Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

Soil Conservation and Sedimentation of Reservoirs

India's massive programme to develop Himalayan water resources for hydroelectricity as well as irrigation, and concurrent soil conservation measures, has produced a large amount of data on rates of sedimentation of reservoirs. These actual rates almost invariably have been much higher than hydroelectric engineers had predicted prior to dam construction. The discrepancies between predicted and actual sedimentation rates are so great that serious concern has been expressed in view of the multi-million dollar expenditures on river channel and discharge modifications on the Ganges Plain and in the foothills; even more critical from this point of view is the current trend to accelerate high dam construction rates in Nepal and India. Before discussing sedimentation rates in the context of the assumed causes (deforestation and mountain land mismanagement) some of the actual data will be presented.

Dhruva Narayana (1987), Director of the Central Soil and Water Conservation Research and Training Institute, Dehra Dun, has provided a valuable overview of the downstream impacts of soil conservation in the Himalayan region. He indicates that of India's 328 million ha of land area, approximately 175 million ha are suffering from intense soil erosion. He maintains that the Himalayan and lower Himalayan regions have deteriorated extensively because of deforestation, large-scale road construction, mining, and cultivation on steep slopes. In the northeastern Himalaya alone he attributes the serious degradation of 3 million ha to shifting (slashand-burn) cultivation. Siltation rates in twenty-one Indian river valley projects were higher by 182 percent than the originally projected rates. He cites the Tehri catchment in the Garhwal Himalaya (total area 7,511 km², of which 2,328 km², are snow- and icecovered) as producing 14.6 million tonnes of silt annually. The annual silt load from the Karnali catchment is estimated at 75 million m³, equivalent to a denudation rate of 1.7 mm/yr. Singh and Gupta (1982) have determined that 28.2 tonnes/ha/yr are removed from the entire area of the Indian Himalaya. These are colossal figures and merit careful examination. Nevertheless, these figures also lie within the upper limits of derived denudation rates that have been attributed to geological, or natural, causes, as discussed above (pp. 105 to 109). This is an extremely important point. It implies that, assuming both sets of figures are within reasonable limits of accuracy, at least at a reconnaissance level, when considered on the regional scale (as distinct from that of a specific small catchment), the natural processes are so predominant that there is no requirement to seek human intervention as the cause of siltation. In other words, artificial reservoirs will silt up rapidly in this very dynamic region regardless of human influences - negative or positive.

On a more site-specific scale, Dhruva Narayana (1987) shows that soil losses of 80-156 tonnes/ha/yr have been recorded from the Chandigarh-Dehra Dun region (Lesser Himalaya and Siwalik small watersheds). These are absolutely unsustainable and destructive losses that can be drastically reduced by soil conservation measures. Such measures include contour cultivation, intercropping in maize, and planting of slopes with grass species.

In the Doon Valley, watershed gauging has demonstrated that transformation of naturally forested small watersheds to agricultural use results in a 72 percent increase in peak river discharge rates. Narayana goes

Table 5.2 Sedimentation rates for major Indian Himalayan reservoirs after Tejwani, 1986, unpublished).

Reservoir and date1

Annual Rates of Sedimentation

  Assumed Observed and year of observation % increase in observed value
Sutlej (Bhakra) (1971) 4.29 6.00 (1975) 39 9
    6.20 (1979) 45.0
Beas (1974) 4.29 15.10 (1975) 251.0
    23.59 (1981) 449.9
Ramganga (1974) 4.29 17.30 (1973) 303.3
Pohru²   7.41 (1973) -
Giribata²   11.60 (1973) -
Teesta²   98.20 (1973) -
Gumti²   3.56 (1973) -

¹Year of completion.
²No reservoir constructed.

On to show that significant reductions can be achieved through conservation practices, especially terracing and bunding. Tables 5.2 and 5.3 give some of the Indian experimental watershed data. Bunding, in particular, is claimed to have reduced soil losses in experimental watersheds by 94 percent. Similarly, the construction of brushwood check dams in a forested watershed reduced soil loss from 4.7 tonnes/ha/yr to 2.8 tonnes/ha/yr (a 54 percent reduction).

Table 5.3 Sediment load calculations for selected major Indian rivers (after Tejwani, 1986, unpublished).

River Site Sediment yield in acre feet/mi²/yr
Ganga Farraka 85
Arun Tribeni 242
Karnali Chisapani 600
Saptakosi Sunarambh 330
Sunkosi Tribeni 574
Tamur Tribeni 1,255
Trisuli T-Bridge 585
Brahmaputra Pandu 160
Buridhirang Rhowang 350
Dibang Tiagaon 564
Dikrang Dikrang Ghat 545
Lohit Digaru Ghat 710
Manas Mathauguri 145
Noadhing Namsai 310
Ranga Nadi Ranganadighat 180
Teesta Anderson Bridge 2,070
Chenab Ranihan 525
Ujh Panchtirthi 1,650

Table 5.4 Ranking of selected major rivers by sediment load: Table 5.4a) Major rivers of the tropics (after Tejwani, K C., 1986, unpublished).

