Hydrol. Earth Syst. Sci.

, 14, 2343–2354, 2010

[Link]/14/2343/2010/ Hydrology and
doi:10.5194/hess-14-2343-2010 Earth System
© Author(s) 2010. CC Attribution 3.0 License. Sciences

Modeling for transboundary water resources planning and

allocation: the case of Southern Africa
D. Juı́zo1 and R. Lidén2
1 Universidade Eduardo Mondlane, Faculdade de Engenharia, Av. de Moçambique, CP 257, Km 1.5, Maputo, Moçambique
2 Lund University, Department of Water Resources Engineering, P.O. Box 118, 22100 Lund, Sweden
Received: 9 January 2008 – Published in Hydrol. Earth Syst. Sci. Discuss.: 19 February 2008
Revised: 6 October 2010 – Accepted: 15 November 2010 – Published: 26 November 2010

Abstract. International water resources agreements for allocation and prioritization principles for the specific river
transboundary rivers in southern Africa are generally basin and the model developer’s experience and integrity are
founded in system analysis models for water planning and al- more important factors to find the optimal and equitable al-
location. The Water Resources Yield Model (WRYM) devel- location.
oped in South Africa has so far been the only model applied
in official joint water resources studies aimed to form water-
sharing agreements. The continuous discussion around the 1 Introduction
model performance and growing distress over it being South
African, where it was originally developed, while South Water resources systems are generally complex with both
Africa is one of the interested parties in the process, results qualitative and quantitative factors governing water resources
in an increased controversy over the system analysis results availability (Raju and Pillai, 1999). In southern Africa the
that are often only meant to guide in selecting the options high climatic variability further adds the complexity. Both
for water resources management in a given set of scenarios. the large seasonal variations and the interannual variations,
The objective of this study was therefore to assess the model with long dry spells, make infrastructural solutions necessary
performance of two other models; WAFLEX and WEAP21 to assure water availability at all times. The role of reservoirs
in the Umbeluzi River Basin system where the WRYM was in water resources management in Southern Africa is high-
previously applied as part of a Joint River Basin Study. A lighted by van der Zaag and Bolding (2005).
set of basin development scenarios was equally tested in the However, building infrastructures will not on its own en-
three models and the results compared. The results show that sure water availability and equitable water allocation be-
the three models all are possible tools for system analysis tween countries and users of a particular river basin. The
of river basins in southern Africa, although the structure and system of natural and regulated water resources should be
complexity of the models are different. The obtained level utilized and managed appropriately for sustainable use of
of satisfaction for specific water users could, however, vary the water resources. In southern Africa, decision on wa-
depending on which model was used, which causes uncer- ter resources permit allocation between different users and
tainties. The reason for the diverse results is the structurally between different riparian countries is therefore commonly
different ways of describing allocation and prioritization of based on system analysis tools. System analysis tools are ad-
water in the three models. However, the large degrees of free- equate to evaluate and propose the best management strate-
dom in all system models cause even larger uncertainty in the gies towards maximization of benefits for a given number of
results since the model developer can, intentionally or unin- users under given objective functions in the catchment. As
tentionally, direct the results to favor certain water user. The pointed out by Dent (2001) “... a model is a tool to help or-
conclusion of this study is therefore that the choice of model ganize a negotiation or learning process in which its primary
does not per se affect the decision of best water allocation function is to provide a framework for thinking by enabling
and infrastructure layout of a shared river basin. The chosen participants to make their implicit assumptions explicit in a
systematic manner”. Despite being a simplification of a com-
plex water resources system, a system analysis model gives
Correspondence to: D. Juı́zo the opportunity for decision makers to know beforehand the
(juizo@[Link]) consequence of a chosen management option or scenario,

Published by Copernicus Publications on behalf of the European Geosciences Union.

