Chapter 12 Groundwater Modelling

CHAPTER 12 GROUNDWATER MODELLING

12.1 Groundwater Basin Management and Groundwater Modelling

Groundwater basin management aims at an effective utilization of groundwater that

improves the living condition of the people, controls drying-up, contamination and other
problems associated with the use of groundwater, and preserves safe environment
(Shibasaki et al., 1995). To achieve this aim, it is important to determine the qualitative
and quantitative goal of management. Numerical value of quantitative goal is
determined based on the concept of permissible yield.

12.1.1 Permissible Yield

1) Permissible Yield and Constraint Factors

The permissible yield can be defined as the “permissible amount of groundwater

withdrawal for the residents of the area determined to compare the benefits resulting from
the pumpage of groundwater and the risks that might arise from it” (Research Group for
Water Balance, 1973).

Factors that determine the permissible yield generally include the following (Shibasaki et
al., 1995):

(1) Recharge factor (natural scientific factor): Water balance is maintained.

(2) Economic factor: Cost of pumpage is below a certain level.

(3) Legal factor: No violation of water rights of water laws.

(4) Geo-environmental factor: Not causing dry-up of springs, land subsidence,

seawater intrusion, or groundwater contamination.

Geo-environmental factor can be included under economic factor. All the factors above
except recharge factor are considered as socio-scientific factors. Though each of these
factors is not completely established in the present, it is pragmatic to combine these
factors, and determine the permissible yield.

The relative importance of these factors differs depending on the changes in the natural
conditions of groundwater or social conditions within communities (Fig. 12.1). The
importance of determining factors for permissible yield depends on the circulating
velocity, or renewability of the groundwater (Shibasaki et al., 1995). For humid areas
with fast circulation and high renewability of the groundwater, the water balance is

12 - 1
 Chapter 12 Groundwater Modelling

important and the concept of sustainability is given a priority. On the contrary, for arid
regions with a slow circulation and low renewability of the groundwater, the possibility of
drying-up is increased, and the economic factor becomes more important.

2) Permissible Critical Groundwater Level

The permissible yield can refer to the groundwater level, which must be retained so that
these factors are not violated. In other words, it can be called the permissible critical
groundwater level. Groundwater level can be monitored easily by using observation
wells and other existing wells. Therefore, if the permissible critical groundwater level is
set up, and groundwater level is observed by using a monitoring system, the groundwater
pumpage of a basin can be controlled to maintain the permissible yield.

The groundwater modelling is an effective tool to determine the permissible critical

groundwater level or the permissible yield, showing an actual example in the following
sections.

12.1.2 Work Elements of Groundwater Basin Management

In managing a groundwater basin, the following basic components should be taken into
account: Monitoring system, Database, Prediction system, and Decision-making system
(Shibasaki et al., 1995) (Fig. 12-2).

1) Monitoring System

Various parameters including groundwater levels, pumpage, groundwater quality should

be monitored. Observation wells are used to measure groundwater levels, and their
screens must be suitably equipped so that the water level can be measured separately for
each aquifer.

2) Database System

A database system is necessary to effectively use and analyze large volume of data and
information collected during monitoring. A database system is particularly needed for
data on groundwater levels, groundwater withdrawals, and water quality. A database
system is also used for geological logs in order to comprehend the structure of a
groundwater basin.

3) Prediction System

Based on the findings of monitoring, the conditions of groundwater basin are predicted.
The prediction results are used in determining appropriate measures for groundwater

12 - 2
 Chapter 12 Groundwater Modelling

utilization. Groundwater modelling (groundwater simulation) is used for prediction.

Modelling (Simulation) has been used effectively in predicting the permissible limits of
groundwater withdrawal in areas of land subsidence or seawater intrusion in Japan.

It is not sufficient to plan a project by means of predicting once at the beginning of it.
Prediction has to be reviewed and repeated every few years, and its accuracy must be
enhanced every time.

4) Decision-Making System

Appropriate measures to be taken for groundwater management are determined on the

basis of monitoring and prediction. The criterion on which these measures are decided
depends upon the concept of permissible yield.

12 - 3
 Chapter 12 Groundwater Modelling

12.2 General Procedure of Groundwater Simulation

The groundwater simulation study is composed of the development of the aquifer model,
calibration, prediction and evaluation as shown in Fig. 12-3. An aquifer model is
developed with field data manipulated for computer input. After calibrating the model
with observed data, prediction of the water level is executed by inputting future
abstraction scenario. The predicted results are evaluated for optimum abstraction, taking
the water balance, environmental impacts and economics into consideration.

12.2.1 Aquifer Modeling

Aquifer modeling is carried out by numerical groundwater model. The aquifer model
simulates a three-dimensional groundwater flow in the phreatic or confined aquifer. The
aquifer model is developed by using mainly topography, aquifer distribution, hydraulic
characteristics, groundwater discharge and recharge, and groundwater quality.

12.2.2 Data Manipulation

Groundwater model study uses various kinds of input data such as aquifer characteristics,
water levels, abstraction, recharge, etc. These data are processed for model inputs.

1) Hydrogeological Data

Hydrogeological data consists of geological and topographical maps, geophysical survey

results, composite loggings of test boreholes and pumping tests. Based on these data,
contour maps on the surface of aquifer, aquitard and hydrogeological basement are
prepared. The contour maps including ground elevation are converted into computer
codes. Also, distributions of aquifer characteristics are prepared and converted into
computer codes.

2) Water Level

Water levels measured in field are modified to groundwater level elevation based on the
ground height of observed points. Then contour maps are converted into computer codes.

3) Groundwater Use and Recharge

Based on the results of the well inventory (Hydrocensus), present groundwater use is
estimated. Variation in groundwater use in the past is estimated from the statistical data.
Groundwater recharge is estimated by using recharge equations and rainfall data.

12 - 4
 Chapter 12 Groundwater Modelling

12.2.3 Calibration of Model

The model is test-operated by inputting recharge and pumpage data and outputting the
calculated groundwater level in each node. The output is compared with the observed
water level records, and then the model will be modified/adjusted until final agreement
between the calculated and the observed water level is achieved. Thus the model is
calibrated and ready to predict water level fluctuations under various abstraction
scenarios.

12.2.4 Model Prediction and Evaluation

On the basis of future groundwater abstraction plans, a tentative proposal for groundwater
withdrawal plan is prepared. Then the plan is inputted into the calibrated model. The
response of the model is evaluated by the following criteria, which should be decided
upon the basis of the socio-economic and environmental conditions in the study area:

Evaluation Factor Criteria

1. Water Balance Abstraction should be less than the recharge
Influence to springs and dug wells should be allowable
2. Environmental Impact
Salinity should be less than the water quality standard
3. Economics Total pumping head should be less than the criterion

If the model response is not acceptable even for one criterion, then the withdrawal plan
should be modified in terms of withdrawal, pumping pattern and borehole location. The
model is operated until the final agreement with all the criteria is achieved. With the
above steps of aquifer modelling, the optimum abstraction plan (the permissible yield) is
fixed and groundwater management plan will be examined.

12.3 Description of Numerical Model

12.3.1 Introduction

It is considered that a model is a tool designed to represent a simplified version of reality,

and properly constructed groundwater models, as they are also representations of reality,
can be valuable predictive tools for groundwater resources management. Of groundwater
models, a mathematical model is commonly used to study groundwater system.

A mathematical model consists of a set of differential equations that describe

groundwater flow. Since the assumptions to solve analytically a mathematical model are

12 - 5
 Chapter 12 Groundwater Modelling

fairly restrictive, numerical techniques are generally required to solve the mathematical
model approximately under realistic situations.

There are many numerical techniques to solve differential equations. Among them, finite
difference and finite element methods are representative (Fig. 12-4). The finite difference
method is superior in less computer memory requirement and simplicity of data
manipulation. The finite element method has excellence in flexibility of mesh formation.
In practical aspect of groundwater flow simulation, there is little distinction between both
methods. The finite difference method is applied for the proposed groundwater
management plan of the study area, in consideration of data availability and accuracy of
aquifer parameters.

