Remote Sensing Applications: Society and Environment 9 (2018) 107–115
Contents lists available at ScienceDirect
Remote Sensing Applications: Society and Environment
journal homepage: www.elsevier.com/locate/rsase
Mapping the spatial distribution of small reservoirs in the White Volta Sub- T
basin of Ghana
Benjamin Ghansaha, Eric Kwabena Forkuoa,⁎, Emmanuel Frimpong Oseib, Richard Ko cfi Appoh ,
Mensah Yaw Asarea, Nana Ama Browne Klusteb
a Department of Geomatic Engineering, Kwame Nkrumah University of Science and Technology, Kumasi, Ghana
b Remote Sensing and Climate Centre, Ghana Space Science and Technology Institute, Ghana
c Institution: Institute of Environment and Sanitation Studies, University of Ghana-Legon, Ghana
A R T I C L E I N F O A B S T R A C T
Keywords: Small reservoirs provide vital ecosystem services in the semi-arid part of Ghana. Their main role is the provisioning
Ghana of water for irrigation in this water-scare part of the country. They also serve as source of water for fishing,
Histogram thresholding recreation, drinking, and other domestic uses and play crucial roles in evaporation, water productivity, water
Landsat 8 images scarcity mitigation and climate modulation in this area. As such, their accessibility by the populace is vital to their
Small reservoirs well-being. Also, their locations, surface area and capacity (volume) are required for water resources assessments
Spatial distribution
White Volta sub-basin and hydrological modelling. This work was done to map the distribution of the small reservoirs in the White Volta
sub-basin in Ghana. The work assessed the accessibility of artificially built small reservoirs in the relevant com-
munities using satellite images and ground information. The study used six Landsat 8 Operational Land Imager
images covering the study area. Histogram thresholding technique was used to delineate the waterbodies from
their adjacent uplands. Accuracy assessments were done through field observed data and Google Earth images. The
method used produced a positional accuracy of 94%. The study estimated that there are approximately 254 small
reservoirs of surface areas between 1 and 53 ha in the sub-basin. The combined surface area and volume of the
small reservoirs were estimated to be 1585.265 ha and 21.154×106 m3 respectively. The study estimated only
35% of communities within the sub-basin have access to small reservoirs. The study concluded that cluster of
communities and population density are the general determinants of the distribution of small reservoirs in the sub-
basin. However, rainfall distribution and suitability of soil for large-scale crop production may also be some im-
portant factors dictating the distribution of small reservoirs in the White Volta sub-basin in Ghana.
1. Introduction To overcome water scarcity alongside creating new income sources
for the steadily growing populations of the sub-basin, hundreds of
The White Volta Basin in Ghana is a semi-arid land in sub-Saharan small-multi-purpose reservoirs were built in the colonial era up to the
Africa with highly variable climate (Mul et al., 2015). The area ex- 1960s (Venot, 2011). The priority of the development agenda in the
periences a unimodal rainfall pattern with the rainy season starting basin-wide scale during the 1970s and 1980s shifted towards medium
from April/May to September/October and dry season beginning No- and large scale dams with little attention to small reservoirs (Venot
vember and ending in March (Gyau-Boakye and Dapaah-Siakwan, et al., 2011). The growing disenchantment with the costs involved and
1999). Most of the inhabitants of the area are farmers who cultivate the social and environmental consequences of large-scale multi-purpose
crops and rear farm animals (Ofosu et al., 2010) and as such heavily dams (WCD, 2000) led to a renewed interest by donors in small re-
depends on water for their livelihood. However, apart from the fact that servoirs development in the 1990s (McCully and Pottinger, 2009;
the rainfall pattern does not supports all year water access, climatic Douxchamps et al., 2012), after the droughts of the 1980s which lead to
changes has coupled the uncertainties associated with access to water water scarcity, food shortages and hunger (Leemhuis et al., 2009).
for both domestic and agricultural use. As such, the development of Small reservoirs have since become a way to develop small-scale irri-
irrigation schemes is the only way farmers can cultivate in the dry gation (Venot and Krishnan, 2011), especially in areas prone to pro-
season and during the dry spells in the rainy season. blems with rainfall. They allow mitigation of negative impact of inter-
⁎ Corresponding author.
E-mail addresses: pecku2000@yahoo.com (B. Ghansah), eric.forkuo@gmail.com (E.K. Forkuo), osfremma@yahoo.com (E.F. Osei), rkappoh@gmail.com (R.K. Appoh),
mensah.yasare@gmail.com (M.Y. Asare), amabrowne@gmail.com (N.A.B. Kluste).
https://doi.org/10.1016/j.rsase.2017.12.003
Received 12 May 2017; Received in revised form 27 November 2017; Accepted 26 December 2017
Available online 09 January 2018
2352-9385/ © 2017 Published by Elsevier B.V.
