Proposta de regionalização hidrológica para a bacia hidrográfica do rio Paranaíba

Abstract

The regionalization of flows is a technique that allows transference of information between similar river basins in order to calculate the fluviometric variables of interest in sub-basins without data, in this case, the reference flows Q7,10; Q90 and Q95. Likewise, it is characterized as being a useful tool in obtaining the grant of water rights, an instrument provided for in Brazilian Law 9433/97. Although there are information and studies about Q7,10 Q90 and Q95 for the State of Minas Gerais in relevant works, such as Run-off of Minas Gerais, Digital Map of Water in Minas Gerais, Water Resources Management Plans available and officially published on Hydrographic Basins and academic studies, for the purpose of interacting these with other segments that adopted other references to the physical space, this study aimed to investigate, with the purpose of expanding as a tool for management agencies, information and, consequently, a new model of hydrological regionalization for Q7,10, Q90 and Q95 in Paranaíba River Basin. The precision of the two models obtained, by the inverse of the distance to the power (IDW) and by ordinary kriging (KO), with respect to the prediction of flows used as Grant criterion was obtained by cross validation, together with Coefficient of Nash-Sutcliffe (COE), using the flows generated by the models as estimated values, and as a standard value the calculation using historical series of fluviometric stations. The new models presented the efficiencies: for Q7,10 in MG COE Map (0.49) <Run-off (0.70) <KO (0.82) <IDW (0.84). Q7,10 for GO, MS, DF COE remained KO (0.87) <IDW (0.81). For modeling Q95 in MG was COE Map (0.75) <IDW (0.84) <KO (0.85); Q95 in MS, GO, DF was COE KO (0.56) <IDW (0.84). The reference flow Q90 was for MG COE Map (0.83) <IDW (0.94) KO (0.95). While for MS, GO and DF COE KO (0.96) <IDW (0.98). It can be concluded that IDW and KO models have presented very good efficiency, except for Q95 in GO, MS, DF, which has presented only as satisfactory to the KO model. The models proposed in this study have shown better efficiency in the estimation of the results for all circumstances when compared with the other models compared herein. Therefore, the new methodology presents potential when compared to those commonly used in this type of study.