Trigger thresholds and propagation mechanism of meteorological drought to agricultural drought in an inland river basin

Quantifying the thresholds and processes of drought propagation is of great significance for early drought warning and ecosystem management. Our understanding of their spatial patterns and driving mechanism remains unclear. In this study, based on Copula functions, we quantified the thresholds and process of meteorological drought to agricultural drought in an alpine-oasis-desert inland river basin of China for the period of 1980–2020. Furthermore, the main factors driving drought propagation were identified using the Random Forest model. The results showed that: (1) significant spatial heterogeneity exists in the propagation of meteorological to agricultural drought, with longer propagation time and higher propagation risk in the upstream; (2) from upstream to downstream, the percentile-based average cumulative precipitation deficit threshold for triggering agricultural drought ranged from 18.5 % to 45.0 % under moderate probability conditions (greater than 0.6); (3) the response of agricultural drought to meteorological drought was characterized by intensity amplification and duration attenuation in the upstream, while the opposite occurred downstream. This response is mainly driven by the interactions of actual evapotranspiration ( ET a ) and vapor pressure deficit (VPD). Specifically, ET a and VPD contributed 15.7 %–54.8 % and 8.7 %–39.5 %, respectively. Additionally, irrigation also plays an important role in drought propagation, contributing 6.5 %–9.6 %. This study provides important implications and valuable insights for understanding the mechanisms of agricultural drought formation. Furthermore, the results can provide scientific guidance for watershed water allocation, drought preparedness and risk management. • Stronger land-atmosphere interactions lead to higher propagation risk upstream in the SRB (Shiyang River Basin). • Precipitation deficit threshold triggering agricultural drought spatially ranged from 18.5 % to 45.0 % in the SRB. • ET a and VPD contributed 15.7 %–54.8 % and 8.7 %–39.5 % of the strength of drought propagation, respectively.