Multiple temporal scale variation characteristics and driving factors of arid inland runoff: A case study of Urumqi River, China

Urumqi River, Xinjiang, China Understanding the periodic evolution and influencing factors of runoff in arid regions is crucial for accurate hydrological modeling and effective water resource management. To explore the difference of runoff periodic characteristics between annual and monthly scales data and evaluate the mutation's impact on periodic evolution and driving factors, the classical Mann-Kendall test and cumulative anomaly curve method were used to identify Urumqi River’s runoff trends and mutation characteristics from 1978 to 2016. Multivariate Empirical Mode Decomposition and Wavelet Coherence Transform identified runoff periodic scales. While Gradient Boosting Regression Trees quantified key factors driving runoff variability. Monthly runoff of the Urumqi River exhibits a stable 12-month dominant period, reflecting the persistent influence of seasonal snowmelt and precipitation. At the annual scale, mutation events shortened and altered the dominant periodicity, shifting from a 2.8-year period in the overall stage to periods of 4.5, 2.7, and 2.5 years during different mutation stages. Before mutations, precipitation, and vapor pressure are the primary drivers; after the mutations, the influence of atmospheric pressure and sunshine duration increases, suggesting that climate variability and human activities are key factors to long-term runoff variations. In managing hydrology for alpine-inland rivers in arid regions, neglecting time scales and mutations can lead to misjudgments of trends and reduce the reliability of predictions. • Consider mutations when analyzing periodic characteristics in studies. • Mutations increase the periodicity frequency by approximately 50 %. • Monthly data captures both seasonal and short-term periodic fluctuations. • Annual data is more effective for detecting long-term periodic trends.