Análise espaço temporal da chuva no estado da Paraiba com base na transformada wavelet

Abstract

Rainfall is one of the most significant hydrological variables in water resources management. Investigation of rainfall trends is therefore essential since excess rainfall can cause flooding and enormous property and crop damage, yet a deficiency of rainfall can cause drought and crop failure. This study uses monthly rainfall data from Climatic Research Unit – (CRU), with a 0.5° × 0.5° horizontal resolution. In this work, one hundred years of time series (1920 - 2019) in all points distributed over Paraiba State were used. Through the wavelet spectrum from wavelet transform analyses, it was possible to perform the signal decomposition, providing not only the frequency but also its location in time, thus identifying existing trends. The wavelet analyses showed an annual frequency pattern in all time-series. In the majority of the 43 points, it was also observed a quarterly and semester periodicity. In a few points, a low-frequency pattern was noticed. Lastly, three different clusters analyses were made splitting the state into four groups. The first cluster analysis used the global spectrum from wavelet transform. The second analysis utilized analysis of monthly spatiotemporal variability of rainfall bands 8-16 over the state, and the third analysis considered all monthly rainfalls in the study area. The clusterization result showed a huge similarity between the Climatic research Unit precipitation data set and the rainfall gauge data set. Also, when monthly spatiotemporal rainfall from 8-16 bands were compared the result from both datasets were alike, noting a predominance of a cluster group in the sertão of Paraíba and in the upper part of Borborema, in addition to a greater variation of groups in the east of the agreste of Paraíba and on the coast. Considering the global spectrum, the result was similar, however, a little more distant than when analyzing the other two parameters.