Impacto da assimilação de dados de Radar na previsão de tempo sobre a região Metropolitana de Manaus

Abstract

The Amazon region has high rainfall, reaching more than 3000 mm.year-1. Over the city of Manaus, there are frequent occurrences of meteorological systems that generate expressive volumes of rain, causing great impacts on the population. With computational advances, meteorological models have become an increasingly indispensable tool for forecasting these systems. The use of the data assimilation (AD) technique for short-term forecasting, nowcasting, has shown significant improvements in the simulations of these more extreme events, where the use of weather radar data stands out. The present work aims to evaluate the application of AD using Censipam - AM weather radar data together with the WRF model for short-term forecasts (nowcasting), where 5 events were selected for the simulations. The simulations were carried out by applying the AD technique in a cyclical and non-cyclical way, assimilating reflectivity and radial wind speed data, being used individually and concurrently. From the results obtained, it was observed that the use of reflectivity only during the AD, whether in cycle or not, presented similar and inferior results compared to the control experiment (without assimilation). When using reflectivity and radial velocity together during the AD, significant improvements were observed compared to the control experiment, both in cyclical and non-cyclical forms, mainly in the first 3h of the forecast. It was observed that the radial velocity assimilation provided a more satisfactory representation of convection, which contributed significantly to improving the representation of the systems studied here.