The Three Gorges Dam is a large multipurpose water-control project on the Yangtze River at Sandouping, near Yichang in Hubei province, China. The project comprises a concrete gravity dam, a reservoir extending upstream toward Chongqing, hydroelectric generating stations, and navigation facilities. Its principal purposes are flood control, electricity generation, and improved inland shipping. With an installed generating capacity of 22,500 megawatts, it is the world's largest hydroelectric power station by installed capacity.
Proposals to harness the Three Gorges developed during the twentieth century, and China's National People's Congress approved the project in 1992. Construction formally began in 1994; the first generating units entered service in 2003, the main dam structure was completed in 2006, and the final main generating units began operating in 2012. A ship lift opened in 2016. These milestones distinguish completion of individual facilities from the longer process of reservoir filling, commissioning, and project acceptance.
The project is a major component of China's infrastructure and electricity system, and its operation links flood management, power production, and river transport. Creation of the reservoir also required the relocation of more than a million people and extensive archaeological and heritage-preservation work. Its long-term effects on sediment movement, river habitats, water quality, and reservoir-bank stability make it an important subject of engineering, environmental, and social research.
River Setting and Reservoir Geography
The Three Gorges Dam crosses the Yangtze River at Sandouping, in Yiling District of Yichang, Hubei Province, central China. The site lies in the lower part of Xiling Gorge, downstream of Qutang and Wu gorges and most of the Three Gorges corridor, rather than beyond the entire gorge system. The separate Gezhouba Dam stands approximately 38 kilometres downstream, closer to central Yichang. Between the two structures, water released from Three Gorges enters the Gezhouba reservoir; they are distinct hydraulic projects, not sections of a single dam.
From upstream to downstream, the Three Gorges are Qutang Gorge, beginning near Fengjie; Wu Gorge, extending through the Wushan–Badong area; and Xiling Gorge, extending toward Yichang. These narrow reaches alternate with broader valleys and tributary junctions. Impoundment raised the river through the gorge corridor and extended reservoir conditions upstream toward Chongqing. The reservoir is commonly described as approximately 600 kilometres long, with figures around 660 kilometres also used for its extent at the normal operating water level of 175 metres above sea level. Its upstream limit is not a fixed shoreline: it changes with reservoir elevation, incoming river discharge, and whether the measurement refers to continuous impounded water or the farther-reaching influence of backwater on river levels.
The drainage basin above the dam covers approximately one million square kilometres, slightly more than half the Yangtze basin. It extends westward to the Tibetan Plateau and includes the Jinsha River, the upper main stem of the Yangtze, together with major tributaries such as the Min, Tuo, Jialing, and Wu rivers. The Jialing joins the main stem at Chongqing, while the Wu enters farther downstream at Fuling. Smaller tributaries, including the Daning and Xiangxi rivers, discharge directly into the reservoir. Flooding of their lower valleys created branching embayments, so the reservoir is a long, irregular network of channels and bays rather than a uniformly wide artificial lake.
Seasonal inflow reflects the East Asian monsoon: most basin precipitation and runoff occur during the warmer months, and large floods commonly develop during summer. Snowmelt contributes to headwater flow, but rainfall over the extensive upstream catchment is central to the flood volumes reaching the dam. Below Yichang, the Yangtze leaves the confined gorge landscape and enters the comparatively low-gradient middle-river plains, including the Jianghan Plain and reaches connected with Dongting Lake. These densely settled floodplains, and cities farther downstream such as Wuhan, form an important part of the geographical setting for the project's flood-control role. Tributary rainfall and lake inflows below the dam remain additional sources of downstream flooding.
At Sandouping, the foundation is predominantly strong crystalline granitic rock associated with the Huangling massif. This bedrock, together with a valley cross-section offering space for the dam, powerhouses, and navigation works, was an important consideration in site selection. Foundation conditions should not be generalized to the entire reservoir perimeter. Upstream shorelines cross varied sedimentary formations, including limestone, sandstone, shale, and mudstone, with differing bedding orientations, fractures, and weathering characteristics. The engineering properties of the dam foundation therefore differ from those of many reservoir banks, where geological structure and changing groundwater conditions influence slope stability.
Proposals, Investigations, and Authorization
Sun Yat-sen included development of the Three Gorges’ water-power resources in his programme for China’s industrialization, set out in The International Development of China, published in English in 1922. His proposal linked river improvement and electricity generation, rather than specifying the dam eventually built. Republican-era investigations subsequently examined the engineering possibilities. In 1944, the American dam engineer John L. Savage surveyed the gorges and prepared a preliminary proposal combining power generation, flood control, and navigation. Cooperation with the United States Bureau of Reclamation led to further design work and training of Chinese engineers, but this programme ended in 1947 amid civil war and financial difficulties. These investigations established an engineering precedent, not a continuous construction programme.
