%0 Conference Proceedings
%4 cptec.inpe.br/walmeida/2004/10.14.17.00
%2 cptec.inpe.br/walmeida/2004/10.14.17.00.52
%T Verificação de um modelo discriminadamente de previsão das precipitações sazonais no nordeste
%D 1998
%A Sansigolo, Clovis Angeli,
%@affiliation Instituto Nacional de Pesquisas Espaciais (INPE)
%B Congresso Brasileiro de Meteorologia, 10.
%C Brasília
%8 26-30 out.
%S Anais
%1 SBMET
%K Precipitações sazonais, nordeste, precipitation, forecasting.
%X The most commonly used skill scores, including that of Heidke, Kuippers, average distance, Brier and LEPS, were used to evaluate a linear discriminant forecasting model of northeastem Brazil seasonal rainfall. The Heidke and the Kuipers scores overestimated the model performance. The Brier and probabilistic LEPS scores agreed conceming to wet and average categorie forecastings, as the average distance and probs. LEPS for the dry ones. The crossvalidation results of the forecasting model were, according to the probs. LEPS scores, good for wet categories (38% improvement over climate), reasonable for dry categories (29%), and poor for average (5%).
%@language pt
%3 Sansigolo_Verificacao de um modelo.pdf