2344 D. Juı́zo and R. Lidén: Modeling for transboundary water resources planning and allocation

though, within the limits imposed by the uncertainty in data 2 Material and methods
and gaps in the knowledge of the processes involved in natu-
ral systems. These tools are therefore important instruments Because the WRYM is the preferred model tool for sys-
for authorities and governments to adopt policies for water tem analysis of international river basins in the SADC re-
resources management in both national and transboundary gion (Carmo Vaz and van der Zaag, 2003; van der Zaag and
river basins. However, because the stakeholders or countries Carmo Vaz, 2003) a comparative analysis of model perfor-
objectives often are conflicting it is necessary that the model mance was conducted through applying two other models
or software used is considered unbiased. in a river basin where the WRYM had been applied. As a
In southern Africa the countries have not yet agreed on the case study it was chosen to do the assessment in the Um-
set of decision support tools that should apply in the water beluzi River in which the WRYM has previously been set
allocation process. The experience from the already imple- up as part of the Joint Umbeluzi River Basin Study (JU-
mented joint system analysis studies in Umbeluzi and Inco- RBS) that aimed to prepare the baseline for a new water re-
mati rivers in southeast Africa (Consultec and BKS Acres, sources sharing agreement between Swaziland and Mozam-
2000; SWECO and Associates, 2005) has shown preference bique (SWECO and Associates, 2005). The Umbeluzi River
to the Water Resources Yield Model (WRYM) as the tool basin was chosen because it is a relatively simple water re-
adopted for system analysis. The experience from these stud- sources system which is faced with some of the core chal-
ies, however, shows that the results obtained from WRYM lenges of transboundary rivers in southern Africa: high cli-
are not easily understood by the stakeholders, and govern- matic variability, scarce water resources, upstream versus
ment representatives of different countries bear some suspi- downstream users, large-scale versus small-scale users and
cion about the results from the system analysis. The lack significant environmental flow requirements. The challenges
of trust in the system analysis tool has prevented a smooth for water resources management in the Umbeluzi River basin
negotiation on transboundary water resources allocation, fo- have previously been studied by Juizo et al. (2006).
cusing the discussion on the nature and properties of the tools The basis for the study was that the different models would
rather than the strategies that can be applied to improve water apply the same inputs and configuration as was used in the
resources allocation between countries. study by SWECO and Associates (2005). As a first step the
Water resources system analysts are challenged to produce WRYM model runs by SWECO and Associates (2005) were
a system analysis model that considers both the legal frame- reproduced to simulate a set of different development sce-
work and the true behavior in the system. Stakeholders of- narios for the Umbeluzi River basin. Secondly a spreadsheet
ten share water resources at very local level and rely mostly based system analysis model, WAFLEX, and a correspond-
on customary principles embodied in local level institutions ing GIS based software, WEAP21, were set up for the Um-
(van der Zaag and Bolding, 2005). Integrating small stake- beluzi River and applied for the same development scenarios.
holders in a large-scale river basin management thus needs The WAFLEX model has been used previously in analyzing
harmonization of local and regional interests. In the case a other river basins in SADC region, namely the Incomati, Ma-
water release in a large basin targets the most downstream puto and Save (Nkomo and van der Zaag, 2004; Sengo et al.,
user, e.g. environment flow to the estuary, all the local wa- 2005; de Groot and Oosterwijk, 2006; Symphorian et al.,
ter users along the river must have a corresponding objec- 2003). While the WEAP21 was used in the Olifants river a
tive not to use this water, otherwise the reserved water will tributary to Limpopo river (Lévite and Sally, 2002; Le Roy,
gradually be used by the local users starting with the most 2005).
upstream community. This situation is even more difficult in
transboundary rivers. In such a situation the question is: how 2.1 The Umbeluzi River basin
complex should a system analysis model be to handle this
scale problem but at the same time be sufficiently transpar- The headwater of the Umbeluzi River is located in Swazi-
ent and uncomplicated to give stakeholders information for land close to its western border with South Africa (Fig. 1).
setting and accepting allocation criteria? The river flows in an easterly direction and discharges into
This study examines the role that decision support tools the Indian Ocean via the Espirito Santos estuary south of
play in the agreement process by evaluating three differ- Maputo City in Mozambique. The total catchment area of
ent system analysis model packages for water allocation for the Umbeluzi River basin is 5400 km2 (SWECO and Asso-
the same river basin, the Umbeluzi River in Swaziland and ciates, 2005). 40% of the area is in Mozambique, 58% in
Mozambique. The objective is to compare these models in Swaziland and only 2% in South Africa. Two major tribu-
terms of complexity, reliability of results, transparency and taries join the main river, the White Umbeluzi in Swaziland
to assess whether the model selection may affect the deci- and the Movene in Mozambique.
sion of best water allocation and infrastructure layout of a The altitude increases from the sea level to almost
shared river basin. 2000 m a.s.l. in the western part. Rainfall varies from
500 mm/year in the lower parts to 1500 mm/year in the
mountainous part. The basin experiences two distinct

Hydrol. Earth Syst. Sci., 14, 2343–2354, 2010 [Link]/14/2343/2010/

 D. Juı́zo and R. Lidén: Modeling for transboundary water resources planning and allocation 2345

Table 1. Water demands in Umbeluzi River (MCM/yr). Source:

SWECO and Associates (2005).

Swaziland Mozambique
2005 2025 2005 2025
Irrigation 229 292 17 39
Urban 12 21 75 182
Other 12 21 5 32
Total 253 334 97 252

diversion channels. Scattered water users have been aggre-

gated according to their location. Return flows from irriga-
tion areas have been included for the large-scale schemes.

2.2 System analysis models

Water allocation in a river basin is usually a two step process

as given in Fig. 3. In the first step (A) hydrological model-
Figure 1 ing is applied to simulate the distributed rainfall and runoff
Fig.
[Link]
The UmbeluziRiverRiver
basinbasin.
processes in the catchment. This model is normally cali-
brated against observed flow sequences at selected stations
along the basin. Given that, in most cases, observed runoff
seasons; the rainy season from November to April and the is already influenced by water uses the hydrological model
dry season between May and October. Two major dams are is used to obtain naturalized or virgin flows of the catchment
Subbasin U14

located in the
UMBELUZI basin. The Mnjoli Dam, with total capacity
RIVER that represent the natural runoff conditions.
of 152 million m3 was built in 1978 with purpose to secure
Subbasin U15
The naturalization of runoff is done by removing the
ATER RESOURCES
water for the sugarSYSTEMcane estates in eastern Swaziland. The known historic water uses from the model. For the Umbeluzi
Movene
Dam site
Pequenos Libombos Dam in Mozambique, with total capac- these values were obtained from the PITMAN rainfall-runoff Farm Small-scale
ity of 385 million m3 , was constructed in 1987 mainly to se- E14 model set up by SWECO and Associates (2005). The sec- dams farming

Subbasin U13
cure the urban water supply for Maputo City. The intake and
Sugar Cane
Factory
Large-scale
ond level (B) of the system analysis concerns testing devel-
Sugar Cane Irrigation
Subbasin U02-04 Subbasin U05 Small-scale
water treatment plant for Maputo City is located some kilo- Movene
opment scenarios of river basin development and water use.
farming
Subbasin U17

meters downstream of the Pequenos Libombos and the dam The model
E12 itself uses input of naturalized flow, river chan-
Farm
dams

is therefore constantly releasing a minimum flow to allow for Pequenosnels, dam characteristics, environmental flow requirements
Release Return flows
To
GS4 water supply.
Mjoli In addition, a small dam in theGS32
GS20 upper basin
E10 in
Libombos
and scenarios ofestuary
water use and afforestation. Different mod-
Spill
Swaziland, the Hawane (2.75 million m3 ), supplies the capi- E395els approach water allocation differently but are all based on
Spill

buluzi Black Mbuluzi Black Mbuluzi Umbeluzi

tal Mbabane with fresh water. water balance in a network of nodes symbolizing the physical
Release
Maputo
The largest water user in the Umbeluzi basin is irri- components of a river basin and links representing conduits Water
supply

gation as can be deduced from the water demands (Ta-

Small-scale
farming
Subbasin U06 of water between nodes (Wang and Hipel, 2003). Alloca-
Subbasin U10 Subbasin U11 Subbasin U12 Subbasin U16
White
ble 1). The sugar cane estates in eastern Swaziland
Mbuluzi
Small-scale tion principles and priorities for different uses are normally
farming Small-scale
stand for GS10
more than 70% of the present water demand. set in the models based on national and international policies farming

