While the damming of the Niger produced conditions in the basin very much more lacustrine than the Niger itself, Kainji Lake retains many characteristics that are riverine. These especially include the flushing of the reservoir basin at least annually, a pronounced flood cycle, and enhanced vertical and horizontal transport of water into the deeper portions of the lake. These factors reduce storage processes in the lake and speed up responses of the lake itself to changes in larger environment. It is therefore, tempting to assume that the description of the lake presented here is that of a reservoir nearly reaching equilibrium, at least physically and chemically, with the present climatic and geographic features of the lake and its drainage basin.
The lake is readily divided into three major sections for the purposes of study and discussion (Fig. 1). From the dam northward, the basin is narrow and deep for a distance of about 40 km, with a substantial reduction in depth beginning at Garafini (15 km from the dam) where there had been rapids. At the site of the old Bussa Rapids, the lake widens into a large shallow basin formed by branching of the old river. This section of the lake comprises nearly 70 percent of the total area, and is the most important from the point of view of the fishery. The surrounding topography is low, in contrast to the lower arm, and the lake shore falls very gently to the depth of the former river.
At the northern tip of Foge Island the lake narrows again and becomes more and more riverine toward Yelwa. This section was given low priority in the limnological studies owing to its riverine character, distance from the base of operations and recent diminishing relative significance in the fishery.
4.1.1 Shore Development
The perimeter of Kainji Lake, including all three sections, is estimated to be about 720 km. With a surface area of 1 280 km2, the shore development factor at high water is 5.65. This is the ratio of the actual shoreline to the shoreline of a perfectly circular lake of the sane area and hence is a measure of the amount of shoreline extension produced by bays and other indentations of the lake margin. While most natural lakes approach the idealized circular form much more closely than Kainji (values around 2.0), the values for others of the African reservoirs are much higher. In an approximate sense, the significance of the shallow littoral zone of a lake in relation to the open pelagic zone increases with the shore development factor. The lower southern arm of the lake, because of its shallow zone, contributes disproportionately to this index in Kainji.
4.1.2 Depth
The maximum depth of Kainji Lake occurs at the dam itself and is about 50.4 m at high water level. The lake is drawn down annually 9 to 10 metres. The total volume at high water is about 1.5.8 km3 and at low water about 5.8 km3. The corresponding areas are 1 280 km2 and 660 km2. The mean depths are, therefore, 12.3 m and 8.8 m respectively.
The ratio of mean depth to maximum depth is often used as an index of basin shape (Hutchinson, 1957). Calculated on the data above, a value of about 0.23 is obtained. Most lakes have values above 0.33, and in this case the low value is again accentuated by inclusion of the lower arm.
Unfortunately the reservoir basin was mapped by aerial photography prior to flooding without thorough ground survey. Contouring could only be accomplished at intervals of 40 ft. There remains, therefore, a considerable uncertainty in the above calculations.
The flooded area in the central basin, particularly over Foge Island, was quite flat. Along the western margin of the lake and throughout the lower arm, the flooded area was heavily dissected by runoff and the bottom topography is, therefore, very complex. Detailed hydrographic mapping now that the lake is formed is likely to be quite difficult, requiring careful positioning and extensive coverage.
4.1.3 Storage
A storage graph was prepared by the Joint Consultants based, presumably, on the aerial survey. This graph, converted to metric units, is reproduced in Figure 11. Volume differences for one metre depth intervals were taken from this graph and smoothed to prepare a chart of area as a function of reservoir height (Table 2). The latter has been used in obtaining weighted means of temperature, primary production and other depth-dependent variables when totals for the whole lake were desired.
4.1.4 Islands
There are three major islands in the lake, one at Garafini in the middle of the lower arm, another at Bussa, and the most significant at the point where the upper arm joins the central lake. The latter island is much extended toward the south at low water and appears to exert a substantial influence on water circulation in the main lake (see Section 4.3.4).
4.1.5 Wind Patch
Short steep waves develop quickly on most parts of the lake in response to increasing wind. The north-south directions of the prevailing winds tends to coincide with the longer axis of all of the major parts of the lake, while the middle basin and certain bays in the south arm also permit long fetches to the easterly squall winds. A maximum fetch of more than 30 km is available in the middle basin to south-southeast winds.
