Aosta Valley, an Alpine region in north-western Italy, has an early warning system (EWS) that issues hydrogeological alerts based on hydrological modelling and rainfall thresholds that identify the possibility of shallow landslides being triggered in different areas of the region. The high headwater catchments are characterized by the presence of permafrost and glacial sediments, and they are frequently prone to debris flows. The summer debris flows are initiated by short-duration, high-intensity rainstorms, which are associated with high meteorological uncertainty; therefore, they are not always detected by the early warning system of shallow landslides. In this study, the hydro-meteorological and permafrost conditions related to the debris-flow events from 2013 to 2018 are investigated in order to determine the variables affecting the triggering of debris flows. Debris-Flow Indicator (DFI), an early warning system specific for debris flows, was developed using recorded air temperatures, thunderstorm alerts and forecast rainfall. Two alert levels of the DFI were defined by two thresholds (S1 and S2) of the freezing level determined from performance metrics. The performance of the DFI was then studied with a back-analysis from 2013 to 2019, using observed air temperatures and forecast rainfalls. This analysis showed that the experimental implementation of the DFI in the EWS of the Aosta Valley region resulted in detecting most of the events with some false alerts (for the lower threshold, S1) or detecting only major events, but without generating false alerts (for the higher threshold, S2). Consequently, the DFI can be applied for issuing debris-flow alerts for large areas in mountain regions based only on meteorological data and forecast.

奥斯塔山谷（Aosta Valley）是意大利西北部的高山地区，拥有预警系统（EWS），该预警系统根据水文模型和降雨阈值发布水文地质警报，以识别在该地区不同区域引发浅层滑坡的可能性。高水位集水区的特征是存在多年冻土和冰川沉积物，它们经常容易产生泥石流。夏季泥石流是由短期，高强度的暴雨引发的，这与高度的气象不确定性有关；因此，浅层滑坡预警系统并不总是能够检测到它们。在这个研究中，为了确定影响泥石流触发的变量，研究了2013年至2018年与泥石流事件相关的水文气象和多年冻土条件。泥石流指示器（DFI）是专门针对泥石流的预警系统，它是使用记录的气温，雷暴警报和预测降雨量开发的。DFI的两个警报级别由根据性能指标确定的冻结级别的两个阈值（S1和S2）定义。然后，使用观测到的气温和预测的降雨量，通过对2013年至2019年的回溯分析研究了DFI的性能。此分析表明，在奥斯塔山谷地区EWS中DFI的实验性实施导致检测出大多数事件，并带有一些错误警报（对于较低的阈值，S1）或仅检测重大事件，但不生成错误警报（对于较高阈值，S2）。因此，仅基于气象数据和预报，DFI即可用于发布山区大面积的泥石流警报。