Water Resources and Irrigation in Africa
Calibration of the water balance model
Calibration objective and data used for calibration
The objective of the calibration is to obtain annual runoff estimates from the model as close as possible to
measured runoff. The runoff values calibrated correspond to the natural flow without irrigation; the effect of this on the river flows is usually negligible for Africa with the exception of the Nile Basin. Calibration was performed independently for each drainage basin with all available runoff data. The figures used for the calibration originate from the FAO study on irrigation potential in Africa (FAO, 1997) and from UNESCO’s publication on African rivers (UNESCO, 1995). These data are also published on-line as part of the Global River Discharge Database.
It should be noted that the figures from FAO (1997) come from various sources which are not always comparable amongst each other. The figures from the UNESCO report are mean values over different years: the periods of measurement differ from one station to another and vary from 100 years to 2 years only. These data should therefore also be considered as very indicative figures.
Calibration was made by adjusting the following parameters:
- soil moisture;
- reference evapotranspiration (ETo);
- open water evaporation.
These parameters were multiplied by correction factors. The corrected parameters are included in the data file belonging to the data layer with the drainage basins. The soil moisture related data as derived from the Digital Soil Map of the World are originally presented in classes and are considered to be only indicative values. Therefore, higher deviations from the original value were allowed for this input layer than for other input layers to which correction factors were applied. Reference evapotranspiration (ETo) calculated according to the FAO Penman-Monteith method with limited climatic data result in values with an accuracy range of about 20 percent from the figures as calculated with a complete climatic dataset. With the applied correction factors, an attempt has been made to maintain this range as much as possible. Precipitation is considered the most reliable input datalayer. Therefore only in exceptional cases have correction factors been applied. The correction factor for open water is applied to the ETo data layer on areas which are indicated to be open water or wetlands on the Digital Soil Map of the World.
During the calibration it appeared that the data on the ETo data layer were systematically too high for open water in wetlands. There are several reasons for the overestimation of open water evaporation. First of all the data layer contains data on ETo of vegetated areas instead of open water evaporation. Secondly, in the model, open water evaporation is also applied to wetlands, on the basis that a lot of runoff water is lost in swamps. The soil water balance does not account for these losses. In reality, however, the evaporation of swamps can be described better as open water evaporation in the centre of the swamp and a soil water balance on the edges. In addition, it should be noted that in the model the open water balance is applied over the whole area indicated as wetland on the Digital Soil Map of World. In reality, the area of the swamps can vary considerably with the seasons. Assuming open water evaporation over the whole area mapped as wetland therefore results in an overestimation of wetland evaporation.
The statistical characteristics of calibration factors for soil moisture, evapotranspiration, precipitation and open water for the whole continent are shown below:
| Standard deviation
A detailed analysis of the calibration factors shows that in general higher correction factors for soil moiture and evapotranspiration have been applied to humid areas (i.e. more water storage in the soil and higher evapotranspiration) and lower correction factors have been applied to the dry areas (thus reducing the evapotranspiration and soil storage resulting in more runoff).
The high correction factors that have been applied to humid areas could be explained by the fact that no vegetation factors have been applied to the ETo. The humid areas are generally covered with dense vegetation with a high transpiration capacity. Interception of precipitation by the canopy may also help increase evaporation.
In dry areas low correction factors have been applied. The main reason for these low correction factors is probably the use of monthly time steps in the model. In dry areas, precipitation is often intensive during a short period of time. In such periods runoff takes place which cannot be modelled using mean monthly values, because evapotranspiration exceeds precipitation on a monthly basis. Furthermore, the runoff during these intensive storms may occur without saturating the soils. The soil water balance model as applied in this project does not describe this natural situation accurately, resulting in a systematic underestimation of the runoff in dry areas. In order to obtain reasonable runoff values, a correction factor of 1.1 for precipitation has been applied for some of these dry areas.
Results of the model calibration - comparison of model results with major river discharge
The differences between calculated and observed runoff at the selected hydrologic stations are presented in Table 4 as absolute and relative figures. For most rivers the absolute differences are small and below 20 percent. The total runoff (avoiding double counting for different stations in the same river) related to the selected stations in Table 4 correspond to more than 50 percent of the total runoff of Africa - generally estimated between 3 500 and 4 000 km3/yr. The final results therefore show a good match between model output and measured runoff.
Results of the model calibration - comparison between computed internal renewable water resources and country statistics
FAO (1995a) and the values computed by the model. The statistics of water resources have been established by FAO as part of the AQUASTAT-programme, in which water resources data have been estimated on the basis of literature review and country statistics.
The calculated values shown in Table 5 are generally lower than the country statistics values. This is especially apparent in the more arid countries. In these countries the model calculates hardly any runoff while the country statistics still show some renewable water resources.
In the water balance model used for this study, the internally generated water resources (Internal Renewable Water Resources) are computed by subtracting the total inflow to the
country from the total flow accumulation leaving the country, disregarding water leaving the system by evaporation from large lakes and wetlands. In arid areas, this method leads to underestimates when compared to the results of conventional studies consisting of estimating the water potential through the recharge of groundwater and the river discharge at the points where the runoff is maximum. However, it should be noted that the production of internal renewable water resources in semi-arid and arid countries is very unreliable, since it varies considerably from year to year. It is therefore difficult to judge the reliability of the model results in the semi-arid and arid countries.
Your access to AQUASTAT and use of any of its information or data is subject to the terms and conditions laid down in the User Agreement.