River Drainage basin

Average annual suspended load

Estimated annual soil erosion from field

  (sq. km 000) (million tonnes) (tonnes/sq. km) (tonnes/sq. km) (tonnes/ha) Rank
Congo 4,014 65 16 320 3 13
Niger 1,114 5 4 80 0.8 14
Nile 2,978 111 37 740 8 12
Chao Praya 106 11 107 2,140 21 9
Ganges 1,076 1,455 1,352 27,040 270 3
Damodar 20 28 1,420 28,400 284 2
Irrawaddy 430 299 695 13,900 139 5
Kosi 62 172 2,774 55,480 555 1
Mahanadi 132 62 466 9,320 93 7
Mekong 795 170 214 4,280 43 8
Red 120 130 1,083 21,660 217 4
Caroni 91 48 523 10,460 105 6
Amazon 5,776 363 63 1,260 13 11
Orinoco 950 87 91 1,820 18 10

(From Tejwani, 1980, unpublished)

Table 5.4b Major rivers world wide (after Qiang and Dai, 1980).

River Drainage area Annual run-off Annual sediment load Average sediment concentration  
(other) (km²) (109 m³) (million tonnes) (kg/m³)
Colorado 637,000 4.9 135 27.5
Ganges 955,000 371.0 1,451 3.92
Missouri 1,370,000 616.0 218 3.54
Indus 969,000 175.0 435 2.49
Brahmaputra 666,000 384.0 726 1.89
Nile 2,978,000 89.2 111 1.25
Red 119,000 123.0 130 1.06
Irrawaddy 430,000 427.0 299 0.70
River Drainage area Name of the gauging station   Average sediment concentration Maximum sediment concentration Erosion modulus
(China) (km²)     (kg/m³) (kg/m³) (t/km²/yr)
Yellow 752,400 Sanmenxia   37.6 666 2,480
Yangtze 1,807,200 Datong   0.54 3.24 280
Haihe 50,800 Guanting   60.8 436 1,944
Huaihe 261,500 Bangbu   0.46 11.0 153
Liaohe 166,300 Tieling   6.86 46.6 240
  23,200 Dalinghe   21.9 142 1,490
Pearl 355,000 Wuzhou   0.35 4.08 260

Finally, when fast-growing Eucalyptus trees were raised in a denuded watershed (for fuel and paper pulp in a short rotation of ten years) total runoff was reduced by 28 percent and peak discharge by 73 percent. Similar advantages were obtained with coppiced plantations of eucalypts indicating the importance of a protective ground vegetation cover under a light crown of introduced trees.

From these data it is apparent that human intervention, both positive and negative, can result in large changes in both soil loss and river discharge on a small scale.

Table 5.2 provides sedimentation rates for some of the major Indian Himalayan reservoirs (Tejwani, 1986, unpublished, 1987). Table 5.3 provides sediment load data for eighteen major rivers and tributaries with a Himalayan origin and similar data for fourteen non-Himalayan rivers (Tejwani, 1986, unpublished, 1987). The very high sediment loads of the Ganga and Brahmaputra main channels, and the (Sapta) Kosi, Dibang, Teesta, and Lo hit tributaries, corroborate the proposition that these rivers reflect the high rates of uplift and denudation in the Himalaya (pp. 98-99). The generally much lower figures for the non-Himalayan rivers are striking. On a worldwide comparison, the Ganges, Kosi, Damodar (and Brahmaputra), in tonnes per km², of catchment (estimated annual soil loss from the field), rank extremely highly. Table 5.4 shows the ranking of a series of rivers in different parts of the world. Two observations are noteworthy: one is that the figures indicate that extremely serious problems of erosion, sediment transfer, and siltation are facing the Ganges-Brahmaputra system and large parts of China; the other is that there are wide discrepancies in the different data sets (Qian and Dai, 1980, also included in Table 5.4). While the cause for alarm is real enough, this issue also falls squarely within our theme of 'uncertainly on a Himalayan scare' (Thompson et al., 1986).

Table 5.5 shows loss of Indian reservoir capacity as percentage of dead storage to live storage in ten reservoirs. While the data are incomplete and little is available after 1974 the development of a very serious situation is apparent.

Table 5.5 Loss of Indian reservoir capacity as a percentage of dead storage to live storage in ten reservoirs (after Tejwani, 1986, unpublished).