The total estimated present water demand for surface wa-

SWAZILAND
Subbasin U08-09 and, in the transboundary case, on water sharing agreements
ter is 350 million m3 /year but is forecasted to increase if existing.
MOZAMBIQUE
to 586 million m3 /year by the year 2025. The avail-
able water, under natural conditions, is estimated to be 2.2.1 Water Resources Yield Model (WRYM)
535 million m3 /year (SWECO and Associates, 2005). The
Subbasin U07

two countries have a number of small-scale users distributed WRYM was developed in South Africa by the BKS consult-
Figure
in the2catchment
Simplifiedand Umbeluzi River
because basin
of the system
water model
scarcity schematic
many pro- ing firm based on the Canadian ACRES Reservoir Simula-
posals exist to build storage infrastructure. tion Program (Mackenzie and van Rooyen, 2003). It relies
A simplified schematic of the Umbeluzi water resources on a solver that optimizes the water allocation in a river sys-
system, for the present infrastructure development, is shown tem based on a set of penalties for storage, channels and
in Fig. 2. In the schematic is shown the subbasins generating demands at various nodes and links. It minimizes a cost
natural runoff, main water users, channels links, dams and function based on storage and allocation deficit cost but also

[Link]/14/2343/2010/ Hydrol. Earth Syst. Sci., 14, 2343–2354, 2010

24
 Figure 1
The Umbeluzi River basin

2346 D. Juı́zo and R. Lidén: Modeling for transboundary water resources planning and allocation

Subbasin U14

UMBELUZI RIVER Subbasin U15

WATER RESOURCES SYSTEM Movene

Dam site

Farm Small-scale
E14 dams farming

Subbasin U13
Large-scale
Sugar Cane Sugar Cane Irrigation
Factory
Subbasin U02-04 Subbasin U05 Small-scale
farming Movene
Subbasin U17
Water supply
Farm
Mbabane
dams E12

Release Return flows

Pequenos
Hawane Libombos To
GS4 Mjoli GS20 GS32 E10
Spill estuary
Spill
Subbasin U01 E395
Black Mbuluzi Black Mbuluzi Black Mbuluzi Umbeluzi
Release
Maputo
Water
supply
Small-scale
farming Subbasin U10 Subbasin U11 Subbasin U12 Subbasin U16
Subbasin U06
White
Legend
Small-scale Mbuluzi Small-scale
farming
Subcatchment farming
GS10
Reservoir
Subbasin U08-09 SWAZILAND
Irrigation scheme
River flow MOZAMBIQUE
Water outtake
GS3
Runoff station
Border Subbasin U07

Figure
Fig. 2. Simplified Umbeluzi River basin 2system
Simplified
model Umbeluzi
schematic. River basin system model schematic

Rainfall Data &

Calibration parameters

Historical
development &
Observed flow for
Hydrological
rainfall/runoff model
A
calibration

Agreements/Conventions/
24
B
Law allocation strategies

Naturalized or virgin runoff Water balances

Rainfall and evaporation Dam storage trajectory

reservoir data System Yield

System Analysis Flow regime compliance
Dam and river characteristics
Model
Water use scenarios Environmental requirements

Environmental flow requirements met?

User satisfaction

Fig. 3. Modeling
Figurefor
3 water resources
Modeling allocation
for water in international
resources allocation in rivers. (A) and
international (B) Adenote
rivers. the two
& B denote thesteps normally
two steps usedused
normally to obtain the
information for water allocation.
to obtain the information for water allocation

between different forms of storage in a catchment (Macken- nodes and across the catchment. Grossmann et al. (1995)
zie and van Rooyen, 1999). At the core of this model is a gives details of the use of network theory in linear program-
penalty structure used in the decision about storing or allo- ming to solve water allocation problems of water resources
cating water in the system. PenaltiesD are assigned to links systems.
supplying water
FRC
reservoirs and other sources to users and to In the example shown in Table 2, the storage in the dam
distinguish between different forms of storage in reservoirs
D has been divided into four zones. The upper zone is above
C A
URC M
Hydrol. Earth Syst. Sci., 14, 2343–2354, 2010 W [Link]/14/2343/2010/
B A
DSC L
A L
 reservoir data
S
Dam and river characteristics
Water use scenarios
D. Juı́zo and R. Lidén: Modeling for transboundary water resources planning and allocation 2347
Environmental flow requirements

Table 2. Example of reservoir zones and penalty structure in the Table 3. Example of link penalty in the WRYM. Source: WRYM
WRYM. Source: WRYM User Guide – 4.1.1. User Guide – 4.1.1.

Storage Penalty Elevation Figure 3 Modeling

Rangefor waterPenalty
of flow resources allocation in internati
to obtain the information for water allocation
100% 1000 1548.96 0 to TD 250
90% 2 1547.99 TD 0
0% 20 1533.00 TD to infinity 20
0% 10 000 1533.00
Bottom 1521.95