12.3.2 Groundwater Flow Model

1) Basic Equation of Groundwater Flow

The unsteady-state, three-dimensional movement of groundwater through heterogeneous

and anisotropic porous media is described by the following partial-differential equation.

∂  ∂h  ∂  ∂h  ∂  ∂h  ∂h
 K xx  +  K yy  +  K zz  −W = Ss (12 − 1)
∂x  ∂x  ∂y  ∂y  ∂z  ∂z  ∂t

Where, Kxx, Kyy and Kzz: values of hydraulic conductivity along the x, y and z coordinate
axes, h: the potentiometric head, W: a volumetric flux per unit volume and represents
source and/or sink of water, Ss: the specific storage of the porous media, t: time.

Equation (12-1), together with specified flow and/or head conditions at the boundaries of
an aquifer system and specified initial-head conditions, constitutes a mathematical
representation of a groundwater flow system. Numerical method, such as the finite
difference method, is usually applied to solve equation (12-1).

2) Derivation of the Finite Difference Equation (McDonald and Harbaugh, 1988)

An aquifer system is discretized with a mesh of blocks called “cells” as illustrated in Fig.
12-5, the location of which are referenced with a row(i), column(j) and layer(k)
coordinate system parallel to the x, y and z directions, respectively. The width of cells in
the row direction, at a given column j, is defined rj; the width of cells in the column
direction at a given row i, is defined ci; and the thickness of cells in a given layer k, is
defined vk. At center of each cell, there is a point called a “node” at which head is to be
calculated.

From the application of the continuity condition, the sum of all flows into and out of the

12 - 6
 Chapter 12 Groundwater Modelling

cell is equal to the rate of change in storage within the cell, expressing as follows

ΣQi = ∆S (12 − 2) .

Where, Qi: flow rate into the cell, S: storage change within the cell over a time interval
t. For convenience flow entering cell is defined positive and outflow is defined
negative.

A cell i,j,k and six adjacent cells are illustrated in Fig. 12-6. Applying finite difference
method, groundwater flow is approximated by the node-to-node flow. Flow into cell i,j,k
in the row direction from cell i,j-1,k is given by Darcy’s law as

hi , j −1,k − hi , j ,k
Qi , j −1 / 2,k = KRi , j −1 / 2,k ∆ci ∆v k (12 − 3)
∆r j −1 / 2

Where, Qi,j-1/2,k: volumetric fluid discharge through the face between cells i,j,k and i,j-1,k,
hi,j,k: head at node i,j,k, hi,j-1,k: head at node i,j-1,k, KRi,j-1/2,k: hydraulic conductivity along
the row between nodes i,j,k and i,j-1,k, cj vk: area of the cell faces normal to the row
direction, rj-1/2: distance between nodes i,j,k and i,j-1,k. The hydraulic conductivity
between the nodes is normally calculated as a harmonic mean.

2 KRi , j −1,k KRi , j ,k

KRi , j −1 / 2,k = (12 − 4)
KRi , j −1, k + KRi , j ,k

Similar expressions can be written approximating the flow into the cell through the
remaining five faces. Flow in row direction through the face between cells i,j,k and
i,j+1,k is expressed as

hi , j +1,k − hi , j ,k
Qi , j +1 / 2,k = KRi , j +1 / 2,k ∆ci ∆v k (12 − 5)
∆r j +1 / 2

While for the column direction, flow into the cell through the forward face is

hi +1, j ,k − hi , j , k
Qi +1 / 2, j ,k = KC i +1 / 2, j ,k ∆r j ∆v k (12 − 6)
∆ci +1 / 2

And flow into the cell through the rear face is

hi −1, j ,k − hi , j ,k
Qi −1 / 2, j ,k = KC i −1 / 2, j ,k ∆r j ∆v k (12 − 7)
∆ci −1 / 2

For the vertical direction, inflow through the bottom face is

12 - 7
 Chapter 12 Groundwater Modelling

hi , j ,k +1 − hi , j ,k
Qi , j ,k +1 / 2 = KVi , j ,k +1 / 2 ∆r j ∆ci (12 − 8)
∆v k +1 / 2

While inflow through the upper face is given by

hi , j ,k −1 − hi , j ,k
Qi , j ,k −1 / 2 = KVi , j ,k −1 / 2 ∆r j ∆ci (12 − 9)
∆v k −1 / 2

The notation can be simplified by combining grid dimensions and hydraulic conductivity
into a single constant (“hydraulic conductance”). Hydraulic conductance between nodes
i,j-1,k and i,j,k can be written as

∆ci ∆v k
CRi , j −1 / 2,k = KRi , j −1 / 2,k (12 − 10)
∆r j −1 / 2

Storage change within the cell over a time interval t is expressed as follows.

∆hi , j ,k
∆S = Ss i , j ,k ∆r j ∆ci ∆v k (12 − 11)
∆t

Where, Ssi,j,k: specific storage of cell i,j,k, hi,j,k: head change over a time interval t,
rj ci vk: volume of cell i,j,k. Assuming t = tm – tm-1, an approximation to the time
derivative of head at time tm can be written as

 ∆hi , j ,k  h m − him, j−,1k

  = i , j ,k (12 − 12)
 ∆t m t m − t m −1

Where, hmi,j,k: head of cell i,j,k at time tm, hm-1i,j,k: head of cell i,j,k at time tm-1 which
precedes tm.

Substituting equation (12-3), equations (12-5) through (12-9), and equations (12-11) and
(12-12) into equation (12-2), and applying relationship in equation (12-10), finite
difference approximation for cell i,j,k can be obtained as

( ) (
CRi , j −1 / 2,k him, j −1,k − him, j ,k + CRi , j +1 / 2,k him, j +1,k − him, j ,k + )
CC i −1 / 2, j , k (hm
i −1, j , k − him, j ,k ) + CC i +1 / 2 , j , k (h m
i +1, j , k − him, j , k )+
CVi , j ,k −1 / 2 (hm
i , j , k −1 − him, j ,k ) + CV i , j , k +1 / 2 (h m
i , j , k +1 − him, j ,k ) + OS i , j ,k
m −1
m
−h
= Ss i , j ,k (∆r j ∆ci ∆v k )
h
(12 − 13)
i , j ,k i , j ,k

t m − t m −1

Where, QSi,j,k: source or sink term of cell i,j,k.

12 - 8
 Chapter 12 Groundwater Modelling

An equation of this form is written for every cell in the calculation domain, and the
system of equations is solved simultaneously for the heads at time tm. Solving techniques
of simultaneous equations are classified into direct and iterative methods. Finite
difference approximation of three-dimensional groundwater flow produces a large system
of simultaneous equations. Iterative method is superior to direct method in requirement
of computer memory, and is usually used to solve such large system. Strongly Implicit
Procedure method (SIP), Slice Successive Overrelaxation method (SSOR), and
Preconditioned Conjugate Gradient method (PCG) are typical of iterative method.
Precise explanation of solving techniques should be referred to McDonald and Harbaugh
(1988), and Hill (1990).

3) MODFLOW

MODFLOW (Modular Finite-Difference Ground-Water Flow Model) developed by U.S.

Geological Survey (McDonald and Harbaugh, 1988; Harbaugh and McDonald, 1996), is
widely used numerical model which can simulate groundwater flow in a
three-dimensional heterogeneous and anisotropic medium. As mentioned in the
following section, MODFLOW was applied to simulate groundwater flow systems in the
Stampriet Artesian Basin.

In a three-dimensional finite difference model, vertical discretization can be seen as a

sequence of horizontal layers. In complicated hydrogeological condition, this grid system
causes a cell contain material from different stratigraphic units (Fig. 12-7(b)). In
MODFLOW, vertical discretization can be viewed as an effort to represent individual
aquifers or permeable zones by individual layers of the model (Fig. 12-7(c)). This
distortion can be generated by giving the elevation of the top and bottom of the layer. It
allows flexibility in discretizing, however, introduces small error into the finite difference
approximation.

