Although tropical cyclone rainfall (TCR) is common over Puerto Rico, the factors that cause this rain to vary from one storm to another have not been studied. The aim of this article is to understand how storm-specific characteristics including storm location, duration, storm centre proximity to land, intensity, horizontal translation speed (HTS) and environmental factors like moisture and vertical wind shear affect TCR variability over Puerto Rico. TCR was determined at rain gauge locations for days when storms were within a 500 km radius of Puerto Rico. The station data were then used to calculate an island-averaged total rainfall value for 86 storms during 1970-2010. The maximum observed rainfall was also examined. Correlation analyses of the individual predictors, principal component regression (PCR) procedures and Mann-Whitney U tests identified precipitable water, storm centre proximity to land, mid-level relative humidity (MRH), duration, HTS and longitude as the predictors with the strongest influence on rainfall. The PCR showed that a component comprised of precipitable water, MRH and longitude accounted for more than 60% in TCR variability. When an additional component comprised of storm duration, storm centre proximity to land and translation speed was considered, the PCR model explained 70% (52%) of the variability in mean (maximum) TCR. Key threshold values for high rainfall across Puerto Rico are a storm centre distance of 233 km or less and moisture exceeding 44.5 mm of precipitable water and 44.5% of relative humidity with forward speeds of 6.4 m s −1 or less. Extreme rainfall at a single location can occur when a TC's centre is over 450 km away.

No abstract

An extreme value analysis (EVA) point process approach has been implemented to examine the flood characteristics of Puerto Rico when tropical cyclones (TCs) are present in the discharge series and when they are removed from it. Mean daily discharge values that exceeded the 99th percentile thresholds were used in both the TC and non-TC data series. In nine of the 12 stations the maximum discharge was associated with a TC, with hurricanes Hortense (1996), Georges (1998) and Eloise (1975) responsible for most of the maximum peaks at each site. Percentage changes in the generalized extreme value parameters, which include location (central tendency), scale (variance) and shape (skewness), between the TC and non-TC data exhibited a decrease in the majority of stations. Stations in the eastern interior and central region of the island showed the largest decrease in all parameters, in flood occurrences and in return periods when TCs were removed from the series. ARTICLE HISTORY