After 1949, the Yangtze planning organization under the Ministry of Water Resources expanded hydrological observation, mapping, geological investigation, and river-basin planning. The catastrophic 1954 floods on the Yangtze River gave renewed urgency to proposals for storing floodwater upstream of the middle-river plains. Mao Zedong publicly invoked the project in his 1956 poem “Swimming,” imagining a barrier across the western river and a lake rising among the gorges. Political attention did not immediately produce construction authorization. A March 1958 Communist Party resolution called for continued investigation and preparation, while leaving the timing of construction open. Subsequent work compared dam sites and development schemes, with Sandouping emerging as the selected location.
Technical appraisal centred on interdependent choices rather than simply whether a dam could be built. A higher reservoir offered greater hydraulic head, storage, and improvement of upstream navigation, but also inundated more settlements and farmland and increased relocation requirements. Flood-control studies examined how much storage could be reserved during the summer flood season and how releases would affect downstream flood levels. Power studies considered the scale and timing of electricity demand and the transmission infrastructure needed to serve distant consumers. Navigation studies assessed the backwater reach toward Chongqing and ways of passing vessels across the dam. The eventual scheme distinguished a normal reservoir elevation of 175 metres from a lower flood-season operating level of 145 metres; these were operating elevations, not measurements of the dam’s structural height.
A renewed feasibility assessment began in 1986, organized by the Ministry of Water Resources and Electric Power, and its principal work was completed in 1989. Specialist groups examined hydrology, geology, hydraulic structures, machinery, flood control, power systems, navigation, sedimentation, environmental effects, and resettlement. The resulting feasibility report underwent further review before submission for national authorization. Population relocation was a central planning constraint: estimates depended on the proposed inundation boundary, the population reference year, and projected growth before impoundment. Contemporary estimates therefore should not be treated as interchangeable with the final number of people relocated.
On 3 April 1992, the Fifth Session of the Seventh National People’s Congress adopted the resolution approving construction of the Three Gorges project. The recorded vote was 1,767 in favour, 177 against, and 664 abstentions. The resolution approved inclusion of the project in the national economic and social development programme and authorized the State Council to arrange implementation at an appropriate time. In 1993, the State Council established the Three Gorges Project Construction Committee, chaired by Premier Li Peng, to coordinate major implementation matters. China Yangtze Three Gorges Project Development Corporation was established in September 1993 as the project’s corporate owner, separating enterprise-level implementation from central administrative coordination.
Construction, Financing, and Commissioning
Formal construction began on 14 December 1994, following preparatory works and the project’s authorization in 1992. Construction proceeded in three main stages rather than through a single closure of the Yangtze River. The first stage established access roads, construction services, a diversion channel along the right bank, and temporary navigation arrangements. On 8 November 1997, workers closed the main river channel, directing its flow through the diversion channel. This closure enabled construction of the principal spillway and left-bank generating facilities within an isolated work area, while river discharge and shipping continued through the bypass.
Temporary cofferdams—barriers separating construction areas from flowing water—were central to this sequence. Within the enclosed areas, contractors pumped out water, excavated riverbed deposits and rock, and prepared the bedrock foundation before placing concrete. The dam was built in separate blocks and successive lifts; cooling arrangements and controlled placement limited the temperature differences and contraction that can crack large masses of concrete. In 2002, closure of the diversion channel initiated the transition to the final construction stage. The completed spillway sections then carried river flows while work continued on the right-bank powerhouse. Navigation likewise shifted between temporary facilities and the permanent locks, which entered trial operation in June 2003. These overlapping arrangements allowed construction, flood passage, navigation, and early power production to proceed concurrently.
Financing combined a government-administered Three Gorges construction fund, borrowing, and revenues from electricity sales. The construction fund drew principally on surcharges on electricity consumption, with rates and exemptions varying by region and period. Revenues associated with the existing Gezhouba project also supported construction; after 2003, electricity produced by the new generating units provided another source of project income. Loans supplied capital before the full generating system was earning revenue, with the China Development Bank serving as a major lender. The frequently cited original estimate of approximately 90.09 billion yuan at 1993 prices covered the dam complex, resettlement, and transmission works. It was a constant-price planning estimate, not a final cash-expenditure total. By the end of 2008, reported cumulative investment had reached approximately 148.4 billion yuan, including construction, resettlement, and financing expenditure. These figures refer to different price bases and accounting dates and therefore cannot, by themselves, establish the amount of a cost overrun.
Initial reservoir filling began in June 2003, raising the water surface to approximately 135 metres above the project’s elevation datum. The first generating unit began producing electricity in July 2003, well before completion of the entire complex. On 20 May 2006, placement of the final concrete in the main dam marked completion of its principal structure. That milestone did not mean that every generating unit, navigation installation, or associated work was complete: equipment installation and testing continued while finished sections were already operating.
