Challenges in African Hydrology and Water Resources (Proceedings of the
Harare Symposium, July 1984). IAHSPubl. no. 144.
Sediment deposition in major reservoirs in the
Zambezi basin
P. BOLTON
Hydraulics
Oxfordshire
Research Ltd,
OXlO 8BA, UK
Wallingford,
ABSTRACT
Fragmentary information from various sources
can be used to give order of magnitude estimates of the
present rates of sediment deposition in Lakes Kariba and
Cabora Bassa. Such estimates are used to identify the
possible implications for future reservoir operation and
for long-term regional planning. This information may be
used to plan appropriate sediment monitoring programmes
within the limitations of available resources. The study
suggests that sediment deposition in Lake Kariba will
have a negligible effect for many centuries whereas in
Lake Cabora Bassa its effect may be appreciable within a
few decades.
Dépôt de sédiments
dans les grands réservoirs
du bassin
du Zambèze
RESUME
Des renseignements fragmentaires provenant de
diverses sources peuvent être utilisés pour obtenir des
évaluations des ordres de grandeur des taux actuels de
dépôt de sédiments dans les Lacs Kariba et Cabora Bassa.
Ces évaluations sont utilisées pour identifier les
conséquences possibles, aussi bien pour l'exploitation
future des réservoirs que pour la planification régionale
à long terme. Ces renseignements peuvent être utilisés
pour planifier des programmes appropriés de contrôle de
la sédimentation dans la mesure des ressources disponibles.
L'étude suggère que le dépôt de sédiments dans le Lac
Kariba aura un effet négligeable pendant de nombreux siècles
tandis que dans le Lac Cabora Bassa, son effet pourra
être appréciable en l'espace de quelques dizaines d'années.
INTRODUCTION
Very few reservoirs in Africa have been adequately studied prior to
impoundment for the purpose of determining the probable rate of
sediment accumulation and its implications. The reservoirs of the
Zambezi basin are no exception. The neglect of this question arises
in part because the effects of sediment deposition are long term and
considered to be negligible within the "economic life" of most
projects and in part because the effects are difficult to quantify
from the fragmentary information generally available. Nevertheless,
order of magnitude calculations based on information and data drawn
from a variety of sources can be useful in identifying where future
problems may arise, so that account may be taken of these effects in
the planning of regional economic development and in the organization
559
560
P.Bolton
of cost effective sediment monitoring programmes.
In this paper estimates are made of the rate of sediment deposition
in the two largest reservoirs in the Zambezi basin, Lake Kariba and
Lake Cabora Bassa. These reservoirs have been selected for study not
because they are in the most immediate danger from siltation (many
smaller reservoirs will be affected sooner) but because of their
importance within the economies of Zambia, Zimbabwe and Mozambique
and their dominant influence over the future development of the lower
Zambezi valley. In addition, comparison of the results reveals
significant differences between the two cases arising from important
contrasts in their design characteristics and operating procedures.
LAKE KARIBA
Estimated
rate
of sediment
deposition
The rate of sediment deposition in Lake Kariba was briefly considered
in the pre-impoundment studies (see Central African Council, 1951).
A useful life of 1000 years was foreseen on the basis of a limited
number of samples taken at Kariba Gorge between 1948 and 1950. The
calculation of this figure has been examined by Bolton (1983a) who
concludes that when allowance is made for an increase in reservoir
capacity, in the project as finally built, and when an apparent
computational error is rectified, the data from these samples suggest
that the period necessary to completely fill the reservoir's "dead"
storage capacity should be about 5000 years.
To provide independent verification of this estimate, the general
characteristics of the drainage basin were studied alongside specific
information and data derived from a number of published sources.
The area of Lake Kariba's drainage basin is approximately 650 x 10
km . Of this, 480 x 10 km lies upstream of the Victoria Falls.
It may be assumed, with reasonable justification, that the Barotse
Plain and Chobe Swamps act as sediment deposition zones for virtually
all the sediment from this part of the basin. Tributaries downstream
of the Victoria Falls, draining directly into the Gwembe Trough,
3
2
comprise the remaining 170 x 10 km of Lake Kariba's drainage basin
(140 x 10 km lying in Zimbabwe and 30 x 10 km lying in Zambia)
and it is from these that the bulk of the sediment deposited in Lake
Kariba originates.