Possible maximum wave heights were computed by the Joint Consultants to be about 1 metre. While direct observations have not been made, experience on the lake suggests that waves of 50 to 75 cm are common and waves up to 1 m indeed occur during storms. Corresponding wavelengths are from 6 to 14 metres. A maximum setup of about 0.2 m was also predicted but has also not been directly measured.
On days of moderate windiness, striking differences are from place to place in the lower arm. One passes from windy sections to wind shadows at each major bend. On some occasions blooming algae and floating debris form large patches in the shadowed areas.
4.1.6 Drowned Trees
A significant factor in the morphology of the lake is the extensive area where trees are partially or wholly submerged. While 40 percent of the lake area was cleared of trees prior to impoundment, large areas remain. These trees reduce the effect of water and wind action on the morphology of the lake.
4.1.7 Depth of Discharge
The spillway of the reservoir lies about 15 m below the normal high water level. The intakes to the turbines have their centres at a depth of about 28 m below the high water level and are about 8 m in diameter. Deep discharge is a major factor in the stratification regime of the lake and an important difference between Kainji and natural lakes.
Extensive hydrological studies were undertaken in preparation for construction of the dam, and are continuing in connexion with its operation. These data, provided by the Niger Dam Authority, form the basis for the following descriptions.
4.2.1 Inflow
The Niger is unusual in having two flood regimes well separated in time. The first flood, beginning in August, is derived from rains in the region south of Naimey, and carries substantial quantities of colloidal clay, chiefly kaolinitic. This water is highly turbid (Secchi disc transparency about 0.3 m) and greyish, giving the name “white flood” to this water* The rise to a highly variable peak is rapid and the flood starts to diminish in late September or early October (Fig. 5). The second flood, originating in rains of the upper Niger, is considerably delayed and spread in intensity by its passage northeastward to the interior delta near Timbuktu and thence southeastward to Nigeria. This water is relatively clear, though carrying appreciable amounts of detrital material, and is given the name “black flood”. Although rising soon after the white flood, it does not reach its peak until February and does not cease until the latter part of April* The annual flow may reach about 80 000 million m3 in a year of high flow, of which 47 000 million is obtained from the white flood and the balance, 33 000 million, in the black flood. The latter is relatively constant from year to year while the white flood may be reduced by as much as one third in dry years.
4.2.2 Discharge
While there are several lateral tributaries to the lake, these are all small (less than a few percent) relative to the Niger itself. Only the Malendo (peak discharge 500 cumecs) is included in water balance calculations.
The reservoir is operated both for power generation and flood control. This necessitates careful control of the lake level and a post-flood drawdown of almost 10 metres. Uncertainty in predicting flood levels has resulted in a substantial variability in discharge in adjusting the filling rate to design requirements during September of the last two years.
At present most of the discharge is taken from the spillway as only 400–500 cumecs are used in power generation. As capacity is increased, however, turbine discharge may be expected to triple for most of the year.
4.2.3 Evaporation
Evaporation has been estimated at about 150 to 200 cm per year or about double the local rainfall. The evaporation is a small item in the annual water balance and difficult to estimate directly.
4.2.4 Horizontal Transport within the Lake Basin
Very little information has been obtained on currents within the lake owing to the difficulties of direct measurement. During the rise and fall of the white flood, the distribution of turbidity provides some information on water movements (Henderson, 1971). Generally, the incoming turbidity spreads rather uniformly down lake, with rather more rapid dispersal along the eastern half of the central basin. Southerly winds appear to push surface water toward the northwestern bay resulting in clockwise surface currents around the northern tip of Foge Island.
The distribution of heat in a lake is one of the most significant factors in determining the dynamic properties of the lake, as the distribution and intensity of most of the internal conditions are directly affected either by temperature or gradients in temperature.
4.3.1 Mean Temperatures
The mean temperature of the water column has been calculated for station 1, three mi above the dam, for each cruise (roughly 6-week intervals). The temperature for each 1-metre stratum was multiplied by the volume of the stratum to obtain a mean representative of the total volume of water in the lake. This calculated mean temperature is shown in Fig. 12.
Peak temperatures are reached in April and May at about 29.5°C (1970) and minima in January–February (26.2°C, 1969; 23.6, 1970). A minor peak in temperature occurs in October while the humidity is high and cloud cover has diminished.
4.3.2 Heat Budget
A heat budget analysis was made for Kainji for the period August 1969 through December 1970 in order to seek information on the sources of variability of the mean temperature from season to season and year to year. The much lower temperatures during December and January of 1971 compared to the previous season raised questions about conditions leading to thermal variability in Kainji Lake.
