Accepted Manuscript

Assessment of water availability for competing uses using SWAT and WEAP in South
Phuthiatsana catchment, Lesotho

Motlatsi Maliehe, Deogratias M.M. Mulungu

PII: S1474-7065(16)30045-6
DOI: 10.1016/j.pce.2017.02.014
Reference: JPCE 2582

To appear in: Physics and Chemistry of the Earth

Received Date: 13 May 2016

Revised Date: 24 November 2016
Accepted Date: 17 February 2017

Please cite this article as: Maliehe, M., Mulungu, D.M.M., Assessment of water availability for competing
uses using SWAT and WEAP in South Phuthiatsana catchment, Lesotho, Physics and Chemistry of the
Earth (2017), doi: 10.1016/j.pce.2017.02.014.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

Assessment of water availability for competing uses using SWAT and WEAP in South
Phuthiatsana catchment, Lesotho

Motlatsi Maliehe and Deogratias M.M. Mulungu*

University of Dar es Salaam, College of Engineering and Technology
Department of Water Resources Engineering
P.O. Box 35131 Dar es Salaam, Tanzania

PT
*Corresponding Author email: dmulungu@udsm.ac.tz; deorgm@yahoo.co.uk

Abstract

RI
The study assessed the quantity of surface water in the South Phuthiatsana catchment,
estimated flows in ungauged catchments using Soil and Water Assessment Tool (SWAT) and

SC
allocated the resources in the catchment using Water Evaluation And Planning (WEAP)
model. SWAT model was calibrated from 1979 to 2001, the p-factor was 65%, r_factor 0.58,
NS 0.59 and R2 0.59 for calibration and for validation from 2002 to 2013, the p-factor was
57%, the r_factor was 1.34, the NS was 0.52, and R2 was 0.66. The results show the water

U
balance as: 26% of precipitation form streamflow, 41% of the total flow comes from
baseflow, while surface runoff accounts for 59%, 14% of precipitation percolates to shallow
AN
aquifer, 1% percolates to deep aquifer and 68% of precipitation is lost through
evapotranspiration. The WEAP model was calibrated using CG024 and CG084 stations and
historical demands. For CG024 calibration (1972 2002) NS was 0.72 and R2 was 0.84 and
for validation (2003 2014) the NS was 0.73 and R2 was 0.74. For CG084 calibration (2007
M
2011) NS and R2 were 0.55 and 0.64 and for validation (2012 2014) the NS and R2 were
0.63 and 0.89 respectively. Two scenarios were evaluated. First for the reference scenario, the
Metolong industrial demands of 1.46 Mm3 and environmental demands of 2.29 Mm3 were
both not met. Secondly, for the irrigation expansion scenario, increasing irrigation land by
D

12.3%, a total of 4.44 Mm3 demands were not met (irrigation accounts for 65.65% of the
unmet). Therefore, the study recommends an irrigation plan for the catchment. The irrigation
TE

plan has to include: irrigation systems designed for the site, meteorological stations and an
irrigators association with experts forming part of the board.

Keywords: South Phuthiatsana catchment; SWAT model; Uncertainty; Water availability;

EP

Water demands; WEAP model

C
AC

Page 1 of 20
 ACCEPTED MANUSCRIPT

1. Introduction

Integrated water resources management (IWRM) is an approach, which promotes

development of water, land and related resources in order to maximize the resultant economic
and social welfare equitably without compromising the sustainability of the environment
(Global Water Partnership, 2002). IWRM recognises that there are competing water uses that
are often connected and interrelated. The problems that arise from social aspects typically
affect water resources and the environment is always at the receiving end.

PT
Lesotho has abundant water resources (5.23 km3/year) exceeding the countrys requirements
(2 km3/year) but seasonality affects water availability to users (LMS, 2013). The water

RI
availability is affected because of rainfall seasonality that there is wet (rainy) and dry seasons.
The rainy season of the country starts from October to February and after this period, the
surface water resources availability slowly declines. During winter, the precipitation is in the

SC
form of snow and mostly in the mountainous areas. As a result, water is predominantly in the
highlands and yet most people prefer to live in the less water abundant western areas
lowlands (TAMS, 1996). The competing users for water resources in Lesotho are:
hydroelectricity, agriculture, domestic, industrial and environment (which was recognised

U
recently following the Lesotho Water Act of 2008).

AN
In the capital city, Maseru, the growth of textile industries as well as peri-urban population
has been of great concern in recent times. The population living in urban areas is expected to
increase by 20% for the next ten years and this will lead to an increase in domestic water
demands (GoL, 2012). This has led to the construction of the Metolong Dam through the
M
Lesotho Lowlands Water Supply Scheme. The dam was constructed in May 2012
(ORASECOM, 2013) and the first impoundment was completed in February 2014 (WASCO,
2015). The dam capacity is 53 Mm3 with a supply capacity of 63.5 M/day until 2020 and
105 M/day by 2035 (available yield period for the dam). The dam catchment is about 20%
D

of the South Phuthiatsana Catchment. It a major water supplier to Maseru and other areas
outside the South Phuthiatsana Catchment (WASCO, 2015).
TE

The South Phuthiatsana catchment used to support irrigation schemes but most of these
schemes have collapsed due to fragmented institutional arrangements. On the same note, the
performances of these schemes were very low and this led to low crop production. Inadequate
EP

knowledge of the hydrology of the catchment has led to failure in understanding of the water
supply which meets the irrigation water demand. Moreover, water for the environment is also
of concern. The socio-economic and environmental conditions have to be maintained
downstream. Thus, assessment of surface water resources can help understand the supply of
C

water resources particularly during dry seasons, and allocation and management strategies of
the same. The basin scale assessments provide vital information since water management
AC

decisions are very often determined by the river basin management authorities (Gain and
Giuppopni, 2015). Prediction of water availability and impact assessment of water supply and
demand sites is required for appropriate decision-making processes for water resource
development (Hishinuma et al. 2014). In order to assess the spatial and seasonal impact of
water demand and allocation of water in a river basin, distributed models are appropriate.
Watershed models are standard tools used to generate continuous estimates of streamflow and
other hydrological variables. However, most of the watershed models lack built-in scenario
tools, which are attractive for planning and management of water allocation.

Page 2 of 20
 ACCEPTED MANUSCRIPT

Scenario testing is a larger aspect for planning and management, and for competing water
uses, this can be done well by planning models such as Water Evaluation And Planning
System (WEAP), which freely available. The technical details on the WEAP model can be
obtained from Yates et al. (2005). WEAP model is an effective tool to advice on water
resources development and management alternatives (Skaggs et al., 2012). WEAP has been
used widely and successfully in the world and in Africa, e.g., Mulungu and Taipe, 2012;
Rodriques et al., 2005; van Loon and Droogers, 2006; Arranz and McCartney, 2007; Haji,
2011; Droogers et al., 2014. It can be applied from community to catchment and to basin

PT
level (Yates et al., 2005; Sieber et al., 2005). Accordingly, WEAP model was selected in this
study for the South Phuthiatsana catchment as a decision support tool for water resources
management at catchment level in Lesotho.