B. Ghansah et al. Remote Sensing Applications: Society and Environment 9 (2018) 107–115
annual flooding, supply of water for livestock, domestic and irrigation north of the country namely; Northern Region, Upper West Region and
purposes (Jean-françois et al., 2015). Apart from the above mentioned Upper East Region. It is located between latitudes 8⁰42' N and 11⁰01' N
usage, it also provides water for fishing and construction of buildings, and longitudes 0⁰13' E and 2⁰30' W. The sub-basin covers a land area of
roads, etc. (Sally et al., 2011). about 49,583 km2 with all the river systems draining into the White
In the White Volta Sub-basin in Ghana, mapping of small reservoirs Volta River (Fig. 1(b)). The terrain is generally flat with few highlands.
using satellite images has mainly been done for the Upper East Region. The climate of the area is variable, both spatially and temporally
Although most of these studies used data from different satellite sensors with climatic zones spanning from Guinean Savanna Zone, Sudanian
and different images classification algorithms to delineate the water- Savanna Zone and a small portion of Sudano-Sahelan Zone (Kranjac-
bodies, the studies has been successful in providing statistics on the Berisavljevic et al., 1999). Annual rainfall in the Guinean Zone is be-
distribution, status and storage capacity of the small reservoirs in the tween 1000 and 1300 mm whiles the Sudano-Sahelian Zone and the
region. Liebe et al. (2005) used Landsat images to map the distributions Sudanian Zone receives annual rainfall between 500 and 900 mm and
and surface area of small reservoirs in the Upper East of Ghana. The between 900 and 1100 mm respectively (Kranjac-Berisavljevic et al.,
study provided quantitative values for the number and surface area of 1999) (Fig. 1(c)). About 70% of the annual rainfall in the area occurs
small reservoirs within the region. The study then used bathymetric during the months of July, August and September, with little or no
survey of some of the reservoirs to derive a relation for estimating the rainfall between November and March. Rainfall is often erratic with
volume of the reservoirs as a function of the satellite derived surface considerable variations between successive seasons, with regard to the
area. Liebe et al. (2008), Annor et al. (2009); Eilander et al. (2014) all time of onset, duration and amount (Obuobie, 2008). The rainy season
used images from active sensors to map the distribution of small re- is also characterized by dry spells of varying duration (Kadyampakeni
servoirs in the Upper East Region of Ghana. Although each of the three et al., 2017).
study applied different image classification algorithm to the radar
images used, the studies proved that radar images could also be used to
2.2. Data used
delineate small reservoirs from their adjacent uplands with high accu-
racy. Ofosu et al. (2014) also employed satellite images to map the
2.2.1. Field data
distribution of small reservoirs in the Volta Basin.
Field observed data used in this study include flood inundation
However due to demographic differences, size and scale of agri-
extent and coordinates of water-land transition points around some
cultural production as well as environmental and financial demands,
selected small reservoirs distributed within the study area. A handheld
building of small reservoirs within the White Volta Sub-basin has not
GPS and a digital camera were used to record these features during a
been well organized, and the collective number, size and capacity of
field visit to the study area in October 2016. The handheld GPS was also
some are not known. Poor maintenance culture and high run-offs in
used to measure the distance of each selected small reservoir to the
areas where these reservoirs are located has given way for the invasion
community/communities it serves. Each measurement was done by
of weeds (Annor et al., 2009) and sediments which often results in
tracking the approximate linear distance between the embankment of
siltation and eventual dysfunctioning of some of the reservoirs
the reservoir and a location within the community. This location is
(Andreini et al., 2009). High abstraction of water from these reservoirs
usually the second or third building of the community from the re-
as well as the extreme and prolonged drought events that characterize
servoir. Community members’ knowledge and participation in defining
the recent climatic conditions of the sub-basin (McCartney et al., 2012)
this distance was greatly utilized.
