Meu SciELO
Serviços Personalizados
Artigo
Indicadores
- Citado por SciELO
Links relacionados
- Similares em SciELO
Compartilhar
Revista Cubana de Meteorología
versão On-line ISSN 2664-0880
Resumo
FUENTES, A.; SIERRA, M. e MORFA, Y.. Quantitative precipitation forecast correction using neural network. Rev. Cubana Met. [online]. 2020, vol.26, n.3 Epub 01-Set-2020. ISSN 2664-0880.
In this paper, a model of neural networks is proposed as an effective technique for the correction of the quantitative precipitation forecast provided by the WRF model. For this, a Multi-Layer Perceptron is used, with the aim of using the output (observations provided by the surface stations) to establish a relationship with the input elements (WRF outputs). Model training is carried out with real rainfall accumulation data corresponding to 2017; and the evaluation is carried out with the period between November 4, 2018 and February 28, 2019. The correction of the quantitative precipitation forecast in the analyzed stations was achieved, the improvement for the mountain station was more significant and, in cases where the WRF overestimates the accumulated rainfall.
Palavras-chave : artificial neural networks; quantitative forecast precipitation; WRF; bias correction.