Subsequent commissioning extended over more than a decade. On 26 October 2010, the reservoir first attained its normal pool elevation of 175 metres. The last of the 32 main 700-megawatt generating units, installed in the underground powerhouse, entered service in July 2012, completing the main generating-unit installation. The ship lift opened for trial operation on 18 September 2016, adding a separate transit option alongside the locks. Overall project completion acceptance was announced in November 2020 following the prescribed inspections and assessments. This administrative acceptance was distinct from both the 2006 structural milestone and the earlier commissioning of individual operational systems.
Dam Structure and Hydraulic Engineering
The Three Gorges structure is a concrete gravity dam: its weight provides the principal resistance to the forces exerted by the reservoir. Its cross-section is broad at the base and narrower toward the crest, placing sufficient concrete above the foundation to resist sliding and overturning. Water pressure increases with depth according to , where is water density, is gravitational acceleration, and is depth below the water surface. Consequently, the greatest upstream pressure acts near the base. The dam transfers these loads into its bedrock foundation through compressive contact and shear resistance. Foundation grouting limits seepage through rock fractures, while drainage reduces uplift pressure beneath the structure; uplift matters because it reduces the effective downward force available to resist sliding.
The dam is approximately 2,335 metres long, with a maximum structural height of 181 metres and a crest elevation of 185 metres. These measurements describe different quantities. Structural height is the vertical distance from the lowest foundation level to the crest, whereas crest elevation is the crest’s height above the project’s adopted elevation datum. Neither figure is the depth of water throughout the reservoir. At the normal pool elevation of 175 metres, the reservoir surface is nominally 10 metres below the crest, although hydraulic design must also account for waves and exceptional high-water conditions.
A central spillway section is flanked by the principal surface powerhouse sections. Its discharge system includes 22 surface outlets and 23 deep outlets, allowing water to be released from different elevations. Gates regulate the effective opening area and therefore the discharge under a given reservoir level. Deep outlets can pass water while the reservoir is below the surface-outlet crests, providing flexibility during changing operating conditions. Downstream, shaped discharge passages and flip-bucket energy dissipators direct high-velocity jets away from the dam’s base. The resulting turbulence and impact in the downstream channel dissipate energy that would otherwise produce concentrated erosion close to the foundation.
Powerhouse intakes provide a separate, controlled route through the structure. Trash racks intercept large debris before water enters the turbine waterways, and intake gates allow passages to be closed for isolation and maintenance. Water travels through pressurized conduits to turbines and then returns to the downstream river through draft tubes and tailraces. The principal surface powerhouses occupy the two sides of the spillway; an additional underground powerhouse lies within the right-bank rock mass. Spillway discharge and turbine discharge thus perform different functions: the former passes water without generating electricity, while the latter extracts energy from the elevation difference between the reservoir and downstream water.
Reservoir elevations also distinguish operating conditions from structural limits. The 175-metre normal pool corresponds to approximately 39.3 cubic kilometres of total storage, not to a volume entirely available for flood retention or power generation. Before the main flood season, the reservoir is normally drawn down toward 145 metres, creating space to accommodate incoming floods; its level can then rise as inflow temporarily exceeds releases. Exceptional flood-check calculations allow for levels above normal pool, including a check-flood level of approximately 180.4 metres. Such a level is a hydraulic safety-design condition, not a routine operating target. The difference between these water levels and the 185-metre crest provides freeboard, but that vertical margin is not equivalent to freely usable storage: wave action, discharge capacity, and structural loading remain design constraints.
Flood Control and Reservoir Operation
The reservoir follows a seasonal cycle designed to make storage available before the main summer floods on the Yangtze River. Water levels are progressively lowered from the normal pool elevation of 175 metres toward the flood-season operating level of 145 metres, normally reached by early June. Both figures are elevations above the project’s reference datum, not water depths. During the flood season, the reservoir can rise above 145 metres as it retains floodwater, then release that water when downstream conditions permit. Refilling toward 175 metres generally begins in September and continues into autumn, subject to incoming flows and flood forecasts. These are operating targets rather than an unvarying annual timetable: drought, late floods, navigation requirements, and downstream water needs can alter the schedule.
Approximately 22.15 cubic kilometres of designated flood-control storage lie between the 145-metre and 175-metre levels. Flood attenuation occurs when reservoir inflow exceeds the combined releases through turbines and other outlets; the difference accumulates in storage rather than immediately passing downstream. The basic water balance is , where is stored water and the two terms are inflow and outflow, neglecting smaller contributions such as direct rainfall and evaporation. A temporary difference of 30,000 cubic metres per second would add about 2.59 cubic kilometres in one day. This illustrates why operators must consider a flood’s volume and duration, not only its peak discharge: prolonged high inflows can consume substantial storage even after the peak has passed.
Release decisions combine rainfall forecasts, observations across the upper drainage basin, predicted reservoir inflows, available storage, and downstream river conditions. The Changjiang Water Resources Commission and national flood-management authorities coordinate reservoir scheduling with upstream projects and downstream flood defences. Upstream reservoirs can retain part of an approaching flood, while Three Gorges releases must also be coordinated with the separate Gezhouba project immediately downstream. Gauges at Yichang, Shashi, Hankou, and other locations indicate how the flood wave is moving through the middle river. Operators may release water ahead of an expected flood to restore storage, but this also increases downstream discharge; the timing must therefore account for tributary floods and existing high water levels.
