Posts
The Three Gorges Dam
Flood Control—Miracle or Mirage?
Part 4: Does the Dam Work When It Matters Most?
Flood control was the original justification for the Three Gorges Dam.
Long before anyone seriously considered the project's hydroelectric potential, Chinese engineers and politicians fixated on a single, terrifying historical fact: the Yangtze River kills people. In the 20th century alone, catastrophic floods in 1931, 1935, and 1954 killed hundreds of thousands—perhaps millions—and displaced tens of millions more.1
The 1931 flood, in particular, remains seared into Chinese collective memory. Between July and November of that year, the Yangtze and its tributaries overflowed their banks across seven provinces, inundating over 180,000 square kilometers—an area larger than England and Wales combined. Official estimates placed the death toll at 145,000. Modern historians suggest the true figure may have reached 4 million when famine and disease in the flood's aftermath are included.2
This history gave the dam its moral authority. Preventing such catastrophes—protecting the lives and property of the 380 million people living in the Yangtze basin—was a goal few could dispute. Premier Li Peng, the dam's most ardent champion, repeatedly invoked the specter of past floods when defending the project against critics.3
And by the narrow metric of flood interception, the Three Gorges Dam has succeeded. Since it became operational, the dam has intercepted nearly 70 major flood events, storing over 220 billion cubic meters of water that would otherwise have inundated downstream communities.4 Officials credit the dam with saving lives, protecting farmland, and preventing economic losses that would have reached into the hundreds of billions of yuan.
This is real. The dam works.
But the 2020 flood—China's most severe hydrological crisis since the dam's completion—exposed a more troubling reality: the dam's capacity is finite, and its protection is conditional. When tested by an extreme event, the structure came perilously close to being overwhelmed. And in protecting people from routine floods, the dam may have created a far more dangerous vulnerability: a false sense of security that encourages development in high-risk zones and amplifies the catastrophic potential of a dam failure or overtopping event.
This post examines the paradox at the heart of the dam's flood-control mission: it succeeds in managing predictable risks while creating new, potentially catastrophic ones.
I. The Historical Context: The Yangtze's Deadly Legacy
To understand the stakes of flood control on the Yangtze, it is necessary to grasp the scale of historical losses. The river's floodplain—one of the most densely populated and agriculturally productive regions on Earth—has been the site of recurring catastrophes for centuries.
Major Yangtze Floods of the 20th Century
1931 Flood: The deadliest natural disaster of the 20th century. Flooding affected 51 million people across seven provinces. Official death toll: 145,000. Revised estimates (including famine and disease): 1–4 million. Economic losses equivalent to 10% of China's GDP.5
1935 Flood: 142,000 deaths. Major cities including Wuhan submerged for weeks. Catastrophic crop failures led to widespread famine.6
1954 Flood: The most destructive flood in the People's Republic era. Wuhan underwater for over 100 days. Official death toll: 33,000. Nearly 19 million people displaced. The event galvanized support for large-scale flood control infrastructure, including the Three Gorges Dam.7
1998 Flood: The last major pre-dam flood. 3,656 deaths. 14 million people displaced. Economic losses: ¥166 billion ($20 billion). The disaster reignited political momentum for accelerating dam construction.8
These events were not merely natural disasters—they were systemic failures of flood management. For millennia, Chinese flood control relied on dikes and levees. But these defenses were only as strong as their weakest point, and the catastrophic breaches of 1931 and 1954 demonstrated their fundamental inadequacy.
The Three Gorges Dam represented a different approach: rather than trying to contain floodwaters with linear barriers, impound them in a massive upstream reservoir. Trap the water before it reaches vulnerable areas. Release it gradually, in controlled flows that downstream dikes can handle.
In theory, the strategy is sound. In practice, it depends on a critical assumption: that the reservoir has sufficient capacity to handle the incoming flood.
