Antecedent rainfall patterns and situational vulnerability of victims of rainfall-induced fatal landslides in Pakistan

Abstract. Fatalities from rainfall-induced landslides depend not only on rainfall conditions but also on the time of occurrence, the exposure setting and demographic vulnerability. In data-scarce mountain regions such as Pakistan, event-scale rainfall histories and individual casualty records have long been disconnected. This study places event-scale rainfall and individual victim circumstances within a single event catalogue, characterises antecedent rainfall patterns and tests whether they explain fatality magnitude, in order to inform scenario-based early warning in data-scarce mountains. Based on a 2004–2024 catalogue of rainfall-induced fatal landslides in Pakistan (positioned as a lower-bound estimate because flood–landslide–debris-flow–building-collapse cascades mask landslide-attributed deaths), 213 of 250 events with spatial accuracy levels I–II were matched to CHIRPS 0.05° daily rainfall, and individual attributes of 309 fatalities during 2018–2024 were compiled. Results show that 53.5 % of events have zero event-day rainfall in CHIRPS; among four antecedent-rainfall types defined by event-day and 7-day antecedent rainfall, the low-event-day-rainfall types C and D together account for 86.4 %, and sensitivity tests using 15/40, 20/50 and 25/60 mm boundaries give 81.2–89.7 %, indicating robustness to the classification boundaries. The intensity–duration quantile relations share a common exponent of −0.59, yet no rainfall metric correlates significantly with per-event fatality. Victim records show that 66.0 % of deaths with known setting occurred indoors and 74.5 % of those with known timing occurred at night, making the night–indoor combination the most concentrated exposure; both gender and age are significantly associated with exposure setting (χ² = 28.61, Cramér's V = 0.367; χ² = 35.34, V = 0.362; p < 0.001), and the setting–timing association is also significant (χ² = 70.87, df = 2, p < 0.001, V = 0.594). Rainfall therefore characterises the hydro-meteorological background of slope instability, whereas fatality consequences additionally depend on timing, exposed space and demographic attributes. Early-warning systems should consider both short-duration and antecedent cumulative rainfall, and translate warnings into scenario-based actions such as night-time evacuation of dwellings, mountain road control and management of children's activity spaces.