Documented events show the distinction between inflow and release. On 20 July 2010, inflow reached approximately 70,000 cubic metres per second, while outflow was held at about 40,000 cubic metres per second. The roughly 43 per cent reduction described the difference between those flows at that time, not a corresponding percentage reduction in downstream water level or total flood damage. On 20 August 2020, inflow reached approximately 75,000 cubic metres per second, the highest recorded since the reservoir began operating in 2003; releases were approximately 49,000 cubic metres per second. Retention raised the reservoir level, and the stored water subsequently had to be released or accommodated within later operations.
Protection is limited by both storage capacity and geography. Rainfall over the middle and lower basin, tributaries entering below the dam, and inflows from the Dongting and Poyang lake systems are not intercepted by Three Gorges. Downstream water level also depends on channel geometry, backwater effects, lake levels, and the timing of several flood waves, rather than dam outflow alone. Consequently, severe flooding can occur around Wuhan and elsewhere while the reservoir is substantially reducing the flood arriving from the upper Yangtze; levees, flood-diversion areas, forecasting, and emergency measures remain necessary components of basin flood management.
Hydroelectric Generation and Grid Connections
Water entering the powerhouses passes through Francis turbines, which convert hydraulic energy into shaft rotation; the shafts drive generators that produce alternating-current electricity. Francis turbines are reaction turbines: pressure changes and the movement of water through the runner both contribute to torque. Their guide vanes regulate flow and direct it onto the runner blades. Electrical power can be approximated by , where is water density, is gravitational acceleration, is turbine discharge, is net hydraulic head, and is combined turbine–generator efficiency. Net head is the difference between upstream and downstream water levels after hydraulic losses, not the dam’s structural height. Rising downstream water can therefore reduce head even when river discharge is high.
The station’s installed capacity is 22,500 megawatts, comprising 32 main generating units rated at 700 megawatts each and two smaller units rated at 50 megawatts each. The main units are distributed among three facilities: 14 in the left-bank surface powerhouse, 12 in the right-bank surface powerhouse, and six in an underground powerhouse on the right bank. The two smaller units form the station-service power facility. The main-unit installation was completed in 2012, when the underground units brought the total number of 700-megawatt machines to 32. Multiple independently controlled units allow operators to change the number of machines running and distribute the required load among them, subject to equipment limits and maintenance requirements.
Installed capacity describes rated generating capability, not continuous production. Instantaneous output depends on available flow, net head, equipment availability, and dispatch instructions. During low-flow periods, water availability limits generation; during floods, some water may have to pass through spillways rather than turbines. Flood-control requirements, navigation needs, environmental releases, and permissible reservoir levels also constrain operation. In 2020, Three Gorges reported generation of approximately 111.8 terawatt-hours. This is an annual energy total, equivalent to an average output of about 12.7 gigawatts over that leap year, rather than a claim that the station operated continuously at its 22.5-gigawatt rating. Annual generation varies with hydrological conditions and operating decisions.
Transformers raise generator voltage for transmission through a combination of high-voltage alternating-current and direct-current connections. The associated transmission network supplies Central China and carries electricity to eastern and southern receiving systems. Dedicated ±500-kilovolt direct-current links include the Three Gorges–Changzhou and Three Gorges–Guangdong schemes. Direct-current transmission provides controllable transfers over long distances and helps connect the generating complex with major demand centres. Plant operation is undertaken within China Yangtze Power, while power scheduling and interregional transfers are coordinated through State Grid’s dispatch system and the relevant receiving-grid arrangements. Dispatch requirements must remain compatible with reservoir-management instructions; electricity demand alone does not determine releases.
The station contributes to grid balancing because hydroelectric units can adjust output comparatively quickly within their permitted operating ranges. Operators can vary guide-vane openings, redistribute load, or start and stop units to follow demand and provide frequency support and operating reserve. This flexibility does not imply unlimited stored energy or pumping capability: Three Gorges is a conventional reservoir hydropower station, not a pumped-storage plant. Its metered electricity production must also be distinguished from estimates of displaced coal consumption or avoided emissions. Such estimates require assumptions about which generators would otherwise have operated, their efficiencies, transmission losses, and the accounting boundary; they are not direct measurements of the station’s generation.
Navigation Facilities and Inland Transport
Two parallel five-stage ship-lock flights on the dam’s left bank connect the downstream Yangtze River with the reservoir. Opened to navigation in June 2003, they normally separate upstream and downstream traffic. Each chamber has usable dimensions of approximately 280 metres long and 34 metres wide, with a minimum water depth of 5 metres. Vessels enter a chamber, gates close behind them, and water is admitted or discharged until the chamber level matches the next stage. Repetition moves vessels through a maximum overall water-level difference of about 113 metres. The locks accommodate groups of vessels, including large freight vessels and barge formations, rather than lifting each ship individually. Passage through the complete flight generally takes several hours, excluding the wait for admission.