D
the full supply level (FSL) and has therefore a very high
penalty. The second zone is between FSL and the 90% ca- FRC
D
pacity. This zone has a penalty of 2 units. The third zone rep- C A
resents the water between the 90% level and the dead storage
level (DSL). Water in this zone has a penalty of 20 units and URC M
represents the main working storage for the reservoir. The W
B A
water between the DSL and the bottom of the dam has a rel-
atively very high value of 10 000 units. Thus the model will DSC L
never draw water from this zone to meet the downstream de- A L
mand. The four penalties used in the example have the effect
of restricting the working storage of the dam to the second Fig. 4. Storage forms in WAFLEX.
and third zones. In the model there are possibilities to have Figure 4 Storage forms in WAFLEX
more zones than shown in the example.
Table 3 shows a typical penalty structure for a channel with The storage is divided in four zones. If the level of stor-
a specified target draft (TD). Failing to meet the target draft age is under the Dead Storage Curve (DSC), no water will
results in a penalty of 250 units. If the target draft is supplied, be released. If the level of storage is between the DSC and
there is a zero penalty and if it is exceeded, the excess flow the Utility Rule Curve (URC) the release will be rationed as
results in a penalty of 20 units. For further explanation of the a factor of the demand. This is also known as hedging rules
penalty structure of the WRYM model, see the User Guide – (Draper and Lund, 2004). The reduction factor which is set
4.1.1 (Mackenzie and van Rooyen, 1999). by the model developer is triggered at the URC threshold and
In WRYM the network is analyzed for each time period is fixed throughout the deficit period. If the level of storage is
and solved with the selected penalty structures. The net- between URC and Flood Rule Curve (FRC) the demands will
work solver will minimize the penalties for each time step by be fully satisfied. At last, if the level of storage is above FRC
choosing the best allocation of water to the different users. the reservoir spills. In the WAFLEX, subsequent to reservoir
It will also choose the most attractive route (i.e. minimum release, the water is available to satisfy the demands in first
penalty) for transferring the water from the storage zones to come first serve order. It is, however, also possible to build
the demand centers. more complex models by developing Macros that are rou-
The WRYM is widely used in southern Africa and is the tines designed to impose a designed sequence of allocating
chosen tool by the South African Department of Water Af- water to downstream users.
fairs and Forestry for system analysis of all the river basins The network built up in the spreadsheet consists of a sup-
in South Africa (Carmo Vaz and van der Zaag, 2003). ply and a demand module. The demand module calculates
in upstream direction to determine the demands for the reser-
2.2.2 Water Allocation Flow model in Excel (WAFLEX) voir release. The supply module calculates in downstream
direction of the flow. Reservoir releases are calculated using
The WAFLEX model was first introduced by Savenije (1995)
macros in which the operation rule curves are used and the
for water resources simulation. It is a simple water balance
demands are determined by the demand module (Sympho-
model that uses the basic spreadsheet capacity to simulate
rian et al., 2003). This computation also includes reservoir
complex systems. It is easy to build and very transparent as
losses through evaporation.
everything takes place in EXCEL spreadsheets. Water allo-
cation strategies of this model are based on a comparison of
demand and availability. Reservoir operation is based on the
division of the storage in different forms as given in Fig. 4.

[Link]/14/2343/2010/ Hydrol. Earth Syst. Sci., 14, 2343–2354, 2010

2348 D. Juı́zo and R. Lidén: Modeling for transboundary water resources planning and allocation

2.2.3 Water Evaluation and Planning system In the case where reservoirs are located in tributaries feed-
(WEAP21) ing to a downstream canal leading to a user, the WRYM
penalty structure allows for reservoirs to conjunctively oper-
WEAP21 was developed by Stockholm Environment Insti- ate to satisfy these demands in the catchment. The reservoirs
tute (SEI) and is widely used as a system analysis model with lowest penalties will be mobilized first to meet deficits
(Sieber et al., 2002). WEAP21 operates on the basic prin- of supply by another concurrent reservoir. WEAP21 on the
ciple of water balancing, accounting between supply and de- other hand is mostly based on comparative value of the reser-
mand at various system nodes. WEAP21 also has a GIS func- voir and the user in question. Normally, users should have
tion allowing the user to visualize the network on the screen high priority value associated with their demands in order to
and interactively modify or update the inputs of the model. profit from water in storage in a given reservoir. The order in
In recent years this software has become popular in various which dams will be mobilized to meet the different deficits
research related to climate adaptation within UN organiza- in downstream nodes is given by the priority in filling that is
tions and others especially in research on effect of climate assigned to the reservoirs.
change on agriculture (Rosenzweig et al., 2004; Yates et al., Table 4 gives a comparison of the main features and dif-
2005; Joyce et al., 2006). ferences of the three system analysis models used in the Um-
The reservoir operation is similar to that of WAFLEX. beluzi River basin.
During the rationing period only a fraction of the storage is
available for release. However, an added complexity in this 2.3 System configurations
model is the possibility of using a priority rule to give prefer-
ence for use of certain reservoirs for satisfying the demands. The schematic in Fig. 2 together with identified sites for new
A reservoir with lower priority will empty faster than those dams and expected future water outtakes was used as basis
with higher priorities. Furthermore, in the WEAP21 model for all models. Naturalized monthly inflows from 1925 to
there is a possibility of controlling the water that is supplied 1999 for all the subbasins of the Umbeluzi River, produced
to different users once it is released from the reservoir. This is by SWECO and Associates (2005), were used as input to
done through a priority rule that set preferences to higher val- all three models. As a basis for the water allocation, priori-
ued uses such as urban water supply and environmental wa- ties were generally set according to the policies in southern
ter requirements as compared to other water uses; the same Africa (SADC, 2000). All other inputs values such as reser-
range of priority values applies. In essence the mix of the set voir net-evaporation, precipitation and demands for different
of supply and storage priority assigned to the network will scenarios are equally based on the study by SWECO and As-
drive the model and the water allocation. sociates (2005).
SWECO and Associates (2005) used a calibration method-
2.2.4 Comparison of the models ology to set the penalty structure of the model. The basis
for the calibration was the knowledge of present water man-
All three models are explicit system analysis models and de- agement in the river basin and the observed runoff records.
pend on input of distributed naturalized inflow calculated by Realistic penalties could thus be found through an iterative
other methods. process where simulated and observed river runoff was com-
The main differences of the three models are how the de- pared. In a similar manner the priorities for reservoirs and
cision to allocate water from storage is done. The WRYM is water uses were set in the WAFLEX and the WEAP21 (Ta-
flexible in the way it approaches water allocation to different ble 5).
users in a catchment. The numerical optimization is essen- Table 5 shows that to enable a description of the historic
tially limited to the users downstream each reservoir. How- and present water management in the Umbeluzi River basin
ever, in case there are complex interconnections of reser- the model parameters were set to values that partly contradict
voirs in the basin these can be incorporated in the model. the international and national water policies. For instance,
The WEAP21 essentially follows a priority rule of allocation. for failing to meet the demands of large-scale scale irrigation
The system water allocation is driven by water user priority in Swaziland the WRYM penalties are set even higher than
versus reservoir priority allowing downstream users to bene- failing to meet urban water supply demands. On the other
fit from any given upstream reservoir in the catchment. Also hand, the WRYM has included an equally high penalty for
the allocation between users along the system will follow a failing to meet the set minimum border flow determined by
priority rule given high preference to high value users. In the 1976 bi-lateral agreement between the countries (Juizo et
WAFLEX water is allocated to downstream nodes connected al., 2006) aiming at supporting the water supply of Maputo
to reservoirs. In order to enable an upstream reservoir to sup- City.
port a downstream it is necessary to build auxiliary functions In WAFLEX a central demand and supply module was in-
that can only capture information from the time step before troduced, which enabled to put the abstraction nodes down-
the one in computation. In other words it is more a compen- stream of the reservoirs in the order of preference. By us-
sation function than a direct support between the reservoirs. ing this module, irrigation was prioritized downstream of the

Hydrol. Earth Syst. Sci., 14, 2343–2354, 2010 [Link]/14/2343/2010/

D. Juı́zo and R. Lidén: Modeling for transboundary water resources planning and allocation 2349

Table 4. Summarized comparison of the three system analysis models used in the study. All models run on a monthly time step.