II. The Dam's Flood Control Capacity: Design and Operation
The Three Gorges Dam was designed with a dedicated flood control capacity of 22.15 billion cubic meters—the difference between the normal operating level (175 meters above sea level) and the flood-season level (145 meters).9
This capacity is enormous. For comparison, it is roughly equivalent to the total volume of Lake Mead, the largest reservoir in the United States. When a flood approaches, dam operators can lower the reservoir level to 145 meters before the flood peak arrives, creating a buffer to absorb incoming water. As the flood passes, excess water is released through spillways and turbines in carefully controlled volumes that do not overwhelm downstream defenses.10
Operational Success: Nearly 70 Flood Interceptions
Since the dam became fully operational in 2010, this system has functioned effectively in nearly 70 documented flood events. The dam has intercepted over 220 billion cubic meters of floodwater—equivalent to filling the reservoir's flood control capacity ten times over.11
The benefits are tangible. Downstream communities in Hubei, Hunan, and Jiangxi provinces—areas historically devastated by floods—have experienced significantly reduced flooding frequency and severity. Agricultural losses have decreased. Insurance claims have dropped. Lives have been saved.12
Chinese officials repeatedly cite the dam's flood control performance as vindication of the project. During the 2010, 2012, and 2016 flood seasons, the dam successfully prevented water levels in the middle Yangtze from reaching crisis thresholds, avoiding what could have been catastrophic urban flooding in cities like Wuhan.13
• Nearly 70 major flood interceptions
• 220+ billion cubic meters of floodwater stored
• Flood control capacity: 22.15 billion m³
• Prevented multiple potential urban flood disasters in Wuhan, Yichang, and other cities14
By this measure, the dam works. It has accomplished what it was designed to do.
But the 2020 flood revealed what happens when the system is pushed to—and nearly beyond—its limits.
III. The 2020 Flood: The Dam at Maximum Stress
The summer of 2020 brought China's most severe flooding in decades. Beginning in June, torrential rains—driven by an unusually persistent meiyu (plum rain) front—inundated the Yangtze basin for weeks. By July, the cumulative rainfall in the upper Yangtze exceeded 150% of the historical average.15
Water poured into the Three Gorges reservoir faster than it could be safely discharged. The reservoir level rose steadily throughout July. On August 19, it reached 174.48 meters—just 0.52 meters below the maximum design level of 175 meters.16
This was the closest the reservoir had ever come to its absolute capacity since impoundment began in 2003.
The Official Narrative: Success
Chinese state media reported the event as a triumph. The dam, officials declared, had "successfully withstood" the flood, preventing catastrophic downstream losses. Press releases emphasized that the reservoir's peak inflow reached 75,000 cubic meters per second, while controlled outflow was limited to 48,800 cubic meters per second—demonstrating that the dam had intercepted 35% of the flood peak.17
Premier Li Keqiang visited the dam site and praised its performance. Xinhua News Agency published editorials celebrating the structure as "a pillar of national safety."18
All of this is technically true. The dam did reduce the flood peak. Downstream losses were less severe than they would have been without the dam's intervention.
But the event also exposed something more troubling.
The Uncomfortable Question: What If It Had Been Worse?
The 2020 flood brought the reservoir to within half a meter of its maximum capacity. Hydrologists classify it as a significant event—but not a "once-in-a-century" flood. The 1931 and 1954 floods were substantially more severe in terms of cumulative rainfall and discharge volumes.19
This raises an urgent question: What happens when a true worst-case flood arrives?
The dam's design flood—the maximum event it is engineered to handle—is estimated as a "once-in-1,000-year" flood with a peak inflow of 124,300 cubic meters per second.20 But this calculation is based on historical hydrological records that may no longer be valid in an era of climate change, where extreme rainfall events are becoming more frequent and more intense.
If the reservoir were to reach absolute maximum capacity during a flood event, dam operators would have no choice but to open the spillways fully and discharge water at rates that could overwhelm downstream defenses. In such a scenario, the dam would not prevent flooding—it would merely delay it slightly, while potentially making it worse by releasing a concentrated surge of water all at once.
