Annual Runoff Forecasting Based on Multi-Model Information Fusion and Residual Error Correction in the Ganjiang River Basin

Abstract

Accurate forecasting of annual runoff time series is of great significance for water resources planning and management. However, considering that the number of forecasting factors is numerous, a single forecasting model has certain limitations and a runoff time series consists of complex nonlinear and nonstationary characteristics, which make the runoff forecasting difficult. Aimed at improving the prediction accuracy of annual runoff time series, the principal components analysis (PCA) method is adopted to reduce the complexity of forecasting factors, and a modified coupling forecasting model based on multiple linear regression (MLR), back propagation neural network (BPNN), Elman neural network (ENN), and particle swarm optimization-support vector machine for regression (PSO-SVR) is proposed and applied in the Dongbei Hydrological Station in the Ganjiang River Basin. Firstly, from two conventional factors (i.e., rainfall, runoff) and 130 atmospheric circulation indexes (i.e., 88 atmospheric circulation indexes, 26 sea temperature indexes, 16 other indexes), principal components generated by linear mapping are screened as forecasting factors. Then, based on above forecasting factors, four forecasting models including MLR, BPNN, ENN, and PSO-SVR are developed to predict annual runoff time series. Subsequently, a coupling model composed of BPNN, ENN, and PSO-SVR is constructed by means of a multi-model information fusion taking three hydrological years (i.e., wet year, normal year, dry year) into consideration. Finally, according to residual error correction, a modified coupling forecasting model is introduced so as to further improve the accuracy of the predicted annual runoff time series in the verification period.

A Hybrid Model Integrating Elman Neural Network with Variational Mode Decomposition and Box–Cox Transformation for Monthly Runoff Time Series Prediction

“…And the standalone SVM, which is based on the risk minimization principle, has more robustness and generalizability for estimating non-stationary and skewed time series than the Elman does. The Elman model provides better predicting results in training period and SVM provides better predicting results in testing period, which is consistent with other studies(Song et al 2020;Feng et al 2020). …”

Simulation and Prediction of the Impact of Climate Change Scenarios on Runoff of Typical Watersheds in Changbai Mountains, China

“…Currently, in making predictions and assessments based on the evolution of the hydrological cycle under different future climate change scenarios [4][5][6][7][8], medium and longterm runoff forecasting has become increasingly important [9,10]. Research usually selects future climate scenarios and climate models or uses mathematical model calculations and downscaling methods to process the data and then analyzes the evolution of future climate and hydrology [11][12][13], while carrying out prediction and response research on hydrological elements [14,15].…”

Evaluating the impact of climate change on surface water resources in the upper Ganjiang River Basin, China

“…As the longest river of the PLB, the Ganjiang reaches a length of 766 km and mainly lies in the Jiangxi Province. The GRB has four distinctive seasons, that is, summers are hot and rainy while winters are dry and cold, with annual precipitation ranging from 1,400 to 1,800 mm and mean temperature of 17.8 °C (Song et al 2020).…”