Model Allocation methodology Optimization User friendliness Transparency

methodology
WRYM Minimizes system penalty based on A numerical Software available at Because of the complex
penalties per unit water set for releases routine is run DWAF, South penalty structures and
from reservoirs and for failing to meet optimization Africa. system optimization
minimum flow requirements or for each time Difficult to use routine transparency is
demands. step without thorough limited.
Higher penalties are given for failing training.
to meet prioritized water use. Different
penalties for same user type can be
given.
WAFLEX Releases from reservoirs are reduced No Uses EXCEL For basic use the
according to storage rates. optimization spreadsheets. spreadsheet
When released, as a base rule: first Lecture notes on the methodology makes it
comes, first served. However, development of the transparent. However,
possibilities to include routines to models are available for non skilled users, if
prioritize types of water uses. from IHE-Delft. special Macro functions
Easy to use. are applied it limits
transparency.
WEAP21 Priorities are set for which reservoirs No GIS based interface Fairly transparent
to draw water from. Amount of water optimization which gives good through its GIS
for releases is reduced depending on overview of the river interface and straight
storage situation allowing for rationing system. forward priority system.
in times of deficit. Software available
When released allocation downstream from SEI webpage.
is made according to given priorities Easy to use.
for each type of water use.
For equal priorities upstream users are
provided first.

Table 5. Configuration used for the three system models applied for the Umbeluzi River basin. PQL = Pequenos Libombos, IFR = Instream
Flow Requirements, EFR = Estuarine Flow Requirements.

WRYM WAFLEX WEAP21

Reservoir penalties for Penalties for failing Penalties for failing Allocation Allocation Allocation Dam User priorities in User priorities in
working storage to meet demand to meet min channel priorities priorities priorities priorities Swaziland Mozambique
flows downstream downstream downstream PQL
Hawane Mnjoli
Hawane 40 Water supply 350 Border flow 400 IFR 1 IFR 1 IFR 1 Hawane 2 Water supply 1 Water supply 5
Mnjoli 10 Industry 340 Releases from Water supply 2 Irrigation 2 Water supply 2 Mnjoli 4 IFR 1 IFR 5
PQL 30 Livestock 300 Hawane 250 Water supply 3 Irrigation 3 PQL 9 Industry 3 Industry 6
MBF 30 Large-scale Releases from Industry 4 Industry 4 MBF 5 Livestock 3 Livestock 6
Isilele 10 Irrigation 400 Mnjoli 250 Livestock 5 Isilele 5 Irrigation 4 Irrigation 7
Farm Dam 10 Small-scale Releases from Farm Dam 5
Movene 20 Irrigation 100 PQL 250 Movene 10
EFR 250
Afforestation 400

Mnjoli Dam in favor for domestic water supply, while the set for the Mozambique users and the Pequenos Libombos
opposite was done downstream of Pequenos Libombos (Ta- dam (Table 5) give the effect that no water from the Swazi-
ble 5). land reservoirs is allocated to Mozambican users.
The WEAP21 parameters were also set according to the It should thus be noted that the internationally and nation-
present praxis that the dams as first priority support the local ally adopted policies for allocation (e.g. SADC 2000) are not
users. The relatively lower priorities (higher penalty values) strictly followed by the system analysis model, despite that

[Link]/14/2343/2010/ Hydrol. Earth Syst. Sci., 14, 2343–2354, 2010

2350 D. Juı́zo and R. Lidén: Modeling for transboundary water resources planning and allocation

Table 6. Scenario development for Umbeluzi River basin.

Scenario Description Comment

1 – 2005 for all users demands including environmental requirements; Assessment of current situation.
– 2005 infrastructure (Dams at Hawane, Mnjoli and Pequenos Libombos).
3 – 2025 demand for all users including environmental requirements; Impact of additional infrastructure in
meeting future demands.
– Additional infrastructure, dams at Mbuluzi Falls, Isilele and Movene.
4 – 2005 irrigation demand in both countries; The existing IncoMaputo Agreement
– 2005 industrial demand Maputo proposes that countries should reorient
– 2025 demand for domestic water supply. there high demand uses to other system
– Environmental requirements; with available water. In this scenario
– Additional infrastructure, dams at Mbuluzi Falls, Isilele and system performance is tested for Movene.
increased demands except for
irrigation that is assumed to be
developed in the Maputo River basin.

they are stated as a basis for the modeling. The reason is that Table 7. Level of satisfaction for the main users.
water resources have not, and are presently not, allocated ac-
cording to these policies on the catchment scale in the Um- WRYM WAFLEX WEAP21
beluzi River. As a choice the model developer, in this case
Scenario 1 (2005 demand level)
SWECO and Associates (2005) and the authors, thus chose
to describe the river systems as it works today rather than Mbabane water supply 100% 99% 98%
how it would work if the SADC Protocol (SADC, 2000) was Large-scale irrigation 100% 95% 99%
strictly followed. Swaziland
The three models were developed to simulate three sce- Small-scale irrigation 97% 94% 97%
nario developments in the basin (Table 6) targeting mainly Mozambique
three main demands in the system (i) urban water supply; Maputo Water supply 99% 97% 98%
(ii) irrigation development and (iii) environmental flow re-
Scenario 3 (2025 demand level, new infrastructure)
leases to the estuary for present (2005) and future condi-
tions (2025). Scenario 1 uses the present infrastructure, while Mbabane water supply 98% 100% 100%
the other two scenarios include development of new dams. Large-scale irrigation 88% 86% 88%
The specific demands for all present and future users were Swaziland
Small-scale irrigation 48% 41% 47%
set equal in the three system analysis models.
Mozambique
Maputo Water supply 69% 69% 68%

