Regionalização e riscos de índices de chuvas intensas no Nordeste do Brasil.

Abstract

The most common natural disasters in northeastern Brazil are the droughts, landslides and floods. The last three are partly consequences of extreme rainfall events. When landslides occur in urban areas during the rainy season cause material damage and deaths. Not coincidentally, are also consequences of the inefficiency of the drainage system and the indiscriminate removal of vegetation cover. These actions combined with the rains increase the water saturation in the soil, reducing their resistance and causes rupture. Landslides are due to misuse of land, poor income distribution, lack of decent housing and poor quality of public education. Floods are usually accumulations of water slides that can invade the interior of the buildings and cause inconvenience to urban mobility of pedestrians and vehicles. These events depend on structural and non-structural measures. This research, conducted in northeastern Brazil, was to identify the spatio-temporal variability of heavy rainfall and determine the risk of their occurrence above the threshold of 30 mm a day, based on data of daily maximum rainfall of 133 rain gauges, relatively well distributed in the study region. From them, the daily average weather for the entire region was obtained, the long-term trend for extreme precipitation values for the Mann-Kendall test and the percentage of heavy rainfall in relation to the climatological average. The Cluster Analysis was applied to daily precipitation data, and identified three homogeneous sub-regions based on total daily average precipitation. The Gumbel probability distribution function was set to intensity rains. This adjustment estimates of risk for a specific magnitude of rain. The results based on return periods of 50 years and 100 showed that RH2 region had periods of higher rainfall returns, represented by the city of Igarassu - PE. The city of Itabaiana - SE, in RH1 region, was the city where he registered the lowest rainfall on the study.