%0 Audiovisual Material
%4 sid.inpe.br/mtc-m16c/2020/11.05.16.47
%2 sid.inpe.br/mtc-m16c/2020/11.05.16.47.36
%T Combining rainfall and water level data for multistep very short-term empirical hydrological forecasting
%D 2020
%A Freitas, Cintia Pereira de,
%@affiliation Instituto Nacional de Pesquisas Espaciais (INPE)
%@electronicmailaddress cintia.cpfreitas@gmail.com
%B Workshop dos Cursos de Computação Aplicada do INPE, 20 (WORCAP)
%C São José dos Campos
%8 8-11 e 14-17 set. 2020
%I Instituto Nacional de Pesquisas Espaciais (INPE)
%J São José dos Campos
%S Vídeos
%X This work proposes a MLP to forecast time series of water level at the outlet of a watershed in a mountain region. Input data was collected by a set of 8 hydrologic monitoring stations, acquired with a 15-minute resolution. Our results show a good agreement between predicted and observed values of the water-level time series, combining water level and rainfall data in the input layer: Nash-Sutcliffe coefficient reached 0.995 for prediction antecedence of 15 minutes, and 0.913, for 120 minutes.
%9 modelagem computacional
%@language pt
%3 Rainfall and water level data for very short-term hydrological forecasting - Cintia Freitas.mp4
%O (15 min)