CHAPTER XX: Huanghe, the Yellow River

"Whoever controls the Yellow River controls China," said the Great Yu, who is credited with the first "taming" of the river around 2200 BC. Well he might: according to legend, his reward was to be made Emperor.

The Yellow River flows across a great blanket of loess, deposited by winds that blew out of the Siberian tundra during the Ice Age. The river erodes the loess easily, carrying along sixty times the sediment load of the Mississippi, until it reaches the plains of North China, still 1400 river km from the sea. The river carries more sediment than any other major river in the world when it reaches the North China Plain. Then the sediment is deposited as fine silt, building up the river bed until it is high above the surrounding plains. In flood times the river may burst its banks, turning the plain into a flood plain, and unable to return to its breached course as the flood recedes. The Yellow River now finds a new way to the sea, and the cycle begins again.

The changes of course have been spectacular, and the river mouth has sometimes changed catastrophically by hundreds of kilometers. It has had 26 major changes in course in the past 2000 years. The river is so treacherous that there was no permanent bridge across it until 1905.

As China has become ever more populous, the floods of the Yellow River have become increasingly dangerous. The Yellow River watered the cradle of Chinese civilization, but also rose in tremendous floods. A Minister of Works was appointed to take preventive measures after a disastrous flood in 2297 BC, but to little avail, because there was another great flood a few years later. The Minister was banished, and another was appointed, and it has been a feature of Chinese administration that for over 4000 years there has almost always been a major official charged with flood control. But flood followed flood: that of 2079 BC caused millions of deaths.

The Cheng-kuo canal was completed by the state of Qin in 246 BC. It irrigated 200,000 acres in the Wei valley north of Xian, and led to such prosperity and population increase that the Qin state eventually became the first to unify China, under the Emperor Qin Shi Huang. The Qin capital was Xi'an, and Qin Shi Huang's tomb was guarded after his death in 210 BC by the famous terracotta army, 8000 strong. However, both the Qin dynasty and the Cheng-kuo canal were shortlived: the canal quickly silted, and was relatively useless after about 150 years. Like all early canals, the system required constant maintenance, and often complete renovation in the face of Yellow River silt. Furthermore, the Qin population pressure induced settlers to move more into the loess plateau, beginning a cycle of deforestation, overcultivation, soil erosion, and increased silt run-off that has accelerated land-use problems in the Yellow River basin ever since.

The Han dynasty that followed the Qin continued to promote large flood control, irrigation, and navigation projects on the Yellow River. They too encountered disasters on a monumental scale. An earthquake sequence from 193 to 176 BC in the loess country not only killed thousands of people directly, but blocked the river with landslide debris that eventually failed catastrophically, flooding downstream areas. 179 BC is still remembered as occurring in "The Year of the Great Flood." Another great flood followed in 138 BC.

From Han times, it has been clear that major investment of human effort would be required to build and maintain levees and canals: and usually the choice has been accepted. The major effort has always gone into flood control on the plains, with irrigation and navigation perceived as additional (but subsidiary) benefits. Heroic efforts have been made to control the damage wreaked by the Yellow River, so that the people could receive the benefits. All these projects used technology that did not change for 2000 years, that is, massive numbers of people working with simple means such as hand shovels and wheelbarrows. This tradition constrasts strongly with practices in the West, where civil engineering schemes often led to rapid advances in technology and labor-saving devices. But whatever the technology, the methods for flood control evolved in parallel, with the Chinese centuries ahead of Western engineers.

In 8 BC an advisory committee led by the engineer Jiarang suggested three steps toward control: to "channelize" the river, improving its rate of flow to the sea; to divert enough water down irrigation canals and into diversion basins that floods would be mitigated; and last, to build higher levees. All three methods have been applied, with less than complete success. A Han successor Wangjing, working from about 58 AD to 76 AD, was able to stabilize the Yellow River with levees in a way that lasted for centuries. He was lucky, however, because the river had shifted its course about 11 AD, and had been wandering uncontrolled across the flood plain. Wangjing therefore inherited river flow across low land, and low levees were enough for temporary control while higher ones were built. It took the river a long time to silt its bed back to the critical levels that Jiarang had faced.