The vertical ship lift, which began trial operation in September 2016, provides a separate route for suitably sized vessels. Its water-filled chamber has usable dimensions of approximately 120 metres by 18 metres and a water depth of 3.5 metres; it was designed for vessels of up to approximately 3,000 tonnes displacement. These limits concern the vessel’s dimensions, draft, and total displaced weight, not simply its cargo capacity. Counterweights and a rack-and-pinion drive system move the chamber vertically while the vessel remains afloat inside it. A passage can take about 40 minutes, substantially less than passage through the locks, although admission, operating conditions, and queues affect total journey time. The lift serves passenger vessels and smaller cargo ships; its chamber cannot substitute for the locks’ ability to process large freight formations.
Reservoir impoundment submerged rapids and shoals and increased channel depth along much of the approach to Chongqing. Reduced gradients and slower currents made upstream navigation easier and supported the use of larger vessels and formations, including 10,000-tonne-class fleets. That description does not mean that every berth or tributary channel can receive an individual 10,000-tonne ship throughout the year. Available draft depends on reservoir elevation, channel conditions, and port approaches. Navigation also remains dependent on channels farther upstream and on the separate Gezhouba navigation facilities downstream.
Freight figures distinguish actual traffic from engineering capacity. The locks’ original design benchmark was approximately 50 million tonnes annually in the upstream direction, rather than a ceiling of 50 million tonnes for both directions combined. By 2022, reported annual cargo throughput through the locks exceeded 150 million tonnes in both directions combined. Increased traffic reflects deeper navigation channels, but also larger vessels, improved scheduling, port investment, and the expansion of manufacturing and trade throughout the Yangtze basin. Chongqing’s river terminals connect inland shipping with road and railway networks, allowing bulk commodities, vehicles, and containerized goods to move between the upper river and downstream industrial centres.
The locks nevertheless impose a concentrated transport bottleneck. Vessels must register and assemble for scheduled lockages, with chamber allocation influenced by dimensions, destination, cargo, and operating rules. Queues can grow during traffic peaks or when maintenance removes one flight from service; the remaining flight may then handle alternating directions with reduced capacity. High river flows, fog, and other unsafe conditions can interrupt movements, while seasonal reservoir drawdown changes available depths and approaches. Consequently, greater navigable depth upstream does not translate automatically into shorter or more predictable end-to-end journeys: port handling, lock scheduling, downstream navigation, and onward rail or road connections all affect freight transit times.
Population Relocation and Resettlement
Impoundment submerged inhabited valley floors and lower slopes along the Yangtze River and its tributaries in Hubei and Chongqing. Relocation therefore involved not only households but also farmland, factories, shops, roads, water supplies, schools, and health facilities. County seats and smaller towns had to be rebuilt above the reservoir’s operating levels, while transport and utility networks required replacement connections. The administrative task extended beyond moving people out of the inundation zone: agricultural households needed replacement livelihoods, industrial enterprises required relocation or closure, and receiving communities needed additional public services.
Official accounts associated with the project’s overall acceptance in 2020 reported the relocation of approximately 1.3 million people. This is a cumulative project-resettlement figure, covering the staged clearance of the reservoir area and associated construction sites, rather than the number moved during a single filling operation. Relocation proceeded before and alongside successive impoundments beginning in 2003. The headline total should not be interpreted as a complete count of every subsequent move in the reservoir region: later evacuations or permanent relocations prompted by unstable slopes and other shoreline hazards were administered through additional programmes, and some affected households moved more than once.
The resettlement system combined nearby reconstruction with organized migration to more distant destinations. Many urban residents moved into replacement districts uphill from their former homes; rural households could be assigned land elsewhere within their county or municipality. Others were transferred to receiving communities outside the reservoir region, including coastal and central Chinese provinces. During the late 1990s, policy placed greater emphasis on outward resettlement because the remaining uplands could not provide equivalent farmland for all displaced farmers. The “development-oriented resettlement” approach sought to combine compensation with agricultural improvement, enterprise development, vocational training, and non-agricultural employment rather than treating replacement housing alone as sufficient.
Compensation and reconstruction operated through several distinct channels. Households received payments or replacement arrangements for housing and other eligible assets, while land-acquisition compensation and resettlement subsidies were administered through local authorities and rural collectives. Public funds also financed replacement infrastructure and the rebuilding of municipal facilities. Outcomes depended partly on household registration, eligibility rules, the quality and location of replacement land, and access to employment. A larger or better-equipped dwelling did not necessarily compensate for the loss of productive fields, customers, or a conveniently located workplace. The national reservoir-resettlement support policy introduced in 2006 provided eligible rural resettlers with support calculated at 600 yuan per person annually for 20 years, through direct payments, development projects, or a combination of the two.