12.4 Conceptual Model

12.4.1 Hydrogeologic Condition

From a geological point of view, there are major four formations. They are the Damara
Sequence, the Nama Group, the Karoo Sequence and the Kalahari Beds. The
relationships among four formations are unconformity. The Damara Sequence and the
Nama Group consist of sandstone, shale and metamorphic rocks. They form the
hydrogeological basement. The Karoo Sequence is divided into the Dwyka Group and
the Ecca Group. The Dwyka Group is composed of tillite and mudstone, and is also a
hydrogeological basement. The Ecca Group is divided into the Nossob, Mukorob, Auob

12 - 9
 Chapter 12 Groundwater Modelling

and the Rietmond Members. The Nossob Member is composed of sandstone and is a
confined aquifer. The Lower Mukorob Member consists of shale and is an aquitard/
aquiclude. The Upper Mukorob Member consists of sandstone and is a confined aquifer.
The Auob Member is composed of sandstone and shale. It contains three confined
aquifers and two aquitards. However, it is difficult to clarify the distribution of these
aquifers and aquitards over the whole study area. Therefore, the Auob Member is
assumed to be one confined aquifer. The Lower Rietmond Member is composed of shale,
and is an aquitard. The Upper Rietmond Member consists of sandstone, and is an
unconfined or confined aquifer. The Kalkrand Basalt and the Karoo Dolerite intrude into
the Auob and Rietmond Members and are considered to be an unconfined-confined
aquifer or aquitard. The Kalahari Beds are composed of sand, gravel and calcrete. It is an
unconfined aquifer. Between the Kalahari Beds and the Upper Rietmond Member, there
is no distinct aquitard or low-permeability layer.

12.4.2 Aquifer System

As a result of the above discussion, an aquifer system consisting of one unconfined

aquifer and two confined aquifers with two confining layers (aquitards) was selected for
the groundwater model in the Stampriet Artesian Basin. The unconfined aquifer
represents the Kalahari and the Upper Rietmond aquifers. The first confined aquifer
represents the Auob and the Upper Mokorob aquifers. The second confined aquifer
represents the Nossob aquifer. The upper unconfined aquifer and the first confined
aquifer are connected due to leakage through the first confining layer (the Lower
Rietmond Member). The first confining layer is absent and the unconfined aquifer
directly overlies the first confined aquifer with a large area of the Stampriet Basin. In that
area, the lower part of the unconfined aquifer is assumed to be an aquitard. The second
confined aquifer is considered to be isolated from upper aquifers (Table 12-1).

12.5 Data Manipulation

12.5.1 Grid Design

Considering the hydrogeological structure of the study area, modeled domains were
defined as shown in Fig. 12-8. The model area covers all of the Stampriet Artesian Basin.
The area of the domain is about 125,000 km2 (303 413km). The domain was divided by
a grid at intervals of 6.3 km in the longitudinal and 6.9km in the latitudinal directions (see
Fig. 12-9). Numbers of cells in row and column directions are 60 and 48 respectively.
The application of a fixed grid system causes a slight error in the cell size, but it is
negligible in practical application of the model (see Fig. 12-10).

12 - 10
 Chapter 12 Groundwater Modelling

12.5.2 Elevation of Top and Bottom of Layers

As mentioned previously, a deformed grid can be used in MODFLOW. Therefore, the

study area was divided into 6 layers in vertical direction, and the elevations of the top and
bottom of layers were given. The Model layers basically correspond to the Kalahari Beds,
the Upper Rietmond Member, the Lower Rietmond Member, the Auob Aquifer, the
Lower Mukorob Member, and the Nossob Aquifer. However, individual aquifers or
aquitard cannot be represented by individual layers of the model, because of the
complicated hydrogeological condition of the study area. The elevations of the top and
bottom of layers were determined according to the following procedure.

1) Digitization of Isodepth and Isopach Maps

The isodepth and isopach maps shown in Section 8.3 were digitised for each cell;
l Surface Elevation
l Bottom of Kalahari Beds (Fig. 8.3-2)
l Isopach of Lower Rietmond Member (Fig. 8.3-3)
l Top of Auob Aquifer (Fig. 8.3-4)
l Isopach of Auob Aquifer (Fig. 8.3-5)
l Isopach of Lower Mukorob Member (Fig. 8.3-6)
l Isopach of Nossob Aquifer (Fig. 8.3-8)

Also, averaged value of “USGS Satellite 30 seconds Elevation Data” was used for the
surface elevation of each cell.

The difference between the surface elevation and the bottom of the Kalahari Beds was
assumed to be the thickness of the Kalahari Beds. If the value of bottom of the Kalahari
Beds was higher than the surface elevation, it was revised to be lower than the surface (5
to 10 meters in thickness). In the area where the Kalahari Beds directly overlay the Auob
Aquifer, the top of the Auob Aquifer was revised to be the same as the bottom of the
Kalahari Beds.

The value obtained through subtracting the top of the Auob Aquifer from the bottom of
the Kalahari Beds was assumed to be the thickness of the Rietmond Member. The
thickness of the Upper Rietmond Member was calculated to subtract the thickness of the
Lower Rietmond Member from that of the Rietmond Member.

The bottom of the Auob Aquifer was obtained to subtract the thickness of the Auob
Aquifer from the top of it. The bottom of the Lower Mukorob Member was calculated to
subtract the thickness of it from the bottom of the Auob Aquifer. Then, the bottom of the
Nossob Aquifer was calculated to subtract the thickness of it from the bottom of the

12 - 11
 Chapter 12 Groundwater Modelling

Lower Mokorob Member successively.

Digitized isodepth and isopach maps of aquifers and aquitards. These digitized isodepth
maps were used for summing up groundwater use by aquifer.

2) Elevation of Model Layers

The abovementioned aquifers and aquitards do not sequentially lie on top of another,
since the hydrogeological condition of the study area is complex. There are 22
combinations of overlaying of them as shown in Table 12-2. The combination of each
cell is shown in Table 12-3.

Allotment of aquifers and aquitards to the model layers was executed according to the
procedure given in Table 12-4. The bottom elevations of layers are shown in Table 12-5
to Table 12-10. Fig. 12-11 shows the N-S cross-sectional grid design at column 25. Fig.
12-12 shows the E-W cross-section at row 30 (Locations of sections are shown in Fig.
12-8).

12.5.3 Boundary Condition

The model area is bounded by basement rocks on the north. However, the Kalahari Beds
overlying basement rocks continue to the outside of the study area. Groundwater inflow
can be expected across the border. Therefore, constant head condition was assumed at
this boundary to approximate the groundwater inflow. On the western side, the
distribution of the first and second confined aquifers is restricted at the border. This
boundary was considered as a no flux condition. Near Kalkrand, groundwater level
contours show outflow from the study area, therefore discharging condition was set on
this border. On the southern border, the model area is bounded by basement rocks. No
flux boundary condition was assumed. On the eastern side, the aquifers are continuous to
the territory of Botswana and South Africa. Since groundwater level contours are almost
perpendicular to the boundary, it was considered to be no flux condition. On the right
bottom corner of the model, groundwater outflow can be expected across the border
through the Kalahari Beds. Constant head condition was assumed to approximate the
groundwater outflow.

The boundary conditions of the modelled area are shown in Fig. 12-8.

12.5.4 Aquifer Constants

There are very few pumping test data, excluding the JICA test boreholes as mentioned in
section 2.5. Distributions of the permeability of aquifers are shown in Figs. 12-13 to

12 - 12
 Chapter 12 Groundwater Modelling

12-15. The study area was divided into several zones based on these figures. The
permeability of a zone is assumed to be homogeneous and isotropic. These values were
modified in the process of model calibration.

All pumping tests, except one, were executed without observation boreholes, therefore
accurate storativity values were not able to obtain. The storativity values of the aquifers
set uniformly and then these values modified in the model calibration process.