RI
However, in many developing countries, availability of hydrological data is a challenge and
as such, WEAP model may not be applied directly and it needs assumptions and input data

SC
from other models. This is the case with Lesotho and the South Phuthiatsana catchment in
particular, where most of the river sub-catchments are ungauged. In South Phuthiatsana
catchment, observed river flow data was only available at the downstream point or river outlet
while WEAP model requires water flows at the river tributaries or sections known as head

U
flows. Accordingly, in WEAP set-up there was a need to get water supply data from another
model. Therefore, the approach of the study was first to quantify the water resources
AN
especially for the ungauged catchments using the Soil and Water Assessment Tool (SWAT)
hydrological model and then allocate water resources to different water demands using the
WEAP model. In this case, the sub-basin structure and river network obtained during SWAT
model set-up were adopted in the WEAP model set-up. SWAT model can be downloaded
M
freely at the website (http://swat.tamu.edu/) and there are different and current versions of the
SWAT model, which are documented and distributed. The model has been used successfully
in the African region (e.g. Mulungu and Munishi, 2007; Ndomba et al., 2008; Githui et al.,
2009; Sood et al., 2013) and also coupled with WEAP model (e.g. McCartney and Girma
D

2012; Chinnasamy at al., 2015). Essentially, SWAT model was used in this study to estimate
the catchment hydrology on daily time step and in ungauged (nested) catchments upstream of
TE

a gauged river outlet. Therefore, the study aimed at assessing the quantity of surface water in
South Phuthiatsana catchment and allocate it for current and for the year 2035, which is the
period for available yield of the dam. The study attempted to determine whether the new
development (Metolong Dam) will meet the South Phuthiatsana catchment demands, which
EP

were expected to be changing with time. The varying water demands were determined for
each sector in the catchment and evaluated with the water supply using the scenario approach
built-in the WEAP model. However, following data scarcity in the study area and other
limitations in hydrological modeling in the southern Africa region as indicated by Hughes
C

(2008), the SWAT model application in this study was a challenge and there was need to
combine observed and reanalysis input datasets.
AC

2. Methods

2.1 SWAT Model Description

SWAT is a physically based and semi-distributed model developed by Agricultural Research
Services of United States Department of Agriculture (Fadil et al., 2011). It is a large scale
model used to simulate: the hydrology of a catchment, water quality, climate change, crop
growth, sediment yield, nutrient transfer, impacts of land management practices (Setegn et
al., 2008; Mulungu and Munishi, 2007; Fadil et al., 2011; Zhang et al., 2007). In SWAT, a

Page 3 of 20
 ACCEPTED MANUSCRIPT

watershed is divided into sub-basins and the sub-basins are further divided into Hydrologic
Response Units (HRUs). The HRUs are units with similar land use, slope and soil type
(Mulungu and Munishi, 2007). The model calculates the water balance for each HRU. SWAT
uses GIS interface (ArcSWAT) which makes it user friendly.

The model use water balance equation (Eq. 1) to simulate the catchment hydrology in daily
time step:

PT

 =
 + 
      (1)


Where:

RI
SWt is the final soil water content (mm)
SW0 is the initial soil water content (mm)
Rday is precipitation in day i (mm)

SC
Qsurf is the amount of surface runoff in day i (mm)
Wseep is the amount of water entering the vadose zone from soil profile in day i (mm)
Qqw is the amount of return flow in day i (mm)
Ea is the amount of evapotranspiration in day i (mm)

U
t is the time in days

AN
In the South Phuthiatsana catchment, the SCS curve number method was used to estimate
runoff and the Penman-Monteith method was used to estimate potential evapotranspiration
(Arnold et al., 1998). Channel flood routing was estimated using the Muskingum method
(Zhang et al., 2007). The data input for SWAT are: Digital elevation model (DEM), land use
M
map, soil type map, meteorological data and streamflow data. All the maps were projected to
UTM projection. The SWAT model efficiency and reliability has been reported in several
places around the world, and in Africa and South Africa (Fadil et al., 2011).
D

2.2 WEAP Model Description

The Water Evaluation And Planning (WEAP) model is designed to assist policy makers to
TE

evaluate water supply policies and suitable water resources plans (Hatcher, 1995). WEAP
was originally developed by Stockholm Environment Institute at Boston, USA (Van Loon
and Droogers, 2006).
EP

WEAP uses scenario approach (answering what if questions) to evaluate water demands,
associated priorities and water supply - for current and future periods (Rodrigues et al., 2005).
WEAP functions are (Sieber et al., 2004):
C

Catchment hydrology simulation, (i.e., surface runoff, evaporation and infiltration)

and assess water availability.
AC

Simulation of interactions of socio-economic activities with water resources and their

allocation as well.

WEAP operates on a monthly time step water balance accounting: total inflows equal total
outflows, net of any change in storage (in reservoirs and aquifers). A linear programming is
used to maximize the satisfaction of demand site and user-specified instream flow
requirements, subject to demand priorities, supply preferences, mass balance and other
constraints. The mass balance equation constraint in the linear programming is shown as Eq.
2:

Page 4 of 20
 ACCEPTED MANUSCRIPT

 = !"#$% (2)

The data required for the model are (Levite et al., 2003): water use (demand site); reservoir
location, capacity and operation rules; flow gauging station (flow requirements and ecological
reserves) and river head flows. It also overcomes data shortages by using data from public
domains and satellite information (Droogers et al., 2014). WEAP also allows the user to
determine the level of details for a given study - the user can lump similar demand sites

PT
together or can present them as they appear from different sub-basins.

During the model set up, the current accounts, key assumptions and scenarios were defined.
Current accounts are viewed as calibration step and provide insights of actual demands,

RI
pollution loads, resources and supply within the catchment. Key assumptions may be built in
current accounts to represent policies, costs and factors affecting demands pollution, supply
and hydrology. Scenarios built on current accounts allow one to explore the impacts of

SC
alternative assumptions (Van Loo and Doorgers, 2006). The scenario in which the data is
available is called the reference scenario (Van Loo and Doorgers, 2006).

Essentially calibration is done by estimating historical water demand patterns and simulating

U
resultant flow (McCartney et al., 2005). The WEAP streamflows can also be calibrated
manually by altering the system demand historical patterns to have the best fit between the
AN
simulated and observed flow (Mulungu and Taipe, 2012). This was the approach adopted by
the study. The Nash Sutcliff and coefficient of determination were used to measure the
efficiency of the model.
M
2.3 Study Area
The South Phuthiatsana catchment (Figure 1), is found between the latitudes 291250 to
293750 and longitudes 27255 to 2825 with an elevation between 1469 m.a.s.l and
2987 m.a.s.l. The catchment has area of 1,116 Km2 with an elongated shape. It is located in
D

the capital city of Lesotho, between Maseru district (covers 20.89% of Maseru district) and
Berea district (covers 16.74% of Berea district). The South Phuthiatsana River flows from
TE

Thaba-Ntsonyane south-westerly to Mohokare River. Figure 1 shows the South Phuthiatsana

catchment with streamflow, SWAT meteorological stations and rainfall gauges.
EP

Insert Figure 1: Map of the study area

2.4 Meteorological Data Preparation

2.4.1 Rainfall
C

The rainfall and temperature data were acquired from the Lesotho Meteorological Services.
The collected rainfall data is presented in Table 1 and it includes those stations outside the
AC

catchment as well. It also shows the: location, elevation, available data period, percentage of
missing data and the station identity number of the eleven collected rainfall stations. It is
observed that for stations within the catchment, the gaps are very significant. The gaps range
from 7% to 48.3%.

Table 1: Rainfall data description.

