Padrões de distribuição temporal das precipitações intensas no Rio Grande do Sul

Abstract

Rainfall design can be estimated by several methods. The most common approach is the time redistribution of plots obtained from intensity-duration-frequency (IDF) relationships. This redistribution may follow a synthetic or critical pattern, as suggested by the alternating blocks method, or based on the actual precipitation behavior from dimensionless curves, as proposed by Huff (1967). Methods that simulate critical conditions are usually simple and versatile but often result in very different flow rates than those observed in a given river basin. In this sense, methodologies based on the actual precipitation behavior are more advantageous, which is important mainly for hydraulic structures projects and mapping of flood areas. Despite the importance and relevance of using a well-designed rainfall estimation methodology for local precipitation characteristics, there is no detailed and comprehensive study of the temporal distribution of Rio Grande do Sul (RS) precipitation. This would be very important by the fact that the region is heavily affected by extreme rainfall events, which represents a constant need for investment in flood control and flood warning systems infrastructure. Therefore, this research aimed to contribute to the understanding of the temporal distribution of intense rainfalls in RS. For this, the hourly database available in the state was synthesized in the form of dimensionless time distribution curves of the precipitation. Due to the sampling interval of the data and the short extension of the historical series, it was necessary to adapt the elaboration methodology proposed by Huff (1967). In order to evaluate the similarlity of patterns of different RS locations, point curves were elaborated from the data of each station. With regard to investigating whether RS patterns are similar to those of other locations in the world, regional curves have been developed that represent the average behavior of all the stations. A theoretical probability distribution was adjusted in order to give greater flexibility to the estimation of design rains from these dimensionless curves. Correlation analyses between the punctual curves indicated that there is a strong tendency of the temporal distribution of precipitation to be similar thorough the state. Moreover, the regional curve has been similar to that developed for other localities of the world, indicating that different climatic regions may also present similar temporal distribution characteristics of precipitation. Finally, the curves developed in other countries from sub-hourly data discretization presented a similar pattern to those developed in this study, indicating that regardless of the sampling interval of the data adopted, the temporal distribution of precipitation is maintained. Hence, it can be concluded that the dimensionless curves developed in this work can be used to estimate design rainfall with lower sampling intervals.