While the available data are of marginal significance for heat budget analysis, annual evaporation from the lake surface, estimated as a difference in the total budget, agreed rather well with other estimates made by the Joint Consultants (1961). The results of the analysis are, therefore, taken as roughly indicative of the distribution of the components of heat flow in spite of the inadequacies of the data. Conventionally, contributions to the heat budget are divided into incoming short-wave radiation less reflection, net longwave (thermal) radiation, evaporation loss, convective exchange between the water surface and the air above, advective gain or loss through inflow and outflow, and heat storage (change in lake temperature). Their sum, sign considered, is assumed to be zero. All were estimated directly except for evaporative and convective losses. The ratio of the latter (Bowen's ratio) can be estimated more satisfactorily than their absolute magnitudes, hence these are determined by difference after estimates of the other terms have been obtained.
Heat storage was estimated from the temperature records (average column temperature weighted by the volume-depth relation) at Station 1, in the lower arm. As the actual measurements did not coincide with calendar months, the observations were converted to monthly intervals by interpolation. In view of the small differences among stations and uncertainties in the meteorological observations, it did not seem worthwhile to average the storage estimates over all stations.
The radiation terms were calculated by formulae and tables given in Budyko (1956), using 10 as a reference latitude for clear-sky radiation. Cloudiness was estimated as the complement of sunshine duration (Campbell-Stokes recorder at Yelwa and Mokwa) expressed as percentage of possible sunshine. Air temperature and humidity data were obtained from records of the Niger Dam Authority Station about 1 km below the spillway (Faku). Air temperature was taken as the mean of the daily maximum and minimum, while the humidity was that of 10.00 h daily observations. Monthly means were computed from these daily observations.
Inflow and outflow were obtained from the hydrological records of the Niger Dam Authority, while their temperatures were estimated from observations of river temperature near Yelwa (inflow) and from the temperature profile at Station 1 at depths corresponding to spillway and turbine discharge (outflow).
The Bowen ratio was also calculated from the Faku records and hence must be regarded as the least satisfactory of all the estimates. The wide spacing of the lake cruises precluded use of the meteorological data obtained on the lake except to indicate that the daily air and lake temperatures followed each other closely.
The results of the analysis suggest that the monthly evaporative heat loss from the lake is almost equal to the net incoming radiation or “radiation balance” (Table 2). The latter varies over the seasons primarily in response to changing cloudiness, while evaporation in turn responds to the temperature of the lake surface and humidity. Small differences between evaporative loss and the incoming radiation are significant in determining heat storage in the lake. The overall effect of both inflow and outflow is to raise the temperature of the lake. While the effect of inflow is dominant, discharge of deep, rather than surface water through the spillway and the turbines add heat to the lake during periods of stratification. The lower temperatures of the lake in late 1970 and 1971 seem to have resulted primarily from reduced inflow during the low flood of 1970 and somewhat increased evaporation of December 1970.
While it remains difficult to separate these effects, it seems likely that the lake temperatures will continue to vary substantially from year to year, more or less as the local rain (and flood) varies.
4.3.3 Thermal Stratification
Kainji Lake stratifies each year in February (Fig. 14), and a pronounced thermocline is established and maintained to the middle of May. Drainage of the cool water below the thermocline (hypolimnion) rapidly lowers the level of the thermocline thus eventually destroying the stratification. Until November the lake is filled with warm water which shows small decreases in temperature with depth but with no evidence of stability. That vertical mixing is reduced during the period is particularly evident in October and November when there is deoxygenated water (less than 1 ppm) below 30 metres. In November and December rapid cooling occurs, apparently with cool flood water and strong differences in air temperature and radiation between day and night. Cool clear nights induce rapid cooling of the surface water which then sinks established strong vertical mixing. At this time there is complete circulation of the water mass.
While only data from Station 7 are shown, the vertical temperature distribution is very similar at all stations. Further details can be found in Henderson (1971)
During the period of local runoff, bays fed by the small streams flowing laterally into the lake show occasional stratification as reported by the fisheries officers. These conditions are presumably established by difference in temperatures of the lake itself and the inflow and appear to be localized and temporary.
The transparency of the water of Kainji Lake is quite variable, primarily as a result of the colloidal turbidity brought to the lake by the white flood. The depth of disappearance of a standard white disc (the Secchi disc) is as great as 3 m in May and June and as little as 0.1 m in September at the rise of the white flood (Fig. 15).
