Forecasting the Yellow River runoff based on functional data analysis methods

• Published in: Environmental and ecological statistics Vol. 28; no. 1; pp. 1 - 20
• Main Authors: Wang, Ting, Zhou, Yingchun
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.03.2021; Springer Nature B.V
• Subjects: Biomedical and Life Sciences; Chemistry and Earth Sciences; China; Computer Science; Data analysis; Ecology; Health Sciences; Hydrometric stations; Life Sciences; Math. Appl. in Environmental Science; Medicine; Methods; Physics; prediction; principal component analysis; Principal components analysis; Regression analysis; River discharge; River flow; Rivers; Runoff; Statistics for Engineering; Statistics for Life Sciences; Theoretical Ecology/Statistics; Time series; time series analysis; Two dimensional analysis; Yellow River; Yellow River; China; One-dimensional functional principal component analysis; SARIMA; Yellow River runoff; Two-dimensional functional principal component analysis
• ISSN: 1352-8505; 1573-3009
• DOI: 10.1007/s10651-020-00469-x

This study examines the runoff prediction of each hydrometric station and each month in the mainstream of the Yellow River in China. From the perspective of functional data, the monthly runoff of each hydrometric station can be regarded as a function of both time and space. A sequence of such functions is formed by collecting the data over the years. We propose a new approach by combining the two-dimensional functional principal component analysis (FPCA) and time series analysis methods to predict the runoff. In the simulation, we compared the proposed method with two others: one based on one-dimensional FPCA and the seasonal auto-regressive integrated moving average (SARIMA) method. The method combining standard two-dimensional FPCA and time series analysis outperforms others in most cases, and is used to predict the runoff of each hydrometric station and each month in the Yellow River in 2018.