Official oversight identified implementation problems as well as completed construction. The National Audit Office’s 2013 audit of the Three Gorges project reported irregularities in the management and use of resettlement-related funds, including diversion of funds and improper payments. Such findings concern the delivery of compensation and reconstruction programmes; they do not, by themselves, measure the long-term welfare of the entire relocated population. Conversely, administrative records of completed housing and infrastructure cannot establish that household incomes or agricultural productivity had been restored.
Peer-reviewed research likewise shows variation rather than a uniform outcome. Brooke Wilmsen’s 2016 longitudinal study, published in World Development, found that improved housing and aspects of material living conditions could coexist with continuing livelihood difficulties. Research by Wilmsen, Michael Webber, and Duan Yue on rural-to-urban resettlement also examined the difficulty of replacing agricultural livelihoods with secure urban employment. Differences in age, education, available labour, compensation, and proximity to expanding labour markets affected households’ ability to adapt. Rebuilt settlements could offer better roads, utilities, and access to services, while farmers allocated limited or less productive land, and workers without suitable skills, faced different economic constraints from households able to enter salaried work or establish businesses.
Archaeology and Cultural Heritage
Preparations for reservoir inundation included surveys of archaeological sites, historic buildings, inscriptions, tombs, and old town centres along the Yangtze River and its tributaries. Rescue archaeology expanded during the 1990s, with excavation schedules governed by construction and successive rises in reservoir level, beginning with the initial impoundment in 2003. Teams from national, provincial, and municipal institutions recorded settlement plans, excavated graves and occupation layers, and collected pottery, stone tools, metalwork, and environmental samples. Work at sites such as Zhongba in Zhong County revealed long sequences of settlement and salt production, while excavations at the Neolithic Daxi site in Wushan added evidence about prehistoric communities in the gorges. These investigations recovered information about ordinary settlements and production activities as well as conspicuous monuments.
Preservation methods depended on the material and its relationship to the site. Movable finds could be conserved and transferred to archaeological repositories and museums, including the China Three Gorges Museum in Chongqing, which opened in 2005. Buildings selected for relocation required measured drawings, photographs, identification of components, dismantling, and reconstruction on higher ground. Stone inscriptions and architectural fragments could sometimes be removed separately. Where relocation or protective engineering was impracticable, excavation and documentation preserved a record rather than the site itself. Such records retained plans, stratigraphic relationships, and selected objects, but could not reproduce all the evidence contained in unexcavated deposits or preserve their accessibility for future investigation.
The Zhang Fei Temple at Yunyang illustrates the relocation approach. Dedicated to Zhang Fei, the third-century military figure associated with the Three Kingdoms, the temple complex stood on the south bank opposite the old county town. Its relocation approximately 30 kilometres upstream accompanied the rebuilding of Yunyang away from the inundation zone. Historic architectural components, sculptures, inscriptions, and other objects were transferred to the replacement site, and the reconstructed complex opened in 2003. This preserved substantial material fabric and continued public access, but changed the temple’s relationship to its original riverbank, surrounding settlement, and historic approaches.
At Shibaozhai in Zhong County, protection instead retained the monument at its original location. Its conspicuous wooden pavilion rises against a steep rock outcrop and provides access to the temple above. A surrounding embankment protected the lower part of the complex from the raised reservoir, creating an enclosed area linked to the shore by a pedestrian bridge. The monument consequently acquired an island-like setting. This solution avoided dismantling the principal structure, but replaced its former relationship with the riverbank and nearby streets with a purpose-built protective enclosure.
The Baiheliang inscriptions at Fuling required another solution. The sandstone ridge bore carved fish, inscriptions, and records associated with low river levels, with dated inscriptions extending back to 763. Because their significance depended partly on their position relative to the river, selected carvings were protected underwater rather than transferred wholesale to a museum on land. The Baiheliang Underwater Museum opened in 2009: a water-filled protective enclosure covers the principal viewing area, while visitors descend through access tunnels to observe the carvings through windows. Keeping water inside the enclosure reduces the pressure difference across it. Reservoir inundation nevertheless ended the ridge’s former seasonal emergence during low water. Across these cases, physical preservation, continued visitor access, and preservation of original landscape or archaeological context were distinct outcomes, not interchangeable measures of survival.
Sediment Transport and Downstream River Change
Impoundment from 2003 reduced water velocity through much of the reservoir, allowing sediment carried by the Yangtze River and its tributaries to settle. Coarser sand deposits preferentially in upstream and backwater reaches, while finer silt travels farther toward the dam; some fine material remains suspended and passes through the outlets. Deposition is therefore uneven, concentrating in submerged channels, tributary mouths, and other low-velocity areas rather than forming a uniform layer. Incoming sediment load, retained sediment, and downstream sediment discharge are distinct quantities: retention is estimated from the difference between sediment entering and leaving the reservoir, with allowances for tributary inputs and measurement uncertainty. Retention percentages also vary with the observation period, flood conditions, and sediment supplied by the upstream catchment.