Station ID Lat Lon Elevation Period %Missing
LESMAS25 -29,380 27,750 1775 1985-2013 26.6
LESMAS24 -29,444 27,725 1690 1979-2013 16.7
LESMAS27 -29,449 27,560 1628 1985-2013 7.3

Page 5 of 20
 ACCEPTED MANUSCRIPT

LESBER04 -29,250 27,920 1700 1979-2013 16.5

LESMAS35 -29,327 27,789 1829 2000-2013 48.3
LESMAS06 * -29,337 27,528 1575 1979-2012 8.4
LESBER06 * -29,107 27,974 1725 1984-2012 17.3
LESMAS29 -29,394 27,554 1600 1998-2005 9.5
LESMAS20 -29,420 27,630 1690 1979-2005 16.5
LESBER07 * -29,137 27,766 1690 1993-2013 12

PT
LESMAS22 * -29,606 27,736 1775 1979-2000 12.2
* Stations outside the catchment

2.4.2 Temperature

RI
The temperature data (Table 2) for the stations include maximum and minimum variables
indicated with a prefix -max and -min respectively. The stations had a lot of gaps and their
data lengths were not corresponding. The maximum and minimum temperature data at the

SC
same station had different data periods. LESMAS 25-min has the highest percentage (31.8%)
of missing data followed by LESMAS 27-max (28.1%). The SWAT stations presented the
reanalysis data from SWAT website (globalweather.tamu.edu/), which was obtained to
supplement the observation data following data gaps or missing variables. The South

U
Phuthiatsana catchment had no solar radiation and relative humidity data. The available
reanalysis data were: minimum and maximum temperature, solar radiation, wind speed,
rainfall and relative humidity. AN
Table 2: Temperature data of South Phuthiatsana catchment
Station ID Lat Lon Elevation Period %Missing
M
LESMAS25-max -29,380 27,750 1775 2007-2013 18
LESMAS25-min -29,380 27,750 1690 1980-2013 31.8
LESMAS27-max -29,449 27,560 1628 2003-2013 28.1
D

LESMAS27-min -29,449 27,560 1700 1985-2013 7.5

LESBER07-max -29,137 27,766 1829 2003-2013 12.8
TE

LESBER07-min -29,137 27,766 1575 1993-2013 13.1

LESMAS35-max -29,327 27,789 1725 2008-2012 0.5
LESMAS35-min -29,327 27,789 1600 2008-2013 0.2
LESMAS29-max -29,394 27,554 1690 No data No data
EP

LESMAS29-min -29,394 27,554 1690 1998-2005 11.6

LESBER06-max * -29,337 27,528 1775 No data No data
LESBER06-min * -29,337 27,528 1715 1984-2013 10.1
292278-max * " -29,193 27,813 1889 1979-2013 0.1
C

292278-min * " -29,193 27,813 1715 1979-2013 0.1

292281-max * " -29,193 28,125 1889 1979-2013 0.1
AC

292281-min * " -29,193 28,125 1889 1979-2013 0.1

298275-max * " -29,818 27,500 1821 1979-2013 0.1
298275-min * " -29,818 27,500 1821 1979-2013 0.1
298278-max * " -29,818 27,813 2890 1979-2013 0.1
298278-min * " -29,818 27,813 2890 1979-2013 0.1
* Stations outside the catchment
" SWAT stations

2.4.3 Wind

Page 6 of 20
 ACCEPTED MANUSCRIPT

The Metolong Authority measure: rainfall, temperature, wind direction and wind speed on a
daily basis within the vicinity of Metolong dam. The station does not have a code so, it was
referred to as MDC in the study (from the name Metolong Dam City). The MDC station has
records from 2012 to 2015. The variable used from this station was wind speed and 7.4% of
the data was missing. The largest gap was on the 1st February 2014 to 2nd March 2014 from
there the other gaps were at least ten days. The available period of the reanalysis data was
1979 to 2014 and was used to: fill the missing data and extend the observed data of other
meteorological variables except rainfall.

PT
2.4.4 Correlation between stations' data

RI
The coefficient of correlation weighted method (CCWM) proposed by Teegavarapu and
Chandramouli in 2005 was used for data infiling (Teegavarapu and Chandramouli, 2005).

SC
The method quantifies the strength of spatial autocorrelation, and the correlation coefficient is
the weighting factor (Marteau et al., 2011). CCWM is a conceptually superior method for
infilling missing precipitation data and can be better a deterministic surface estimation
method and data-sensitive in any climate.

U
According to Teegavarapu and Chandramouli (2005), four stations with the highest
AN
correlation coefficient are considered. The correlation of a minimum of 730 concurrent days
is required (TWINLATIN, 2009). Kajornrit (2012) compared coefficient of correlation
weighted method and inverse distance weighted method and found that coefficient of
correlation weighted method provides better accuracy but is not much better than inverse
M
distance weighted method.

In the current study, the Pearson correlation coefficients were calculated on daily basis and
the maximum coefficient was 0.76 and the minimum used was 0.50. The method was then
D

used to simulate the rainfall of a certain period with known values at random and the
simulated and observed rainfalls R2 ranged from 0.9869 to 1.
TE

2.5 Streamflow Data

The catchment has five stream gauges with records between 1972 and 2014. Table 3 shows
EP

the data period and location of these gauging stations. The data gap analysis shows that
CG024 had no gaps and had the longest record. Generally, the data had few gaps except for
CG075, which stopped working in 2009.
C

Table 3: South Phuthiatsana streamflow gauging stations

Stations ID Stations Name Lat Lon Period % Missing
AC

CG024 Masianokeng -29,40 27,56 1972-2013 0.00

CG060 Khoshane at Toll gate -29,42 27,82 1989-2014 0.58
CG075 S/Phuthiatsana at Pulane -29,38 27,76 1990-2009 10.09
S/Phuthiatsana at
CG84A -29,25 27,92 2007-2014 1.42
Metolong
CG061 Mohlakakuta at Ha Ntsi -29,33 27,79 2008-2014 0.12

2.6 SWAT Model Setup

2.6.1 Spatial data preparation

Page 7 of 20
 ACCEPTED MANUSCRIPT

A 30 x 30 m resolution ASTERGDEM was used for delineating the watershed, creating sub-
basins, stream network and generating longest reaches, calculating terrain slope and channel
slope. The DEM shows that the Eastern side of the catchment has the highest elevation of
2987 m.a.s.l and the western has the lowest elevation of 1469 m.a.s.l (Figure 2a). SWAT uses
the land use map as either a shapefile or a raster file. The land use map was acquired from the
Ministry of Forestry and Land Reclamation of Lesotho. The land uses in the catchment are:
agriculture (48.08%), bare soil (25.07%), grasses (13.71%), bushes (9.49%), forest (0.36%),
urban (3.28%) and water bodies (0.01%). These were respectively reclassified and coded in

PT
SWAT as: urban-URBN agriculture-AGRR, barren - BARR, forest - FRSD, bushes-RNGB,
grassland-RNGE and water-WATR (Figure 2b) during Land use/soil/slope definition.

RI
The catchment had ten soil series when using the Lesotho soil association classification.
However, the soil properties (chemical and physical) required by SWAT were not available
for these soil series. The Harmonized World Soil Database (FAO/IIASA/ISRIC/ISS-

SC
CAS/JRC., 2009) was therefore used. The database had two soil types for the whole country
and thus for the study area and for the study area as well. These were: EutricPlanosols (We
FAO 74 classification) and Lithosols (I- FAO 74 classification) (Figure 2c). The databases
did not have some of the physical properties such as: saturated hydraulic conductive constant,

U
soil albedo, KUSLE and soil hydrologic group classification. These were calculated according
to Gies and Merwade (2013) tutorial. To calculate the albedo, the soil colour was taken from
AN
the World Inventory of Soil Emission Potential database (Batjes, 2008).

The slopes were classified into three categories: 0% 15%, 15% 30% and 30% 9999%.
These slopes were calculated using ArcGIS and the catchment had the highest distribution at
M
15% (Figure 2d). The threshold percentages for the HRU definition were: 20% land use, 10%
soil and 20% slope. These are default thresholds suggested by SWAT manual.

