Abstract: A systematic quantification of the thresholds for the propagation from meteorological drought to different levels of hydrological drought is useful in guiding the early warning and delicacy management of hydrological drought. In this paper, a standardized precipitation index and a standardized runoff index are used to characterize meteorological drought and hydrological drought, respectively, for the Qinhe River basin of Loess Plateau; the run theory is applied to identify drought events, and merge or reject drought events to match both types of drought events. Then, Bayesian networks combined with Copula functions are used to construct models to solve for the propagation thresholds. Results show that when the level of hydrological drought is raised, the threshold is elevated and the intensity of a meteorological drought is weakened in the propagation. For the Qinhe River basin, the duration thresholds of meteorological droughts are 12.8, 21.8 and 30.9 months for triggering moderate, severe and extreme hydrological droughts, respectively, corresponding to the intensity thresholds of 14.2, 22.4, and 30.0, respectively, and its drought tolerance ability is related to the basin meteorological and underlying surface conditions.

Key words: meteorological drought, hydrologic drought, copulas functions, Bayesian network, drought threshold, Qinhe River basin