Seasonal Patterns of Relationship Between Rainfall and Streamflow in Brazil (PAP014784)

Código

PAP014784

Autores

Caluan Rodrigues Capozzoli, ANDREA DE OLIVEIRA CARDOSO

Tema

Impact of climate change on flood risk

Resumo

In this work the multivariate analysis technique known as Canonical Correlation Analysis (CCA) on quarterly averages of flow and quarterly cumulative rainfall of Brazilian historical series was used with the objective of identifying spatial and temporal variability of river flow patterns possibly associated with variability climate. The period of analysis is from January 1970 to December 2005, being used to calculate the quarterly average streamflow and cumulative rainfall quarterly seasonal quarters December- January-February (DJF), June- July-August (JJA) , September- October-November (SON) and March-April-May (MAM). CCA concerns the interdependence between two groups of variables. Therefore, it is useful to determine the dominant modes of linear covariability between two sets of data (in this case , rainfall and streamflow). The modes obtained are orthogonal to each other and can characterize the physical processes that are associated with the coupling between the two variables being able to identify a sequence of pairs of patterns in multivariate data of two fields and construct transformed variables projecting the two data fields documents on a new pair of axes so that they start to be maximally correlated (WILKS , 1995). Generally, the canonical correlation analysis shows to be a good tool to identify patterns of variability between rainfall and stream flow in large scale, allowing the identification of different modes of rainfall and streamflow variability consistent with some physical phenomena related to the process of rainfall-streamflow. This paper highlights the main modes obtained for each of the seasonal quarters, where it was possible to identify different modes of climate variability that interfere with the flow of Brazilian rivers for each of the seasonal quarters. The first mode of the DJF and SON have similar spatial configurations with the findings of Grimm (2009) for the total summer precipitation and spring respectively. The first mode of precipitation for DJF highlights a positive (negative) dipole of Central Brazil (Southern Brazil), the flow variability is observed and extends along the basins of the São Francisco River and the eastern Atlantic . The first mode of the SON shows negative correlations (positive) over most of the Southeast , Midwest , and the states of Goiás and Tocantins (western Paraná and Rio Grande do Sul). Consider the large climate variability over Brazilian, it is important to know the main weather patterns that drive the hydrological response of Brazilian rivers to best planning and management of water resources in important sectors of economy and society like flood management.