Published studies relate only to the drainage basins of tributaries
in Zimbabwe. It will be assumed that the sediment yield from the
tributaries in Zambia is similar. Ward (1980) and Chikwanha (1980)
studied two tributaries of Lake Kariba, the Gwaai and the Umsweswe.
between 1975 and 1979. Despite significant differences between the
characteristics of the two basins upstream of the sampling sites the
results suggest that the sediment yields of the two were similar at
—2
—1
about 40 t km year . The mean sediment concentrations differed
significantly (approximately 1600 mg 1" for the Gwaai and 500 mg 1~
for the Umsweswe). In seeking to derive mean values of sediment
yield from these results, account must be taken of two important
characteristics of the drainage basins: firstly, they lie on the
plateau rather than on the steeper escarpment slopes of the Gwembe
Trough; and, secondly, neither are undergoing accelerated erosion due
Sediment
deposition
in
Zambezi
reservoirs.
561
to population pressures.
Zimbabwe has, for many years, had an active research programme
into the implications of soil erosion for arable agriculture. Of
particular interest is the study by Stocking & Elwell (1973) of
potential erosion hazard throughout Zimbabwe based on various
physical and land use parameters. Although it is unlikely that their
approach can provide quantitative predictions of the rate of erosion
in a given area, it has been useful in identifying regions of the
country where erosion rates are potentially high. In Fig.l a
simplified version of their results is presented for the tributary
basins to Lake Kariba and Lake Cabora Bassa within Zimbabwe. The
drainage basins studied by Ward (1980) and Chikwanha (1980) which
0
100
flake Cabora Bassa
Drainage basin of
Lake Cabora Bassa
Drainage basin of
\Lake Kariba
Potential erosion hazard
based on Stocking and Elwell
Drainage basins studied
by Ward and Chikwanha
Very low to low
S = Gwaai
U = Umsweswe
H = Hunyani
J Below average to average
Above average to very high
FIG.1
Potential
Zimbabwe.
erosion
hazard
in
tributary
basins
in
are outlined in Fig.l lie largely in regions where the potential
erosion is considered to be below average or lower. The value of
mean yield found by Ward and Chikwanha is, therefore, likely to
provide only the lower limit of the estimated mean yield for the
whole area under consideration. Significant parts of the escarpment
of the Gwembe Trough have potential erosion rates classified by
Stocking & Elwell (1973) as above average or higher. Such regions
are likely to have an appreciable effect on the mean yield of the
whole area and, since local erosion rates may vary by several orders
of magnitude, a safe estimate for the upper limit of the mean yield
for the whole area, in the absence of further information, would
probably be a factor of 10 greater than the lower limit. In other
words, the mean yield of the drainage basin of Lake Kariba downstream
of the Victoria Falls lies in the range 40-400 t km~2year-1.
On the basis of the foregoing, admittedly broad, range of values
562
P.Bolton
for sediment yield, the mean annual input of sediment to Lake Kariba
lies in the range 7 to 70 x 10 t, assuming that tributary reservoirs
in Zimbabwe trap only a small proportion of the total yield. Ward
(1980) suggests that the dry density of submerged sediment deposits
in reservoirs in Zimbabwe probably lies in the range 0.64-1.28 t m~ .
In the case of Lake Kariba the long time-scale involved suggests
that most of the sediment will reach a high degree of consolidation.
For this reason a dry density of at least 1.0 t m
is likely to
occur. Thus, the estimated annual rate of loss of storage capacity
in Lake Kariba lies between 7 and 70 x 10 m .
Effects
of sediment
deposition
The "dead" storage capacity of Lake Kariba is reported to be 116 x
9
3
10 m , which represents over 60% of the total reservoir capacity.
In view of the concave shape of the reservoir's longitudinal profile
and the relatively small drawdown (9 m maximum) it is anticipated
that the bulk of the incoming sediment will reach the "dead" storage.
At the present rate of input this storage would be filled in 160016 000 years. It is, therefore, reasonable to conclude that the
effect of sediment on the operation of the project can safely be
ignored.
On the other hand, siltation will probably occur in some localized
areas principally at tributary inlets. Although those deposits may be
small relative to the capacity of the reservoir, their effect on
fisheries and navigation may be appreciable. For this reason it maybe
necessary to undertake periodic surveys of areas which are at risk.
LAKE CABORA BASSA
Estimated
rate
of sediment
deposition
No quantitative estimate of the rate of sediment accumulation in Lake
Cabora Bassa was published in the numerous reports which formed the
basis for the design of the project. Furthermore, none of the
engineers or officials, contacted by the writer, who were concerned
with the project's construction or its present operation considered
siltation to be a significant problem. Siltation is widely believed
to be occurring at the same rate as at Kariba.
