Metodologia de curvas envoltórias probabilísticas para a estimação de cheias de projeto em bacias hidrográficas não monitoradas no estado de Minas Gerais

• Ano de defesa: 2010
• Autor(a) principal: Jose Alexandre Pinto Coelho Filho
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Meio Ambiente; Recursos Hídricos; Saneamento; Engenharia sanitária
• Link de acesso: http://hdl.handle.net/1843/BUOS-8E4NXC

Regional envelope curves have been used since the beginning of the 20th century as convenient summaries of the largest flood discharges observed at all gauging stations located in one or more geographical regions. Usually the envelope curve for a specific region relatesthe upper bound on all observed floods of record to the corresponding drainage areas. Regional envelope curves have been used by practitioners worldwide as a simple and expedite procedure to provide preliminary estimates of design floods for hydraulic structures, in spiteof the fact that they have been usually regarded as deterministic graphical tools and, as such, they were not associated to probabilistic statements. Only recently Castellarin et al. (2005) and Castellarin (2007) proposed a new interpretation of regional envelope curves according to which it is possible to assign an exceedance probability p to the expected envelope curve valid for a given region. The method described by Castellarin (2007) relies on the assumptions of regional homogeneity in the sense of the index-flood hypothesis and a GEV (Generalized Extreme Value) regional parent distribution, allowing at the end the estimation of the T-yearflood for the ungauged site draining a given catchment area comprised within the interval of areas of the expected envelope curve, the excedance probability of which being p = 1/T . The method proposed by Castellarin (2007) was extended to the estimation of the set of the 3 GEV parameters, and thus the entire flood frequency curve, for any gauged and/or ungauged site within a given region. This extension has been performed by combining the methods assumptions and its main result into a system of 3 equations, having the GEV parameter estimates as the 3 unknowns. Then, with the possibility of estimating the entire GEVdistribution for any site in the region, a validation procedure was performed for the gauged sites by comparing the results of the proposed extension with the ones obtained by L-momentbased regional frequency analysis.