Estudo da precipitação no Estado da Paraíba usando análises de séries temporais

Abstract

The objective of this work was to study the dynamics of the climatic variability of precipitation in the state of Paraíba using Multivariate Analysis, Descriptive Analysis, Rainfall Anomaly Index (RAI), Empirical Mode Decomposition (EMD) and Spectral Analysis. For this, monthly average data from more than 170 rainfall stations were used, obtained from 1910 to 2016. For the multivariate analysis, Principal Component Analysis (PCA) and Cluster Analysis (AA) were used in order to obtain the spatiotemporal distribution of precipitation and its respective regionalization of homogeneous sub-regions. From PCA, the mean precipitation index was selected and represented in space and time by two principal components that explained about 85% of the total precipitation variance. From the cluster analysis it was possible to select three homogeneous regions, which were called sub-regions G1, G2 and G3. For the descriptive analysis, the average precipitation of each sub-region found in the cluster analysis was used. The RAI was used as a tool, the indices obtained (both positive and negative) were compared with the years under the action of the El Niño Southern Oscillation (ENOS) phenomenon. Comparing the years under the influence of the ENSO phenomenon with the values of the RAI, it was found that there was a greater correspondence with the years under the El Niño effect than under the La Niña effect. When comparing the RAI in the series with the positive and negative cycles of ENSO, a strong correspondence was found between the hot and cold phases, including their extreme events. Finally, the RAI proved to be an important tool in the analysis of average rainfall in the state of Paraíba to identify wet years, dry years and normal years. The Wavelet Analysis was used in order to obtain the energy associated with different periodicities over time in each homogeneous group G1, G2 and G3. The EMD was the technique used to separate the oscillations and trends of the time series of each sub-region, the three sub-regions showed a negative trend. The interannual variability found in the analyzed series showed correspondence with the ENSO cycle and the decadal variability (~11 and 22 years) indicates a possible relationship with the sunspot cycles, through the modulation of the position of the Intertropical Convergence Zone, which is shifts further north at maximum solar activity. It was also concluded that the statistical tools were extremely useful for understanding the climatology of the study region and showed which possible climate variability events exert the greatest influence on regional and local rainfall.