Calibração de um modelo chuva-vazão em bacias sem monitoramento flluviométrico a partir de curvas de permanência sintéticas

Abstract

The determination of some characteristic values of hydrologic variables and related functions at a given river cross, along with the correct assessment of how the associated streamflow records fluctuate over time, are key elements for planning, designing and operating water resources systems in gauged catchments. However, in catchments with scarce (or even none) at-site streamflow records, resorting to methods that allow transferring hydrological information from gauged to ungauged locations is most needed and absolutely crucial for developing and managing the local water resources. This MSc dissertation proposes and evaluates a novel approach for the automatic calibration of the parameters of a rainfall-runoff model at an ungauged catchment, using the long-term synthetic flow duration curve, estimated on the basis of a statistical regional model, as a paradigm for the local hydrology. In this context, the estimated long-term flow duration curve is regarded as a useful synthesis of the local hydrological regime which may be reasonably approximated from the discharges simulated by an appropriate conceptual rainfall-runoff model with an adequate set of calibrated parameters. Once the set of calibrated parameters is obtained for a given ungauged catchment, the hydrologic model can be used to transform existing daily rainfall series into simulated daily discharges, thus enabling one to perform standard hydrologic studies such as the frequency analysis of floods and droughts, reservoir water balance, flow routing and the assessment of the local water resources, among others.The approach proposed herein consists basically of two phases. The first relates to the construction of a regional statistical model for estimating the long-term flow duration curve at an ungauged catchment, by spatially transferring hydrological information from existing gauged sites within a hydrologically homogenous region. The second phase refers to the calibration of the parameters of the RIO GRANDE simulation model, a conceptual model largely based on the well known Xinanjiang water balance structure, having the estimated flow duration curve as the paradigmatic object to be mimicked. The proposed approach has been applied and verified for the Pará and Paraopeba river basins, both located in the state of Minas Gerais, in southeastern Brazil. In both phases, a jackknife. experiment has been devised in order to calculate the associated performance indexes and finally assess the overall reliability attained by the proposed approach.