Analysis of extreme rainfall events hydro-through variables such local maxima of annual or seasonal rainfall measured on time of day not gives outstanding value for some rainy years and temporal distribution of rainfall. Probabilistic models used in hydrology for the analysis of univariate extreme values were presented, in order to adapt the samples to the laws for their study. We were able to adapt the sample sizes of measurement series of 58 year rainfall and extreme events obtained ASECNA certain laws for the study of extreme values.
The maximum annual or seasonal hydrologies are often modeled by the Gumbel distribution.
But more and more of skepticism toward the modeling of seasonal or annual maxima by the Gumbel distribution are developed in the literature hydrology. In our case several laws have been tried, this is how Extreme values were investigated with estimation return periods through the laws of Gumbel, Normal and Fuites. But it is difficult to draw a single conclusion, because the results depend on whether we apply a Gumbel, Normal or Fuites. For this we compared the 2010 rainfall event in a cycle with a return period of between 10 and 20 years.
We also sought to establish a relationship between the exceptional events of 2010 and Climate Change. The annual values of the extremes are not made daily extreme events, but the accumulation of events with low and middle temporal distribution. Changes cycle for rainfall are minimal.
We also studied the Hydrometic and Piezometry fluctuations , but the series are too short for scientic results quality. Despite this, we could establish a relationship between the cyclic flooding and the impact of human activities on water flow through the anarchic occupation of the soil which reveals an important characteristic of natural disasters in Cotonou about.
Key words : Extreme Values, Hydrometic fluctuations, Piezometric fluctuations , Climate Change, Natural Disaster.