Flood risks from failure of infrastructure

When preparing an editorial, members of the Editorial Board sometimes discuss cases of floods that are topical at the time of writing. In this editorial, I have chosen to reflect briefly on three catastrophic cases of flooding and consequential impacts following the failure of dams, which illustrate why the scope of the Journal of Flood Risk Management includes flood risks associated with dams. Although the impacts are mainly quantified as numbers of deaths, building damaged, etc., it is important for all involved in flood risk management to recognise the emotional distress that flooding causes to the survivors. These human impacts can persist for years and permanently change behaviour. When I started to consider composing this editorial, there had been international news reports of flash flooding in Southern Europe in Bulgaria, Greece and parts of Turkey from storm Daniel where in some areas over 750 mm of rain had fallen on 5 September, see NASA (2023). In the subsequent days, this weather system passed over the warm waters of the Mediterranean Sea to the south and developed in intensity into a ‘Medicane’, a storm of hurricane force winds that struck the coast of Libya. Although not in the location of the most intense rainfall over Libya, European Commission (2023), extreme flooding occurred without warning in Derna on the Mediterranean coast at the outfall of the Wadi Derna in the early hours of September 10. The immediate cause of the disaster was the cascade failure of the Derna and Mansour dams, which had been constructed on the wadi for flood control. The impacts for the town of Derna were catastrophic, with the dam-break flood causing around 4000 fatalities, internally displacing over 16,000 people, destroying 876 buildings and damaging 3100; further consequences of the flooding include the disruption of health services and the lack of safe and affordable water, see United Nations Office for the Coordination of Humanitarian Affairs (2023). Governance in this part of Libya has been contested since the Spring uprising in 2011, most likely leading to poor maintenance of the dams; moreover, the dams were designed using climatic conditions prevailing in the 1970s. An initial assessment of the influence of climate change has indicated that an event as extreme as the one observed over Libya has become up to 50 times more likely and up to 50% more intense compared with a 1.2°C cooler climate, see World Weather Attribution (2023). Early on the morning of 6 June 2023, the failure of the Kakhovka Dam in the Russian occupied area of Ukraine led to widespread flooding downriver on the Dnipro River. Being in an active war zone, the cause and responsibility for the failure of the dam are contested. However, the impacts are clear: extensive inundation, loss-of-life, destruction of buildings, internal displacement of those affected, loss of hydro-electric power generation at the site and loss of water resources in a large area. The UN Country Team in Ukraine (2023) has published an early assessment of the long-term consequences of the destruction of the dam including impacts on the internal displacement of people; health, water and sanitation; chemical hazards; agriculture, fishery and forestry; community infrastructure; energy; housing; and culture. In January 2019, the Dam ‘B1’ of the Córrego do Feijão mine in Brazil, 9 km from the town of Brumadinho, collapsed without warning. Prior to collapse, the dam was about 80 m high and its crest length was 700 m; the failure released about 9.7 million m3 of mining waste, causing between 270 and 320 deaths and significant environmental pollution; see Robertson et al. (2019). Tailings dams are generally not constructed in the same way as permanent earth embankment dams used for flood control, water resources, hydropower, etc., instead they are raised progressively as the mining waste consolidates during the life-time of the operations. At this site, construction started in 1976 and tailings disposal in the dam ceased in July 2016, 30 months before the failure. The post-event expert report into the disaster, see Robertson et al. (2019), concluded that the design and construction of the dam contributed to its failure, in particular, the design resulted in a dam that was steep with a lack of sufficient drainage. The expert report also concluded that none of the monitoring devices at the site detected precursors to failure; rather, the dam's failure was sudden and abrupt. However, Lumbroso et al. (2020) show through modelling that if a warning had been issued even as the dam failed the number of fatalities could have been reduced significantly. These three dams differed in design and primary function, but their failures led to immediate catastrophic human and environmental impacts, which will continue in the long term. Although these incidents had differing and complex causes—design, construction, inspection and maintenance, hydrometeorology, and governance—the consequences and risks have many similarities. This is why the Journal of Flood Risk Management seeks to publish advances in knowledge on all such aspects of flood risks and their management. Another category of infrastructure that can pose some similarity in the risk of dam failure, albeit less extreme, includes raised flood defences—embankments and walls that impound water level temporarily above the surrounding ground level during river floods or coastal surges. Four papers in this issue of the journal relate to the management of flood risks in areas protected by embankments; the first covers the ‘source’, the second and third the ‘pathway’ and the fourth the ‘receptor’ in the well-known ‘source-pathway-receptor’ framework for analysis of flood risk. In the first paper, Sarwar and Borthwick (2023) consider the uncertainty in the future hydrodynamic loading on coastal embankments in Bangladesh. Extensive embankment construction has been undertaken along the Bangladeshi coastline to protect people and properties in nearby cities and villages from flood inundation and land erosion. Sarwar and Borthwick (2023) develop a derived distribution approach to assess uncertainty in extreme water levels. The analytic version of the derived distribution approach, where the dependence of the output parameter on the input parameter is known, has been used for 50 years in flood frequency and water resources estimations. However, the innovation in the paper is that it is the first to report on the application of the numerical analogue of the derived distribution approach to maximum tidal elevations in the context of an actual large-scale bay subject to sea level rise. The second paper is from Pennisi et al. (2023), who discuss the use of ground penetrating radar as a non-invasive survey technique to assess the presence and extent of animal burrows in flood embankments (levees). Animal burrows can provide preferential seepage routes within and below raised flood defences. The authors present a case study of the Dirillo River in Sicily where several failures of the levees occurred during a flood in 2012. The site investigation reported in the paper illustrates the depths and paths of dens inside the levee. Moreover, they reveal that the tunnels start as a single straight tunnel and, inside the levee, they split into more than one tunnel connecting the riverside slope to the landside slope. This new evidence highlights the potentially destructive impact of animal dens inside the levees. In the worst case, such tunnels may trigger intense erosion in floods that could trigger the failure of the overall structure. For several years, fragility curves have been used to assess the failure probability of flood defences when undertaking probabilistic flood risk assessments. In the third paper, Mainguenaud et al. (2023) propose a probabilistic method to assess levee failure probability by integrating three failure mechanisms relevant to fluvial levees: sliding, internal erosion and overflowing. This aggregation of failure probabilities from the three mechanisms provides a single fragility curve to represent the failure probability and avoids making a biased interpretation of the results due to distinct fragility curves for the same levee segment. The authors illustrate the method on a reach of the Bow River in Canada and recommend the use of Monte-Carlo aggregated fragility curves in practice. Drawing on the national conditions in Aotearoa, New Zealand (A-NZ), the fourth paper by Fu et al. (2023) discusses the management of residual flood risks for the population behind raised defences. The so-called ‘levee-effect’ leads to a false sense of security and underestimation of flood risk. Fu et al. (2023) discuss empirical findings from a survey carried out with flood risk practitioners from the public and private sector. They argue that, without a change in approach—through a national directive for A-NZ and the provision of resources—residual flood risk is likely to continue to rise in flood-prone communities, which will cause all stakeholders to underestimate their flood risk as well as the necessary insurance coverage. Finally, I wish to thank colleagues on the journal's panel of editors, the associate editors and especially all the anonymous peer reviewers for their contribution to the success of the Journal of Flood Risk Management over the past year.