Name of reservoir River system Dead storage(%) Live storage(%) Remarks
Sutlej (Bhakra) Indus 16.0 2.5 up to 1974
Maithon Ganga 26.0 12.0 up to 1974
Panchet Ganga 35.2 17.8 up to 1974
Mayurakshi Ganga 21.8 7.7 up to 1974
Nizamsagar Godavari 100.0 60.4 up to 1974
Lower Bhawani Cauveri 100.0 3.8 no date
Matatila Yamuna/Ganga 15.0 8.5 no date
Gandhisagar Yamuna/Ganga 27.2 11.5 in 14years
Girna   33.8 7.2 in 14 years
Hirakund Mahanadi 29.9 9.0 in 22 years

Two points must be made concerning reservoir siltation, estimated suspended sediment load, and small river basin conservation response data. First, there are inadequate long-term data. Second, the data that are available do not give any reliable information about the relative importance of human landuse interventions and natural processes, except on a scale of tiny experimental watersheds. All that can be concluded is that the Ganges and Brahmaputra systems carry a higher sediment load than most other major rivers world wide. Some of the Himalayan tributaries, for example, the Tamur, (Sapta) Kosi, and Teesta, produce still higher rates in proportion to the area of their catchments.

Kattelmann (1987) has produced a valuable overview of hydrological problems in terms of development of Himalayan water resources. He stipulates that 'many watershed management projects have failed to moderate reservoir sedimentation except where there was little problem to begin with and [where] upstream cultivation and forestry were eliminated.'The former is irrelevant to our discussion of the highland-lowland sediment transfer linkage; reducing upstream cultivation and forest utilization is virtually impractical where a growing highland population is facing severe land shortage.

Impact of Road Construction on Sediment Production

Tejwani (1987), quoting data from Bansal and Mathur (1976), indicates that in the Indian Himalaya for each linear kilometre of mountain road, ten small to medium landslides occur as the direct result of slope instability caused by the road construction. Prior to the 1962 border war with China, the Indian Himalaya were, for the most part, accessible only on foot along trails; the only roads led to famous hill stations such as Mussoorie, Simla, Nainital, and Darjeeling, products of the British Raj. The shock of the Chinese military presence on the Himalayan frontier prompted a massive road construction programme that was put into effect after 1962 in great haste, for military expedience outweighed concern for careful planning and sound engineering.

According to Tejwani, post-1962 road construction has produced more than 10,000 km of highways in the Indian Himalaya. He estimates that poor alignments and ill-considered design are resulting in a total 'veil loss' of 1.99 million tonnes annually. This he equates to slope movement of 1.99 tonnes of sediment per linear metre of road per annum (recalculation indicates a misplacement of the decimal point, so this figure should read 0.199 tonnes/ m/ yr - authors'note). Valdiya (1985, 1987) uses a figure of 44,000 km and calculates that during the construction phase an average kilometre of road required the removal of 40,000-80,000 m³ debris.² After construction, the extensive slope instability resulted in the production of enormous volumes of debris, usually dumped on the road bed and further downslope during heavy monsoon rainstorms in the form of debris flows, rockfalls, rockslides, and mudflows. Valdiya provides the following estimates of annual debris production per linear kilometre of road bed for three specific highways:

Western Himalaya: Jammu-Srinagar 724 m³
Central Himalaya: TanakpurTawaghat 411 m³
Eastern Himalaya: Arunachal Pradesh 719 m³

From these figures he derives an average annual debris production of 550 m³ per km and a total debris production of 24 x 105 m³ for the 44,000 km road network.

In practical terms, these circumstances, regardless of the somewhat different data derivations of Tejwani and Valdiya, have induced frequent highway blockages and enormous maintenance costs, especially during the summer monsoon periods. And since the standard method of road clearance is to dump the debris over the road side and down the slope below, this in turn further extends the area of instability. It causes destruction of downslope vegetation cover as well as the agricultural terraces of local subsistence farmers, who are usually not compensated for their losses.

Narayana and Rambabu (1983) accredit the enormous annual production of debris to unsatisfactory highway alignment and poor design. Valdiya (1985:24) states that 'The damage to the ecological balance [of the Himalaya] is mostly man-made or is caused by human negligence' end cites three main causes: road construction, overgrazing, and reckless deforestation. He, like Narayana and Rambabu, believes that road construction in seismically and tectonically unstable bedrock is 'the most important factor.' Haigh (1982a and b, 1984a and b) has undertaken detailed investigation of landslides along the MussoorieTehri road in Uttarakhand. He measured 470 debris outfalls in 1977 and 1978 and concluded that the frequency of movement was related to depth of road cut, steepness of slope, degree of forest cover, and geological structure and Ethology. He showed that annual costs for road clearance could be calculated from a simple measure of outfall width of the twenty largest landslides for this particular highway, data that could be obtained readily from air photographs.