Inevitably the dyke building had to continue, but was eventually a losing battle against silting of the river bed. In fact, the dykes have failed in one place or another 1500 times in the last 2000 years, and the Yellow River has made further catastrophic course changes. Nevertheless the irrigation schemes based on the river have increased land productivity enough to feed one of the densest rural populations in the world, with surplus usually remaining to feed large cities.

Attempts to control the Yellow River can be categorized by different strategic approaches: it is ironic that the same discussions were replayed centuries later in attempts to control floods on the Sacramento drainage in northern California.

One strategy is active control of the river: to confine it within a narrow channel by high levees. The narrow-channel concept carries the danger of active erosion of the levee, but it encourages fast flow that keeps sediment in suspension, and therefore allows only slow silting of the river bed: in some circumstances the river may actually cut its bed deeper. However, there is little reserve capacity for absorbing a major flood crest, and even the high levees will inevitably be overtopped.

On the other hand, one might adopt a strategy of confining the river in a wider flood plain, between lower levees. This is cheaper to construct, but requires that more land be sacrificed to river control. It also permits a slower flow, and promotes silt deposition. Over time the river will inevitably build up its bed. However, there is much more reserve capacity for flood water in a wide channel, and there is room to build small diversion dams to encourage the river to keep to the center of the channel, avoiding the problem of scouring against the levee foundations. Either strategy can be combined with large diversion basins, into which some of the river flow can be diverted until a flood crest passes (this had been one of the strategies of the Great Yu).

Chinese scholars have sometimes seen the close confinement of the river as a "Confucian" solution of discipline and order imposed upon nature: this contrasts with the "Taoist" solution of allowing the river a more "natural" course within lighter constraints. In either case, however, river engineering represented a tremendous interference with any "natural" regime: and the contrasting solutions were more opposites of engineering than philosophical approaches. As in northern California, the "Taoist" solution was eventually preferable because the river's flow was so variable and its flood crests in particular so awesome. For example, in July 1958 flow along the river tripled in only 24 hours.

The discussion between the two styles of engineering had been played out two thousand years ago between followers of Jiarang ("Taoist," 8 BC) and Wangjing ("Confucian," 60 AD). As we have seen, Wangjing's methods worked, probably because he was able to begin from a state of chaos, so that it took centuries for the silting problem to overtake his levees.

After a devastating flood in 1917 the Chinese government invited outside advice on Yellow River flood management, and the debate was replayed. Some engineers advocated high narrow levees, while others recommended low wide river channels. Political turmoil, war with Japan, and civil war prevented any large-scale river control projects: in fact, the river underwent more war damage than new construction. On June 9, 1938, the Chinese blew up the levees in an attempt to stop the invading Japanese. The river was close to the peak of its annual flood, and swept over 9000 square miles of the plain. Close to a million Chinese peasants died or starved, and over 12 million were made homeless, and the Japanese army was not stopped. The flood of 1943 destroyed crops in Henan, and 3 million people starved to death.

At first, the new Communist government that came to power in 1949 built the traditional way, by strengthening and raising the levees. But they did build two tiers of levees: an outer set to confine large floods, and an inner one to concentrate the low-flow river and encourage it to scour away its load of silt. This was based on the Confucian dictum of the great Ming engineer Panjixun, "Build dykes to hold water in check, and let the water carry away the sand." Traditional methods of levee building continued, spurred by promises that when the upstream projects were completed, the problems would be reduced.

