径流序列相空间重构的水文学含义及应用OA北大核心CSTPCDEI

To determine the hydrological meaning of the reconstructed phase space of the runoff series and improve the accuracy of mid-to long-term runoff prediction,the phase space of the runoff series was reconstructed based on chaos theory,and correlation analysis was conducted between the influencing factors of runoff and the reconstructed phase space column vectors.On this basis,a runoff prediction model coupled with Chaos theory and artificial neural network(Chaos-BPNN)was established,and applied to the Yingluoxia and Zhengyixia hydrological stations in the upper reaches of the Heihe River.The results indicate that the reconstructed phase space column vectors of the runoff series have clear hydrological meaning.The Chaos-BPNN runoff prediction model only requires runoff sequence data for modeling and prediction,avoiding the problems of difficult determination and quantification of main control factors in the runoff prediction process.The precipitation,sediment transport,water level,and temperature in the upper reaches of the Heihe River are highly correlated with the reconstructed phase space columns 1,3,6,and 7,respectively.Wind speed is not correlated with any column,and it is speculated that factors such as snow line elevation,vegetation coverage,and land use type are correlated with columns 2,4,and 5.The Chaos-BPNN runoff prediction model constructed has a runoff prediction accuracy of over 86％at the Yingluoxia and Zhengyixia hydrological stations in the upper reaches of the Heihe River.