4.4.1 Spectral Distribution of Transmitted Light
On two occasions, the transmission of natural light was measured with a photometer equipped with red, green, and blue filters. With relatively clear water (Secchi disc 2.2 m, 3 February 1970) the peak of transmission occurs in the green region of the spectrum, the more or less normal condition of natural freshwaters. During a period of turbid water (Secchi disc 0.5 m, 9 September 1969) transmission of blue and green light is greatly attenuated and the peak lies in the red region of the spectrum. Swimmers report that the underwater lighting is reddish during such periods.
A similar effect seems to occur in the illumination entering the lake during period of “harmattan” dust as observed by comparing the spectrum of low angle (30°) solar illumination with high angle (850) solar illumination during the “harmattan” period and under clear skies. While very rough, these results show a substantial reduction of blue and of green light by the dust.
4.4.2 The Nature of the Colloidal Particles of the White Flood
As the colloidal material might contain appreciable amounts of ferric iron, and also because of the question of the rate of clearing turbidity from the lake, an effort was made to establish the nature of the colloidal particles. It was demonstrated that the particles carry negative charge bound to the surface by placing turbid lake water in an electric field and noting accumulation and deposition of the particles of the positive electrode. Ferric colloidal particles are typically positively charged while clays are negatively charged. While chemical tests for iron were also negative, small amounts of iron are normally present in the clays washed from the soils of the area and are presumed present in the turbidity of the lake water in, however, very small amounts (clays about 1 mg/l, iron less than 0.001 mg/l). As the dominant clay in the local soils in kaolinite (Joint Consultants Report, Vol. VII), it is assumed that this is the major constituent of the colloidal particles of Kainji.
Strong acidification precipitates the colloidal particles by removing the charge stabilization. The range in the lake does not exceed 6.2 to 8.5 and is ineffective in precipitation. It should, however, be determined whether or not appreciable decomposition of the clay particles occurs under the strong reducing conditions during stratification.
The settling time of these particles is very low under laboratory conditions, less than 50 percent per month. Under the turbulent conditions of the lake it must be negligible. Nevertheless the lake loses the turbidity rapidly as evidenced by the rapid increase in transparency at the end of the white flood. It seems likely that the clearer black flood water displaces the white flood water prior to the onset of stratification in late February, and calculations suggest that the normal black flood volume is just sufficient to do so before stratification sets in in mid-February.
Isotopic analysis of the waters of the Kainji basin and the Niger (cooperative programme with the University of Ifs, the International Atomic Energy Commission and FAO) confirm that the concentration of colloidal material (as measured by colorimetric densitometer using blue light) is directly related to the proportions of black and white flood water. The water of the black flood has been subjected to heavy evaporation and differential removal of the lighter isotopes of hydrogen and oxygen in the water molecules. Unfortunately it has not been possible to determine the concentration of the colloidal material in situ without a wide-range transmissometer.
4.4.3 Other Sources of Turbidity
The black flood appears to carry appreciable amounts of finely divided organic detritus as evidenced by accumulation on netting and anchor ropes but concentrations are low and no attempt has been made to make quantitative estimates.
The higher turbidity of the mid-lake stations April–July would appear to result both from the suspended organic materials and higher phytoplankton concentrations and later from increased stirring of the bottom sediments as the lake becomes shallow and with frequent severe wind storms. Much of the old Foge Island region lies near the surface at depths less than 10 m in June and July.
The water of Kainji Lake is low in dissolved materials and only slightly modified from the composition of the flood water that feeds the lake. Analyses of individual components have been made by several investigators (Imevbore, 1970; White et al., 1965; NDA, data on file) and the present study has concentrated on seasonal variation in gross characteristics of the chemical composition.
4.5.1 Dissolved Oxygen
The concentration of dissolved oxygen in the surface waters (1 m in depth) at Kainji does not fall below about 70 percent of saturation. Even during the initial filling when decomposition was high, oxygen never fell below 20 percent (imevbore. unpublished manuscript). The lowest values occur in September just as the lake level begins to rise.
During much of the year the concentrations of oxygen in deep water remain above 4 mg/l. There are, however, two periods when oxygen concentrations reach values less than 1 mg/l in deep water (Pig. 14). Shortly after stratification, in early March oxygen concentrations below the thermocline drop to near aero in the entire hypolimnion. This event is clearly marked each year at the dam site by the appearance of a strong odour of sulphide owing to the escape of hydrogen sulphide by aeration at the spillway. Sulphide concentrations may reach 0.6 mg/l in the deepest water, but are generally much lower.
