Non-stationarity analysis of flood extremes under climate change and human activities in the Yuan River Basin, China

Abstract

Study region: Yuan River Basin in China. Study focus: This study aims to identify temporal variations and non-stationarity in the annual maximum Peak Flow (PF) and Peak Water Level (PWL) series, along with their occurrence dates and intervals series at three stations (Anjiang, Pushi, and Taoyuan). Temporal variations were detected and non-stationary models were developed by incorporating physically-based covariates under climate change and human activities using the Mann-Kendall test, Pettitt test, and the Generalized Additive Models for Location, Scale, and Shape (GAMLSS) framework. New hydrological insights for the region: (1) The intervals between Pushi and Taoyuan stations for both PF and PWL series have significantly shortened over the past 40 years, while the Taoyuan station showed no significant upward trend over the past 70 years. (2) Covariates associated with precipitation, Normalized Difference Vegetation Index (NDVI), Reservoir Index, and Impervious Area (IA) showed a significantly increased trend within the basin, particularly the annual maximum daily precipitation (P1) at the Pushi station. (3) The non-stationary models performed best when incorporating either the 7-day or 15-day accumulated antecedent precipitation before the flood occurrence date for the flood extremes series, while the inclusion of the IA and mean NDVI three months before the flood occurrence month provided superior fitting for the occurrence dates series. (4) The variability of P1, 3-day accumulated precipitation and their overlap with 7-day and 15-day accumulated precipitation were most likely key factors in triggering typical flood extremes.

Keywords

• Climate change
• Flood non-stationarity
• GAMLSS
• Human activities
• Physically-based covariates
• Yuan River Basin