Computer Simulation of Streamflows with GAR(1)-Monthly and GAR(1)-Fragments Models
Authors: Nguyen Van Hung, Huynh Ngoc Phien, Tran Quoc Chien,
Number of views: 435
in streamflow simulation, the first-order gamma autoregressive (GAR(1)) model  has been found to be very effective for annual data. This paper presents some attempts to apply the GAR(1) model to the simulation of monthly streamflows. To this aim, we propose two models, namely the GAR(1)-Monthly and GAR(1)-Fragments models that will be compared with the popular Thomas-Fiering model. Based on actual data of monthly streamflows at three stations and generated series of monthly data for 1000 years, it was found that both GAR(1)-Monthly and GAR(1)-Fragments models can reproduce very well all statistical descriptors, namely mean value, standard deviation and skewness coefficient, of the historical monthly series. Moreover, the GAR(1)-Fragments model was found to perform very well in reproducing those statistical descriptors of historical annual flows also.