Drought is a pervasive natural hazard that propagates through the hydrologic cycle, leading to water deficiencies across key hydroclimatic variables such as precipitation, streamflow, and soil moisture. However, understanding the intricate dynamics of drought propagation remains challenging because of spatiotemporal variability and interactions among multiple drought types, including meteorological drought (MD), agricultural drought (AD), and hydrological drought (HD). Traditional analytical methods struggle to capture these complexities, particularly when using high-dimensional time series data characterized by nonlinear causality. To address these challenges, we propose a novel cause–effect-based framework that integrates the process-based Community Land Model (CLM5), a causal inference method (Peter–Clark momentary conditional independence (PCMCI)-CMIsymb), and causal effect (CE) analysis to investigate drought propagation across multiple time scales (1, 3, 6, and 12 months) in the Yellow River Basin (YRB), China. CLM5 provides high-resolution spatiotemporal drought estimations, PCMCI-CMIsymb detects nonlinear causal relationships, and CE analysis quantifies propagation probabilities across varying time lags. The results demonstrate that (a) drought propagation patterns vary by time scale, with short-term (1 month) propagation occurring primarily from MD to AD, whereas long-term (12 month) propagation extends from MD to both AD and HD; (b) subregional HDs tend to influence each other across regions, with stronger interconnections observed at long-term time scales; and (c) the likelihood of drought propagation differs spatially, with higher probabilities of MD propagation into AD and HD in the middle and lower YRB at the 6 month time scale, mostly peaking at 40%–60% with a 1 month lag. This study presents an advanced analytical framework for drought assessment, providing insights for improving drought mitigation and water resource management, particularly in regions affected by spatial hydroclimatic variability.

Read original article

Read original article