Spatiotemporal Variability and Prediction of Rainfall over the Eastern Caribbean

Recent, recurrent, and extreme weather events have been a cause for concern over the Eastern Caribbean (EC). Given the dependence on rainfall of agriculture, the main stay of the fragile economies throughout the region, accurate and timely forecasts of seasonal rainfall need to be issued to facilitate decision making in Water Resource Management. Understanding the causes of climate variability can lead to the development of more robust models for climate prediction. So as a diagnostic approach, different techniques are employed. Empirical Orthogonal Function (EOF) analysis is performed in order to isolate the different modes of rainfall variability as well as investigating their amplitudinal modulations. The evolution of external forcing mechanisms that impact on precipitation extremes is also investigated with the use of composites. Based on the strength of the relationship between Sea Surface Temperature Anomalies (SSTA) and EC rainfall, a statistical model is subsequently developed using multivariate Canonical Correlation Analysis (CCA) to predict rainfall over the region on seasonal time scales. The CCA model demonstrated useful skill in predicting seasonal rainfall over the EC up to six months lead. The highest average predictive skill is realized for the June-July-August (JJA) season at one-month lead, while the lowest average skill is realized for the March-April-May (MAM) season at five months lead. The December-January-February (DJF) season maintained steady skill throughout six months lead. Below normal conditions are forecasted by the CCA model for the 2004/2005 dry season (DJF/2004-05, MAM/2005). This outlook is in part, verified from seasonal rainfall totals at two stations within the EC. The outlook for the coming rainy season is for above normal conditions.