Each of the above-cited researchers assumes that landslides produced by road construction are responsible for recent massive increases of suspended load in the local headstreams of the Ganges. Furthermore, a significant proportion of this is also assumed to be carried downstream to the Bay of Bengal, contributing to the development of New Moore Island (24 x 11 km) offshore of the Sundarban delta (Valdiya, 1985: 20-4).

Regardless of the weight of argumentation and the relative precision of the estimates of debris dumped onto specific sectors of the road bed each year, no accurate assessment of the proportion of the debris which enters the local rivers is available. Many of the landslides run part-way down the slope below the road bed and do not even reach the third-order stream channels. We feel obliged to insist, therefore, that while road construction and road maintenance problems in many parts of the Himalaya, and the attendant destruction of local slopes, are economically catastrophic, no data are available to support the claim that any of the debris so produced reaches the plains, let alone the Bay of Bengal. However, the landslide scars associated with the Himalayan roads are very evident even to the casual observer, and it is reasonable to conclude that sediment yield will be proportional to the density of the road network. The fact that Laban (1979) concludes that only 5 percent of Nepalese landslides are attributable to highway construction is presumably a reflection of the much lower mountain highway density in Nepal.

The data on road construction and induced slope stability introduced above are significant. The broader interpretations of the effects of roadworks on flooding and sedimentation on the Ganges Plain and on the seaward portion of the delta in the Bay of Bengal are supported by many workers (Haigh, 1982a and b, 1984a and b; Tejwani, 1984a and b, 1987; Valdiya, 1985; Narayana, 1987). We believe, nevertheless, that our general conclusions to the effect that man-made erosion (accelerated erosion) is negligible on a regional scale are not invalidated. At issue, however, is the 'uncertainty theme.' The seriousness of reckless road construction from a physical point of view cannot be doubted. Much avoidable damage has been, and is being, perpetrated. The costs of keeping the roads open are significant in themselves, as are the economic losses of repeated and widespread road closures. The development of between 10,000 and 50,000 km of road has also produced a major socioeconomic impact on the Himalaya which includes greater accessibility of hitherto remote forests to commercial logging, ease of movement of people both from the mountains to the cities of the neighbouring plains and from the plains to the mountains.

Geomorphic processes and human interventions: summing up

We have tried to demonstrate in this chapter that mountain geomorphology faces many problems in its attempts to identify and to rank specific slope processes in terms of their relative importance or absolute amounts of 'work' accomplished. These problems confound both determination of the manner of evolution, over geological time, of the mountain landscape as we see it today. They also limit understanding of what is currently happening in natural, or nearnatural, situations. The crux of this issue is the representativeness (or lack of) of collected data in both space and time. Especially difficult is assessment of the relative and absolute importance of high magnitude events with very long recurrence intervals compared with the continuous slow-moving downslope material transfer. Regardless of these problems a broad and extremely valuable understanding of denudation rates, orogenesis, and sediment transfer on a regional scale and within a geological time-frame has emerged.

We conclude from the foregoing discussion that natural processes in this dynamic region virtually obliterate the effects of human intervention, in so far as they can be gauged from the available data. Human intervention takes many forms; we have emphasized, within the Himalayan context, deforestation and general changes in land use as influenced primarily by the needs of a rapidly growing subsistence mountain farming population to sustain themselves. Other interventions have been introduced, including road construction and the installation of hydroelectricity facilities.

In taking our discussion from a rather academic review of geomorphology as a science into considerations of soil erosion and landslides that are widely presumed to be caused by bad land-use practices, we have inferred that the two sets of processes, geophysical and human, are probably several orders of magnitude apart. Furthermore, we believe we have demonstrated that the adherents of a theory stipulating that human intervention is the primary culprit of the large-scale environmental damage (as perceived from a human point of view) have failed to prove their case. Landslides and gullies, bare soil, floods, and silted reservoirs and irrigation works are visually graphic. At a local, site specific, scale they are serious and need to be corrected or reduced. However, we have found no reliable data to indicate that a specific mountain road, for instance, however badly constructed, is contributing a single grain of silt to the Ganges-Brahmaputra delta. We believe that, until there is very strong evidence to the contrary, the geophysical and climatic processes of our region, as reflected in some of the most rapid uplift estimates (orogenesis) and high rates of mountain denudation and concomitant accumulation of vast thicknesses of sediments, in the form of the Ganges and Brahmaputra plains, give adequate explanation for the workings of this extremely active landscape. The next major issue - large-scale processes on the plains - is the focus of Chapter Six.

Contents - Previous - Next