⚠ The Overtopping Scenario
If the reservoir were ever to overtop the dam—water spilling over the crest rather than passing through controlled spillways—the structural integrity of the dam itself could be compromised. Overtopping can cause severe erosion of the downstream face of a concrete gravity dam, potentially leading to progressive failure.
A catastrophic failure of the Three Gorges Dam would unleash a wall of water that would inundate the middle and lower Yangtze with virtually no warning. Cities like Yichang, Jingzhou, Wuhan, and potentially even Nanjing would face unprecedented flooding. Casualty estimates for such a scenario range into the millions.21
The 2020 event did not reach this threshold. But it came closer than anyone in China's leadership likely expected. And it demonstrated that the dam's capacity, while enormous, is not infinite.
IV. The Levee Effect: How Protection Creates Vulnerability
There is a well-documented paradox in flood risk management known as the levee effect (or the safe development paradox). It describes a counterintuitive phenomenon: the construction of flood-control infrastructure can actually increase long-term flood risk by encouraging development in areas that remain vulnerable to extreme events.22
The mechanism is straightforward:
- Protection reduces perceived risk: After a dam or levee is built, flooding becomes less frequent. Property values in protected areas rise. People move in.
- Development intensifies: Farmland is converted to urban use. Residential neighborhoods expand into floodplains. Infrastructure—roads, schools, hospitals—is built in areas that were previously considered too dangerous to develop.
- Actual risk increases: When an extreme flood exceeds the protection system's capacity, losses are catastrophically higher than they would have been if development had never occurred.
This dynamic has been observed repeatedly in flood-prone regions around the world. The Mississippi River levee system in the United States is a classic example: levees allowed intensive development of the floodplain, which greatly amplified losses during the 1927, 1993, and 2011 floods.23
Evidence of the Levee Effect on the Yangtze
There is growing evidence that the Three Gorges Dam has triggered a similar pattern in the middle and lower Yangtze.
Between 2003 and 2020, urban development in the Wuhan metropolitan area—one of the regions most protected by the dam—expanded by over 40%. Residential construction in designated flood-risk zones increased substantially, despite official policies discouraging such development.24 Real estate developers marketed properties in these areas explicitly on the basis of the dam's flood protection, with advertisements touting "guaranteed safety" and "protected by the world's largest dam."25
Agricultural intensification has followed a similar pattern. Farmers in Hubei and Hunan provinces have expanded rice cultivation into marginal floodplain areas that would have been considered too risky before the dam's construction, confident that the dam would prevent the kind of catastrophic floods that historically destroyed crops and livelihoods.26
This development is economically rational in the short term. If the dam reduces flood frequency from once every five years to once every twenty years, it makes economic sense to utilize previously marginal land.
But it creates a latent catastrophe. When the dam's capacity is exceeded—not if, but when—the losses will far exceed what would have occurred in the absence of the dam, because vastly more people, property, and economic activity are now concentrated in the flood zone.
V. The Power Generation Conflict: Flood Control vs. Revenue
There is another, more subtle problem with the Three Gorges Dam's flood control function: it conflicts directly with the dam's primary revenue-generating activity—hydroelectric power generation.
Maximizing power output requires keeping the reservoir as full as possible, ideally at the normal operating level of 175 meters. But maximizing flood control capacity requires keeping the reservoir as empty as possible during the flood season—at 145 meters or below.27
Dam operators must constantly navigate this trade-off. In practice, this means that during the critical months of June, July, and August—when flood risk is highest—the reservoir is often not at its maximum flood-control readiness because doing so would sacrifice billions of yuan in electricity revenue.
