Season-to-Season Cyclone Frequency Prediction

Bruce P. Hayden Department of Environmental Sciences, University of Virginia, Charlottesville, 22903

Search for other papers by Bruce P. Hayden in
Current site
Google Scholar
PubMed
Close
and
William Smith Department of Environmental Sciences, University of Virginia, Charlottesville, 22903

Search for other papers by William Smith in
Current site
Google Scholar
PubMed
Close
Restricted access

Abstract

Winter and summer half-year cyclone frequencies for eastern North America and the western North Atlantic were tabulated for 2.5° latitude by 5° longitude grid cells for the years 1885–1980. Correlation matrix eigenvectors were calculated for matrices of both the winter and summer data for a dependent set of years (1885–1959). The first four eigenvectors for winter are highly similar in form to their summer counterparts and are correlated with them in the time domain. These correlations permit a season-in-advance estimation of half-year cyclone frequencies. Forecast skill for independent years (1960–1980) in both winter and summer averages 75% relative to the 1885–1959 means as forecasts. The multivariate statistical forecast scheme outperforms both simple and damped persistence in addition to raw climatology. The climatological persistence from which the forecast skill results is shown to have both intermediate- and low-frequency components.

Abstract

Winter and summer half-year cyclone frequencies for eastern North America and the western North Atlantic were tabulated for 2.5° latitude by 5° longitude grid cells for the years 1885–1980. Correlation matrix eigenvectors were calculated for matrices of both the winter and summer data for a dependent set of years (1885–1959). The first four eigenvectors for winter are highly similar in form to their summer counterparts and are correlated with them in the time domain. These correlations permit a season-in-advance estimation of half-year cyclone frequencies. Forecast skill for independent years (1960–1980) in both winter and summer averages 75% relative to the 1885–1959 means as forecasts. The multivariate statistical forecast scheme outperforms both simple and damped persistence in addition to raw climatology. The climatological persistence from which the forecast skill results is shown to have both intermediate- and low-frequency components.

Save