Reducing Forecast Errors of a Regional Climate Model Using Adaptive Filters

In this work, the use of adaptive filters for reducing forecast errors produced by a Regional Climate Model (RCM) is investigated. Seasonal forecasts are compared against reanalysis data provided National Centers Environmental Prediction. The used to train based on Recursive Least Squares algorithm in order reduce error. K-means unsupervised learning obtain number employ from climate variables. proposed approach applied some variables such as meridional wind, zonal and geopotential height. Eta RCM at 40-km resolution domain covering most Brazil. Results show that capable errors, according evaluation metrics normalized mean square error, maximum absolute thus improving seasonal forecasts.