Insert Figure 2: Map of South Phuthiatsana set up in SWAT: (a) DEM (b) Land
D

use/land cover (c) Soils (d) Slope classes

TE

2.6.2 Model sensitivity analysis, calibration and validation

Sensitivity analysis, calibration, validation and uncertainty analysis for SWAT 2012 were
done outside the ArcSWAT interface using SWAT-CUP - SUFI-2 optimization algorithm.
The model warm up period was two years and the calibration period was from 1981 to 2001
EP

at CG 024 station, which is the outlet of the catchment. The validation was done from 2002 to
2013. Twenty two flow parameters (Table 4) were then run in the model and the sensitive
parameters were determined afterwards. The Nash Sutcliffe (NS) function was used as the
optimization function during calibration and validation periods with a minimum threshold for
C

the behavioral solution at 0.5. In SUFI-2, the uncertainties are quantified by the p-factor,
which is the percentage of data bracketed by the 95% prediction uncertainty (95PPU). The
AC

95PPU is calculated at the 2.5% and 97.5% levels of the cumulative distribution of an output
variable (Abbaspour et al., 2007). Another factor used to quantify the uncertainty is the r-
factor, which is the average thickness of p-factor divided by the standard deviation of the
observed data. Ideally, the p-factor should be close to one and the r-factor close to zero.

Table 4: SWAT parameters for sensitivity analysis

Lower Upper
Parameter Description
limit limit
CN2 Initial SCS CN II value 35 98

Page 8 of 20
 ACCEPTED MANUSCRIPT

SLSUBBSN.hru Average slope length. 0 400

SURLAG Surface runoff lag time (days) 0 24
EPCO Plant uptake compensation factor 0 1
ESCO Soil evaporation compensation factor 0 1
CANMX Maximum canopy storage (mm) 0 100
ALPHA_BF Baseflow alpha factor (days) 0 1
GW_DELAY Groundwater delay (days) 1 450
Threshold water depth in the shallow aquifer for flow

PT
GWQMN 0 2
(mm)
GW_REVAP Groundwater "revap" coefficient -0.02 0.2
Threshold depth of water in the shallow aquifer for

RI
REVAPMN 0 500
"revap" to occur (mm).
RCHRG_DP Deep aquifer percolation fraction 0 1
SOL_Z Depth from soil surface to bottom of layer. -0.8 0.8

SC
SOL_AWC Available water capacity (mm H20/mm soil) 0 1
SOL_K Saturated hydraulic conductivity (mm/hr) 0 100
HRU_SLP Average slope steepness -0.4 0.4
OV_N Manning's "n" value for overland flow. 0 12

U
LAT_TTIME Lateral flow travel time 0 180
CH_K2 Channel erodibility factor -0.01 500
CH_S2
CH_L2
AN
Average slope of main channel
Length of main channel
-0.001
-0.05
10
500
CH_N2 Manning's "n" value for the main channel. 0 0.3
M
There are two sub-basins ((Koro-koro and Thupa-kubu ), which are ungauged (Figure 3).
Their sizes are respectively 312.65 km2 and 90.04 km2, and representing 27.98% and 8.06%
of the whole catchment respectively. The land cover/use are: agriculture, barren land, bushes,
D

forest, grassland and urban while soils are: EutricPlanosols (We FAO 74 classification) and
Lithosols (I - FAO 74 classification). The land cover/use in these accounts: 33.76%
TE

(agriculture), 37.63% (barren land), 37.09% (bushes), 79.02% (forest), 42.60% (grassland)
and 28.90% (urban) of the corresponding land cover in the whole catchment.
These ungauged catchments are inside the South Phuthiatsana catchment and have: similar
soils, similar land use, similar climate and they are geographically close to each other. Since
EP

the ungauged catchments are nested within the larger South Phuthiatsana catchment, which
was calibrated at the outlet (CG 024), it was assumed that the processes in the catchment
(ungauged and gauged) were captured. This comparison warrants the use of the approach
adopted in this study.
C

Accordingly, the calibrated SWAT model parameters (spatially varying) at the outlet of the
AC

catchment (CG 024) were applied to the ungauged catchments. However, Heuvelmans et al.
(2004) indicated that transfer of parameters within a catchment and to neighbouring
catchment results in small decrease in the model performance than transfers to catchments at
greater distance.

Insert Figure 3: The ungauged South Phuthiatsana catchment

2.7 WEAP Model set-up for South Phuthiatsana

Page 9 of 20
 ACCEPTED MANUSCRIPT

For the WEAP model for the South Phuthiatsana, the water system was characterised by:
water demand sites, reservoir (location, operation rules, water balance from SWAT model
and dam capacity), flow gauging station (streamflow including flows generated from
SWAT model and ecological reserves), river head flows. The study area boundaries were
used to describe the spatial location of the water system. The demand site, wastewater
treatment plant, reservoir were represented with nodes. These nodes were linked to the river
through transmission links and return flow links. The demand site nodes were created in the
schematic view at relative positions. The demands were then named accordingly and demand

PT
priority set according to GoL (2008) domestic 1 and environment 2. However the Act does
not provide priority for industry and irrigation rather a communication with the Department
of Water Affairs personnel gave a rough priority of irrigation 3 and industry 4. The supply

RI
sources were then connected to the demand sites using transmission links and waste from
demand sites was returned using return flow links. A period where all or most the data is
available is termed current accounts and this is the period where simulation begins. For South

SC
Phuthiatsana the year 2010 had more data and thus was set as current accounts. The last year
of scenario was 2035. This is the planning horizon of the Metolong Dam. The Metolong dam
was set to be active in the year 2014 as that was the year for completion of impoundment.

U
The SWAT sub-basin map and South Phuthiatsana River shapefiles were then added into the
WEAP study area and the project was saved. The SWAT model provided the streamflow data
AN
for the sub-basins, which was calculated for the catchment including the ungauged areas.
These data provided the water supply sources for the WEAP model. The hydrology (water
supply) was not simulated within WEAP but with SWAT. Therefore, the SWAT model
generated flows that were used to determine river head flows in WEAP model.
M
2.7.1 Defining catchment water demands
In the study, there were three identified domestic demands: Metolong dam domestic
demands, Roma domestic demands and catchment domestic demands. The Metolong
D

domestic demands are those that were supplied directly by the Metolong dam (Parkman,
2005). The Roma domestic demands are the demands in the Roma valley supplied from the
TE

Liphiring River by Water and Sewage Company (WASCO). The majority of people use
groundwater for domestic purposes, but due to the limitations of the study, this consumption
was not dealt with fully. A small portion (10%) of catchment domestic demands were
supplied from the main river. This comes from the understanding that people still depend on
EP

the river for such things as: bathing, laundry, and to a small degree, drinking purposes. These
demands were given the name catchment domestic demands. The demands were calculated
first by estimating the catchment population from the national village census data (2006
census) and then multiply the population with the water consumption per capita. The census
C

provided population for constituents and councils. However, there were instances where the
councils in the same constituents were not inside the study area. The councils were then
AC

located using the GeoWikia platform (Driskell, 2008). According to BoS (2007), the
population growth rates for Berea district and Maseru district, between 1996 census and 2006
census, were 0.06 and 0.09, respectively. These growth rates were used to project the future
demands for the respective districts.

