Análise multivariada das condições atmosféricas em mês chuvoso extremo na área de Petrolina, semiárido do Nordeste do Brasil.

Abstract

The present work analyzes the atmospheric conditions found over the area of Petrolina during the rainy month of April, 2014. These conditions have been analyzed focusing on the extreme rainfall event of April 8th. The database used consisted of reanalysis data, enhanced weather satellite images, daily and hourly rainfall totals, wind hourly values taken at the surface and upper air daily soundings. Weather analysis techniques have been used to diagnose the environment at synoptic scale and mesoscale all associated to extreme event. Data based on isobaric standard levels and stability indices have been submitted separately to multivariate analysis techniques in order to identify physical factors and verify the appropriateness of indices such as those of indicative of rain occurrences in the semiarid region. Rainfall was recorded between the 7th and 12th, and also the 27th of April related to weak southeast winds at the surface.The rain extreme event on April 8th was caused by a multicellular convective system which develops along large-scale conditions, characterized by the Inter-tropical Convergence Zone (ITCZ) acting at the far north and north of Northeastern Brazil, the Bolivian high moved east of its climatological position, and a high amplitude trough in the upper troposphere, east of South America. The use of factorial analysis for the principal components (PCA) reduces a set of nine variables along three principal components (PCs) which retain over 90% of the total data variance at all isobaric levels analyzed. The first retained PC accounts for the variations of temperature levels at 925 hPa, 850 hPa and 200 hPa, and also the moisture content variations at 500 hPa level. The application of cluster analysis (CA) to retained factors has helped in the identification of six groups of homogeneous days, considering the atmospheric conditions at each isobaric level. The application of PCA to a set of ten stability indexes results in three retained PCs holding 82.8% of the data total variance. CA application allows us to identify days that belongs to groups on which the indices represent either high or low instability condition. The indices CAPE, LI, SI, K, TT and SWEAT are the ones that best represent the thermodynamic and dynamic features of the atmosphere in the month studied. They have presented the largest variations on the days before rainfall events; an indication that may be useful in forecasting these events in the semiarid region of Northeast Brazil.