12.5.5 Groundwater Levels

The groundwater levels have been measured in the water level survey (section 2.8). The
elevation survey of boreholes has been also done (section 2.7). Based on the results of
these surveys, groundwater level contour maps were drawn. These maps were used in the
model calibration.

There are 22 observation boreholes of the DWA. At these boreholes groundwater level
changes have been measured monthly. Also, groundwater levels have been observed
monthly in 30 NamWater boreholes. Observed groundwater level changes were used in
the model calibration.

12.5.6 Pumpage

1) Data Source and Estimation Method

Data source and estimation method of present groundwater use is shown in Table 12-11.
Production rate of NamWater is monthly, and report of irrigation permit holders is also
monthly. The results of the Hydrocensus are average values at the surveyed date.

If the number of people in a farm was unknown, it was estimated by using farm area and
unit population. Also, number of stocks was estimated from farm area and carrying
capacity, if it was unknown. Average value of the Hydrocensus is used in the ratio of
small stocks and large stocks.

Information of farms, which have not been surveyed by the Hydrocensus, is only farm
area in the DWA database. The number of people and stock was estimated in the same
way as mentioned above.

2) Results of the Hydrocensus

Present groundwater use was estimated, mainly based on the results of the Hydrocensus.
The results are summarized in Table 12-12.

12 - 13
 Chapter 12 Groundwater Modelling

3) Summary of NamWater Scheme

Production rates of NamWater scheme are summarized in Table 12-13, and variations of
each scheme are shown in Fig. 12-16. Monthly production rates of each scheme are given
in Appendix Table B-1 to B-28.

Total production increased from 482 thousand m3/year in 1986 to 604 thousand m3/year
in 1999. The production of the Aranos scheme is largest and about 40% of total in 1999.

4) Summary of Irrigation Permits

The irrigation permits are valid for five years, and water allocation is prescribed in annual
production per farm. Boreholes must be equipped with water gauges, but some are not
equipped. Monthly productions should be reported, however the DWA records on each
farm are incomplete.

Using values of same month in the previous year or in the following year, lacking records
of the monthly production reports were estimated. The results are shown in Table 12-14.

Irrigation uses of the permit holders have a tendency to increase. However, it is 64% of
the allocated amount in 1999. The relation to rainfall is not distinct. The production in
1997 with a lot of rain is smaller than those in 1996 and 1998 with little rainfalls. But, the
production in 1995 is smaller than that in 1996, although the rainfall in 1995 is less than
that in 1996 (Fig. 12-17). Also, the relationship between monthly production and
monthly rain is not clear (Fig. 12-18).

The groundwater abstraction for irrigation use affects the groundwater level change. For
example, the water level change of the Spes Bona observation borehole (WW32457) is
almost consistent with the variation of irrigation use. The water level of this borehole is
considered to represent the change of irrigation use (Fig. 12-19).

The unit consumption for irrigation was calculated based on the groundwater use of the
permit holders and irrigated areas of the Hydrocensus. Using the allocated amounts, the
unit consumption becomes 42.0 m3/day/ha. Using the reported products, the unit
consumptions are 28.6 m3/day/ha in 1998 and 26.0 m3/day/ha in 1999 respectively.
Excluding farms with very small consumption values, the unit consumptions become
34.0 m3/day/ha in 1998 and 31.0 m3/day/ha in 1999 (Table 12-14).

5) Present Groundwater Use by Usage Type

Basically using the unit consumptions of the Hydrocensus (Table 12-16), present
groundwater use was estimated. Results are shown in Table 12-17 and Fig. 12-20. Total

12 - 14
 Chapter 12 Groundwater Modelling

groundwater use in the study area is estimated to be 40,912 m3/day (14.9 million m3/year).
Irrigation use amounts to 18,868 m3/day (6.9 million m3/year) and is 46% of total use.
Stock watering use is 15,589 m3/day (5.7 million m3/year) and 38%. Domestic use is
6,455 m3/day (2.4 million m3/year) and 16%.

The distributions of present groundwater use by usage in each cell are shown in Figs.
12-21 to 12-24. The estimated groundwater use in a farm was divided among boreholes
belonging to that farm. Then, the groundwater uses of boreholes were summed up for
each cell. The location of boreholes was based on the results of the Hydrocensus and the
DWA Database.

Domestic use is generally distributed over the study area. Pumping rate of almost cells is
less than 10 m3/day. Near Aranos, there is only one cell with large pumping rate that is
more than 500 m3/day (Fig. 12-22). Stock watering use is also generally distributed over
the study area except the Aminuis region. Pumping rate of almost cells is less than 30
m3/day. There is no cell whose pumping rate is more than 500 m3/day (Fig. 12-23).
Irrigation use centres on the Stampriet region, and expands along the Auob River and the
road from Stampriet to Aranos. There is one cell whose pumping rate is more than 2000
m3/day. The ratio of cell with large pumping rate is high comparing with domestic or
stock watering use (Fig. 12-24). Total groundwater use coincides with irrigation use in
heavy pumping area. Generally groundwater use is dense in western half of the study area
(Fig. 12-21).

6) Present Groundwater Use by Aquifer

(1) Present Groundwater Use by Aquifer

The estimated groundwater use in each borehole was divided into the aquifers based on
the borehole construction information (i.e. screen depth, borehole depth, water strike
depth) and the isodepth maps of aquifers. Unfortunately there are few boreholes with
complete information. Therefore, dividing of the groundwater use into aquifers was
done through the procedure shown in Table 12-18.

Results are shown in Table 12-19 and Figs. 12-25 to 12-31. Groundwater use from the
Kalahari Aquifer amounts to 26,739 m3/day (9,8 million m3/year) and occupies 65% of
total use. Groundwater use from the Auob Aquifer is estimated to be 13,622 m3/day
(5.0 million m 3/year) and 33%, and that from the Nossob Aquifer is merely 551 m3/day
(0.2 million m3/year) and 1% (Fig. 12-25).

The domestic use from the Kalahari Aquifer amounts to 69% of total domestic use.
These from the Auob Aquifer and the Nossob Aquifer are 24% and 7% respectively

12 - 15
 Chapter 12 Groundwater Modelling

(Fig. 12-26). For the stock watering use, groundwater from the Kalahari Aquifer
occupies 81%, and that from the Auob Aquifer is 19% (Fig. 12-27). The irrigation use
is equally shared between the Kalahari Aquifer and the Auob Aquifer (Fig. 12-28).

In the Kalahari Aquifer, the stock watering use is 47%, and the irrigation and domestic
use are 36% and 17% respectively (Fig. 12-29). In the Auob Aquifer, the irrigation use
is prominent and is 67%. The stock watering and domestic use are 22% and 11%
respectively (Fig. 12-30). In the Nossob Aquifer, the domestic use is 88%, and the
irrigation and stock watering use are only 10% and 2% respectively (Fig. 12-31).

(2) Distribution of Present Groundwater Use by Aquifer

The distributions of present groundwater use by aquifer in each cell are shown in Figs.
12-32 to 12-34. The distributions of each usage by aquifer are shown in Figs. 12-35 to
12-43.

Groundwater use of the Kalahari Aquifer is distributed over almost all study area.
Heavy pumping cells are concentrated in the Stampriet region, and there is a cell with
pumping rate of 1,000-2,000 m3/day. Along the Auob River, cells with relatively large
pumping rate expand (Fig. 12-32). Groundwater use of the Auob Aquifer is centred at
the Stampriet region, and there is a cell whose pumping rate is more than 2,000 m 3/day.
Cells with small pumping rate spreads in the area to the southeast of Aranos (Fig.
12-33). Groundwater use of the Nossob Aquifer is dotted around the study area (Fig.
12-34).

The domestic use of the Kalahari Aquifer extends over the study area. The pumping
rate of cells is generally less than 10 m3/day (Fig. 12-35). The distributed area of the
domestic use of the Auob Aquifer is smaller than that of the Kalahari Aquifer. The
pumping rate is almost less than 10 m3/day (Fig. 12-36). The domestic use of the
Nossob Aquifer is very similar to total groundwater use of it (Fig. 12-37).