In Metolong domestic water demands calculations, the population growth rates catered for
HIV/AIDS pandemic as it reduces the population (Parkman, 2005). The high income
populations were given 100 liters per day per person and low income settlements were given
30 liters per day per persons. The losses were assumed to be 30% based on WASCO water

Page 10 of 20
 ACCEPTED MANUSCRIPT

losses (Parkman, 2005). In the Roma valley, the population supplied by WASCO was 1200,
and WASCO supplies about 1.3 M/day.

Monthly water irrigation requirements were calculated using CROPWAT. The model requires
the meteorological data. LESMAS 25, a meteorological station was used as it is within the
area. Firstly, the evapotranspiration was calculated using Penman-Monteith method built in
the model. In Lesotho, infiltration is about 4% of precipitation and 25% forms runoff
(Bonney, 1975). The effective rainfall was calculated using the fixed percentage of 75%. The

PT
crops used were maize and wheat and accounted for 60% and 40% area coverage (Ntai,
2011). The planting date was taken to be 01 October. This month is seen to provide the best
yield in Lesotho for both wheat (Tolmay et al., 2000) and maize (Bruns and Ryan, 2012).

RI
The livestock census for the two districts of Maseru and Berea was collected from the Bureau
of Statistics. The average unit livestock water consumptions were adapted from (TAMS,

SC
1996). The livestock population growth rate between 1999/2000 and 2009/2010 was -0.4%
(Central Bank of Lesotho, 2011). The population in the catchment was calculated with Eq. 3;

&' = & )*' /* , (3)

U
Where; Pcat is the population of livestock in the catchment

Acat is the catchment area

Adist is the district area
AN
Pdist is the district livestock census
M
The only industry withdrawing water directly from the river is Tikoe Crushers and it abstracts
500 /sec. The industrys life span is unknown because it is limited by the unknown amount
of quarry to be mined. The study therefore expects no growth and the current consumption is
not expected to change.
D

The Metolong industrial demands (addition of institutions and industries together) are
TE

supplied with potable water through WASCO pipelines from Metolong dam in this set-up.

The environmental flow requirements were calculated using Downstream Response to

Imposed Flow Transformations (DRIFT) method (GoL, 2009). The DRIFT method
EP

recommended 21% (10 - 11 MCM per annum) of mean annual runoff at Metolong IFR site
and had four flood classes. The floods are set to flush the stream so as to mimic the natural
stream floods. The IFR site is 20 km downstream of the dam wall (GoL, 2009). The low
flows were summed with flood flows when they existed in a given month. This represented
C

the total amount of water allocated to the environment in that particular month. The
applicable flood classes (1-3) and are released for 3 days. Flood class 1 (1.6 m3/s) are
AC

expected to be released from October to November then from April to June. Flood class 2
(3.73 m3/s) are to be released from December through out March. Lastly, the class 3 floods
(7.12 m3/s) are to be released in October and from December to February. The class 4 floods
are not applicable (GoL, 2009). There are catch dams in the catchment but there is no data on
these dams: the levels, location and capacity.

The demands are presented in Table 5. The WEAP model can project water demands when
given the time step, annual activity level, growth rate and annual water use rate per unit.

Table 5: Calculated South Phuthiatsana Demands for 2010

Page 11 of 20
 ACCEPTED MANUSCRIPT

Demands Demands Growth

Sectors
(M3/annum) (M3/unit/annum) rate (%)

Crop 18 6459.80 7 411.00 0.90

Agriculture
Livestock 59 8297.30 5.05 -0.40
Domestic 17 033 455.00 27.14 2.80

PT
Metolong
Industry 12 100 480.00 - 2.70
Environment 17 442 432.00 - 0.00
Industry 182.50 0.18 0.00

RI
Roma Domestic Demands 50 370.00 41.98 0.00
Catchment Domestic Demands 268 953.90 10.95 0.08

SC
2.7.2 Supply
The rivers are made up of nodes connected by river reaches which have to be drawn. A river

U
node was made at the mouth of each sub-basin and on tributaries. The SWAT model output
generated flows for the ungauged catchments, which were used in nested catchments with no
observed streamflow data. The SWAT text output file was converted to the csv format and
used in WEAP model. AN
Arcus Gibb and Jeffares and Green consulting companies undertook a study in 2010 to
M
determine the impact of climate change on the Metolong long-term yield. The study
concluded that the long-term yields will not be affected by climate change and this estimation
was for 30 to 50 years within the planning horizon of the dam (Jeffares and Green, 2010). The
research was based on the general circulation models (GCM) and the A2 greenhouse-gas
D

(GHG) emission scenario. Therefore, the hydrology of the catchment is not expected to be
affected by climate significantly within this period. The current streamflow variation was
TE

mirrored to the future (2035) period. The future flows were modelled in WEAP using the
Cycle Method within the ReadFromFile procedure. When cycles are selected, WEAP will
wraparound from the end of the file back to the beginning. These are done for every ten years
of the available data. This method was adopted in the study.
EP

2.7.3 WEAP calibration and validation

The WEAP model was calibrated at the upstream using station CG 084 and downstream
C

using CG 024 (Figure 1). The CG 084 calibration period was from 2007 to 2011 and the
validation period was from 2012 to 2014. The downstream station , CG 024 was calibrated
AC

from 1972 to 2002 and the validation period was from 2003 to 2014. These different time
periods were related to streamflow data availability (Table 3). The model was calibrated
manually by altering the demand supply priorities and return flows. This was to maximise the
fit between the observed and simulated flow.

2.7.4 Scenarios evaluated

Reference scenario (2010 2035)

This was the business as usual scenario which used the actual data and represented the
prevailing current situation.. It was used to understand the future situations in case the

Page 12 of 20
 ACCEPTED MANUSCRIPT

prevailing situation does not change in future. It uses current trends of the water system
studied. The reference scenario includes the domestic water supply by Metolong dam from
2014 to 2035. The Metolong Dam was activated in 2014 and its water uses begun in this year
because of the completed impoundment.

Possible irrigation expansion

The contribution of agriculture to the countrys GDP has been decreasing from 20% in 1983

PT
to about 14% in 1999 and 7% in 2011. The crop-agricultural declines were due to a lot of
causes and to mention a few: climatic variability (drought and erratic rainfalls), collapsing
institutional arrangements and low profits (Ntai, 2011). The Government of Lesotho and
irrigation donors have realised that the previous policies have failed but so far there is no

RI
comprehensive alternative (FAO, 2015). However, one of the objectives of the National
Strategic Development Plan of 2012/2013 - 2016/16 is to promote sustainable
commercialisation and diversification in agriculture. The government intends to develop

SC
water harvesting infrastructure and increase the irrigation capacity (GoL, 2012). If this
happens, the Metolong Dam would have to cater for this irrigation schemes. The catchment
demands are envisaged to increase in dry periods and irrigation water requirements would be
expected to increase as well.

