Previsão da temperatura máxima: análise de dois modelos de previsão através de dados de radiossondagem

Abstract

The temperature plays a crucial role in the physiological processes of animals and plants, particularly in sectors such as agricultural planning and adaptation of crops in different regions. Accurate prediction of maximum temperature (Tmax) is essential for forecasting all adverse meteorological phenomena. The main objective is to evaluate two forecasting models using radiosonde data and provide commendations for improving these forecasts. Two forecasting models were evaluated using statistical parameters and a synoptic analysis for the Natal – Rio Grande do Norte, based on daily data obtained at 00 UTC and 12 UTC for the years 2019 and 2020. Method 1 uses an equation that relates the predicted temperature to the temperature at the top of the boundary layer. In Method 2, the predicted temperature can be calculated using the Skew-T thermodynamic diagram. Data validation was carried out using statistical parameters such as bias, mean square error, root mean square error, Pearson coefficient, and determination coefficient. For the six analyzed cases with the highest prediction errors, a detailed synoptic analysis was conducted using satellite images, streamlines, and thermodynamic profiles. The results revealed two problems contributing to errors: 1) errors in the identification and forecasting of cloudiness, and 2) the presence of intense inversion layers. It was observed that cloudiness prediction errors were associated with synoptic system couplings, such as Intertropical Convergence Zone, South Atlantic Convergence Zone, frontal zones, and Upper Tropospheric Cyclonic Vortex. Solutions to these problems were proposed.