Variabilidade do clima da América do Sul e sua relação com os índices oceânicos e atmosféricos

Abstract

The objective of this study was to analyze and evaluate the dynamics of climate observed in South America. To achieve this aim, we used time series of climatic indices of the Pacific Decadal Oscillation (PDO), Atlantic Multidecadal Oscillation (AMO), Southern Oscillation and North Atlantic Oscillation, as well as the NCAR reanalysis data set of the variable precipitation rate. We applied statistical techniques to these four data sets, namely Hodrick – Prescott filter, wavelet analysis, exploratory data analysis and analysis of empirical orthogonal functions. The Hodrick – Prescott filter applied to the climatic indices time series generated two new time series for each index, the cyclical/random (high frequency variability) and trend (low frequency variability) components. Wavelet analysis was applied to the cyclical/random and trend components time series separately to search for possible apparent periodicities in the components time series. In the second step, we used the exploratory data analysis to identify periods of superposition between the PDO and AMO. Then, precipitation rate anomalies were calculated for the period 1981 to 2010 for each superposition period, namely 1948 to 1964 (PDO negative and AMO positive), 1965 to 1977 (PDO negative and AMO negative), 1978 to 1996 (PDO positive and AMO negative) and 1997 to 2006 (PDO positive and AMO positive). The third step was to calculate the correlation coefficients between the precipitation rate and the PDO and AMO indices for each period. After that, first and second components of empirical orthogonal functions for the rate of precipitation in the PDO cold and warm phases were calculated. It was found that the Hodrick - Precott filter is a good tool for analyzing time series of meteorological variables and separating their trend and cyclical/random components. It was noted that the oscillations may have the influence the lunar nodal cycle, since we found two distinct cycles in the wavelet analysis of the climatic indices trend and cyclical components, that is, the 9 and 18 year cycles. The correlation analysis suggested that the precipitation rate may have different behaviors depending on the phases of PDO and of the modes of AMO. It may increase, decrease or have no correlation with one or both oscillations. The first component of the precipitation rate showed a nucleus of high values in the center of South America in both PDO phases and it is most probably associated with the Bolivian High, resulting from the summer continental heating and convection, and South American Convergence Zone. The second component showed inverted signs between the Northern Region and the Center/Southeastern Region of Brazil, possibly related to El Niño/La Niña events that affect these regions and interact with large-scale atmospheric structures.