has also caused some of the reservoirs to dry-up and are hence no more
Various ancillary data were also used. These included vector dataset
in-use. These challenges affect the spatiotemporal distribution of the
of topographical map of Ghana with layers including contours, roads,
small reservoirs both temporarily and permanently. There is thus the
rivers and administrative boundaries of the country, projected onto the
need to map and update the statistics of these reservoirs to identify the
UTM Zone 30 N coordinates system. This dataset was obtained from the
current collective number, size and capacity of these reservoirs for
Survey and Mapping Division of the Ghana Lands Commission. Vectors
water resources assessment of the sub-basin. Also, none of the small
datasets of the agro-ecological zones and climatic zones were obtained
reservoirs mapping studies so far conducted has reported on the ac-
from the Savanna Accelerated Development Authority (SADA) website
cessibility of these small reservoirs to the communities within which
(http://www.sadagh.org/) and the dataset were already projected onto
they are found. This creates a knowledge gap in the per capita water
the UTM Zone 30 N coordinate system. Boundary shapefile of Africa
accessibility assessment within the sub-basin. It is therefore prudent to
countries was also obtained from the Thematic Mapping API website
assess the accessibility of small reservoirs within the sub-basin in spatial
(http://thematicmapping.org/api/). Though this layer had the WGS 84
context. Moreover, with Ghana government's short-term policy of pro-
coordinate system as its reference, no attempt to project the data onto
viding a water-supply dam for each community in the north of the
the UTM Zone 30 N coordinate system was made as that will distort the
country, long-term policy of creating a national spatial data infra-
data. However, this data was successfully used in depicting the study
structure as well as her goal of producing ‘sustainable data for sus-
location in relation to neighboring countries (Fig. 1(a)). These ancillary
tainable development’ (United Nations, 2015) as required by the in-
datasets provided information about the location, climate, agro-ecology
ternational community, there are ample reasons to assess the
and rivers systems the study area.
accessibility of small reservoirs to the populace within this water-scarce
A flowchart of the methodology used in this study is shown in Fig. 2.
area. This study was therefore conducted to provide statistics and ready
data on the distribution and accessibility of small reservoirs to com-
munities in the White Volta sub-basin in Ghana using Landsat 8 satellite 2.2.2. Satellite images
images and field data. Six Landsat 8 images covering the study area were downloaded from
the USGS website Global Visualization Viewer (http://glovis.usgs.gov/).
2. Methodology and materials These images were acquired in October and November 2016. This in-
terval is the period Leibe et al. (2005) showed that the reservoirs were
2.1. Study area likely to be fully filled with water and hence the satellite images captures
the full surface area of the reservoirs, which can give a good estimation
The area under this study is the part of the White Volta Sub-Basin of the surface area of the reservoirs. The characteristics of these images
within Ghana (Fig. 1(a),). This area spans through three regions in the are summarized in Table 1.
108
B. Ghansah et al. Remote Sensing Applications: Society and Environment 9 (2018) 107–115
Fig. 1. Study area: (a) Geographic location (b) River systems (c) Isohyets and agro-ecological zones of the White Volta Sub-basin in Ghana.
2.3. Image pre-processing proven images from OLI to be geometrically accurate (Storey et al.,
2014). As such, no further geometric correction were done for the
Both ESRI ArcGIS and Erdas Imagine were used to pre-process the images used in this study.
images. Landsat 8 OLI Level 1 T products have been geometrically Radiometric corrections were done by converting all the DN values
corrected using ground control points and digital terrain model cor- of each image band into top-of-atmosphere planetary reflectance with a
rections (Storey et al., 2014; DI/USGS, 2016). This has practically correction factor of the sun angle using Eqs. (1) and (2) below
Fig. 2. Flowchart of the methodology.
109
B. Ghansah et al. Remote Sensing Applications: Society and Environment 9 (2018) 107–115
Table 1 as buildings, large waterbodies and uplands features of the small re-
Characteristics of Landsat 8 data used. servoirs.
Sensor Path Row Date
Landsat 8 194 52 28-10-2016 2.4. Water feature extraction
Landsat 8 194 53 28-10-2016
Landsat 8 194 54 28-10-2016 Mapping of waterbodies from conventional land surveying is a te-
Landsat 8 195 52 20-11-2016 dious task and at times inaccessibility becomes a hindrance. However,
Landsat 8 195 53 20-11-2016 space-borne satellite systems observe the earth surface in entire field of
Landsat 8 195 54 20-11-2016
view of sensors and records ground objects (Duong, 2012). Extraction
of waterbodies from satellite images of varying spatial, temporal and
respectively (Landsat, U.S.G.S., 2016): spectral resolutions have been widely explored by researchers to obtain
and analyze geo-information (Jawak et al., 2015). The technique of
ρλ′ = Mρ*Qcal + Aρ (1) remote sensing involves the measuring of reflected radiance from an
ρλ = ρλ′/cos(θ) (2) object that is captured and this depends on the extent of electro-
magnetic radiation absorbed by the object; that is, the more it absorbs,
where: the less it reflect (Mishra and Prasad, 2015). Optical satellite systems