As with Lake Kariba, a large part of the drainage basin of Lake
Cabora Bassa (which totals 1 x 10 km ) can be neglected as regards
the input of sediment to the reservoir. The Kariba Dam regulates
approximately 65% of the basin, trapping all its incoming sediment.
A further 15% of the area comprises the Kafue basin where sediment
is trapped not only by two storage reservoirs but also in the flood
plain of the Kafue Flats. The parts of the drainage basin which
are likely to contribute the bulk of the sediment input are as
follows: the basins of tributaries in Zimbabwe (42 x 103 km 2 ), the
basins of minor tributaries in Zambia and Mozambique (35 x 10 3 km 2 );
and the Luangwa basin (148 x 103 km 2 ).
Ward and Chikwanha studied one of the tributaries in Zimbabwe
which flows into Lake Cabora Bassa, the Hunyani. However, their
study was in the head reaches of the river at a point where the
Sediment
deposition
in Zambezi reservoirs
563
discharges are partially regulated by a dam. The work of Stocking &
Elwell (1973) suggests that Cabora Bassa's tributaries in Zimbabwe
may have a higher mean yield than the ones flowing into Lake Kariba.
On the other hand, about 10% of their area is regulated by reservoirs,
including the Darwendale Reservoir. Offsetting sediment retention in
such reservoirs against higher yields, the total sediment discharge
from these tributaries into Lake Cabora Bassa probably lies in the
range 2 to 20 x 10s t year-1.
The Luangwa basin, in Zambia, is the principal source of sediment
to Lake Cabora Bassa. Qualitative study suggests that the yield of
this basin will be higher than that of the tributaries considered
above because of its physical and hydrological characteristics and
because erosion rates have been accelerated by changes in land use in
the present century. Although no systematic quantitative study of
sediment transport rates has been published, data from an intermittent
sampling programme undertaken by the National Council for Scientific
Research in Lusaka indicate that sediment concentrations rarely fall
below 1000 mg 1 _ 1 . The foregoing evidence provides little basis for
deriving an estimate of mean annual sediment yield for the basin.
Nevertheless, it seems unlikely that it will be less than 100 t km~
or greater than 1000 t km- . Using these values, the rate of
sediment input to Lake Cabora Bassa from the Luangwa lies in the
range 15 to 150 x 10 t year" .
As with the Luangwa, there is little direct information on the
sediment yield from basins of the minor tributaries in Zambia and
Mozambique, although one of the planning reports for the Cabora Bassa
—2
—1
Project suggested that it may be of the order of 200 t km year
Using a range of values slightly below that used for the Luangwa
basin would give a sediment input rate from these tributaries into
Lake Cabora Bassa of 3 to 30 x 10 t year- .
On the basis of the foregoing estimates, the total rate of
sediment input to Lake Cabora Bassa appears to lie in the range 20 to
200 x 10 t year- . Making the same assumptions about the consolidation
of deposits as for Kariba, the loss of storage capacity therefore
lies in the range 20 to 200 x 10 m year- .
Effects
of sediment
deposition
The sediment input rate to Lake Cabora Bassa appears to be a factor
of 3 greater than that to Lake Kariba, but its storage capacity is
considerably smaller. In the design of the Cabora Bassa Project the
minimum drawdown level of the reservoir was taken to be 295 m a.m.s.l.
9 3
The "dead" storage, below this level, is approximately 12.5 x 10 m .
This would be filled in a period of 60-600 years at the rates of
sediment input estimated above, if all sediment were to be deposited
in the "dead" storage. Mathematical model studies of the project
have shown, however, that because of the loss of efficiency in the
generating equipment as the hydraulic head falls, there is no net
benefit, in terms of energy production, in drawing down the reservoir
to this level. A minimum level of 305 m a.m.s.l. or above would be
preferable in this respect and would more than double the "dead"
storage capacity and hence its siltation life. At the same rate of
9 3
sediment input, the reservoir's total storage capacity (72.5 x 10 m )
would be filled in a period of 350-3500 years.
564
P.Bolton
In the time-scales of economic planning it would appear, at first
sight, that, although the loss of storage capacity is occurring much
more rapidly than in Lake Kariba, sediment deposition in Lake Cabora
Bassa can also be neglected. However, study of the shape of Lake
Cabora Bassa (Fig.2), suggests that a significant proportion of the
incoming sediment will accumulate within the "live" storage of the
Distance from dam (km)
250
FIG.2
level)
200
150
100
Lake Cabora Bassa: outline
and longitudinal
profile.
