The research team, led by scientists at the University of Nottingham Ningbo China and the Chinese Academy of Sciences, screened peer-reviewed literature from the Web of Science and Scopus databases using strict inclusion criteria. Out of an initial search, 1,967 papers met the bar for detailed analysis. Using bibliometric mapping and dynamic topic modelling alongside qualitative synthesis, the team uncovered stark imbalances in where and how flash-flood science is done. Research is heavily concentrated in Europe and Asia, with China and India hosting the largest evidence bases among studies with identifiable locations, followed by Italy, the United States, Egypt, France, Spain, and Saudi Arabia.

That concentration matters because it shapes what the world knows about flash floods in the first place. The authors caution that publication patterns reflect monitoring density, language, database coverage, institutional capacity, and funding priorities rather than a direct measure of where floods strike hardest. Clusters of studies in Ethiopia, Morocco, and South Africa stand against substantial monitoring and reporting gaps across much of Africa and South America, a north-south imbalance the review says ultimately hinders locally tailored risk-reduction strategies. Evidence from the Andes and the East African Highlands remains limited and fragmented, with the East African Highlands represented by only seven studies in the system-resolved subset.

Thematic analysis revealed six interlocking research domains: meteorological forcing and rainfall drivers; arid wadi flash flooding; sediment processes; special geological and topographic settings; risk management and adaptation strategy; and eco-climate impacts. Risk management dominated the corpus, accounting for 683 publications, or 37.4 percent, and has grown steadily since 2014. At the other extreme, special geological and topographic settings, covering hazards such as karst hydrology, glacier-lake outbursts, and volcano-tectonic terrain, remain the least studied despite their high consequence. The review also found that only 72 of the risk-management papers overlapped with sediment-process research, suggesting process understanding and governance remain poorly connected.

At the heart of the review is a technical argument about how flash floods begin. The traditional approach relies on fixed rainfall thresholds, often expressed as intensity-duration-frequency relationships. But the evidence compiled shows that triggering is strongly state-dependent. In Egypt’s Wadi Al-Ambagi watershed, just 10 millimeters of effective rainfall generated roughly 15.7 million cubic meters of runoff, while documented events elsewhere ranged from 43 to 1,631 millimeters of cumulative rainfall. Modelling studies show that raising antecedent soil saturation from 0.20 to 0.50 on a dimensionless scale cut threshold rainfall by 8 to 38 percent across studied catchments. In other words, the same storm can be benign in a dry catchment and catastrophic in a wet one.

Surface conditions and connectivity amplify or dampen these effects. Vegetation acts as what the authors describe as a green reservoir, slowing overland flow; along Ethiopia’s Rift Valley escarpment, peak discharge coefficients correlate strongly and negatively with forest and grassland cover. Wildfire disrupts this protection: post-fire experiments show ash-laden flows forming steep flood fronts that can bulk into debris flows, and in Colorado’s Front Range, susceptibility to infiltration-excess runoff peaked in the first two years after burning before recovering by years five to seven. Landslides and earthquakes leave similar legacies; in the Wenchuan earthquake zone, modelling indicates that when six-hour rainfall exceeds roughly 235 millimeters, peak flows in local sub-catchments can rise by up to 87 percent. Urbanization compounds the problem, with one Mediterranean study estimating that a once-in-50-year rainfall event would produce a natural flood peak of 380 cubic meters per second but 940 cubic meters per second after urban development.

The review then traces how physical hazards become disasters through exposure, vulnerability, and capacity. Mountain settlements cluster on valley floors and alluvial fans where flat land is scarce, and rapid development often pushes construction into hazardous zones, including 1,021 buildings documented on 20-to-45-degree slopes in Petrópolis, Brazil, where 258 millimeters of rain in three hours in February 2022 killed at least 231 people. Recovery is uneven too: in rural Cameroon, only 5 percent of affected households had fully restored livelihoods six weeks after a severe flood, and a rapid assessment after Pakistan’s 2022 floods found 77 percent of female respondents had limited access to sexual and reproductive health services. Cultural heritage, from sections of the Great Wall to Himalayan temples, adds losses that cannot be measured in replacement cost alone.

Perhaps the most striking section deals with cascading hazard chains, where one disaster triggers another. The 2023 South Lhonak event in Sikkim, India, began when a landslide entered a glacial lake, overtopped and breached the moraine dam, and sent a glacial lake outburst flood downstream that contributed to the failure of the Chungthang Dam. In Libya’s Derna disaster, extreme flooding caused the sequential failure of two upstream dams, producing successive destructive flood waves through the city. Simulation studies cited in the review suggest that without the existing dam infrastructure, the same storm would have produced lower maximum flood depths and inundation in Derna, underscoring the need to re-examine aging structures designed for historical climates.

On solutions, the authors argue that effective risk reduction requires coordinated portfolios rather than isolated interventions. Non-structural measures, including dynamic early warning that combines rainfall intensity and duration with soil moisture and catchment conditions, spatial planning, and community preparedness, form one pillar. Structural measures form another: sediment-filled check-dam systems on China’s Loess Plateau can cut downstream peak flows by 60 to 69 percent, and two detention basins modelled for Madeira could reduce peak discharge by 72 percent under a 100-year rainfall scenario. Nature-based solutions complement these, with small wetlands reported to retain 50 to 70 percent of rainfall and green infrastructure in Belo Horizonte, Brazil, reducing outlet peak flows by roughly 60 percent when implemented across all suitable areas, though effectiveness scales sharply with extent and takes years to establish.

The review closes with an adaptive governance framework organized around three functions drawn from UNDRR guidance: preventing new risk through planning and zoning before development occurs, reducing existing risk through monitoring, protection, and preparedness, and managing residual risk through emergency response, insurance, and post-event learning. The authors stress that no portfolio eliminates mountain flash-flood risk, and that governance must avoid simply transferring danger downstream or onto populations with fewer resources. They call for high-resolution rainfall and discharge monitoring, open event inventories, and stronger validation of machine-learning warning systems in the under-monitored mountain regions of Africa and South America, where the evidence base remains thinnest precisely where vulnerability may be greatest.

Subject of Research: Formation processes, cascading impacts, and disaster risk reduction pathways for mountain flash floods

Article Title: Review article: Mountain flash floods: from hazard formation to disaster risk reduction pathways

Article References: Chen, Z., Feng, M., Johnson, M. F., Wright, N., Weng, Y., Chan, F. K. S., & Wu, F. (2026). Review article: Mountain flash floods: from hazard formation to disaster risk reduction pathways. Natural Hazards and Earth System Sciences, 26(10), 5073-5102. https://doi.org/10.5194/nhess-26-5073-2026

Image Credits: AI Generated

DOI: 10.5194/nhess-26-5073-2026

Keywords: mountain flash floods, disaster risk reduction, early warning systems, nature-based solutions, hydrology, sediment transport, climate change, check dams, glacial lake outburst flood, vulnerability, adaptive governance, bibliometric review