The new Communist government had to choose the best plan for basin-wide control of the Yellow River. It was clear that silt control was the key to most of the problems. Reafforestation and/or soil conservation would cut the supply of eroded silt to the river, but it would take several decades to work. In the meantime, any major construction would have to deal with silt. In 1954, the Chinese began to implement more Taoist solutions to the problems of the Yellow River by including engineered spillways and levees set to protect a wider river course.

However, the Chinese made the gigantic mistake of turning to the Soviets for technical help. In 1955 the Soviet plan for the overall engineering of the Yellow River was presented to the 1st National People's Congress. It carried many of the hallmarks of Stalin-era engineering: it concentrated on massive prestige projects, around which lesser details were built. It called for multi-purpose river use, centered on 46 dam projects varying from large to gigantic, and large land reclamation schemes, and many secondary storage reservoirs. Dams upstream would largely serve to generate hydroelectric power (HEP); those in the middle and floodplain section would promote irrigation; and those in western Honan would provide flood control. The Soviet proposals had already been openly questioned at a planning conference in 1953: some Chinese engineers had noted that no Soviet rivers flooded as badly as the Yellow, and no Soviet rivers carried such a load of silt, or such a danger of building up the river bed in the lower stretches: therefore, they doubted whether Soviet advice would be useful. Attempts to build multi-purpose structures might therefore lose sight of the primary objective in the Yellow River, which was flood control.

Nevertheless, the Soviet plan was approved, and work began on its giant centerpieces, the dams at Sanmenxia as the river enters the lower plains, and at Luijiaxia above Lanchow. The Sanmenxia Dam was projected as holding 36 cu km of water, and would have a 1000 MW HEP plant. The design involved the resettlement of 600,000 people. There would be no silting problem in the reservoirs, claimed the Russian engineers (the people who designed the High Aswan for Egypt at about the same time), because the erosion that caused silting would be cured by afforestation and soil conservation. The particular feature of the scheme that turned out to be totally disastrous was the decision to build a high dam at Sanmenxia without any silt traps upstream, and without any mechanism for flushing silt through the dam.

The Russians quarrelled with the Chinese in 1960, and pulled out their technicians before the first major dams were finished. This slowed construction, but eight major projects were eventually completed. Left on their own, the Chinese learned by bitter experience that revolutionary enthusiasm and Russian advice are no substitute for far-sighted, careful engineering. The Sanmenxia dam ran into severe silting problems almost immediately. The dam was not even provided with silt-flushing gates. Some silt was supposed to run harmlessly through the Russian-built HEP turbines with the water, but instead, it clogged them. Massive amounts of silt were deposited at the upstream end of the reservoir, which filled at an alarming rate. In 1960, when the dam was closed, the reservoir held 96 cu km of water, but the river was dumping sediment into it at 10 cu km/yr, so that the reservoir held only 56 cu km of water by 1964, and had lost 41% of its storage capacity. In an act of desperation, some of the generators were scrapped in a redesign effort to get rid of silt. HEP capacity was dropped to 50 MW from the planned 900 MW before the turbines were installed.

Even then, the remedial actions did not work to restore the original plan. Sanmenxia released about 500 million tons of sediment a year after its redesign, but even more accumulated. By the late 1980s the reservoir was down to a flood-reserve capacity of only 1.8 cu km, and silt scouring had badly degraded the retrofitted discharge tubes. Sanmenxia is not used at all now to impound "normal" floods, but is still in place to provide some control of very large floods. It is certainly not the centerpiece of Yellow River control, as had been hoped. The Chinese are beginning to plan a new dam downstream from Sanmenxia that will replace some of its functions.

Six other large dams had been built on the river by 1988, providing between them protection, thought the Chinese, against a 60-year flood. The Longyangxia Dam was begun in 1978 and began to impound water in 1986. It is China's highest dam (178 m) and impounds the largest reservoir (25 cu km).

The Yellow River today.
The Yellow River Basin grows more than 50% of China's wheat, cotton, and tobacco. The average flow of the river in its lower course is 56 cu km/yr. It loses 17 cu km/yr in Mongolia, and then downstream to Zhengzhou, from percolation, evaporation, and diversion for irrigation. Another 10 cu km/yr are drawn off across the North China Plain.

