Frequency analysis, infilling and trends for extreme precipitation for Jamaica (1895–2100)

Study region: The island of Jamaica. Study Focus: Existing intensity duration frequency (IDF) curves for Jamaica were calculated over 20 years ago for two stations using an annual maxima series (AMS) that is 30 years long. The goodness of fit for the IDF curves is no longer available, and trends have not been accounted for in the series. Data recovery efforts undertaken in this study facilitate a re-examination and comparison of the existing IDF curves with new curves derived from an extended AMS from 1895 to 2010. In extending the data, gaps in the short durations (5 min to 12 h) AMS were filled using an empirical formula and gaps in the long durations (2–10 days) were filled by artificial neural network (ANN) downscaling. Analysis included an examination of frequency analysis configuration for varying probability distribution functions (PDF), plotting point functions (PPF) and parameter estimation methods (PEM). The study determines PDF as the most important factor in the configuration of the frequency analysis. Infilling and extension are also shown to reduce the biases, improve the frequency analysis and yield higher predicted intensities. New hydrological insights for the region: Temporal trends in location,scale and shape parameters for each station were also examined and lin-early projected to 2100. Increases in 2100 intensities vary from 27% to 59%for the 100 years return period (RP) as a result of increases in the variability. Keywords: Frequency analysis, Climate change, Extreme precipitation, Artificial neural network, Intensity duration frequency curve, Jamaica