50
0
(at normal
maximum
reservoir and thereby affect its operation much sooner. The
longitudinal profile of the reservoir is convex and the bulk of the
incoming sediment passes into a broad shallow basin, the Mucangadze
Basin, which lies wholly above 295 m a.m.s.l. and which is separated
from the deeper parts of the reservoir by a narrow reach constricted
by islands. These characteristics will tend to prevent appreciable
quantities of sediment from being transported into the "dead"
storage until a later stage in the reservoir's development.
A detailed study of the hydrology of the project (Bolton, 1983a),
has shown that, with the present installed hydroelectric capacity,
the principal constraint in its operation is not the generation of
power but the regulation of flood discharges. The maximum discharge
capacity of the dam's spill gates is less than the mean 3 month
inflow of the 10 000 year design flood. The operating procedures
for the project must, in consequence, follow a flood rule curve
which provides an annual drawdown in advance of the wet season to
provide sufficient flood storage to prevent the dam from being
overtopped. The appropriate value to use for the lowest drawdown
Sediment
deposition
in Zambezi reservoirs
565
level of the flood rule curve is a matter of some dispute; values
between 315 and 325 m a.m.s.l., on 1 February each year, have been
proposed in consultants' reports but rigorous application of British
design recommendations (Institution of Civil Engineers, 1978) would
require even greater values of drawdown.
To illustrate the possible effect of sediment on the project, it
will be assumed that an annual drawdown level of 316 m a.m.s.l. is
adopted and that approximately one half of the incoming sediment is
deposited in the reservoir zone between 316 and 326 m a.m.s.l. Thus,
at the end of 50 years the loss of flood storage will be in the range
0.5 to 5 x 10 m . To compensate for this loss of storage capacity
the annual drawdown would have to be increased. Calculated on the
basis of the pre-impoundment reservoir capacity curve the extra drawdown required would be in the range 0.3-2.6 m. However, values
calculated in this way would in practice need to be increased both
because the incremental capacity provided by the extra drawdown
would itself be affected by sediment deposition and because, if the
drawdown were to be increased, it would be necessary to further
increase the flood storage to compensate for the reduced discharge
capacity of the spill gates at the reduced reservoir levels of the
early flood season. Taking account of these effects, the necessary
increase in drawdown after 50 years, to provide the same degree of
flood protection for the dam, would be in the range 0.5-4.0 m.
It should be stressed that the results of the above calculations
are dependent both on the form of the flood rule curve finally
adopted by the project's operators and on assumptions about the
distribution of sediment in the reservoir. Nevertheless, at the
higher rates of sediment input which may occur within the range
estimated above, sediment deposition is likely to have an appreciable
effect on the project's operation within its "economic life". In
particular the additional drawdown needed to compensate for the loss
of flood storage will, in all probability, restrict the peak power
output of the project during the flood season and may also jeopardize
its firm power production. The effect of sediment accumulation will
be greater if, as proposed, additional generating capacity is
installed in a North Bank Power Station; the increased drawdown will
then also have a significant effect on the maximum guaranteed power
output since the reservoir level will be lower at the start of
critical dry periods.
In the original plans for the Cabora Bassa Project considerable
stress was placed on its potential multipurpose benefits. In the
implementation of the plans the emphasis was changed to that of
optimizing energy production to provide the revenue necessary to
finance the undertaking; although such benefits as navigation,
fisheries, agricultural development and flood mitigation were regarded
as potentially important.
Progressive siltation directly affects fisheries and navigation on
the reservoir and indirectly affects other multipurpose benefits by
reducing the flexibility of operation of the project. In particular
the degree of attenuation of downstream flood discharges which can
be achieved is strongly influenced by the available storage capacity.
Since this capacity will decrease with time as sédiment accumulates
the risk of inundation to development projects in the lower Zambezi
floodplain will gradually increase. This factor must be considered
566
P.Bolton
in planning the long-term development of the region. Likewise the
progressive lowering of the annual drawdown level of the reservoir
will affect the planning of development projects on its shores.
APPROPRIATE METHODS FOR SEDIMENT MONITORING
The shortcomings of the results presented in this paper as a basis
for official policy decisions are readily apparent. The purpose of
the study has been to identify those effects of sediment deposition
which are likely to be most significant in planning the medium and
long-term development of the region. The actual extent of the
effects can only be determined by undertaking appropriate field
surveys.
