Service Communautaire d'Information sur la Recherche et le Développement - CORDIS


PACLIVA Résumé de rapport

Project ID: EVK2-CT-2002-00143
Financé au titre de: FP5-EESD
Pays: Sweden

Precipitation patterns in the Antilles

Statistical analyses were used to characterise rainfall patterns in the Antilles, and their relationship to remote climate patterns. The Antilles archipelago stretches between Florida and eastern Venezuela. Rainfall climate in the Caribbean can be classified as dry-winter tropical with significant subregional variation in total annual rainfall, length of the rainy season, and timing of rainfall maxima. The North Atlantic Oscillation (NAO) and El Niño-Southern Oscillation (ENSO) are known to influence precipitation patterns in the North Atlantic Basin.

Many meteorological stations in the Caribbean have limited temporal scope, and are not complete. The analysis of 35 stations from the time interval 1951-1981 provided an optimum balance between geographic and temporal scope. Time-series from these stations were subject to multivariate-statistical analysis in order to unravel regional precipitation climate, and to investigate the response of each subregion to NAO and ENSO fluctuations.

The two-stage least squares (TSLS) factor analysis (FA) method was applied to identify distinct subregions that are each marked by common patterns of fluctuations through this time-interval, the geographical coherency of these regions, and differences in the seasonal precipitation variability among them. An advantage of TSLS FA is the possibility to compute standard errors for the factor loadings that are useful to assess the significance of the contribution of a variable to a particular factor.
For analyses of the associations between the precipitation patterns and NAO the mean December-March NAO index (NAOI) for the time period 1826-2002 was divided into years with a low NAOI (lowest tercile of the NAOI) and high NAOI (highest tercile). For analysis of the effect of ENSO on precipitation the Niño 3.4 region, which has been shown to essentially characterize the ENSO behaviour, was used. A positive SST anomaly in of >0.4ºC in N3.4 represents a warm ENSO event, and a negative anomaly of >0.4ºC represents a cold ENSO event.

The statistical significance of the differences between low and high NAO events, and warm and cool ENSO events, for different months was tested using t-tests. The problem inherent in multiple testing of the H0 and H1 hypotheses in climatology was considered in evaluations of the results.

Application of TSLS FA indicated that four distinct patterns of precipitation fluctuations existed in the Caribbean through 1951-1981:
Regime 1 - Westernmost Caribbean (western Cuba and western Bahamas), southernmost Dominican Republic, westernmost Puerto Rico. This is the most geographically disjunct region.
Regime 2 - Eastern Cuba and the eastern parts of the Bahamas.
Regime 3 - Northern Dominican Republic, northwesternmost part of Puerto Rico.
Regime 4 - The rest of Puerto Rico, Lesser Antilles (U.S. Virgin Islands, St. Martin, and Barbados).

These regimes have distinctly different seasonal precipitation patterns. Regime 1 is characterized by high rainfall in the 6-month period from May through October. Regime 4 is marked by low precipitation amounts through the winter and early spring (January-April) and a maximum peak in September, October, or November. Regimes 2 and 3 differ from Regime 1 by relatively lesser May-October and greater November-December precipitation and from Regime 4 by lesser May-December precipitation. Regime 3 is marked by more even distribution of rainfall throughout the year in comparison with Regime 2. Regime 2 stands out from the other regimes by lower total annual precipitation.

The analyses of differences in precipitation between low and high NAO years for each of the four precipitation regimes indicated no effect of the NAO on precipitation in Regimes 1 and 2. In contrast, the analysis indicates significantly enhanced May precipitation during low NAO years in Regimes 3 and 4. Similarly, July rainfall is increased during low NAO years in Regime 3. For example, nearly 700 mm/yr more rain fell over all stations in Regime 3 in low NAO years than in high NAO years. A further analysis of stations with more than 30 years of continuous observations indicates that the pattern observed is not just a feature characteristic of the period 1951-1981.

ENSO control of precipitation in the Antilles is also documented by the study. In Regime 1 most of the first five months of the year (all except February) are controlled by ENSO phases during parts of the previous year or the same calendar year. Regime 2 is relatively little affected by ENSO. Similarly to Regime 1, increased January precipitation in Regime 3 is related to warm ENSO phases the previous year. May is wetter during February-May warm ENSO phases. In Regime 4 increased July precipitation is associated with warm ENSO phases during the fall of the previous year and the spring of the same year. July-August cold ENSO phases cause enhanced rainfall in October.


Emeritus Björn MALMGREN, (Professor)
Tél.: +46-0176-10746
Fax: +46-0176-10746