Estudo da variabilidade hidrometeorológica no Nordeste do Brasil utilizando análise multivariada e espectral.

Abstract

The aim of this work was to study the dynamics of climate variability in precipitation and flow over the Brazilian Northeast (BN) using Multivariate Analysis and Spectral. For this, were used monthly data from 11 climates indexes rain (defined by the World Meteorological Organization (WMO)) and flow rate, obtained from daily data (1960-2006) of 258 and 45 stations/posts, pluviometric and fluviometric, respectively. For Multivariate Analysis, was used Principal Component Analysis (PCA) and Cluster Analysis (CA) in order to obtain the spatial and temporal distribution of rainfall and flow and their respective regionalization of homogeneous groups. For spectral analysis was used Wavelet Transform (WT) in order to obtain the frequency and power levels associated with different frequencies in each group homogeneous G1, G2, G3 and G4. Furthermore, correlations and cross spectrals were performed for flow and climatic indexes rain from each group against thermal anomalies of Pacific and Atlantic oceans, and three indexes of ocean climate variability global: Southern Oscillation Index (SOI), Multivariate El Niño Index (MEI) and Pacific Decadal Oscillation (PDO). For PCA, the indexes rain R99p (extremely wet days) and SDII (Simple Daily Intensity Index) were select and represented in space and time by four principal components (PCs) that explained approximate 70% of the total variance of the system, and for other indexes, three PCs explain more than 90%. For the flow, two CPs explained more than 90% of the total variance explained. In CA, the rain climate indexes: R50 (Amount in days with precipitation exceeding 50 mm) and WCD (Wet Consecutive Days) were regionalized by only two homogeneous groups (G1 and G2), SDII for three (G1, G2, and G3) and the other indexes for four (G1, G2, G3 and G4). For the flow, two groups (G1 and G2) were sufficient for represented fluviometric of the BN. The results of applying WT showed that the scale of four years had the highest levels of energy intensity and spectral modulated in all groups homogeneous rain much of the higher frequency oscillations of the biennial at semiannual, highlighting the decade of 1980s. As for the flow, G1 and G2, were the scales one and two years (G1) and one and four years (G2), the highest levels of energy on the BN, with noticeable decay during the 1980s (G1 and G2) and 1990s (G1). In the correlation between rainfall indexes and indexes oceanic/sea surface temperatures (SSTs), G1 (eastern BN) was shown to be significantly more anomalies associated with variations in the Tropical North Atlantic, G4 (northern of the BN) of indexes PDO and SOI, G2 and G3 of indexes PDO and MEI. For the flow, Niño 1+2, SOI, MEI and PDO exerted influence in G1 and G2. The cross spectral showed that in many, the frequency of four years was modulated by the scales of 12 and eight years, and proved quite active when confronted indexes rain in days and in milimeters with the indexes TNAI (G1) and PDO (G4), and when confronted flow rates (G1 and G2) with the indexes Niño 1+2, MEI, SOI and PDO. With that, it became easier to identify temporal patterns of monthly runoff, which greatly affect the river dynamics, since this is the main agent for transporting water fountains. The indexes PDO and TNAI were most affected the rainfall in the region. The interannual variability is linked to the cycle of ENSO and decadal variability eight to 12 years, possible from of the sunspots cycles, influences in variability rainfall modulating the level of four years. It was also concluded that the statistical tools are extremely useful for understanding the climatology of the BN and showed what possible events of climate variability exert greater influence on the pluviometric and fluviometric regional and local.