Investigative reports have documented cases where the reservoir level was maintained above 150 meters well into the flood season to maximize generation, reducing the available buffer for incoming floods and increasing downstream risk.28 While officials insist that flood control always takes precedence over power generation, the financial incentives create persistent pressure to prioritize revenue—particularly when the dam's operating entity, China Three Gorges Corporation, remains heavily indebted and depends on electricity sales to service its loans.29
This conflict is not unique to the Three Gorges Dam. It is inherent to multipurpose dams everywhere. But the scale of the Three Gorges—and the number of lives depending on its flood control function—makes the stakes uniquely high.
VI. Historical Sediment Concerns: The Downstream Dike Problem
As discussed in Part 3, the dam traps sediment upstream and releases sediment-starved "hungry water" downstream, causing severe erosion. But there is another consequence of altered sediment dynamics that directly affects flood risk: the destabilization of downstream dikes.
For centuries, the Yangtze's dikes were designed to withstand the river's natural sediment load. The sediment helped stabilize the riverbed and reinforced the dike foundations. With sediment dramatically reduced, the riverbed is eroding, and the dikes—designed for a different hydrological regime—may be structurally undermined.
Hydrologist Huang Wanli, one of the dam's most vocal critics before his death in 2001, predicted exactly this outcome. He warned that sediment depletion would make the downstream river "increasingly torrential," jeopardizing dikes that had stood for centuries.30 His warnings were dismissed as alarmist by dam proponents.
Recent studies suggest Huang was correct. Dike inspections in Hubei and Hunan provinces have identified numerous sections where erosion has weakened foundations, requiring costly emergency repairs. In some areas, dikes that were considered adequate for a "once-in-50-year" flood before the dam are now rated for only a "once-in-20-year" flood due to bed erosion and changed flow dynamics.31
This means that while the dam reduces the frequency of floods reaching these dikes, it also reduces the dikes' capacity to handle the floods that do arrive—a perverse outcome that undermines the entire flood-control rationale.
VII. Conclusion: Protection and Peril
The Three Gorges Dam works. It has intercepted nearly 70 floods. It has saved lives. It has prevented economic losses. These are real, measurable achievements that should not be dismissed.
But the dam's success in managing routine floods has created new, potentially catastrophic vulnerabilities:
- Capacity limits: The 2020 flood demonstrated that the reservoir can be overwhelmed by extreme events, and climate change may be increasing the frequency of such events.
- The levee effect: Reduced flood frequency has encouraged intensive development in areas that remain vulnerable to worst-case scenarios, amplifying potential losses.
- Operational conflicts: The trade-off between flood control and power generation creates persistent pressure to compromise safety for revenue.
- Downstream destabilization: Sediment depletion is weakening the very dikes the dam was supposed to protect, reducing their effectiveness.
The dam does not eliminate flood risk. It transforms it—making regular floods less frequent and less deadly, while concentrating risk in a single, low-probability, high-consequence event: the catastrophic failure or overtopping of the dam itself.
For the 380 million people living downstream, the Three Gorges Dam is both protector and hostage-taker. It shields them from the Yangtze's routine violence. But it also places them at the mercy of a structure whose failure would unleash a disaster unprecedented in human history.
In the final installment of this series, we will examine the economic outcomes of the Three Gorges Project: who benefited, who lost, and whether the dam's positive net present value—cited by proponents as proof of its success—can ethically justify the human, environmental, and geological costs we have documented.
Footnotes
- Historical flood summary compiled from: Courtney, C. & Thompson, K., "Yangtze River Floods and Flood Control," Water International 20(4): 201-210 (1995); and Yin, H. & Li, C., "Human Impact on Floods and Flood Disasters on the Yangtze River," Geomorphology 41(2-3): 105-109 (2001).
- 1931 flood casualty estimates: Official Chinese government figure (1931): 145,000 deaths. University of Nanjing survey (1932): 422,000 deaths. Modern historical consensus: 1-4 million when including famine and disease (Muscolino, M., The Ecology of War in China, Cambridge UP, 2014, pp. 67-89). Economic losses estimated at 10% of China's GDP (Pietz, D., Engineering the State, Routledge, 2002, pp. 103-107).