There is a less marked period of deoxygenation during October and November more or less coincident with a decrease in the surface concentration. This is assumed to result from increased biological activity associated with the flood and incomplete vertical mixing.
A similar effect seems to occur in the mouths of the streams tributary to the lake where local stratification occurs during flood. These floods are variable and rise much more quickly after the rains than the flood of the Niger.
4.5.2 Anion Concentration
The dominant anion in Kainji water is bicarbonate (Fig. 16). The total alkalinity (methyl orange) is quite constant at about 0.55 me/l or 28 mg/l as calcium carbonate equivalents Other anions (sulphate, nitrate and chloride) average about 0.05 or less me/l. A sample of water from the lake (October 1969) was titrated with .02 N acid and base after aeration with CO2 free air. The results are shown in Pig. 17. Water saturated with CO2 was also used and an approximately comparable curve is shown for the weak acid titration. Chloride concentrations are of the order of .01 me/l and sulphate even less. Total conductivity ranges between 45 and 60 micromhos/cm at 25° C and differs little from the river water before impoundment. The relation between conductivity and total anion concentration reported by White et al. appears to remain valid for the lake, i.e., conductivity (20°C in micromhos) equals 100 times total ionic concentration (me/l) or 2 times total alkalinity.
4.5.3 Cation Concentration
Calcium and magnesium are the dominant cations (roughly 0.2 me/l each) while sodium and potassium are an order of magnitude lower (.01 and .04 me/l respectively). Sodium and potassium appear to have decreased considerably from concentrations measured in the River Niger before impoundment while magnesium has increased (Fig. 16).
4.5.4 Total Dissolved Solids
Total dissolved solids have not been measured for Kainji waters. Estimates based on the above data suggest values between 35 and 40 mg/l. This factor is of interest in computing the morphoedaphic index of Ryder (1965) which has been used to estimate potential fish productivity of various lakes (Section 6.1.2).
4.5.5 Other Constituents
Little reliable information is yet available on the concentrations of biologically active components of the chemical system. Nitrates and phosphates are reputed to be low but quite variable while silica is very high. Iron is also variable reaching detectable levels (greater than .01 mg/l) in the lake only during stratification when ferrous iron remains in solution in the hypolimnion. Iron and manganese form dark reddish deposits on the rocks in the river below the dam where aeration of the water is high.
4.5.6 Inflow of Nutrients
The data of Imevbore (1970) suggest that the lateral tributaries to the lake may be nearly as significant as the Niger in contributing nutrients to Kainji Lake. While their combined flow is small compared to the river the concentrations of nutrients are considerably higher. Rapid deoxygenation of the mouths of these tributaries has already been mentioned and blooms of algae have frequently been noted in these bays. Nevertheless the observed variations in conductivity and total ionic concentration over the year are quite small suggesting either that the contribution of these tributaries is small or that such excess nutrients are rapidly taken up into biological storage.
4.5.7 Nutrients from Savannah Fires
It is well known that the burning of the savannah grasslands during the dry season releases bound nutrients to the air and the soil surface. Nitrates are lost to the atmosphere while phosphates remain in the ash. Much of this ash is carried short distances from the fire by convection currents and some ends up in the lake, either directly or in initial runoff at the start of the rains. There is no evidence at present from which to estimate the magnitude of this source of nutrients. In terms of long-range projections of the productivity of Kainji Lake work along this line should be encouraged.
Kainji Lake is rather different from the other large reservoirs of tropical Africa owing to the complete replacement of its contained volume each flood season and short period of stratification. A large annual drawdown (10 m), and other strong seasonal changes associated with the savannah climate, impart considerable variability to the lake. There is also evidence that variation among years, particularly in temperature, may be rather greater than in other lakes of the tropical region.
In contrast to the above, the chemical composition appears to be fairly uniform throughout the season and annual replacement should reduce the possibilities of gradual enrichment of the lake, while ensuring rapid response to changes in agricultural practice in the drainage basin. It is apparent that the enrichment of the water at initial flooding by decomposition of drowned vegetation was very short-lived in Kainji Lake. This was presumably also accentuated by the burning of the areas which were cleared prior to flooding.