3 Results Scenario 4 (2005 demand for irrigation, 2025

demand for water supply,)
The results of the three models were compared in terms of Mbabane water supply 100% 100% 98%
level of satisfaction for different users and the dam behavior. Large-scale irrigation 88% 89% 88%
The level of satisfaction is the ratio between supplied water Swaziland
and demand for the whole period of modeling. Small-scale irrigation 70% 57% 62%
The models’ natural runoff input covered a period of Mozambique
75 years (1925–1999). Through the long series, the models Maputo Water supply 84% 76% 77%
can thus provide estimates of the available water resources
for different demand and infrastructure scenarios taking into
account the climatic variability of southern Africa.
Despite differences in the level of satisfaction that can Maputo City). It shows that the additional storage capac-
be seen for specific demands and scenarios, all models per- ities introduced are not sufficient to balance the increased
form similarly (Table 7). As expected when comparing 2005 water demand (Table 1) on the catchment scale. Only for
and 2025 (Scenarios 1 and 3) the level of satisfaction de- small users in the upper Umbeluzi and in the tributaries, the
creases for the large users (e.g. large-scale irrigation and additional reservoir storages compensate for the increased

Hydrol. Earth Syst. Sci., 14, 2343–2354, 2010 [Link]/14/2343/2010/

D. Juı́zo and R. Lidén: Modeling for transboundary water resources planning and allocation 2351

demand (e.g. for Mbabane water supply). Even when irriga- Table 8. Total level of satisfaction for all users in the Umbeluzi
tion development is restricted and new infrastructure is put River.
in place (Scenario 4), the level of satisfaction for urban water
supply to Maputo does not reach acceptable levels. WRYM WAFLEX WEAP21
A comparison of the dynamics of the main reservoirs (ex-
Scenario 1 99% 97% 99%
emplified for Mnjoli Dam in Fig. 5) confirms that the three
Scenario 3 84% 79% 79%
models perform similarly. Scenario 4 88% 85% 85%
A more detailed examination of the results of the three
models shows that the WRYM model has a totally larger
level of satisfaction for all users, although the difference is
not large (Table 8). For specific users the models may also
give different results. An example is the small-scale irriga- in Mozambique. So in general terms all models allocated
tion that shows significant differences in satisfaction levels water as intended. But the level of satisfaction calculated
for the three models (Table 7). Since satisfaction levels are with the three models for specific users could vary with up to
closely linked to the production yield the different model re- 10–20% units, especially for smaller users. Such differences
sults may therefore affect the judged feasibility of these irri- may give a significant difference in the long-term feasibility
gation schemes. Also the satisfaction levels for Maputo City of the applied agriculture. On the other hand, the application
show varying levels, although all three models predict lev- of the three models showed that for large-scale users all three
els far below what is acceptable for urban supply for both models gave similar satisfaction levels as seen by Table 7.
Scenarios 3 and 4. The differences in satisfaction levels due to the different
Another way of expressing how the models performed structure of the models are, however, completely overridden
in meeting demands is to calculate the assurance of supply by the different results that can be obtained through changing
(Wurb, 2005), i.e. for how many months of the total period the allocation priorities in the models. The set up of the three
the demand was fully supplied. All three model results for as- system analysis models for Umbeluzi River illustrated that
surance of supply were similar to those for satisfaction levels. the allocation principles stated by the international and na-
A trend was, however, that assurance of supply is generally tional policies are difficult to apply. In the Umbeluzi case the
lower than satisfaction level for the users located in small model developers set up the prioritization according to how
tributaries with limited upstream storages. the water has been allocated historically. The reason was that
the model developers judged this as the most probable sce-
nario in the future. The Mnjoli Dam was constructed by, and
4 Discussion for, the sugar cane estates in Swaziland and the Pequenos Li-
bombos dam was constructed by the Government of Mozam-
The results of the three system analysis models applied for bique for water supply to Maputo City. History shows that
the Umbeluzi River basin in this study revealed two general it is very difficult to change allocation patterns and, faced
and essential findings: with the task to predict future water resources management,
the model developer chose the most likely allocation fore-
1. Different system analysis models may give differences cast, which is that things will be as they always have been.
in satisfaction levels for specific water users although SWECO and Associates (2005) made a sensitivity analysis
the same governing input variables are used. for the Umbeluzi River basin and found that by changing
2. The degrees of freedom in all three system analysis the penalties in the WRYM model to favor the downstream
models are very large and thus the model developper Maputo water supply the assurance of supply could be in-
has huge power to allocate water although the models creased with 30% for the City and that overall satisfaction
are generally described to follow the internationally ac- levels could be slightly increased for the whole catchment.
cepted allocation principles. This of course led to decreased assurances of supply for the
upstream irrigation schemes. If the prioritizations for water
The differences in satisfaction levels for specific water users user types in Swaziland and Mozambique in the WEAP21
found in this study most probably depend on the different (Table 6) were set equal, the results would have been simi-
ways water allocation and prioritization are structurally han- lar. Then water would have been drawn from the Mnjoli to
dled in the three models. In general terms the prioritization support Maputo City before allocating water to the irrigation
between different users were described similarly in all mod- schemes.
els but the exact levels of satisfaction still differed. In the What is essential is whether the stakeholders realize that
water scarce situation, as described by Scenario 3 where total these choices have been made by the model developer. The
demand is higher than the available water resources, all three system analysis models are generally described to follow the
models gave higher level of satisfaction to the large-scale ir- international praxis of water allocation, and even if the input
rigation in Swaziland compared to all the downstream users data and parameters are reported in detail the stakeholders

[Link]/14/2343/2010/ Hydrol. Earth Syst. Sci., 14, 2343–2354, 2010

2352 D. Juı́zo and R. Lidén: Modeling for transboundary water resources planning and allocation

WAFLEX WRYM WEAP21 Mnjoli Dam

million (m )
200

3
150
Stored volume

100

50

0
0 48 96 144 192 240 288 336 384 432 480 528 576 624 672 720 768 816 864
timestep (months)