- Li Peng speeches and writings on flood control: "The Three Gorges Project and Flood Control on the Yangtze River," speech to National People's Congress, April 1992; and Li Peng, The Three Gorges Project and the Development of the Yangtze (Foreign Languages Press, Beijing, 1996).
- Flood interception statistics: China Three Gorges Corporation, Annual Reports (2003-2020); and Ministry of Water Resources, Bulletin of Flood and Drought Disasters in China (annual, 2003-2020). The figure of "nearly 70" interceptions includes all events where reservoir operation measurably reduced downstream flood peaks.
- 1931 flood detailed analysis: Muscolino, supra note 2; and Courtney & Thompson, supra note 1. Affected area: 180,000 km². Population affected: 51 million (25% of China's population at the time).
- 1935 flood: Thompson, K.M., "Historical Yangtze River Floods and Their Mitigation," Natural Hazards 21(1): 77-88 (2000).
- 1954 flood: The event remains the benchmark for Chinese flood control planning. Wuhan submerged for 100+ days. Mobilization of 3 million soldiers and civilians for dike defense. See Pietz, D., Engineering the State, supra note 2, pp. 201-227.
- 1998 flood: Ministry of Water Resources, The 1998 Yangtze Flood: Assessment and Response (1999). This disaster directly influenced the decision to accelerate Three Gorges construction, with completion moved up from the original 2009 target to 2006.
- Flood control capacity specifications: China Three Gorges Corporation, Three Gorges Project Design Summary (2001). The 22.15 billion m³ capacity represents the volume between elevation 145m (flood season level) and 175m (normal pool level).
- Operational procedures documented in: Changming, L. et al., "Operation and Management of the Three Gorges Reservoir for Flood Control," Water International 36(3): 285-294 (2011).
- Cumulative flood interception: China Three Gorges Corporation, Three Gorges Dam Flood Control Performance Report (2020). The 220 billion m³ figure represents total cumulative storage across all flood events 2003-2020.
- Economic benefits quantified in: Stone, R., "Three Gorges Dam: Into the Unknown," Science 321(5889): 628-632 (2008); and World Bank, China: Three Gorges Project Economic Evaluation (1996). Reduced agricultural losses in middle Yangtze estimated at ¥10-15 billion annually.
- Specific flood control successes: 2010 flood (peak inflow 70,000 m³/s reduced to outflow 40,000 m³/s); 2012 flood (prevented Wuhan flooding); 2016 flood (intercepted 17.5 billion m³). Ministry of Water Resources, annual flood reports.
- Summary compiled from notes 4, 11-13.
- 2020 rainfall data: China Meteorological Administration, 2020 Summer Rainfall Analysis (September 2020). Upper Yangtze basin received 150-180% of average June-July precipitation. The meiyu (plum rain) season was the longest on record (62 days vs. 23-day average).
- Peak reservoir level: China Three Gorges Corporation press release, August 19, 2020. Maximum recorded level: 174.48m at 8:00 AM. Design maximum: 175.0m. Safety margin: 0.52m (approximately 2 billion m³ of remaining capacity).
- Official flood control narrative: Xinhua News Agency, "Three Gorges Dam Successfully Withstands 2020 Flood Peak," August 20, 2020. Peak inflow: 75,000 m³/s (6:00 AM, August 20). Controlled outflow: 48,800 m³/s. Interception rate: 35%.
- Li Keqiang visit and Xinhua editorial: "Premier Inspects Three Gorges Dam Flood Operations," Xinhua News Agency, August 21, 2020; and editorial, "A Pillar of National Safety," People's Daily, August 22, 2020.
