<p>Empirical rainfall thresholds are widely used for debris flow early warning due to their simplicity, but they often produce false alarms by ignoring non-triggering rainfall uncertainty and hydrological processes. To address this, we propose probabilistic thresholds based on two years of in-situ monitoring data, incorporating observed soil moisture and calculated water runoff. To our knowledge, this is the first time that calculated water runoff has been used as a factor in generating probabilistic hydro-meteorological thresholds for debris flows. Our findings indicate that the probabilistic hydro-meteorological thresholds slightly outperform the probabilistic cumulative rainfall-duration (<i>E-D</i>) thresholds, with accuracy values improving from 0.92 to 0.94. Compared to the probabilistic hydro-meteorological thresholds derived from observed soil moisture content, those derived from calculated water runoff offer significant potential for improving early warning systems for post-earthquake debris flows, as calculated water runoff can be predicted in advance. We argue that the findings of this study provide valuable insights for enhancing the accuracy of post-earthquake debris flow prediction.</p>