Fig. 5. ModelFigure 5 Model for

performance performance for operation
Mnjoli dam Mnjoli damforoperation
Scenariofor
[Link] 3. Reservoir
Reservoir trajectorytrajectory
during 75during 75 simulation.
years of years of
simulation.

normally lack the knowledge to comprehend them. This is transfers for thermo power cooling that require high reliabil-
why the transparency and user friendliness of the models are ity of supply, the WRYM models may be advantageous to the
important factors in the system analysis. If there is no un- WAFLEX and WEAP21.
derstanding or transparency of the system analysis model- The numerical optimization to find the least penalty for
ing, considerable trust must exist between the stakeholders each time step may be the reason why the WRYM gives
and the model developer (Chapman et al., 1995). Such trust totally slightly higher satisfaction levels than the two other
has been developed between the Department of Water Af- models, as seen by Table 8. On the other hand, the sim-
fairs and Forestry and the consultants running the WRYM in pler allocation algorithm used in the WAFLEX model, basi-
South Africa. Despite that the WRYM model is complicated cally upstream to downstream for each water use type, may
to use and has limited transparency it is therefore chosen as be more close to how water is actually allocated in reality in
the preferred tool in South Africa for water resources alloca- the river basins. It may therefore be more close to the actual
tion. In transboundary rivers in southern Africa the same future satisfaction levels taking into account human’s inabil-
trust does not exist between the model developers, which ity to operate a whole river system optimally. The WAFLEX
are often South African, and the other governments acting and the WEAP21 also have an advantage in being more trans-
as stakeholders. parent and user friendly compared to the WRYM.
It may therefore be necessary to review the process of wa- The use of all three models, however, demands thorough
ter resources planning and allocation in transboundary rivers understanding of system analysis. The experience of apply-
in southern Africa. An important step is to agree on the al- ing the three models to the Umbeluzi River is that mistakes
location principles and the modeling procedure before the are easily made when the individual penalties and priorities
models are applied. This goes beyond the general princi- are set. Neither of the models have any quality assurance
ples already agreed in the SADC Protocol. The principles module, which means that it is up to the model developer’s
and procedures may be different depending on the local or experience and knowledge to identify errors and judge the
regional situation for the different rivers. This step also in- reasonability of the model results. The application of the
cludes the choice of model tool, model developer and a pro- three system analysis models to the Umbeluzi River basin
cedure on how to make the modeling transparent to as many further showed that there is not one single solution for the
stakeholders as possible. Capacity building of the stakehold- water allocation. Depending on the parameters set or the
ers is obviously a key factor in this process. model used, different results can be obtained as seen in Ta-
This study may give advice on the choice of system anal- ble 7. None of these results can be said to be more correct
ysis tool. All three models applied to the Umbeluzi Rivers than the other.
have the functions necessary to make a system analysis of
water resources in a river basin in southern Africa. The
WEAP model, by its priority rule based allocation, seems
to give the best possibility to maximize water use in a whole
river basin by allowing reservoirs to support each other in
times of scarcity. For river basin systems that are even 26
more complicated and where water has a high price, e.g. sys-
tems including large hydropower schemes or large interbasin