The stock watering use of the Kalahari Aquifer is widely distributed except Aminuis
area. Pumping rate is generally less than 30 m3/day. In the south-western part of the
study area, there are several cells with pumping rate of 100-500 m3/day (Fig. 12-38). In
the Auob Aquifer, the stock watering use is distributed around Stampriet, Aranos and
Leonardville. Pumping rate is almost less than 30 m3/day (Fig. 12-39). The stock
watering use of the Nossob Aquifer is scarce (Fig. 12-40).

The irrigation use of the Kalahari Aquifer is concentrated to the Stampriet region, and
expands along the Auob River and the road from Stampriet to Aranos (Fig. 12-41).
That of the Auob Aquifer is only centred on Stampriet region. There is a cell whose

12 - 16
 Chapter 12 Groundwater Modelling

pumping rate is lager than 2000 m3/day (Fig. 12-42).

7) Variation of Groundwater Use

The groundwater use variation from 1990 to 1999 was estimated by using the statistics
and existing data (Table 12-20). In the domestic use, groundwater use variation in
commercial farms and communal land was calculated from the population change. For
commercial farms, annual population growth was estimated to be 0.68% based on the
1991 Population and Housing Census and the Hydrocensus. For communal land, annual
population growth was estimated to be 2.14% in Aminuis and 2.48% in Nama land (see
Chap.10). For village centre, the productions of NamWater (Table 12-13) were used
directly.

The variation of the stock watering use was estimated by number of livestock. Based on
the Hydrocensus 1986-1989 (DWA, 1986; 1987; 1989), the stock watering use in Area
1-3 amounted to 883,600m3/a, using unit consumptions 35 Litter/head for LS and 5
Litter/head for SS (Table 12-21). Based on the Hydrocensus, the stock watering use in
the same area was estimated to be 735,000 m3/a, using the same unit consumptions.
From these figures annual growth was calculated to be –1.82%.

From 1994 to 1999, production of the permit holders (Table 12-14) was used directly
estimating the irrigation use variation. From 1990 to 1993, exponential approximation
equation of productions from 1994 to 1999 was used (y=2.9754exp(0.1017x); y: annual
production, x=year-1993).

The estimated results are shown in Table 12-22 and Figs 12-44 and 12-45. The
groundwater use increased from 11.6 million m3/year in 1990 to 14.9 million m3/year in
1999. The stock water use decreased from 6.7 million m3/year and 58% of total in 1990 to
5.7 million m3/year and 38% in 1999. On the other hand, the irrigation use increased 2.7
million m3/year and 23% of total in 1990 to 6.9 million m3/year and 46% in 1999 (Fig.
12-44).

Groundwater from the Kalahari Aquifer increased from 8.4 million m3/year in 1990 to 9.8
million m3/year in 1999. That from the Auob Aquifer rapidly increased from 3.0 million
m3/year to 5.0 million m3/year (Fig. 12-45).

12.5.7 Recharge

It is difficult to estimate the groundwater recharge based on the runoff analysis (see
section 2.2). Therefore, the groundwater balance analysis was applied to estimate it.

12 - 17
 Chapter 12 Groundwater Modelling

1) Groundwater Balance in An Area

The groundwater balance in an area is expressed as follows.

dh
R−D=S (12 − 14)
dt

Where, R: recharge rate, D: discharge rate, S: storativity, dh/dt: groundwater level change
in unit time. R means the groundwater inflow, D means the groundwater outflow and
Sdh/dt is the storage change of groundwater in the area.

For the period without recharge, a regression curve of groundwater level is expressed by
following exponential-type equation.

= C (h − h0 ) (12 − 15)
dh
dt

Where, h: groundwater level, h0: standard groundwater level (at this level, discharge rate
becomes zero), C: constant. Substituting equation (12-15) into equation (12-14), the
discharge rate (D) can be expressed as follows (where, R=0).

D = − SC (h − h0 ) (12 − 16)

For the recharge period, a recharge rate (R) is expressed following equation, substituting
equation (12-16) into equation (12-14).

R = − SC (h − h0 ) + S
dh
dt
 dh 
= S  − C (h − h0 ) (12 − 17)
 dt 

From equation (12-17) the recharge rate can be estimated reading groundwater level on
the observation borehole records.

2) Recharge Estimation at Olifantswater

To estimate the recharge rate by the above method, it needs to select an observation
borehole that represents natural recharge. Of course the recharge rate cannot be estimated
from an observation borehole influenced heavily by pumping.

The Olifantswater WW21815 was selected for the recharge estimation. It has observed
the groundwater level in the unconfined Kalahari aquifer. The groundwater level change
of WW21815 differs from that of the Spes Bona WW32457, which is strongly effected by
the irrigation groundwater use (see Fig. 12-19). It is considered to represent the natural

12 - 18
 Chapter 12 Groundwater Modelling

recharge-discharge process.

The water level of WW21815 shows a tendency to decrease annually. To calculate C and
h0, it is necessary to make a correction of the water level. As shown in Fig. 12-46, the
correction value is 0.61m for 11 years (i.e. 5.5cm/year). The calculated C and h0 are
0.0167 and -15.245m respectively (Fig. 12-47).

Using equations (12-16) and (12-17), the monthly discharge and recharge rates were
calculated. For the recharge calculation, it was assumed that the recharge occurs when
water level is rising, and it equals zero when water level is declining. The results are
shown in Fig. 12-46. Annual values are summarized in Table 12-24.

The discharge and recharge rates calculated from equations (12-16) and (12-17) are
proportional to the storativity that equals the effective porosity in unconfined aquifer.
The results of neutron logging of the JICA test boreholes show that the porosity of the
Kalahari aquifer is about 25%. Using this value as an effective porosity, the calculated
recharge rate exceeds rainfall. Therefore, the effective porosity is smaller than the
porosity estimated by the neutron logging, and it seems that value from 2% to 5% is
adequate.

These values were used for the initial setting of model calibration, and were revised in
process of the calibration.

12 - 19
 Chapter 12 Groundwater Modelling

12.6 Model Calibration

12.6.1 Procedure of Model Calibration

Model calibration was carried out in the following steps:

a. Calibrating groundwater level distribution by the steady-state calculation

b. Calibrating groundwater level variation by the unsteady-state calculation

c. Calibrating groundwater level change caused by 1/50 years rain (in 2000)

The first step of model calibration is to clarify the hydraulic conductivity and to make
initial heads (1990) for unsteady-state calculation. Calculated heads were compared with
groundwater level distribution based on the groundwater level survey. The second step of
yearly unsteady-state calculation from 1990 to 1999 is to clarify the specific storage and
recharge rate. Calculated head variations were compared with water level records of
DWA and NamWater. Third step is to clarify the recharge rate with 1/50years rain (in
2000). Calculated head was compared with observed water level, then parameters
(mainly hydraulic conductivity) and recharge rate were modified until final agreement
between the calculated and the observed water level is achieved.

12.6.2 Calibrated Model

1) Comparison between Observed and calculated results

(1) Groundwater Level Configurations

Calculated heads were compared with results of the groundwater level survey. The
results of survey show distribution of groundwater level in 2000. Groundwater level
changes between 1990 and 2000 are relatively small, therefore the configuration of
groundwater level is considered to be almost unchanged. Calculated heads were also
verified by observation well records of Jan. 1990.

Fig. 12-48 shows observed and calculated groundwater level configurations of the
Kalahari Aquifer. From the 1300m to 1100m contours, calculated and observed heads
are almost coincided. On the 1000m and 950m contours, differences between
calculated and observed heads are slightly large. In that area, there were few
measuring points therefore strict comparison is difficult. Comparison of the Auob
Aquifer is shown in Fig. 12-49. From the 1300m to 1050m contours, calculated and
observed heads well agree. The 1000m and 950m contours of the both are not in

12 - 20
 Chapter 12 Groundwater Modelling

agreement. Fig. 12-50 shows the configuration of the Nossob Aquifer. From 1250m to
1100m contours, the both are well coincided. In the eastern and southern part of the
model, the both do not entirely agree. In these areas, configuration of observed head is
not accurate since there are few measuring points.

