Modelagem integrada de meteorologia e recursos hídricos em múltiplas escalas temporais e espaciais: aplicação no Ceará e no setor hidroelétrico brasileiro

• Ano de defesa: 2014
• Autor(a) principal: Silveira, Cleiton da Silva
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Recursos hídricos; Precipitação - Previsão; Previsão de vazões
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/11327

This study aims to develop a planning system on multiple spatial and temporal scales, and apply it to the Brazilian electric sector and Ceará State, Jaguaribe Metropolitan System. For realization of this proposal, we have been considered some temporal scales: short-term (up to 1 month), short term (up to one year) and medium to long term (1-10 years and 10-30 years, respectively). To obtain estimates of the flow of short-term rainfall forecasts from atmospheric models for later entry in the hydrological rainfall-runoff model are used. To short term scale were considered stochastic and statistical models, as the Periodic Autoregressive type (PAR), Periodic Autoregressive with exogenous variables (PARx) and K-nearest neighbor models, and the use of global atmospheric models as input to hydrological rainfall-runoff model Soil Moisture Accounting Procedure (SMAP). For the range of the medium term were considered auto regressive models (AR) and Fourier and wavelets. We used data from the Coupled Model Intercomparison Project Phase 5 (CMIP5) as input in hydrological rainfall-runoff model for long-term scale. For the weather forecast, as the rain threshold adopted in the construction of the contingency table increases, the quality of the forecasts decreases, except for the adjustment index. Thus, the system of numerical prediction proves efficient in detecting the occurrence of rainfall of less intensity, with most satisfactory results in the North and Northeast regions of Brazil. On seasonal scale the models feature up better than the climatology. Likewise, in the range of medium-term models based on Fourier series and wavelets have better likelihood than the weather. In multi-scale, there are differences in the future shown by the projections of the CMIP5 models that were analyzed for RCP8.5 and RCP4.5 the XXI century scenarios, but in the North sector of the National Interconnected System (SIN), most models indicate negative trend, diverging only in magnitude.