Gagne de la cryptomonnaie GRATUITE en 5 clics et aide institut numérique à propager la connaissance universitaire >> CLIQUEZ ICI <<


Non classé

The increasing interest raise up these last years to study extreme events such as dryness, heat-waves, floods, etc., is as a result of their unpredictable character and the prejudices raised to the society. Our understanding of the average behaviour of climate and its variability has improved these last decades. On the contrary it is always di cult to understand extreme climatic events and more hard to foresee, because they are rare and follow different statistical laws from those of the average. In this context we carried out study on extreme events, more exactly those of rare values (temperatures and precipitations peaks) and the estimation of their return levels for given return durations. This through statistical models from the Extreme Value Theory (EVT). Block Maxima model (BM) and Peaks-Over-Threshold model (POT), this within the context of simple analysis (stationary and univaried data). We applied these models to monthly average temperatures and to cumulated monthly precipitation in Douala during the periods 1971-2003 and 1960-2005 respectively. These in the viewpoint to determine temperature return levels and extreme precipitations in Douala. This procedure starts with GEV (Generalised Extremes Values) and GPD (Generalized Pareto Distribution) distribution statistic formulations for each extreme value sample. In the end it comes out that temperatures and extreme precipitations in Douala are modeled following limit laws for the two models and the return levels estimated for the two methods are close and augment gradually for long return durations. Thus the two models show us that the extremes values of monthly averages temperatures of 30.7 C observed in february 1998 and monthly accumulation of precipitations of 1240.40 mm observed in august 1966, have one return level of 200 years for temperatures and 80 years for precipitations.

Key words : extreme events, generalized Pareto law, return level, Maximum Likelihood.

Page suivante : Introduction générale

Retour au menu : Etude des Evénements Extrêmes : cas des Températures et Précipitations à Douala