U
The total irrigated area in the South Phuthiatsana catchment is 27.03 ha (Ntai, 2011). The
AN
normal irrigation expansion rate from 2000 provided by FAO was 0.9% and was used in the
current accounts to get 2010 irrigated area. FAO projected an increase in irrigated area for
Southern Africa for the period of 2020-2030 as 12.3% (FAO, 2008). Therefore, the study
adopted for the 12.3% (that result in an area of 491.3 ha by 2035 from 27.03 ha in 2010) as a
M
possible future irrigation expansion scenario.
D

3. Results and Discussions

3.1: SWAT Modelling Results
TE

The sensitivity analysis was done for the entire study period using previously mentioned
parameters (Table 4) for 500 simulations. Table 6 shows the parameters sensitivity analysis
using global sensitivity analysis. The t-stat and p-value were used to rank the sensitivity of
the parameters. The highest absolute value of t-stat gives the most sensitive parameter and the
EP

smallest p-value gives the most sensitive parameter. The six most sensitive parameters
obtained for calibration period were: CN2, SOL_AWC, CH_L2, SURLAG, SOL_Z and
CH_S2 (Table 6). This indicates that the streamflow depends mostly on runoff parameters,
soil parameters and channel parameters. van Griensven et al. (2006) gave ranks 1-6 as
C

significantly very important parameters. Accordingly, the six parameters were considered for
model calibration and also served for model parsimony (Table 7).
AC

Table 6: Sensitivity analysis of SWAT parameters

Parameter Name t-Stat p-Value Ranking
1:R__CN2.mgt -30.97 0.00 1
16:R__SOL_AWC(..).sol 15.73 0.00 2
22:R__CH_L2.rte 15.48 0.00 3
5:R__SURLAG.bsn -9.95 0.00 4
11:R__SOL_Z(..).sol 8.95 0.00 5

Page 13 of 20
 ACCEPTED MANUSCRIPT

Parameter Name t-Stat p-Value Ranking

21:R__CH_S2.rte -7.67 0.00 6
12:R__CH_N2.rte 5.84 0.00 7
13:R__SLSUBBSN.hru 4.47 0.00 8
19:V__GW_DELAY.gw 4.35 0.00 9
14:R__HRU_SLP.hru -3.70 0.00 10

PT
7:R__SOL_K(..).sol -2.91 0.00 11
2:V__ALPHA_BF.gw -2.23 0.03 12
18:R__OV_N.hru 1.49 0.14 13

RI
17:R__CANMX.hru 1.36 0.17 14
4:R__ESCO.bsn -1.34 0.18 15
20:R__CH_K2.rte 1.24 0.22 16

SC
8:R__RCHRG_DP.gw -1.06 0.29 17
3:V__GWQMN.gw -0.99 0.32 18
9:R__GW_REVAP.gw 0.88 0.38 19

U
15:R__EPCO.bsn 0.72 0.47 20
10:R__LAT_TTIME.hru 0.43 0.67 21
6:R__REVAPMN.gw -0.27 AN0.79 22

The fitted model parameters are presented in Table 7, which were found to give better
M
simulated streamflow for South Phuthiatsana catchment.

Table 7: Calibration parameters and calibration values

Fitted value
D

Parameter
Description after Min Max
name
calibration
TE

SCS curve number for moisture condition

CN2 II 35 -25 65
Available water capacity (mm H2o/mm
SOL_AWC soil) 0.13 -2 2
EP

CH_L2 Length of main channel (km) 7.215 0 10

SURLAG Surface runoff lag coefficient 4.2 0 24
SOL_Z Soil depth (mm) 0.1 -0.8 0.8
C

CH_S2 Average slope of main channel (m/m) 0.131 -3 3

AC

The simulated streamflow showed a good match of the observed flow with exception of some
peak flows during calibration and validation periods (Figure 4 & 5). The calibration
hydrograph (Figure 4) showed p-factor = 65%, r_factor= 0.58, NS= 0.59 and R2 = 0.59. For
the validation period (Figure 5) it showed p-factor = 57%, r_factor = 1.34, NS = 0.52, and R2
= 0.66. The p-factors showed that there were uncertainties in the SWAT modelling. Only
65% of the measured data could be bracketed in calibration period and only 57% in validation
period. This could be due to uncertainties inherent in: input data, non-uniqueness of
parameters or some processes that were not captured by the SWAT model.

The SWAT model did not capture extreme events in both calibration and validation. During
the model set-up, the model used a rainfall station LESMAS 25, which was near the centroid
Page 14 of 20
 ACCEPTED MANUSCRIPT

of the South Phuthiatsana catchment. The station does not capture the spatial variability of
rainfall in the catchment, which could be influenced by the ruggedness nature of the
catchment. The covariance at the rainfall station showed that the minimum covariance was
0.22 and the highest was 0.32. The higher covariance showed that the catchment experience
extreme events. This is the case for arid and semi-arid areas. Another cause could be from the
uniqueness of the parameters. The parameters which provided the best results did not capture
all processes in the basin.

PT
Insert Figure 4: SWAT calibration flow series and uncertainty in South
Phuthiatsana

RI
Insert Figure 5: SWAT validation flow series and uncertainty in South
Phuthiatsana

SC
3.1.1 South Phuthiatsana water balance
The calibrated SWAT model was then used to estimate the water balance in the catchment for
1981 to 2001. The annual average hydrological processes are presented in Table 8.

U
Table 8: Average annual hydrological processes for South Phuthiatsana

Units
AN
Hydrological balance components
P ET SW PERC SURQ GW_Q WYLD LAT Q Tlosses
M
(mm) 759.5 518.1 26.8 106.0 115.6 62.1 201.4 18.5 0.0
(%) 100.0 68.2 3.5 14.0 15.2 8.2 26.5 2.4 0.0
D

Note: P=precipitation, ET = evapotranspiration, SW = soil water, PERC = percolation

below root zone, SURQ = surface runoff, GW_Q = groundwater contribution to
TE

streamflow, WYLD = SURQ + LATQ + GW_Q - TLOSSES, LATQ = lateral flow into
stream, Tlosses = transmission losses

The water balances ratios (Table 9) were obtained from SWAT CHECK (White et al., 2012),
EP

which is a program that checks whether the produced results are within typical ranges
(Arnold et al., 2012). The SWAT-CHECK also prompts the user when they are out of typical
ranges. The obtained water balance ratios were within acceptable ranges.
C

Table 9: Annual water balance ratios

AC

Ratios Ratio values

Streamflow/ Precipitation 0.26
Baseflow/Total Flow 0.41
Surface Runoff/Total Flow 0.59
Percolation/ Precipitation 0.14
Deep Recharge/ Precipitation 0.01
ET/Precipitation 0.68

From Table 9, it can be interpreted that on account of surface water supply, the surface
depression storage loss or infiltration is about 6% of precipitation. These results agrees well

Page 15 of 20
 ACCEPTED MANUSCRIPT

with previous studies in Lesotho, whereby Bonney (1975) indicated that infiltration and
runoff were about 4% and 25% of precipitation respectively. The ratios (Table 9) showed the
partition of the precipitation in the catchment. The extreme values of the ratios can be of
interest for intervention or management measures or decisions by river basin managers.

3.2 WEAP Results

3.2.1 Calibration and Validation

PT
The model calibration NS and R2 for CG024 were 0.72 and 0.84 respectively. The validation
NS and R2 were: 0.73 and 0.74 respectively. The CG084 gauging station calibration NS and
R2 were: 0.55 and 0.64 respectively. The CG084 validation NS and R2 were: 0.63 and 0.89

RI
respectively.

3.2.2 Water Supply

SC
The water supply of South Phuthiatsana is shown in Figure 6 where the volumes are long
term monthly averages of the respective rivers. The graph show that the month of February
has the highest water supply and July up to September are dry months.

U
Insert Figure 6: South Phuthiatsana Water Supply

3.2.3 Reference Scenario

AN
This is the business as usual scenario which predicts the likelihood of events in the future if
the current trends continue. From WEAP modelling results, the most significant water
M
demands in the scenario are the: Metolong domestic water demands, Metolong industrial
water demands and catchment domestic water demands. The Metolong domestic demands
contribute 58%, the Metolong industrial demands contribute 34%, livestock contributes 1%
and the catchment domestic demands contribute 7% to the total demands (977.61 Mm3).
D

However, Metolong industrial demands of 1.46 Mm3 were not met. In addition, the
environmental demands of 2.29 Mm3 were not met. For monthly environmental unmet
TE

demands, October had the highest unmet demands of 1.2 Mm3 but for July, August and
September, they were fully met.