Fig.12-51 shows comparison of observed and calculated heads at observation wells. If

the both values coincide, corresponding point is plotted on a line of 45 angle. All
points distributes along the line of 45 angle. The mean error that is the mean
difference between measured and calculated head is 1.0m. The results indicate good
agreement of the two.

(2) Variations of Groundwater Level

Observed and calculated groundwater level variations in 1990-1999 at observation

wells are shown in Figs. 12-52 to 12-59. Observations are monthly, but calculated
results are yearly. At Olifantswater, calculated heads of the Kalahari Aquifer well
agree with the trend of observed groundwater level (WW21815). Calculated heads of
the Auob Aquifer are approximately 1 meter higher than the observed one (WW21784).
However, these well simulates the declining trend of the observed records (Fig. 12-52).
At Gomchanas (WW8399) calculated heads are about 1.5 meters lower than observed
groundwater level. Decline of calculated heads is slightly smaller than that of the
observation (Fig. 12-53).

At Spes Bona calculated heads are approximately 2 meters lower than the observed
heads (WW32457), however these well agree with the trend of observed variations
(Fig. 12-54). Calculated heads at Boomplaas correspond to the observation
(WW10120). These in 1998 and 1999 are slightly lower than the observation (Fig.
12-55).

At Tugela (WW22838) calculated heads of the Kalahari Aquifer are about 3 meters
lower than the observed heads and these of the Auob Aquifer are about 4 meters lower
than the observed ones. Both of calculated heads well coincide with the trend of
declining observation records. Calculated heads of the Nossob Aquifer is
approximately 5 meters higher than the observation. Decline of calculated heads is
slightly smaller than that of the observation (Fig. 12-56).

At Gochas calculated heads of the Kalahari and Auob Aquifers are almost unchanged.
Levels of the both correspond to the observation (WW7491, WW16343)
approximately (Fig. 12-57). Calculated head of the Nossob Aquifer at Aranos well
agree with the observed heads (WW7407) (Fig. 12-58). At Aminuis calculated heads

12 - 21
 Chapter 12 Groundwater Modelling

are about 1 meter lower than the observed groundwater level (WW26164). Since
observed variation is irregular, it is difficult to compare the trends of both (Fig. 12-59)

Gochas(WW7491, WW16343), Aranos(WW7407) and Aminuis(WW26164) are

NamWater scheme wells. Therefore groundwater change caused by pumping is large,
and it is difficult to clarify the trend of variation.

(3) Groundwater Level Change with 1/50 years rain

Groundwater level recovery in 2000 caused by heavy rain in 1999-2000 was calibrated,
assuming pumping rates in 2000 and 2001 to be the same as that in 1999. Results are
also shown in Figs. 12-52 to 12-59.

Calculated heads of the Kalahari Aquifer agree well with the observed recovery at
Olifantswater. In the Auob Aquifer groundwater level recovery has not observed,
however calculated heads indicate a little recovery (Fig. 12-52). At Gomchanas the
observed records shows slight recovery and calculated head agree with it (Fig. 12-53).

Though observed heads at Spes Bona shows about 3 meters recovery, calculated heads
continue to decline. This recovery is considered to arise from pumping rate change. In
the calculation pumping rate in 2000 is the same as that in 1999, therefore the
calculated heads do not agree with the observed recovery (Fig. 12-54). At Boomplaas,
calculated and observed heads indicate the same situation as Spes Bona (Fig. 12-55).

At Tugela calculated heads of the Kalahari and Auob Aquifers coincide with about 2
meters recovery of the observed heads. Calculated heads of the Nossob Aquifer do not
agree with slight recovery of observation (Fig. 12-56).

At Gochas, Aranos and Aminuis groundwater level recovery in the observed records
are not distinct for the above-mentioned reason (Figs. 12-57 to 12-59).

2) Water Budget in Calibration Period

Water Budget in 1999 and 2000 is given in Figs. 12-60 and 12-61. In 1999, well pumping
rate (38,400 m3/day) considerably exceeded recharge rate (12,600 m3/day). To
compensate this shortage, water released from the storage (24,000 m3/day) declining
groundwater level. Pumping rate of the model doesn’t agree with that of the study area,
since the model area is smaller than the study area (Fig. 12-60).

In 2000, recharge rate (218,800 m3/day) increased by about 17 times that of 1999. With
1/50years rain, groundwater level recovered and the storage increased (181,900 m3/day)
(Fig. 12-61).

12 - 22
 Chapter 12 Groundwater Modelling

3) Fixed Parameters

Hydraulic conductivity, Specific Storage (effective porosity) and recharge rate were
modified to calibrate the model. The distributions of fixed hydraulic conductivity of
model layer 1 to 6 are shown in Figs.12-62 to 12-67. The distribution of fixed specific
storage and effective porosity of model layer 4 and 6 are shown in Figs. 12-68 and 1-69.
Fixed hydraulic conductivity, specific storage and effective porosity are summarized in
Table 12-25. Zone number in Figs. 12-62 to 12-69 corresponds to that in Table 12-25.
Specific storage and effective porosity of model layer 1 to 3, and 5 sets uniformly to be
zone 1.

Recharge rate is summarized in Table 12-26 and distribution of zones is shown in Fig.
12-70. Zone number in Fig. 12-70 corresponds to that in Table 12-26. In ordinary years
(1990-1999), recharge rate is very small except zone 6. Results of Carbon-14 and Tritium
analysis support this low recharge rate. At zone 6 relatively large recharge rate was
needed to prevent cells form drying up, since the aquifers locate at high altitude. In 2000
with 1/50years rain, recharge rate increased to about 1 to 3% of annual rainfall to calibrate
the groundwater level change at observation wells.

12.7 Model Prediction

12.7.1 Prediction Cases

1) Pumpage

To predict the groundwater level change caused by pumping rate change, 6 cases shown
in Table 2-27 were studied. Cases 1 and 2 were assumed to keep present groundwater use.
For case 3 irrigation use was increased to 120% of present. For cases 4 to 6 irrigation use
was decreased to 70%, 50%, 0% respectively. Indicating the calibration results,
groundwater level depletion in Stampriet area is major problem in the basin. Therefore
change of irrigation use that causes the depletion was studied. Prediction period for each
case is 100 years.

2) Recharge

Recharge rate will vary in prediction period, but there is no precise recharge analysis to
obtain probability of recharge at present. Calibrated recharge rate was used as an average
of long period. For Cases 2 to 6, recharge rate with1/50years rain was assumed in 30th
and 80th year (Fig. 12-71).

12 - 23
 Chapter 12 Groundwater Modelling

12.7.2 Constraints on Permissible Yield

As mentioned in section 12.1.1, constraints on permissible yield are social scientific

matter except water balance. Therefore it is difficult to define the constraints. Here,
following constraints are assumed. These are not complete and should be revised with
exhaustive argue.

1) Water Balance Constraint

Groundwater abstraction should be within the possible amount of recharge on arbitrary

spatial and time scales. In this study time scale is determined to be one year, therefore
annual amount of pumpage must not exceed that of the possible recharge. In other words,
the groundwater level must recover to the initial level at the end of one hydrologic year.
Following table shows the criteria for the annual residual drawdown.

Criteria for Water Balance Constraint

Rank Annual Residual Description
Drawdown (m)
A 0.00 – 0.03 Allowable: Not surely safe, but allowable if there
is no alternative plan
B 0.03 – 0.10 Undesirable: The aquifer storage will be possibly
depleted in future
C 0.11 - Not Allowable: The aquifer storage will be
probably depleted in near future

2) Water Quality Constraint

In this simulation groundwater quality is not studied, since groundwater level change is
considered to be small and it would not cause groundwater quality change. However,
there are high salinity zones (“salt blocks”), and groundwater quality is one of the
problems in this basin. Prediction of groundwater quality is the subject for a future study.