In the case of Lake Cabora Bassa two aspects of sediment deposition
require further examination: firstly, to determine more precisely the
rate of sediment inflow; and, secondly, to identify the principal
regions of deposition in the reservoir. In smaller reservoirs it is
possible to evaluate both these factors by undertaking periodic
hydrographie surveys along established survey lines. This approach
would not be applicable in a reservoir the size of Cabora Bassa
because of the large number of sections which would have to be
surveyed for a given level of accuracy and the difficulty of achieving
accurate horizontal fixes on an open surface which frequently exceeds
20 km in width.
In the absence of hydrographie data, the only reliable method of
estimating the total rate of sediment inflow is by monitoring the
discharge of the principal sediment carrying tributaries. For Lake
Cabora Bassa the bulk of the inflow could be measured by establishing
one permanent sediment monitoring station at a suitable location on
the Luangwa River. Hydraulics Research, Wallingford, has evolved a
method of monitoring sediment discharges by exploiting the physical
differences between the transport of "wash load" and "suspended bed
material load" (Bolton, 1983b). Wash load must be monitored
continuously since it is supply dependent; this may be achieved
through the installation of an adequately calibrated silt sensor
(Fish, 1983). By contrast, the discharge of suspended bed material
load and bed load can be correlated with river stage; this may be
achieved by an intensive programme of pumped sampling over a single
period of, say, three months. Such a monitoring procedure provides
reliable estimates of the long-term sediment discharge without
requiring a heavy long-term commitment of resources and staff.
However, in the case of the Luangwa River, its success would depend
on achieving close cooperation between the authorities in Zambia and
Mozambique. At a later stage other tributaries of Lake Cabora Bassa
could be monitored if it were found to be necessary.
The problem of accurately determining the distribution of sediment
in Lake Cabora Bassa remains unresolved although hydrographie survey
of a small number of sections in the Mucangadze basin would provide
some indication of the rate at which the "live" storage was being
filled. A more accurate value could be obtained from study of a
sequence of aerial photographs of this basin taken at, say, 5-year
intervals when the reservoir level is at a minimum. The disadvantages
of this method are its cost and the difficulty of providing accurate
Sediment
deposition
in Zambezi
reservoirs
567
ground control in such an inaccessible region. Alternatively it may
be possible to gain some indication of sediment distribution by
studying the change in shape of the shore line at known drawdown
levels on satellite images. However, with the present level of
resolution and the difficulty of obtaining cloud-free images it is
unlikely that this method will provide the required information.
As with many projects in which it is necessary to quantify the
effects of sediment deposition, the final choice of monitoring
procedure should be based on an assessment of the purpose for which
the data are required. At present such an assessment can only be
made subjectively. To develop a more objective procedure would
require both the development of appropriate methods of economic
analysis and a knowledge of the level of accuracy which can be
achieved in different types of sediment survey. Further study of the
latter question is currently being undertaken at Hydraulics Research.
ACKNOWLEDGEMENTS
The initial research on which this paper is based
was undertaken at the University of Edinburgh. I am indebted to my
supervisor, Mr H.Dickinson, and to the authorities in Mozambique for
their encouragement and assistance. The preparation of the paper
and the development of the final section were carried out at
Hydraulics Research, Wallingford in the Overseas Development Unit
headed by Dr K.Sanmuganathan with funds provided by the Overseas
Development Administration of the British Government.
REFERENCES
Bolton, P. (1983a) The regulation of the Zambezi in Mozambique: a
study of the origins and impact of the Cabora Bassa Project. PhD
Thesis, Univ. of Edinburgh.
Bolton, P. (1983b) Sediment discharge measurement and calculation.
Tech. Note 0D/TN2, Hydraulics
Research,
Wallingford,
Oxfordshire,
UK.
Central African Council (1951) Report on Kariba Gorge and Kafue
River
Hydro-electric
Projects.
Inter-Territorial Hydro-Electric Power
Comm., Salisbury.
Chikwanha, R. (1980) Sediment Yield from Rhodesian Rivers:
1978-1979
Season.
Hydrological Branch, Min. Water Dev., Salisbury.
Fish, I.L. (1983) Partech turbidity monitors. Tech. Note OD/TN1,
Hydraulics
Research,
Wallingford,
Oxfordshire,
UK.
Institution of Civil Engineers (1978) Floods and Reservoir
Safety.
ICE, London.
Stocking, M. & Elwell, H.A. (1973) Soil erosion hazard in Rhodesia.
Rhodesian Agric.
J. 70 (4), 93-101.
Ward, P.R.B. (1980) Sediment transport and a reservoir siltation
formula for Zimbabwe-Rhodesia. Civ. Engr in S.A. 22 (1), 9-15.