Oceanic influence on the precipitation of the south-east of Venezuela

The study of the interannual variability on monthly rainfall in the south east region of Venezuela, where the Caroni river cachment is located, is a big priority for the country since this cachment provides more than 70% of the electric energy for the whole country. A truncated normal model is used as an underlying model for rainfall which has been transformed by raising positive values of the normal variable to a positive power and truncated for negative values which correspond to the dry periods. The normal variable evolves in time as in a dynamic linear model in which the evolution equation comprises a baseline parameter representing the changing mean of the process and the coefficients of a Fourier representation with two harmonic to account for the strong seasonal behaviour of the rainfall in the region. The model is fitted using a Markov Chain Monte Carlo method that uses latent variables to handle both dry periods and missing values. The model is used to predict the amount of rainfall and the probability of dry periods in the region. The baseline component and the amplitude of the Fourier harmonics are investigated in relation to their dependence on the sea surface temperature anomalies for different regions in the Atlantic and Pacific oceans. The results demonstrate the potential use of these variables to improve rainfall forecasting in the region.