3.2.4 Possible Irrigation Expansion Scenario

EP

This scenario analysed an increase of irrigated area by 12.3%, in the South Phuthiatsana
catchment. From WEAP modelling , the 12.3% increase would make a total of 491.3 ha by
2035 from 27.03 ha in 2010. In this scenario, some demands were not met, mostly after 2024.
The total unmet demand were 4.44 Mm3, whereby the Metolong industrial demands
C

accounted for 33.05%, irrigation 65.56% and livestock 1.39%. The unmet environmental
demands had not changed in this scenario.
AC

4. Conclusions and recommendations

This was the first attempt to use SWAT and WEAP models in the South Phuthiatsana
catchment. Regardless of data unavailability, the two models can be used in the catchment.
The results from two models showed that the models can be supported with open sources data
from reliable websites and measured data.

It can be concluded that, the South Phuthiatsana SWAT model depends on soil, runoff and
channel parameters. The model was successfully calibrated at CG 024 with: p-factor of 65%,
r_factor of 0.58, NS of 0.59 and R2 of 0.59 and for validation period p-factor was 57%,

Page 16 of 20
 ACCEPTED MANUSCRIPT

r_factor was 1.34, NS was 0.52, and R2 was 0.66. The study had provided the water balance
of South Phuthiatsana catchment and the following can be concluded: 26% of rainfall form
streamflow, 41% of the total flow comes from baseflow while surface runoff accounts for
59%, 14% of precipitation percolates to shallow aquifer, 1% percolates to deep aquifer and
68% of precipitation is lost through evapotranspiration.

The WEAP model was calibrated using CG024 and CG084 stations. For calibration at
CG024, NS was 0.72 and R2 was 0.84 and for validation period, NS was 0.73 and R2 was

PT
0.74. The calibration at CG084, NS and R2 were 0.55 and 0.64 respectively and for validation
period, NS and R2 were 0.63 and 0.89 respectively.

RI
In the reference scenario, 1.46 Mm3 of Metolong industrial demands were not met and 2.29
Mm3 of environmental demands were not met. In the irrigation expansion scenario, the
environmental demands were still not met with October having the highest month with unmet
demands. Excluding the environmental unmet demands, the unmet demands were 4.44 Mm3

SC
and irrigation accounts for 65.65% of this amount. Beyond the 12.3% increase of irrigation
area, the amount of unmet demands will increase in future.

U
The following recommendations can be drawn from this study:
Following data scarcity in the catchment, the Lesotho Meteorological Services should
AN
improve and update the meteorological database on regular bases.
Assess the SWAT model performance using only available reanalysis data so as to
fully understand the potential of these data in the catchment.
Since SWAT model did not capture most of the peaks, it is recommended to use
M
alternative data inputs or other hydrological models.
Irrigation in the catchment should be carried out through a detailed irrigation plan and
this should include: Meteorological stations, irrigation systems designed for the site
D

and an irrigators association with experts forming part of the board. This can
contribute towards IWRM.
As part of a demand management strategy, improvement from sprinklers used in the
TE

study to drip irrigation. Sprinklers are used for irrigation in the catchment.
The unmet demands in the industrial sector can be managed through demand
management strategies like: recycling and reusing of water.
A further study on the level of uncertainties in the use of coupled SWAT and WEAP
EP

models.
C

Acknowledgements
Sincere gratitude is extended to the SADC's WaterNet for providing a study scholarship to the
AC

first author for the Master of Integrated Water Resources Management.

Page 17 of 20
 ACCEPTED MANUSCRIPT

5. References
Abbaspour, K. C., Vejdani, M. and Haghighat, S. 2007. SWAT-CUP calibration and

PT
Uncertainity programs for SWAT. The fourth International SWAT conference. Delft:
Netherlands, 1596-1602.
Arnold, J. G., Moriasi, D. N., Gassman, P. W., Abbaspour, K. C., White, M. J., Srinivasan,

RI
R., Harmel, R. D., van Griensven, A., van Liew, M. W., Kannan, N. and Jha, M. K.
2012. SWAT: Model use, Calibration and validation. Trans. ASABE (55)4, 1491-1508.
Arnold, J. G., Srinivasan, R., Muttiah, S., and Williams, J. R. 1998. Large area hydrologic

SC
modelling and assessment part 1: model development. Journal of the American Water
Resources Association 34, 73-89.
Arranz, R., and McCartney, M. 2007. Application of the Water Evaluation And Planning
(WEAP) model to assess future water demands and resources in the Olifants catchment,

U
South Africa, Colombo, Sri Lanka. IWMI working paper 116. International Water
Management Institue.
AN
Batjes, N. H. 2008. ISRIC-WISE Harmonized Global Soil Profile Dataset (ver. 3.1), Report
2008/02, Wageningen, http://www.isric.org/isric/Webdocs/Docs/ISIRIC Report 02.pdf.
Bonney, G. 1975. Groundwater exploration and borehole sitting. Department of Mines and
Geology, Report MG GB5. Maseru.
M
Bruns, and Ryan, M. 2012. Optimizing Maize Planting Date, Plant Population, and Fertilizer
Application for Lesotho Subsistence Farmers. Tennessee: Master's Thesis, University of
Tennessee.
Bureau of Statistics. 2007. 2006 Lesotho census of population ans housing preliminary results
D

report. Maseru.
Central Bank of Lesotho. 2011. Annual Report, Maseru.
TE

Chinnasamy, P., Bharati, L., Bhattarai, U., Khadka, A., Dahal, V. and Wahid, S. 2015. Impact
of planned water resource development on current and future water demand in the
Koshi River basin, Nepal, Water International, 40(7), 1004-1020, DOI:
10.1080/02508060.2015.1099192
EP

Driskell, S. 2008. GeoWikia. Available at: https://sites.google.com/site/geowikia/projects/

lesotho-community -councils. [Accessed on 02nd June 2015].
Droogers, P., de Boer, F., and Terink, W. 2014. Water Allocation Models for the Umbeluzi
River Basin, Mozambique. Wetter skip Frystan and ARA-Sul report future water:
C

132, The Netherlands.

Fadil, A., Rhinane, H., Kaoukaya, A., Kharchaf, Y., and Bachir, O. A. 2011. Hydrologic
AC

modeling of the Bouregreg Watershed (Morocco) Using GIS and SWAT Model.
Journal of Geographic Information System 3, 279-289.
FAO. 2008. The challenges of climate change. Ministerial conference on water for agriculture
and energy in Africa. Sirte/08/INF/5. Libyan Arab Jamahiriya Libya.
FAO/IIASA/ISRIC/ISS-CAS/JRC. 2009. Harmonized World Soil Database (Version 1.1).
FAO. Rome and IIASA, Laxenburg.
Gain, A.K. and Giupponi, C. 2015. A dynamic assessment of water scarcity risk in the
Lower Brahmaputra River Basin: an integrated approach. Ecological Indicators, 48,
120-131. doi: 10.1016/j.ecolind.2014.07.34

Page 18 of 20
 ACCEPTED MANUSCRIPT

Gies, L., and Merwade, V. 2013. Creating soil database using FAO soil and terrain database
of East Africa. Available at: http://web.ics.purdue.edu/~vmerwade/tutorial.html.
[Accessed on 06th April 2015].
Githui, F., Mutua, F. and Bauwens,W., 2009. Estimating the impacts of land-cover change on
runoff using the Soil and Water Assessment Tool (SWAT): case study of Nzoia
catchment, Kenya.Hydrological Sciences Journal, 54 (5), 899908.
Government of Lesotho. 2012. Metolong Dam ESIA. Final executive summary, Report E
1925-1. Maseru.

