Previsão de vazões por lógica difusa e ensemble hidrológico

Abstract

Resumo: O conceito de lógica e variável difusa geralmente é utilizado na representação das incertezas de um modelo. O método sugerido previsão pluviométrica por ensemble e o conceito de lógica difusa cnoemstoa otebsjeet iuvtoil idzae eobntseerm ubmle hhididroroglróagmicao . prPeavriast ois qsou e asse jpar emceiplhitaoçr õqeuse por ehviidsrtoags radmoa emnséedmiob ldee suãmo fuzificadas através de funções de pertinência. Através das precipitações previstas fuzificadas, do grau de pertinência da precipitação observada e de regras difusas se obtém as precipitações previstas pelo método proposto que se transforma em vazões previstas através da modelagem chuva-vazão pelo modelo Sacramento. vPaazrõae asv aolibasre or vmadéatosd. oA psr ovpaozsõteo,s oo hbitdidroagsr apmelaa pmreévdisiato déo e netnãsoe cmobmlep atraamdboé cmom s ãaos comparadas com as vazões observadas. Os resultados mostram que apesar do método sugerir uma leve melhora em alguns casos na etapa de calibração, na vaa vzaãzoã op rmevéisdtiaa dfinoa el nos emmébtoled oh ipdrrooplóogsitcoo .n ão teve êxito, sendo equivalente a utilizar.

Abstract: The concept of logic and diffuse variable is usually used in the uncertainties of a model. The method suggested in this thesis uses pluviometric forecast by ensemble and the concept of diffuse logic as objective to obtain a predicted hydrograph that is better than the average hydrograph of hydrological ensemble. For this, the expected precipitations of the ensemble are fuzzified through pertinence functions. By means of the fuzzified predicted precipitations, the pertinence degree of the observed precipitation and of diffuse rules, the precipitations are obtained, foreseen by the proposed method, which becomes the expected flows through the rainfall-flow modeling by the Sacramento model. To evaluate the proposed method, the predicted hydrograph is then compared with the observed flow rates. The flows obtained by the average ensemble are also compared with the observed flows. The results show that although the method suggests a slight improvement in some cases in the calibration stage, in the expected final flow the proposed method was not successful, being equivalent to using the average flow of the hydrological ensemble.