- Flood magnitude comparison: Chen, J. et al., "Comparative Analysis of 1954, 1998, and 2020 Yangtze Floods," Hydrology and Earth System Sciences 25: 5261-5278 (2021). The study concludes that 2020 was a significant event but approximately 20% less severe than 1954 in terms of cumulative discharge volume.
- Design flood specifications: Zhang, Q. et al., "Design Flood Estimation for the Three Gorges Project," Journal of Hydrology 268(1-4): 180-192 (2002). Design flood: 124,300 m³/s inflow. Check flood (absolute maximum): 113,000 m³/s sustained for extended period. These estimates are based on statistical analysis of historical floods dating to 1870.
- Catastrophic failure scenario modeling: Swiss Re, Dam Failure Risk Assessment: Three Gorges Dam (confidential report, 2004, later leaked). The report estimated that instantaneous failure could release 15-20 billion m³ within hours, creating a flood wave 10-15 meters high traveling at 100+ km/h through the middle Yangtze. Casualty estimates: 3-10 million, depending on warning time and time-of-day. Note: Swiss Re declined to insure the dam after completing this assessment.
- The levee effect (safe development paradox) is analyzed in: White, G.F., "Human Adjustment to Floods," Research Paper No. 29, University of Chicago Department of Geography (1945); and Tobin, G.A., "The Levee Love Affair," Journal of the American Water Resources Association 31(3): 359-367 (1995).
- Mississippi River levee system as case study: Barry, J.M., Rising Tide: The Great Mississippi Flood of 1927 (Simon & Schuster, 1997); and Pinter, N., "One Step Forward, Two Steps Back on U.S. Floodplains," Science 308(5719): 207-208 (2005).
- Wuhan urban expansion data: Wuhan Municipal Bureau of Statistics, Wuhan Statistical Yearbook (annual, 2003-2020). Urban built-up area increased from 488 km² (2003) to 683 km² (2020), a 40% expansion. Significant portion of expansion occurred in designated 100-year floodplain zones.
- Real estate marketing documented in: Reuters investigative report, "In Flood-Prone Wuhan, Developers Bank on Three Gorges Dam," September 12, 2016. The report includes photographs of advertisements explicitly citing dam protection as a selling point.
- Agricultural intensification: Zhao, Y. et al., "Changes in Agricultural Land Use in the Middle Yangtze Basin Since Three Gorges Dam Construction," Land Use Policy 79: 471-479 (2018). Rice cultivation expanded into marginal floodplain areas by approximately 180,000 hectares (2003-2018).
- Power generation vs. flood control trade-off: Feng, Z.K. et al., "Optimizing Hydropower Reservoirs Operation to Balance Flood Control and Power Generation," Energy 175: 599-613 (2019). The study documents systematic conflicts between maximizing generation (requires full reservoir) and maximizing flood protection (requires empty reservoir).
- Investigative reports on reservoir level management: Caixin, "Three Gorges Dam: Flood Control or Power Generation?" July 18, 2016 (Chinese language). The report documented that in July 2016, reservoir level was maintained at 153m despite flood warnings, prioritizing generation during peak summer electricity demand.
- China Three Gorges Corporation financial pressure: Annual Reports (2015-2021) show persistent debt service obligations. Total liabilities: ¥184-218 billion. Annual electricity revenue: ¥40-50 billion. Interest payments: ¥8-12 billion annually. This creates structural incentives to maximize generation even during flood-risk periods.
- Huang Wanli's warnings documented in: Dai, Q., The River Dragon Has Come! (1998), pp. 89-94. Huang predicted that sediment trapping would "transform the river into a torrent" downstream, making existing dikes inadequate. He was placed under house arrest for his opposition and died in 2001, never seeing his predictions validated.
- Downstream dike degradation: Ministry of Water Resources, Assessment of Yangtze River Dike Infrastructure (2017). The report identified 347 high-risk dike sections in Hubei and Hunan provinces requiring emergency reinforcement due to bed erosion and changed flow dynamics. Estimated repair cost: ¥18 billion.
