Caribbean hurricanes: interannual variability and prediction

Climatic conditions that affect the interannual variability of Caribbean hurricanes are studied. Composite meteorological and oceanographic reanalysis fields are constructed for active and inactive seasons since 1979, and differences are calculated for spring and summer periods to provide guidance in statistical analysis. Predictors are extracted for areas exhibiting high contrast between active and inactive seasons, and intercomparisons are made. Zonal winds north of Venezuela exhibit westerly anomalies prior to active years, so coastal upwelling and the north Brazil current are diminished. Rainfall increases in the Orinoco River basin, creating a fresh warm plume north of Trinidad. The predictor time series are regressed onto an index of Caribbean hurricanes, and multivariate algorithms are formulated. It is found that atmospheric kinematic and convective predictors explain only ??20% of hurricane variance at 3?C5-month lead time. Subsurface ocean predictors offer higher levels of explained hurricane variance (42%) at 3?C5-month lead time, using 1?C200-m-depth-averaged temperatures in the east Pacific and southern Caribbean. We place the statistical results in a conceptual framework to better understand climatic processes anticipating Caribbean hurricanes.