When the dams broke: uncovering the lessons of Libya’s Storm Daniel tragedy

Marcus Wishart is World Bank’s Lead Water Specialist for the Middle East and North Africa. Speaking at ICOLD’s Annual Meeting in New Delhi last year he said Libya isn’t typically the first place that springs to mind in terms of large dams. However, the country does have a relatively small portfolio concentrated on the north coast which plays an important role in ensuring water security, despite its arid conditions and natural scarcity of water. 

Wishart explained that a lot of these dams were constructed on highly variable ephemeral wadi systems that don’t have any flow for large parts of the year. They’re important not only in terms of flood management but for agricultural production, and are quite often positioned in such a way they support recharging of underlying aquifers. 

Constructed in the 1970s, the Derna and Abu Mansour dams were built to control floods and protect the city of Derna, recharge aquifers, and provide water for the Fataya Agricultural project.

Located around 1km upstream of Derna City, Derna dam was an embankment with a clay core that stood 40m high, with storage of 1.15Mm³. It had a morning glory type of overflow spillway with a capacity of 350m³/sec. Abu Mansour Dam also goes by various different names and Wishart said this caused some confusion during the early stages following the failure. A 73m high embankment dam with a clay core it had a storage of 23.7Mm³ and catchment area of 476km². Located around 11km upstream of Derna Dam it also had a morning glory overflow spillway and spillway capacity of 170m³/sec.

Storm Daniel

Storm Daniel made landfall in Libya around 9 September 2023. Rainfall levels were well in excess of the long-term average, up to 200 times higher than expected during that period. 

Wishart spoke about the impact of climate change and said heavy rainfall is up to 50 times more likely to occur compared with a 1.2°C cooler climate. So there are questions around the role of climate change in this disaster, as well as other considerations around original design parameters, particularly the idea there was a limited hydrological record to inform the initial design.

Storm Daniel had significant far-reaching impacts not just to do with the dam failures. After the event, a rapid damage and needs assessment looked at the extent of the damage. More than 250,000 people were affected by Storm Daniel with damages and losses equivalent to 3.6% of GDP and US$1.65 billion. More than 4000 people died, 10,000 people were unaccounted for and 34,000 people displaced (although more recent estimates suggest the number of lives lost range from 5900 to 20,000, making this the second-deadliest dam failure in recorded history.) And over 7% of the country’s housing stock was destroyed or damaged with initial reconstruction estimated at US$1.8 billion.

Aftermath

A collaborative assessment between ICOLD, UNESCO and the World Bank took place in the aftermath of the Storm Daniel tragedy. Wishart said it was difficult to source a lot of information as there were a number of conflicting reports in the public domain, including information provided by the original design firm.

The country’s dam portfolio itself remains a little uncertain but there are 18 large dams in Libya, 14 of which are registered in ICOLD’s World Register of Dams. This was a useful starting point to understand some of the dam characteristics associated with the failure. 

Another noteworthy point was the highly variable system these dams were constructed on. When looking at the overall portfolio, storage capacity is somewhere in the order of 380-390Mm³ and there are three dams that account for more than 50% of that. With the long-term average sitting around 60Mm³ year to year, Wishart said it is a highly variable system that had implications when looking at the underlying issues associated with failure.

One of key determining factors contributing to the disaster itself was the design of the structures’ morning glory spillway that reaches saturation fairly quickly. In the immediate aftermath there was concern across Libya about the potential for more dam failures upstream of a number of other important cities. To help understand this, initial risk assessments were remotely carried out and an additional 14 dams were identified, bringing the portfolio total to about 28. There were two very high risk dams with 27,000 people at risk, five high risk with around 25,000 people, two medium to high, one medium, one medium to low, and one low risk dam.

Wishart wanted to highlight the challenges associated with ensuring the safety of dams and downstream communities, especially in the context of operation and maintenance, in what are known as Fragile and Conflict Affected States such as Libya. He spoke about similar situations in other FCA countries, such as Sudan, which are facing significant challenges in governance, security, and development, with specific concerns in terms of both financial and human resources availability. 

According to Wishart, there were also other “events playing out in Libya” when it was recognised there was a need for an independent and more objective assessment of the failure. This proved to be “both timely and very important” when 12 officials were detained in the immediate aftermath of the failure in July 2024 and sentenced up to 27 years in prison.

According to news reports some of these officials, who were responsible for managing water resources and maintaining the dams, were charged with crimes including negligence, premeditated murder and wasting public money. 

Detailed assessment

According to ICOLD Vice President Laurent Mouvet, the experience of the Abu Mansour and Derna dams underscore the need to identify national priorities for safeguarding the rest of the national Libyan dam portfolio, mitigating the potential impacts of further dam failures on infrastructure, public health and environment. 

Eight specific actions were proposed following the findings of expert assessment, a comprehensive review of existing literature, and good international industry practices pertaining to measures that can improve safety of dams and downstream communities. These were:

1) Carry out more detailed assessment of the 2023 flood event. There was a relatively high level of uncertainty regarding spatial and temporal distribution of Storm Daniel’s rainfall through the Wadi Dam catchment area, demonstrating difficulty in accurately characterising the return period of the flood event. 

2) Carry out more detailed assessment of dam failures.

3) Improve flood forecasting, early warning systems and emergency preparedness.

4) Determine level of protection for Derna City.

5) Determine appropriate interventions for building back better using climate resilient principles.

6) Develop a regulatory framework for dam safety assurance.

7) Launch a national dam safety programme.

8) Implement stakeholder engagement and communications plan. 

Lessons learned

Was the catastrophic flooding in Derna, Libya – one of the deadliest hydrometeorological disasters on record – an inevitable outcome of rare weather conditions, or did the design of the infrastructure fail to account for probable risks? 

This is the question Moshe Armon et al ask in their research published in Science Advances. By using a combination of atmospheric re-analysis, satellite data, and hydrologic modelling, they claim to have overcome key limitations typical of data-scarce, high-variability regions and revealed that despite the catastrophic impact, the return periods of the rainfall and flood were only a few decades. Hydraulic simulations revealed that the dam failures amplified the damage nearly 20-fold compared to a dam-free scenario. With extensive and timely implications, they say their findings underscore the importance of uncertainty-aware risk assessment and highlight the value of distributed flood prevention and early warning systems in mitigating risks in vulnerable regions.

The authors go on to add that the Derna disaster should serve as an alarming sign that frequency analyses of extreme events in drylands should take into account the very high natural climate variability, the different and evolving approaches to derive extreme value estimates, changes in land use and runoff coefficients, and climate change impacts. “These challenges,” they claim, “cast doubt on the economic and the environmental rationale for dam construction as a strategy to mitigate flood risks in drylands.”

The lessons from Derna are not confined to this region and similar risks are present globally, requiring immediate attention from both the scientific community and policymakers. 

“Our results strongly imply that the complete failure of other dams is not a matter of if, but rather when,” Moshe Armon et al add. “We urge the scientific community, local knowledge holders, and authorities to collaborate on this timely hazard to improve knowledge, data availability, and risk communication and to re-evaluate the planning of habitat areas along dryland streams to prevent the next disaster.”