2004). It attributes to the response variable (annual tropical cyclone counts) a Poisson distribution whose expected value depends on a set of covariates in the following way

Was this article helpful?

0 0
Survival Treasure

Survival Treasure

This is a collection of 3 guides all about survival. Within this collection you find the following titles: Outdoor Survival Skills, Survival Basics and The Wilderness Survival Guide.

Get My Free Ebook

Post a comment