A Multi-Layer Hierarchical Combinative Model for the Prediction of Regional Groundwater Level

In this paper,a new combinative prediction model for thd time-series which has a periodic feature is proposed. In the model,the real measured series is decomposed into an annual average,an annual amplitude and a remains series. The multi-layer hierarchical models are used to model and predict for the average and the amplitude series because of their time-varying feature, the remains series is modelled with ARMA model. The case study shows that the model has good resuits for the prediciton of groundwater level dynamics.