Acoplamento do modelo atmosférico WRF ao modelo hidrológico SCS/HMS para a previsão de cheias na Bacia do Paraíba do Meio em Alagoas

Abstract

The basin of the Paraíba do Meio river located between the states of Pernambuco and Alagoas in the northeast region, totaling an area of 3.148,5km2, distributed among 21 cities, eight of them belong to the state of Pernambuco, and the other 13 to the state of Alagoas. As it happens in most Brazilian cities, the process of the urbanization of the municipalities of the valley of the Paraíba do Meio river happened intensely and erratically on the river, which consequently led to high vulnerability to events of flood. The impacts caused by the floods in the region are: material and social losses as well, disruption of economic activities in the affected areas, and loss of human life. Regarding the problem, this study aimed to evaluate the unidirectional coupling between the atmospheric forecasting mesoscale model WRF (Weather Research and Forecasting) and a hydrological model formulated in HEC-HMS program (Hydrologic Engineering Center - Hydrologic Modeling System) in the perspective of the development of a monitoring and warning system, in order to reduce the damage caused by extreme flow events in the region. Therefore, the FNL data analysis were used (Final Analysis GFS) resulting from the Global Forecast System model (GFS) as initial and boundary conditions of the WRF model for simulation of rainfall, and data on hourly scale precipitation and flow, collected in Data collected in vests Platforms (DCPs), which were used in the phase of adjustment and validation of both models. To assess the results generated by the models during and after calibration, we used seven statistical bias indicators, Bias, to evaluate the existence of systematic errors; the root of mean squared error to estimate the magnitude of the error of the simulated values; Pearson correlation coefficient, to measure the degree of relationship among the simulated and observed series; and the quantifiable measures, Nash-Sutcliffe, Heidke Skill Score (HSS), hit rate (H) and the false alarm ratio (FAR), to assess the predictive ability of the models involved and the unidirectional coupling technique. After the calibration and validation steps, the results were analyzed statistically and visually. The atmospheric model WRF showed underestimation of the observed rainfall; however, its predictability was considered satisfactory according to the indicators. In assessing the hydrological model SCS/HMS formulated in HEC-HMS program, acceptable results were obtained in the phases of calibration and validation. Correlation coefficients were greater than 0.90 in two stages, despite the evidence of the presence of systematic error, and thus, underestimating the flow observed according to the statistical criteria. The forecast of the hydrological model was rated between satisfactory and acceptable by the efficiency coefficients evaluated. The corresponding analysis to the coupling indicated correlations above 0.75, with satisfactory predictability (Nash = 0.572 and Nash = 0.749) for a forecasting system. However, the model underestimated 15% to 44% the flow rates observed in the analyzed points. Overall, the analysis of the method used for the purpose of the development of an early warning system to floods showed promising results in the short-term flow forecasts.