The Chinese have managed to control normal floods on the Yellow River, but they are only buying time. The really fundamental problems of the Yellow River are siltation and its corollary, upstream erosion, especially in the loess belt. The incredible erosion on the loess plateau is a natural phenomenon to some extent, but it has been greatly increased by human activity, especially deforestation, overgrazing, and overcropping. The upstream reservoirs in Shanxi province are filling at 80 million cu m of sediment a year, and are thereby losing not only volume but the ability to absorb flood crests. The problem is getting worse rather than better. Perhaps 3000 years ago, loess erosion was 1 billion tons a year from the plateau: it was 1.6 billion tons in the 1950s and was more like 2.2 billion in the 1970s. However, some Chinese engineers are beginning to wonder whether any conservation measures can reduce loess erosion by a worthwhile amount.

The sediment load of the Yellow River averages 37.6 kg of silt per cu m (compared with 0.07 for the Amazon and 0.6 for the Mississippi). Overall, the historic sediment load of the Yellow River since 1919 has been about 1.6 billion tonnes of silt a year. About 1.2 billion tonnes was swept out into the Gulf of Bohai, but perhaps 0.4 billion tonnes was deposited in the river bed in an average year, raising the bed another 10 cm. In the flood of 1933, 3.7 billion tonnes were deposited, and 0.9 in 1977. Against this background, a figure of only 0.2 billion tonnes deposited in 1986 can be seen either as a triumph of soil conservation or as the result of a dry year.

Silt deposition worsens as the river flow lessens across the flooplain. Ironically, this happens as more and more river water is diverted for irrigation. The irrigation diversions now mean that in some years the Yellow River no longer reaches the sea in parts of June, before the mountain snowmelt reaches the plains in July and August; but the gap was averaging 15 days in the 1970s, and was getting worse. Schemes have been proposed to divert Yangtze water into the lower Yellow River plain, and by 1987 some Chinese engineers were calling such a diversion "inevitable" or "indispensable."

By the time the Yellow River reaches Kaifeng, it runs 10 m above the surrounding plain, where the river bed is above the rooftops of the houses behind the levees. Ming dynasty artefacts 500 years old lie under 5 m of silt at Kaifeng, and Song artefacts 800 years old lie 6 m deep. Archeologists were stunned (and delighted) recently to find the famous Song dynasty Bridge of Zhou, portrayed in one of China's most famous art works, buried 8 m below ground level at Kaifeng! The present height of the river above its floodplain is about the same as was the old course that was breached and abandoned in 1855.

The levees on the Yellow River floodplain have already been rebuilt four times in forty years: in 1950, 1955, 1964, and 1977. Fleets of suction dredges remove silt from the river bed, and pile it up on and outside the levees, raising them, but also slowly raising the level of the floodplain too, and fertilizing it. Small diversion structures try to keep the river's main current in its center, so that it will not undercut the base of the levee. Major irrigation projects have been instituted on the flood plain. There is not much reserve flood storage capacity, however: and the density of the floodplain population has meant that 1,500,000 people have settled in the two largest "diversion basins." Finally, a major oilfield was discovered under one of them!

Altogether, the Chinese feel that the system is secure against a "30-year" flood, but they recognize that a "100-year" flood, like those of 1761 or 1843, would be catastrophic. They believe that the present course of the Yellow River across the flood plain can be maintained for another 100 years. It is not clear what can be done after that. Clearly, one could dig out an alternative route for the river, complete with levees, and then divert it into the new course which it would then follow (perhaps for another few centuries). However, such a scheme would be incredibly expensive. It will not lightly be undertaken before it becomes essential. It seems far more likely to me that there will have to be a catastrophe before China summons the resources for a major renewal of the Yellow River floodplain.