Non linear discriminant based on mixture of distributions: an application in return time of heterogeneous maximum flows observed in the Paranapanema river basin-Brazil

• Autores: Guevara Otiniano, Cira Etheowalda, C. Cruvinel, Evelyn, H. Lima, Carlos
• Formato: artículo
• Fecha de Publicación: 2020
• Institución: Universidad Nacional de Trujillo
• Repositorio: Revistas - Universidad Nacional de Trujillo
• Lenguaje: portugués
• OAI Identifier: oai:ojs.revistas.unitru.edu.pe:article/3079
• Enlace del recurso: https://revistas.unitru.edu.pe/index.php/SSMM/article/view/3079
• Nivel de acceso: acceso abierto
• Materia: MGEV; Discriminant Analysis; Flows; Return Period; Análises de discriminante; Cheia de bacias hidrográficas; Período de retorno

In this work, maximum flows observed in the Paranapanema river basin are modeled by a mixture of two GEV distributions whose components correspond to the summer and winter subpopulations. The nonlinear discriminant function allowed to confirm the assumption of mixture model.The proposed model is then used to determine probabilities of exceedance and return periods associated with extreme flows, which are of fundamental importance for hydraulic projects. The results show significant differences when a mixture model is used and therefore an approach more coherent with the observed data.