The operating principle commonly described as “storing clear water and releasing muddy water” exploits the seasonal coincidence of high river discharge and high sediment concentrations. Maintaining a relatively low reservoir level during the summer flood season increases flow velocity and helps sediment-laden water move through the reservoir; refilling later in the year stores comparatively clearer water. Deep outlets can convey sediment-bearing flows, but they do not enable complete flushing of deposits across the approximately 600-kilometre reservoir. Deposition consumes storage and can raise the bed in upstream navigation reaches, requiring channel surveys, dredging, and adjustments to navigation management. Near Chongqing, changing backwater conditions and sediment supplied by the Jialing River complicate channel behaviour. Since the 2010s, upstream impoundments, including Xiangjiaba and Xiluodu on the Jinsha River, have reduced sediment reaching Three Gorges, changing both deposition rates and the sediment available for downstream release.
Below the dam, water with a reduced sediment load can entrain material from the riverbed and banks. This sediment deficit promotes channel erosion, although its extent depends on discharge, bed composition, channel geometry, and bank protection. Observations downstream of Yichang and through the Jingjiang reach document bed lowering, redistribution of shoals, and changes in individual channels. As finer particles are removed, a coarser surface layer can develop and slow further erosion. Bank retreat is less uniform: it reflects local bank materials, revetments, groundwater conditions, and changes in the position of the main current. Altered seasonal releases also affect the duration of flows capable of moving sediment. Downstream sediment discharge consequently includes material eroded from the intervening channel, not simply sediment that has passed through the dam.
Connections with Dongting Lake and Poyang Lake introduce additional exchanges. Dongting receives Yangtze water and sediment through distributary channels as well as inputs from its own tributaries. Reduced main-stem sediment supply and changes in channel levels can alter both sediment deposition and water exchange. At Poyang, the difference between lake level and Yangtze level near Hukou controls drainage and backwater effects. Main-stem erosion and autumn reservoir filling can influence this gradient, but lake conditions also depend on tributary runoff, rainfall, and sand extraction. Changes in lake sediment budgets cannot therefore be attributed to reservoir trapping alone.
At Datong, a major downstream gauging station, mean annual suspended-sediment discharge fell from roughly 490 million tonnes in 1950–1968 to about 145 million tonnes in 2003–2012. This comparison measures a basin-wide change, not an isolated Three Gorges effect: sediment discharge had already declined before impoundment because of earlier dams, soil conservation, and other catchment changes. Reduced sediment delivery affects deposition and erosion in the Yangtze delta and adjacent estuary, but the response varies among channels, tidal flats, and offshore areas. Riverbed erosion partly replenishes the downstream load, while dredging, sand extraction, reclamation, tidal redistribution, and coastal engineering further modify where sediment accumulates or is removed.
Aquatic Ecology, Water Quality, and Climate Effects
Impoundment beginning in 2003 transformed part of the upper Yangtze River from a predominantly flowing channel into a deep, seasonally regulated reservoir. Increased cross-sectional area generally reduced current velocity and increased water residence time, although the main channel retains substantial throughflow and differs markedly from sheltered tributary bays. These changes altered sediment suspension, light penetration, plankton growth, and aquatic habitats. Greater water depth and heat storage also changed the seasonal temperature cycle: releases can be cooler than the former river during spring warming and warmer during autumn cooling. Thermal stratification varies with location, season, inflow, and operation rather than forming a uniform reservoir-wide condition. Between the usual operating elevations of approximately 145 and 175 metres, the drawdown zone is largely submerged during winter storage and exposed during summer flood-season operation—a reversal of the natural seasonal inundation pattern that affects shoreline vegetation and invertebrate communities.
Fish responses reflect both habitat conversion within the reservoir and altered conditions downstream. The four major Chinese carps—black carp, grass carp, silver carp, and bighead carp—spawn in flowing water, producing drifting eggs whose development depends on suitable currents, temperature, and downstream transport. Downstream observations after impoundment documented changes in spawning timing and reduced egg and larval abundance relative to earlier monitoring periods. Suppression of rising-flow signals and delayed spring warming provide mechanisms linking reservoir operation to these changes, but fishing pressure, pollution, other dams, and loss of connected floodplain habitats also affect populations. Beginning in 2011, ecological flow operations at Three Gorges deliberately created rising discharge pulses during the spawning season. Monitoring has associated some releases with increased spawning activity; these observations do not establish recovery of adult stocks or isolate flow management from all other influences.
Eutrophication is particularly evident in relatively slow-moving tributary reaches, including Xiangxi Bay and the Daning River backwater. Reservoir formation lengthened residence times in these areas, while tributary runoff and wastewater supplied nitrogen and phosphorus. Under suitable light and temperature conditions, weak flushing and water-column stability allow phytoplankton to accumulate into blooms. Exchanges with the main channel, density currents, and changing reservoir levels can either disperse algae or transport nutrients into favourable growth zones. Consequently, a comparatively well-flushed main channel can coexist with bloom-prone bays. Monitoring therefore requires more than a reservoir-wide water-quality classification: nutrient concentrations, chlorophyll-a, dissolved oxygen, temperature profiles, and phytoplankton composition reveal different aspects of ecological condition. Expansion of sewage collection and treatment addresses an important nutrient source, but agricultural runoff and nutrients retained in sediments remain relevant.
