Non-Stationarity Analysis of Rainfall in São Paulo City, Sp. (PAP014802)

Código

PAP014802

Autores

Gustavo de Almeida Coelho, Arisvaldo Vieira Mello Júnior

Tema

Impact of climate change on flood risk

Resumo

Systems for management of water throughout the developed world have been designed and operated under the assumption of stationarity (Milly et al., 2008). Change in rainfall pattern over time or adoption of misguided design parameters can significantly affect design and operation of these systems. Non-stationarity analysis of precipitation has been performed in different regions (Angel e Huff, 1997; Alexander et al., 2006; Seneviratne et al., 2012) and results can be used to improve drainage systems calculation methods. Generally, there is great difficulty to find precipitation data with long duration and good spatial distribution that enable reliable results for a given basin or locality. Such studies are intended to provide representative results for decision-making about flood control measures. The main objective of this work is to perform a statistical analysis of rainfall in the city of São Paulo to check their non-stationarity. Study interest it was to seek trends existence in total annual precipitation and daily annual maximum precipitation obtained in six rain gauges located in São Paulo city. It is also expected to obtain practical recommendations on rainfall design for civil engineering and on non-structural measures and their implications on the cost of the drainage works and measures to control results. After a preliminary analysis of available data, six rain gauges situated in a 12 km ratio in São Paulo city were selected. Rainfall series duration vary between 57 and 114, from 1888 to 2012. Figure 1 shows rain gauges location. Mann-Kendall non-parametric test (1945) was performed to verify trend existence on data series. Stationarity was confirmed for Kendall's tau value between -0,05 and 0,05 considering a 5 % significance level. Results showed different behaviors on total annual precipitation were 3 series confirmed non-stationarity and 3 series did not. When analyzing maximum precipitation, only one serie confirmed non-stationarity. In final paper we intend to show other statistical methods applied to detect trend and an economic analysis about urban drainage systems costs when considering precipitation design increase.