Filtragem adaptativa para a redução de desvios em séries temporais de previsão numérica climática

This article presents results of the application of adaptative filters, using algorithms NLMS (Normalized Least Mean Square) and RLS (Recursive Least Square), for the reduc- tion of deviations in weather forecasts. The climatic parameter that me consider is the meridional wind in the region covered by the atmospheric forecast model Eta/CPTEC with resolution of 40km. Numerical models provide atmospheric forecast that have errors. Were used adaptive filters to correct the forecast error. Results show that the adaptative filter reduces the mean square error of the Eta model with respect to the re-analysis provided by NCEP, improving weather forecasting