3) Environmental Constraint

Influences on existing springs and vegetations are possible environmental constraints to

be considered. According to the environmental analysis (see Chap. 14), springs in this
basin already dried up decades ago. Therefore, the impact on existing springs can be
negligible. Decline or recovery of groundwater level will affects the vegetations in this
basin. However, no quantitative analysis on the relationship between groundwater level
and vegetations has been done. In this study, the influence on vegetations was neglected.

12 - 24
 Chapter 12 Groundwater Modelling

It is the subject for a future study.

4) Economical Constraint

This constraint comes from the limits of pump lift. Average PID (Pump Installed Depth)
is 46 meters and average groundwater level depth is 29 meters based on the Hydorocensus.
As a result, the total drawdown should be within about 10 meters. Otherwise the pump
must be altered to suit deeper water level, which will cost more. Criteria for the economic
constraint are shown in the following table.

Criteria for Economic Constraint

Rank Total Drawdown after Description
100 years (m)
A 0 – 10 Good: No problems in practical use
B 10 – 20 Allowable: Well yield may decrease
C 20 > Undesirable: Pump should be changed

12.7.3 Model Prediction

In Figs. 12-72 to 12-89, predicted results are shown as drawdown after 100 years for each
aquifer and case. Calculation of bottom area of the model was unstable and there were
incomprehensible head differences. These differences were ignored in figures. Also,
groundwater level variations at observation wells are shown in Figs. 12-90 to 12-92.

1) Case 1

Groundwater pumpage is kept at the rate of 1999 (14.9 million m3/ year) in this case.
Maximum drawdown of the Kalahari Aquifer will exceed 30 meters, and the Kalahari
Aquifer will dry up within 35 years at Spes Bona (Figs. 12-72 and 12-91). It corresponds
to the rank C (Not Allowable) of water balance constraint and the rank C (Undesirable) of
economic constraint. In Stampriet area there is no observation well of the Kalahari
Aquifer then the predicted results are not calibrated. However, the results can show the
trend of the variation of groundwater level. Maximum drawdown in the Auob Aquifer is
14 meters at Stampriet (Figs. 12-78 and 12-91). It corresponds to the rank C (Not
Allowable) of water balance constraint and the rank B (Allowable) of economic
constraint. When a cell dries up, the MODFLOW automatically eliminates pumpage in
that cell. Therefore, the groundwater level in the Auob Aquifer is slightly recovered,
when the Kalahari Aquifer dries up and the pumpage of the aquifer is eliminated.

At Olifantswater drawdown in the Kalahari Aquifers is 3 meters and that of the Auob

12 - 25
 Chapter 12 Groundwater Modelling

Aquifer is 3 meters also (Fig. 12-90). These correspond to the rank A (Allowable) of
water balance constraint and the rank A (Good) of economic constraint. At Tugela
drawdown in the Kalahari Aquifer is 1 meter. It corresponds to the rank A of water
balance and economic constraints. Drawdown in the Auob and Nossob Aquifers are 5
meters and 4 meters respectively (Fig. 12-92). These classified into the rank B
(Undesirable) of water balance constraint and the rank A of economic constraint.

2) Case 2

Pumping rate is the same as that of Case 1 (14.9 million m3/ year). In 30th and 80th years
it is assumed to get recharge with 1/50years rain. At Spes Bona drawdown in the Kalahari
and Auob Aquifers are the same as these of Case 1 (Figs. 12-73 and 12-79). The Kalahari
Aquifer will dry up within 35 years (Fig. 12-91). The classifications are the same as these
of Case 1. In the Stampriet area no recharge condition is set (Fig. 12-70), therefore no
groundwater level recovery is calculated in the year with 1/50years rain. The drawdown
of the Kalahari Aquifer is accelerated with increase of groundwater level decline. The
Kalahari Aquifer is unconfined aquifer. When a decline of groundwater level becomes
larger, the transmissivity of unconfined aquifer becomes smaller. Pumping rate in the
aquifer is constant. Therefore, acceleration of decline is observed.

At Olifantswater groundwater level of the Kalahari Aquifer shows the recovery of about
0.4 meters in the year with 1/50years rain. Drawdown for 100 years reaches about 2.6
meters. Groundwater level of the Auob Aquifer shows the recovery of about 0.2 meters
in the year with 1/50years rain. Total drawdown is about 2.2 meters (Fig. 12-90). These
correspond to the rank A of water balance and economic constraints.

At Tugela groundwater level of the Kalahari Aquifer indicates the recovery of 0.7 meters
in the year with 1/50years rain. The recovery of the Auob Aquifer with that rain is 0.6
meters. Total drawdown of the Kalahari Aquifer is 0.8 meters. These of the Auob and
Nossob Aquifers are 3.0 meters and 2.8 meters respectively (Fig. 12-92). The results are
classified into the rank A of water balance and economic constraints.

3) Case 3

Pumping rate of irrigation use is increased to 120% in this case. Total pumping rate is
109% of Case 1 (16.3 million m3/ year). Maximum drawdown of the Kalahari Aquifer
will exceed 30 meters, and the Kalahari Aquifer will dry up within 25 years at Spes Bona
(Figs. 12-74 and 12-91). It corresponds to the rank C of water balance constraint and
economic constraint. Maximum drawdown in the Auob Aquifer exceeds 20 meters at
Stampriet (Figs. 12-78 and 12-91). It corresponds to the rank C of water balance and

12 - 26
 Chapter 12 Groundwater Modelling

economic constraints.

At Olifantswater groundwater levels of the Kalahari and Auob Aquifers are slightly lower
than these of Case 2. The differences of total drawdown between the both are less than
0.1 meters (Fig. 12-90). These correspond to the rank A of water balance and economic
constraints. At Tugela groundwater level of the Kalahari Aquifer is almost the same as
that of Case 2. Drawdown of the Auob and Nossob Aquifers are 0.05 and 0.1 meters
lower than these of Case 2 (Fig. 12-92). These are classified into the rank A of both
constraints.

4) Case 4

In this case the pumping rate of irrigation use is reduced to 70% of Case 1. Total pumping
rate is 86% of Case 1 (12.9 million m3/ year). At Spes Bona the period to dry up in the
Kalahari Aquifer extends for 80 years (Fig. 12-91). However, maximum drawdown in
Stampriet area exceeds 30 meters (Fig. 12-75). It corresponds to the rank C of water
balance and economic constraints. Total drawdown of the Auob Aquifer at Spes Bona is
3.4 meters. Maximum drawdown in Stampriet area is less than 10 meters (Fig. 12-81). It
is classified into the rank B of water balance and the rank A of economic constraint.

At Olifantswater groundwater level of the Kalahari and Auob Aquifers are 0.06 to 0.15
meters higher than these of Case 2, and are classified into the rank A of water balance and
economic constraints (Fig. 12-90). At Tugela groundwater level of the Kalahari Aquifer
is almost the same as that of Case 2. Drawdown of the Auob and Nossob Aquifers are
0.07 and 0.21 meters higher than these of Case 2 (Fig. 12-92). These are classified into
the rank A of both constraints.

5) Case 5

The pumping rate of irrigation use is reduced to 50% of Case 1 in this case. Total
pumping rate is 77% of Case 1 (11.5 million m3/ year). At Spes Bona total drawdown of
the Kalahari Aquifer is almost 0 meter, however total drawdown of 5 to 9 meters remains
in Stampriet area (Figs. 12-76 and 12-91). The water balance constraint is the rank B and
economic constraint is the rank A. Total drawdown of the Auob Aquifer is 3.0 meters,
and it corresponds to the rank A of both constraints (Figs. 12-82 and 12-91).

At Olifantswater groundwater level of the Kalahari and Auob Aquifers are 0.1 and 0.25
meters higher than these of Case 2, and correspond to the rank A of both constraints (Fig.
12-90). At Tugela groundwater level of the Kalahari Aquifer is almost the same as that of
Case 2. Drawdown of the Auob and Nossob Aquifers are 0.12 and 0.25 meters higher
than these of Case 2 (Fig. 12-92). These are classified into the rank A of both constraints.

