Resumo Os riscos de extremos hidrológicos para um local ou região estão associados aos modos de variação do clima, em suas diversas escalas temporais. A compreensão da variabilidade de baixa frequência ganha uma elevada importância em regiões onde eventos de seca são frequentes, por estar associada a longos períodos de secas consecutivas. O presente artigo analisou a relação entre a precipitação média da bacia a montante da estação fluviométrica de Iguatu, com os índices PDO e AMO (Pacific Decadal Oscillation e Atlantic Multidecadal Oscillation) através das metodologias de Análise de changepoint, Transformada de Ondeletas (TO), Transformada de Ondeletas Cruzadas (XTC) e Análise da Coerência das Ondeletas (WTC). Essa estação mensura as vazões afluente ao reservatório de Orós, um dos principais do Estado do Ceará (Brasil). A precipitação média da bacia foi obtida a partir de dados de pluviômetros. Os resultados permitiram estabelecer uma relação entre a precipitação da região e os índices PDO e AMO, indicando que um modelo baseado nos índices pode ter alguma capacidade preditiva do comportamento da precipitação local. Nota-se também que períodos com fases simultaneamente positivas (negativas) da PDO e da AMO resultam em um comportamento mais previsível das precipitações da região, com valores abaixo (acima) do normalmente esperado.
Rainfall-runoff modeling in ungauged basins continues to be a great hydrological research challenge. A novel approach is the Long-Short-Term-Memory neural network (LSTM) from the Deep Learning toolbox, which few works have addressed its use for rainfall-runoff regionalization. This work aims to discuss the application of LSTM as a regional method against traditional neural network (FFNN) and conceptual models in a practical framework with adverse conditions: reduced data availability, shallow soil catchments with semiarid climate, and monthly time step. For this, the watersheds chosen were located on State of Ceará, Northeast Brazil. For streamflow regionalization, both LSTM and FFNN were better than the hydrological model used as benchmark, however, the FFNN were quite superior. The neural network methods also showed the ability to aggregate process understanding from different watersheds as the performance of the neural networks trained with the regionalization data were better with the neural networks trained for single catchments.
Three change-point methodologies were used to detect changes in the mean value of annual streamflow series and analyse simultaneous changes in large-scale global sea surface temperature (SST) oscillations. To verify the relationship between the variables we used wavelet coherence analysis. A preliminary detection skill test was performed using asynthetic series and Pruned Exact Linear Time (PELT) presented the best results among the methods used (Pettitt test, Bai and Perron algorithm) when combined with a penalty selection via the Changepoints for a Range of Penalties (CROPS) method. However, the use of classical penalty functions resulted in a poor performance of PELT. The three methods showed an extremely high convergence rate (> 90%) for the correct change points and a smaller rate for false positives (< 24%). Changes in the streamflow mean value coincided with phase shift of the low-frequency indices Atlantic Multidecadal Oscillation (AMO) and Pacific Decadal Oscillation (PDO), also corroborated by the wavelet results. Most of the changes can be associated with phase shift impacts in the South Atlantic Convergence Zone.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.