Integrated atmospheric and hydrologic modelling for short-term and basin-scale forecasts in a tropical semi-arid context

Precipitation in the Brazilian northeastern semi-arid region is characterized by a convective rainfall regime. Recently, numerical weather prediction models have been run with resolutions fine enough to potentially improve the quality of the simulated rainfall patterns within hydrographic basins and therefore contribute to more reliable flow forecasts. This paper presents a unidirectional coupling between BRAMS - Brazilian Regional Atmospheric Modelling System - and a lumped hydrological model in order to analyse the impact of precipitation uncertainties on runoff forecasts. The results show that forecasts produced one day in advance lead to reliable hydrographs, but for two to five days in advance, precipitation uncertainties impact the runoff forecast skill.