12 - 27
 Chapter 12 Groundwater Modelling

6) Case 6

The pumping rate of irrigation use is reduced to 0% (abolishing irrigation) of Case 1 in

this case. Total pumping rate is 54% of Case 1 (8.1 million m3/ year). At Spes Bona
groundwater level of the Kalahari Aquifer shows the recovery of 21 meters (Fig. 12-77
and 12-91). It corresponds to the rank A of water balance and economic constraints.
Groundwater level of the Auob Aquifer shows the recovery of 15.5 meters, and it
corresponds to the rank A of both constraints (Figs. 12-83 and 12-91).

The model layer 1 and 2 correspond to the Kalahari Aquifer as mentioned before. The
model layer 1 at Spes Bona is dry cell (inactive) in the calibration periods and the bottom
altitude of it is 1155m. The model layer 2 is unconfined until groundwater level recovers
to 1155m. When groundwater level reaches to 1155m, the model layer 2 becomes
confined condition. Storativity of confined aquifer is smaller than that of unconfined
aquifer in two orders. Then groundwater level recovers suddenly in 2014. This recovery
causes the model layer 1 to be active (re-wetting). The model layer 1 is unconfined and
the groundwater level recovery becomes to be gradual.

At Olifantswater groundwater level of the Kalahari and Auob Aquifers are 0.25 and 0.5
meters higher than these of Case 2, and correspond to the rank A of both constraints (Fig.
12-90). At Tugela groundwater level of the Kalahari Aquifer is almost the same as that of
Case 2. Drawdown of the Auob and Nossob Aquifers are 0.25 and 0.7 meters higher than
these of Case 2 (Fig. 12-92). These are classified into the rank A of both constraints.

12.8 Evaluation of Permissible Yield

12.8.1 Permissible Yield in the Basin

1) Water Balance

Results of water balance in prediction period are shown in Figs. 12-93 to 12-103 for
ordinary year (after 10 years) and year with 1/50years rain (after 30 years). After 30 years
pumping rate from wells reduced since pumping rate of dry cells is eliminated by the
MODFLOW. In ordinary year total groundwater outflow (pumping + outflow at
boundary) exceeds the groundwater inflow (recharge + inflow from boundary) and is
compensated by the groundwater storage release, declining groundwater levels. In year
with 1/50years rain, the groundwater inflow exceeds temporarily the groundwater
outflow and increases the groundwater storage, recovering groundwater levels.

12 - 28
 Chapter 12 Groundwater Modelling

2) Results of Prediction

As shown in previous section, predicted results in Stampriet area are different from these
of other area. Therefore, the results were evaluated for Stampriet area and remaining
other area respectively. Results of the model prediction in the Stampriet Basin were
summarized as bellow:

Relation to Expected Constraints

Area Stampriet area Other area
Constraint Water Balance Economic Water Balance Economic
Aquifer
Case Kala Auob Kala Auob Kala Auob Kala Auob
1 NA NA UD A A A/UD G G
2 NA NA UD A A A G G
3 NA NA UD UD A A G G
4 NA UD UD G A A G G
5 UD A G G A A G G
6 A A G G A A G G

Remarks: G=Good, A=Allowable, UD=Undesirable, NA=Not Allowable

Kala = Kalahari Aquifer, Auob = Auob Aquifer

3) Permissible Yield

Present groundwater abstraction (Case1, Case 2) is acceptable in the Stampriet Basin

except Stampriet area. In this area groundwater is mainly used for stock watering and
domestic purposes. It is considered that groundwater use will not increase remarkably. In
Tugela area declines of groundwater level are slightly large with present groundwater
pumpage. Careful monitoring is necessary.

In Stampriet area, the Kalahari Aquifer will dry up in near future, if present groundwater
abstraction is kept. From the above results, pumping plan of Case 5 (reducing irrigation
use to 50%) and Case 6 (reducing irrigation use to 0%) are acceptable in Stampriet area.
Case 4 (reducing irrigation use to 70%) is not allowable since the Kalahari Aquifer will
dry up within 80 years. To prevent the dry-up of aquifer, groundwater pumping for
irrigation use is at least reduced to 50% of that in 1999, which is almost the same as the
irrigation use in 1992.

12 - 29
 Chapter 12 Groundwater Modelling

4) Steps to Permissible Yield

(1) Schedule of Protective Measures

As protective measures differ for each area, their specific schedule needs to be suitably
set respectively for each area. Based on the experience in Japan, the following stages
can be considered in general (Shibasaki et al., 1995).

i) Research Stage

A preliminary reconnaissance based on available information is carried out in order to

learn the general conditions of the basin. For all the boreholes in the basin, registration,
arrangement of well inventory, measurement of groundwater level and pumping rate,
and submission of its report should be ensured.

Base on these data, a preliminary review is carried out, minimum level of facilities for
monitoring is set up, and the permissible (yield) critical groundwater level is tentatively
worked out. In this stage, construction of new boreholes is strictly regulated.

The period for the first stage is desirably within a few years, even for areas without
substantial data.

ii) Observation and Arrangement Stage

Based on the permissible critical groundwater level determined initially in the first
stage for each area, groundwater level and extraction are observed, and groundwater
pumpage for each area is reduced. The deduction of groundwater pumpage is closely
related to the socio-economic factors as well as with the general utilization of water
resources including the possibility of development of alternative water sources such as
dams.

In the second stage, the monitoring system is further improved, and the data are
comprehensively analyzed. Then for each area, each municipal unit (cities and towns),
of for smaller zones, permissible yield are determined. For this second stage, a period
of five years would be practical.

iii) Stage for Intensive Enforcement Measures

As long as measures are strictly enforced in the second stage, the objectives to stop
groundwater problems may be considered to be mostly achieved. If groundwater level
still cannot be regained, a third stage may be considered in which intensified restriction
on pumpage, intensive alteration of water sources, and measures for artificial recharge

12 - 30
 Chapter 12 Groundwater Modelling

should be resorted to.

For the realization of effective groundwater basin management, besides technological

improvements, legal and social issues such as the improvement of groundwater
utilization system should be considered.

(2) Steps to Permissible Yield in the Study Area

For the study area, groundwater management may be in a transition period from the
first stage to the second stage. For the present, observation of groundwater levels
should be continued using the monitoring system that made by this study. Improving
the groundwater model based on the observation results, tentative permissible yield in
this study should be revised precisely.

Urgent measures to be taken are preventing the increase of groundwater abstraction. In

order to achieve this, it is important to inform farmers of the present situation in the
basin, and to make farmers understand the problems that will occur with present
groundwater use. Therefore the farmers may be cooperative to reduce the pumpage.

Meanwhile measures reducing pumping rate to the permissible yield should be

prepared. It is considered to apply permits strictly to irrigation use. Charging system
for groundwater use is worth considering.

12.8.2 Future Improvement in Modelling

The modelling of the Stampriet Artesian Basin is based on the knowledge at present.
Therefore, modification of the model will be expected in future observations and
investigations.

There is no observation well of the Kalahari Aquifer in Stampriet area, where serious
problems are estimated to occur in near future. New Kalahari observation wells are
necessary to monitor the variation of groundwater level and to improve the model
accuracy.

The calculation model in this study was unstable at the bottom area of the model and there
were incomprehensible head differences. Future improvement is necessary to enhance
the model accuracy.

Groundwater pumping rate was estimated by using many assumptions. The reports from
permit holders are incomplete. It is necessary to report monthly abstraction properly. For
domestic and stock watering use, there are no data on the variation of pumping rate. It is
recommended to select sample wells and to install the flow meter.

12 - 31
 Chapter 12 Groundwater Modelling

The recharge rate is unknown at present. A basic hydrologic and hydrochemical study is
necessary to estimate accurate recharge rate.

The model calibration should be improved based on the future studies above mentioned
and the observation results of the monitoring system. It is recommended to model the
groundwater quality, investigating the influence of “salt block”.

12 - 32