Antecedent rainfall could be a critical prerequisite for debris-flow triggering on steep slopes of arid regions

Debris flows are fluidized, unconsolidated sediments that gravitationally flow downslope, and constitute one of the most impactful natural hazards in mountainous regions, with casualties and damage to infrastructures. They are typically triggered by heavy rain or sudden ice melt in mountainous and volcanic areas. In arid regions, where vegetation is sparse and not stabilizing, debris flows are occasionally observed when torrential rain showers hit the steep slopes. This is the case of our study area: the arid slopes of the Eastern Judean Desert, on the western margins of the Dead Sea. In this region, the mean annual precipitation does not exceed 100 mm yr^-1. Currently, debris flows in this area are not considered an important hazard, because they are very rare and they mostly endanger infrastructures of natural reserves and main roads. However, previous studies reported a significant increase in their frequency during a late Holocene dry period, raising the question of whether their future occurrence could be affected by climate change. In this study, we focus on the critical rainfall conditions for debris flow triggering in these arid areas, which were not fully addressed by previous studies due to the small number of reported events. We combine high-resolution digital terrain models, to systematically identify small-size debris flows, with high-resolution weather radar data, to represent rainfall conditions corresponding to the debris flow locations. We identify over 40 debris flows by comparing digital elevation models available for the period 2013-2019. The deposits are relatively small (a few tens of meters) and are usually observed along the steepest slopes of the escarpment, at the outlet of small ephemeral streams. We divide the debris flows into four groups based on their spatial and temporal distribution. Using radar data and witness information, we identify three storms as the most likely triggering events for these groups, and we isolate the convective cells that led to the triggering. In all cases, debris flows were triggered by an intense convective cell (lasting 30 min to 1 hour) which was preceded by significant rainfall amounts (8-12 mm) delivered over relatively long times during the storm. Comparing triggering and non-triggering storms, we observe that rain intensity alone is insufficient to explain the phenomena: we discuss the possibility that antecedent conditions could represent a critical factor for the triggering of debris flows in steep slopes of arid environments.