Parameter selection for phase space reconstruction in hydrological series and rationality analysis of its chaotic characteristics

EasyChair Preprint 516

Abstract

This study performed a rationality analysis of the delay time and embedding dimension value during phase space reconstruction in hydrological series and the effect on their chaotic characteristics. Using a monthly average runoff time series from the Ayanqian station (upstream) and the Jiangqiao station (midstream) in the Nen River Basin, we reached the following regularity conclusions. ① Based on the flood season (4 months) in the Nen River Basin, we can deduce that the correlation sequence length for the runoff is 4~5 months, i.e., the delay time =3 or 4 is a reasonable choice. ② Learn from the predictability experiment results for the monthly rainfall time series, we know that the calculation results of the G-P algorithm for the dimension of runoff series for the Nen River Basin are reasonable, i.e., the embedding dimension is no more than seven. ③ the most suitable parameters for the phase space reconstruction and its chaotic characteristic index in the Nen River Basin are as follows: delay time = 3~4, embedding dimension = 6~7, correlation dimension = 2.90~3.00, maximum Lyapunov index = 0.24~0.32, and the forecast time is 3~4 months.

Keyphrases: chaos theory, delay time, embedding dimension, phase space reconstruction

@booklet{EasyChair:516,
  author    = {Hongyan Li and Shanshan Bao and Yunqing Xuan},
  title     = {Parameter selection for phase space reconstruction in hydrological series and rationality analysis of its chaotic characteristics},
  howpublished = {EasyChair Preprint 516},
  year      = {EasyChair, 2018}}