Hydrol. Earth Syst. Sci., 14, 2343–2354, 2010 [Link]/14/2343/2010/

D. Juı́zo and R. Lidén: Modeling for transboundary water resources planning and allocation 2353

5 Conclusions more important factors to find the optimal and equitable al-
location. This has to be realized by stakeholders and policy
In this study we have investigated the implications of apply- makers in order to improve the water resources planning and
ing three different system analysis models for water alloca- allocation in transboundary rivers in southern Africa. This
tion in an international river basin. The results show that the study also shows the importance of river basin conceptual-
three models, WRYM, WAFLEX and WEAP21, all are ad- ization as a key element in preparation for allocation.
equate tools for system analysis of river basins in southern
Africa, although the structure and complexity of the mod- Acknowledgements. This work received financial support from
SIDA-SAREC. The authors are thankful to all those who provided
els are different. The obtained level of satisfaction for spe-
data for this research. We also wish to thank Stockholm Environ-
cific water users could, however, vary depending on which mental Institute (SEI) for providing a free WEAP21 licence to
model was used, which causes uncertainties if an individ- UEM and DWAF for providing the WRYM software.
ual model is applied directly for studying the feasibility of
water demanding development. The reason for the diverse Edited by: P. van der Zaag
results is the structurally different ways of describing allo-
cation and prioritization of water in the three models. How-
ever, the large degrees of freedom in all system models cause References
even larger uncertainty in the results since the model devel-
Carmo Vaz, A. and van der Zaag, P.: Sharing the Incomati wa-
oper can, intentionally or unintentionally, direct the results
ters: co-operation and competition in the balance: From potential
to favor certain water users. These uncertainties are often conflict to co-operation potential, UNESCO/IHP/WWAP/IHP-
unknown for the stakeholders and policy makers, who use VI/Technical Documents in Hydrology/PC-CP Series/n. 14,
the results of the system analysis models to plan future water 2003.
allocation. Chapman, R. A., Manders, P. T., Scholes, R. J., and Bosch, J. M.:
The water allocation in Umbeluzi River for future develop- Who should get the water? Decision support for water resource
ment scenarios is faced with considerable difficulties as the management, Water Sci. Technol., 32(5–6), 37-43, 1995.
future demands can not be fully met even with an increase Consultec and BKS Acres: Joint Inkomati Basin Study: Phase 2,
in reservoir storages. This study shows that planning of fu- Tripartite Permanent Technical Committee, Kingdom of Swazi-
ture water allocation in this transboundary river is difficult land – Republic of Mozambique – Republic of South Africa,
2000.
through straightforward application of system analysis mod-
de Groot, R. and Oosterwijk, W.: Water Resources Study of the Ma-
els. Prior to modeling, the main stakeholders have to agree puto River Basin: Creating a water resources yield model for the
on the specific allocation principles and the modeling proce- Maputo river basin, [Link]. report, Delft University of Technology
dure to apply for the specific river basin in question. General (TU-Delft), Delft, The Netherlands, 2006.
principles such as the SADC Protocol are not sufficient ba- Dent, M.: Installed water resource modelling systems for catchment
sis for the system analysis. To enable joint agreement on the management agencies, Water SA, 27(3), 333–340, 2001.
principles to apply and to quality assure the results, signifi- Draper , A.J, Lund, J.R.: Optimal Hedging and Carryover Storage
cant capacity of system analysis must exist among the main Value, Journal of Water Resources Panning and Management,
stakeholders. Furthermore, transparency of the system anal- ASCE, January/February, Vol. 130, no 1,pp 83-87, 2004.
ysis modeling is a key for the joint acceptance of the results. Grossmann, W., Guariso, G., Hitz, M., and Wethner, H.: A Min
Alternatively, trust must be built between the stakeholders Cost Solution for Dynamic Assignment Problems in Networks
with Storage Devices, Manage. Sci., 41(1), 83–93, 1995.
and the system analysis model developers, something that is
Joyce, B., Vacuña, S., Dale, L., Hanemann, M., Purkey, D., and
difficult in transboundary river basins. Yates, D.: Climate change impacts on water for agriculture in
This study illustrates the possibility for the introduc- California: A case study in the Sacramento Valley, California
tion of other modeling packages for water resources alloca- Climate Change Center, White Paper, CEC-500-2005-194-SF,
tion in transboundary rivers in southern Africa beyond the California, USA, 2006.
WRYM, which has been adopted as the preferred tool but Juizo, D., Lidén, R., and Carmo Vaz, A.: Remaining Challenges
has been questioned because of its complexity and limited for bi-national agreements on shared water: The Umbeluzi case,
transparency. For complex river basin systems the WRYM Water Policy, 8, 231–253, 2006.
model most probably is the best tool but in many cases the Le Roy, E.: A study of the development of water resources in
WAFLEX and WEAP21 may be good alternatives. The ad- the Olifants catchment, South Africa: Application of the WEAP
model, [Link]. report, Imperial College, London, 2005.
vantage of these models compared to the WRYM is that they
Lévite, H. and Sally, H: Linkages between Productivity and equi-
are relatively user friendly and transparent.
table allocation of water, Phys. Chem. Earth, 27, 825–830, 2002.
The conclusion of this study is that the choice of model Mackenzie, R. S. and van Rooyen, P. G.: Water Resources Yield
does not per se affect the decision of best water allocation Model: User Guide – Release 4.1.1, Department of Water Af-
and infrastructure layout of a shared river basin. The chosen fairs (DWAF), South Africa, 1999.
allocation and prioritization principles for the specific river Mackenzie, R. S. and van Rooyen, P. G.: Management large water
basin and the model developer’s experience and integrity are resources systems, Water Supp., 3(3), 297–304, 2003.

[Link]/14/2343/2010/ Hydrol. Earth Syst. Sci., 14, 2343–2354, 2010

2354 D. Juı́zo and R. Lidén: Modeling for transboundary water resources planning and allocation

Nkomo, S. and van der Zaag, P.: Equitable water allocation in a Sweco in association with Consultec Lda, Impacto Lda, BKS,
heavily committed international catchment area: the case of the Knight Piésold: Joint Umbeluzi River Basin Study: Final Report,
Komati Catchment, Phys. Chem. Earth, 29, 1309–1317, 2004. Report prepared for the Government of the Republic of Mozam-
Raju, K. S. and Pillai, C. R. S.: Multicriterion decision making in bique and the Government of the Kingdom of Swaziland, 2005.
river basin planning and development, Eur. J. Oper. Res., 112, Symphorian, G. R., Madamombe, E., and van der Zaag, P.: Dam
249–257, 1999. operation for environmental water releases; the case of Osborn
Rosenzweig, C., Strzepek, K. M., Major, D. C., Iglesias, A., Yates, Dam, Save Catchment, Zimbabwe, Phys. Chem. Earth, 28, 985–
D. N., McCluskey, A., and Hillel, A.: Water resources for agri- 993, 2003.
culture in a changing climate: international case studies, Global Van der Zaag, P., and Bolding, A.: Water governance in the Pungwe
Environ. Change, 14, 345–360, 2004. river basin: institutional limits to the upscaling of hydraulic
SADC: Revised Protocol on Shared Watercourses in the Southern infrastructure, in: Transboundary water governance in South-
African Development Community (SADC), 2000. ern Africa; examining underexplored dimensions, edited by:
Savenije, H. H. G.: Spreadsheets: Flexible tools for integrated man- Swatuk, L. A. and Wirkus, L., Nomos, Bonn, 163–177, 2009.
agement of water resources in river basins, IAHS Publications, Van der Zaag, P. and Carmo Vaz, Á : Sharing the Incomati wa-
231, 741–744, 1995. ters: cooperation and competition in the balance, Water Policy,
Sengo, D.J., Kachapila, A., van der Zaag, P., Mul, M., and Nkomo, 5, 349–368, 2003.
S.: Valuing the environmental water pulses into the Incomati es- Wang, L. F. and Hipel, W. K.: Cooperative water resources alloca-
tuary: Key to achieving equitable and sustainable utilisation of tion based on equitable rights, Proceedings of the 2003 IEEE In-
transboundary waters, Phys. Chem. Earth, 30, 648–657, 2005. ternational Conference on Systems, Man and Cybernetics, Wash-
Sieber, J., Huber-Lee, A., and Raskin, P.: WEAP: Water Evalua- ington, USA, 5–8 October, 2003.
tion And Planning System User Guide (for WEAP21), Stock- Wurb, R. A.: Modeling river/reservoir system management, water
holm Environmental Institute – Boston, and Tellus Institute, User allocation, and supply reliability, J. Hydrol., 300, 100–113, 2005.
Guide for WEAP21, Boston, MA, 2002. Yates, D., Siber, J., Purkey, D., and Huber-Lee, A.: WEAP21-
A Demand-, Priority-, and Preference-Driven Water Planning
Model Part 1: model characteristics, Water Int., 30(4), 487–500,
2005.

Hydrol. Earth Syst. Sci., 14, 2343–2354, 2010 [Link]/14/2343/2010/