PT
Government of Lesotho. 2009. Environmental Impact Assessment for Lesotho Lowlands
Bulk Water Supply Scheme- Central Region, Project No. 9 ACP LSO 005. Maseru.
Government of Lesotho. 2008. Water Act 2008. Maseru.

RI
Haji, H. T. 2011. Impact of cliamte change on surface water availability in the Upper Vaal
River Basin. Masters Dissertation, Tshwane University of Technology.
Hatcher, K. J. 1995. WEAP: A comprehensive and integrated model of supply and demand.

SC
Preceding of the 1995 Georgia water resources conference. Athens, Georgia.
Heuvelmans, G., Muys, B., and Feven, J. 2004. Analysis of the spatial and variation in the
paramters of the SWAT model with application in Flanders, Norhtern Belgium.
Hydrology and Earth Systems Science 8(5), 931-939.

U
Hishinuma, S., Takeuchi, K. and Magome, J. 2014. Challenges of hydrological analysis for
water resource development in semi-arid mountainous regions: case study in Iran,
AN
Hydrological Sciences Journal 59(9-10), 1718-1737.
Hughes, D.A. 2008. Modelling semi-arid and arid hydrology and water resources: the
southern Africa experience. In H.S. Wheater, S. Sorooshian and K.D. Sharma, eds.
Hydrological modeling in arid and semi-arid areas. New York: Cambridge University
M
Press, 1-20.
Jeffares and Green (Consultancy). 2010. Climate change unlikely to decrease Metolong dam
yields: Engineering News. Available at: http://www.jgi.co.za/news/item/climate-
change-unlikely-to-decrease-metolong-dam-yields-engineering-news. [Accessed on 18th
D

May 2015].
Kajornrit, J., Wong, K., and Fung, C. C. 2012. A comparative analysis of the soft computing
TE

techniques used to estimate missing precipitation records. In. International

telecommunications society 19th Biennial conference. Bangkok Thailand.
Levite, H., Sally, H., and Cour, J. 2003. Testing water demands management scenarios in a
water-stressed basin in South Africa: Application of WEAP model. Physics and
EP

Chemistry of the Earth 28, 779-786.

Marteau, R., Sultan, B., Moron, V., Alhassane, A., Christian, B., & Traore, B. S. 2011. The
onset of the rainy season and farmers' sowing strategy for pearl millet cultivation in
South-West Niger. Agriculture and Forest Meteorology 151, 1356-1369.
C

McCartney M.P. and Girma, M.M. 2012. Evaluating the downstream implications of planned
water resource development in the Ethiopian portion of the Blue Nile River, Water
AC

International, 37(4), 362-379, DOI: 10.1080/02508060.2012.706384

McCartney, M. P., Yawson, D. K. and Huber-Lee, A. 2005. Simulating water resources
development in the Olifants catchment, South Africa. IWMI Theme 1. Working Paper
(Draft).
Mulungu, D. M. and Munishi, S. E. 2007. Simiyu River catchment parameterization using
SWAT model. Physics and Chemistry of the Earth 32, 1032-1039.
Mulungu, D. M. and Taipe, C. I. 2012. Water evaluation and planning in Wami River - Basin:
Application of the WEAP model. In E. Beall (Editor), Bioenergy and Food Security -
The BEFS Analysis for Tanzania FAO 54, 47-69.

Page 19 of 20
 ACCEPTED MANUSCRIPT

Ndomba, P., Mtalo, F. and Killingtveit, A., 2008. SWAT model application in a data scarce
tropical complex catchment in Tanzania. Physics and Chemistry of the Earth, 33 (8
13), 626632.
Ntai, P. J. 2011. Critical factors determining successful irrigation farming in Lesotho. Masters
dissertation, University of Pretoria.
Parkman, M. 2005. Lesotho Lowlands Water Supply Feasibility Study. Funded by the EU
EDF, Maseru.
Rodrigues, L., Asamoah, I., Andah, W., Kempt-Benedict, E., Sieber, J., and Huber-Lee, A.

PT
2005. Water Evaluation And Planning (WEAP), Small reservoir tool kit. Somerville.
Global Water Partnership. 2002. Integrated Water Resources Management in Mediterranean.
GWP-Med and MIO-ECSDE, Athens.

RI
Setegn, S. G., Srinivasan, R., and Dargahi, B. 2008. Hydrological modelling in the Lake Tana
Basin, Ethiopia Using SWAT model. The Open Hydrology Journal 2, 49-62.
Sieber, J., Swartz, C. and Huber-Lee. D. 2005. WEAP Water Evaluation and Planning system

SC
user guide for WEAP21. Boston: Stockholm, Environment Institute.
Sieber, J., Yates, D., Purkey, D. and Huber-Lee, A. 2004. WEAP: A demand, priority and
preference driven water planning model: Part 1: Model characteristics. Submitted to
Water International.

U
Skaggs, R., Janetos, T. C., Hibbard, K. A., and Rice, J. S. 2012. Climate and Energy-Water-
Land systems interactions. Washington United States of America.
AN
TAMS Consultants and Associates. 1996. Water Resources Management: Policy and
Strategies Study, Executive Summary. Final Report. Maseru, 1-6.
Teegavarapu, R. S., and Chandramouli, V. 2005. Improved weighting methods, deterministic
and stochastic data-driven models for estimation of missing precipitation records.
M
Journal of Hydrology 312, 191-206.
Tolmay, J., Rosenblum, M. L., Moletsane, M., Makula, M., and Pederson, T. 2000. The
introduction of disease and pest resistant wheat cultivars to small-scale farming systems
in the highlands of Lesotho. The Eleventh Regional Wheat Workshop for Eastern,
D

Central and Southern Africa . Addis Ababa Ethiopia, 336-340.

TWINLATIN. 2009. Cocibolca lake basin (Nicaragua). WP3: Hydrological modelling and
TE

extremes, 69-114.
van Griensven, A., T. Meixner, S. Grunwald, T., Bishop, Di Luzio, M. and Srinivasan, R.
(2006). A global sensitivity analysis tool for the parameters of multi-variable catchment
models. Journal of Hydrology 324(1-4), 10-23.
EP

Van Loon, A., and Doogers, P. 2006. Water evaluation and planning system, Kitui-Kenya-
Report no 2. Kenya.
White, M. J., Harmel, D. R., Arnold, G. J. and Williams, R. J. 2012. SWAT CHECK: A
screening tool to assist users in the identification of potential modle application
C

problems. Journal of Environmental Quality. Available at:

http://swatmodel.tamu.edu/softwares/ swat-check.
AC

Yates, D., Sieber, J. and Huber-Lee, A. 2005. WEAP21: A Demand-, Priority-, and
Preference-Driven Water Planning Model: Part 1: Model Characteristics. Water
International 30(4), 487-500.
Zhang, X., Srinivasan, R., and Hao, F. 2007. Predicting hydrologic response to climate
change in the Luohe River basin using the SWAT model. American society of
agriculture and biological engineers 50(3), 901-910.

Page 20 of 20
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

Highlights

SWAT showed that out of the precipitation: 26% is streamflow and 68% is evaporation

WEAP calibration and validation at two stations was satisfactory: R2 = 0.64 - 0.89

Reference scenario indicated that industrial and environmental demands were not met

PT
Irrigation expansion scenario accounted for 65.65% of the unmet demands

RI
Following unmet demands, it is recommended to have an integrated irrigation plan

U SC
AN
M
D
TE
C EP
AC