The project's greenhouse-gas implications require life-cycle assessment, not simply treating hydropower generation as emission-free. Relevant sources include cement and steel production, construction, maintenance, and decomposition of organic material in the reservoir. Flooded soils, incoming organic matter, and aquatic production can generate carbon dioxide and methane, released through surface diffusion, bubbles, or downstream degassing. Measurements vary substantially among tributary bays, the main channel, and exposed shorelines, making extrapolation from individual sampling sites uncertain. Gross reservoir emissions must also be distinguished from net changes relative to the river and terrestrial landscape before inundation. Estimates of avoided emissions depend on electricity actually generated and the assumed alternative generation mix; a comparison with coal-fired power yields a different result from one using a progressively lower-carbon grid.
Local climatic effects operate at a different scale from these global emissions calculations. The enlarged water surface changes evaporation and exchanges of heat and moisture, potentially modifying nearby humidity, fog, and day–night temperature ranges. Station observations and remote-sensing studies must separate these effects from elevation, urban expansion, land-cover change, and wider climatic variability. Local temperature or moisture contrasts do not by themselves establish that the reservoir caused a regional drought, major flood, or long-term change in the East Asian monsoon.
Structural Safety, Geological Hazards, and Long-Term Management
Safety monitoring at Three Gorges distinguishes the concrete dam and its foundation from the much larger area of reservoir-bank terrain. Survey networks, displacement instruments and foundation instruments track movement of the structure and underlying rock; piezometers and drainage measurements track seepage and uplift pressure beneath the dam. Uplift reduces the effective downward force available to resist sliding, making foundation drainage an important part of gravity-dam safety. Temperature and joint-opening measurements help distinguish normal thermal expansion and contraction from abnormal concrete behaviour. Measurements are interpreted against reservoir elevation, downstream water level and temperature rather than treated as isolated readings. Inspection and maintenance also cover spillway surfaces exposed to high-velocity flow, gate seals and hoisting machinery, and turbine-generator vibration, bearings and cooling systems.
Reservoir-bank instability involves different geological conditions from those at the dam foundation. Layered sedimentary rocks, weak bedding surfaces, fractured slopes and old landslide deposits occur along reservoir shores. Rising water can saturate previously unsaturated material and increase pore-water pressure, although water against a slope can also provide external support. During rapid drawdown, that external support declines while groundwater within a poorly drained slope may remain high, reducing effective stress and resistance to sliding. Heavy rainfall can further raise groundwater levels. Consequently, slope responses depend on rock structure, permeability and the timing of rainfall and reservoir-level changes, not simply on whether the reservoir is rising or falling.
The Qianjiangping landslide occurred on 14 July 2003 beside the Qinggan River, a tributary within the reservoir area, shortly after the first impoundment raised the reservoir to approximately 135 metres. Investigations identified a large rockslide controlled by geological structure and pre-existing weakness, following reservoir-level rise and sustained rainfall. Its entry into the water generated a destructive displacement wave. The case illustrates a hazard that does not require failure of the dam itself: slope collapse can threaten shoreline settlements and vessels through both direct movement and waves. Reservoir-shore management therefore combines geological inventories, displacement and groundwater monitoring, rainfall observations, warning thresholds and, where necessary, evacuation or relocation.
Seismic safety likewise requires separating ground motion at the dam from earthquake occurrence elsewhere around the reservoir. Design assessment considers earthquake loading of the concrete structure, foundation and associated facilities, including the interaction between the dam and impounded water. Research on reservoir-induced seismicity examines whether added water loading and the diffusion of elevated pore pressure can alter stresses on existing faults or fractures. Seismic networks have recorded small earthquakes in the Three Gorges reservoir region following impoundment, but spatial or temporal association alone does not establish the mechanism of every event. Earthquake occurrence, calculated structural response and observed damage are distinct kinds of evidence; a recorded reservoir-area earthquake is not, by itself, evidence of structural damage to the dam.
Emergency planning links instrument alarms and engineering inspections with operational decisions, communications and downstream warning procedures. Shoreline emergencies also require local geological-hazard authorities because a landslide may demand evacuation even when dam operation remains normal. Operating adjustments must account for competing requirements: flood-storage availability, safe rates of water-level change, navigation, electricity production and ecological releases. Coordination across the Yangtze reservoir cascade affects both incoming flows and the timing of downstream releases. Long-term uncertainties include changing flood and drought sequences, evolving reservoir-bank groundwater conditions, and ageing gates, turbines and monitoring equipment. Continued calibration, replacement of instruments and comparison of new observations with long records are necessary to distinguish gradual deterioration from seasonal or operational variation.