Using the PRISMA framework to conduct a comprehensive study of source stream occurrence in mountain areas

Abstract

Headwater stream blockages present a significant hydrogeomorphic challenge, often linked to landslides, flash floods, and debris flows in mountainous regions. Blockage caused by landslides, debris flows, or human activities can create temporary reservoirs, heightening the risk of sudden dam failure and severe downstream impacts. As climate change intensifies the frequency and severity of extreme rainfall, the threat of stream blockages is rising in Vietnam's mountainous river basins. This study systematically reviews recent advances in the mechanisms, detection methods, and predictive modelling of stream blockage using a structured evidence synthesis approach consistent with PRISMA 2020 guidelines. The review reveals that current research is predominantly fragmented, focusing either on slope instability thresholds or post-failure dam-break hydraulics, with limited efforts to model the full event chain linking extreme rainfall, landslides, headwater stream blockages, temporary impoundment, and dam failure. Based on the synthesis, this study proposes a conceptual integrated framework for early warning that combines dynamic rainfall thresholds, probabilistic slope instability modelling, satellite-based blockage detection, and two-dimensional flood routing simulation. The framework emphasises the need for physics-informed machine learning and multi-source data integration to reduce uncertainty and enhance interpretability. In the context of intensifying extreme precipitation under climate change and increasing anthropogenic disturbance in mountainous catchments, the findings provide a strategic direction for advancing probabilistic, process-based hazard assessment of landslide-induced upstream flow blockage. The proposed framework offers a scientific foundation for developing operational early warning systems in data-scarce, tectonically active monsoon regions such as Northwestern Vietnam.