ρλ′ = Top of Atmosphere Planetary Reflectance are most frequently used in water body extraction research. The parts of
the electromagnetic spectrum covered by these sensors include the
Mρ = Band Specific Reflectance Multiplicative Scaling Factor for the Band Visible and Near- Infrared VNIR ranging from 0.4 to 1.3 µm, the
Qcal DN Values Shortwave Infrared SWIR between 1.3 and 3.0 µm, the Thermal In-=
frared TIR from 3.0 to 15.0 µm and the Long-Wavelength Infrared LWIR
Aρ = Band Specific Reflectance Additive Scaling Factor for the Band from (7–14 µm). A review of these sensors can be found in Nath and
Deb (2010). Various techniques and algorithms such as supervised and
ρλ =Correction Factor for Sun Angle unsupervised classification, feature extraction, and data fusion have
Cos(θ) Cosine of Local Sun Elevation Angle been adopted by researchers to extract water features from satellite=
imagery (e.g. Mcfeeters, 1996; Frazier and Page, 2000; Ji et al., 2009;
The bands were then stacked, mosaicked and clipped to the Du et al., 2014; Ouma and Tateishi, 2006; Bhardwaj et al., 2015; Jiang
boundary of the sub-basin to obtain a complete coverage of the study and Wang, 2016) with each method has its own merits and demerits
area. Though only the band 6 was used as the input data for the his- (Duong, 2012). The use of water indices and histogram thresholding,
togram thresholding (Section 2.4.1), the inclusion of the other bands which form part of feature extraction techniques (Duong, 2012), have
aided the visual identification and interpretation of other features such taken the center stage in waterbody extraction due to their simplicity
Fig. 3. (a) Histogram of the mosaicked L8 band 6 image retrieved from Erdas Imagine (b) Spread of the selected small reservoirs used for the histogram thresholding and the proximity
analysis (c) Scatter plot of the reflectance from the 76 water-land transition points around the selected reservoirs.
110
B. Ghansah et al. Remote Sensing Applications: Society and Environment 9 (2018) 107–115
and the accuracy they give. A detailed review of algorithms for ex- 2.6. Accuracy assessment
traction of water bodies from satellite image can be found in Jawak
et al. (2015) and Jiang et al. (2014). Histogram thresholding techniques In assessing the accuracy of the delineated small reservoirs, both
was used in extracting water features in this study. positional and lateral accuracies were assessed. Positional accuracy
defines the quantitative value that indicates the positional difference
between two geospatial layers or between a geospatial layer and geo-
2.4.1. Thresholding on SWIR
graphical reality (ArcGIS Desktop, 2017a, 2017b) whiles lateral accu-
The thresholding technique involves two parts; decomposition of
racy defines the quantitative value that represent the lateral differences
the image into sub images according to spectral reflectance patterns and
between two geospatial layers or between a geospatial layer and geo-
level slicing into water and land using different threshold values for the
graphical reality. In the positional accuracy assessment, the center co-
SWIR band (Duong, 2012). For this research work, it was done in Erdas
ordinates of 100 delineated small reservoirs were projected onto Google
Imagine where the Modular Tool was used to obtain a water only layer.
Earth image captured in 2016. These 100 coordinates included the
Deka et al. (2011) has shown that this method is effective and has been
known coordinates of the 30 small reservoirs that were visited during
used in many studies to delineate water bodies from their surrounding
the field survey. Manual inspection was done by visually counting the
uplands. The method results in precise grouping, for example, water
number of projected centroids that fell on designated reservoirs on the
pixels from land pixels from which a binary layer containing water-only
Google Earth. The total number of matching coordinates points were
can be created as it was done in Ghansah et al. (2016).
then counted to give the positional accuracy.
Histogram thresholding technique was executed on the Shortwave
Lateral accuracy assessment was done by comparing the surface
Infrared (SWIR) band 6 of the mosaicked images. The resultant histo-
area of 30 of the delineated reservoirs to the measured surface area of
gram displayed a double peak curve of water and land reflectance.
the corresponding 30 small reservoirs recorded using the handheld GPS
Between these double peaks lies the transition point (threshold pixel) of
during the field survey. A t-test computation was then done to find the
water and land, Fig. 3(a). Though accurate determination of this value
differences between the two sets of values.
can be a confusing because of reflection from mixed water-land fea-
tures, precise selection of this threshold value is possible and is essential
for obtaining a good classi cation. To avoid the confusion this re- 2.7. Spatial distribution of the reservoirsfi
presents, the inundation extents of 30 small reservoirs spatially dis-
tributed within the study area (Fig. 3(b)) were marked with a handheld The coordinates and surface areas of the small reservoirs were
GPS just after the peak flooding season in early October 2016. The computed in ArcGIS environment. To estimate the volume of the re-
decision to select these reservoirs were based on authors’ knowledge of servoirs, an existing regional area-volume relation that has been de-
these reservoirs and their accessibility. As such, coordinates of water veloped in previous studies for the area were considered. Leibe et al.
pixels at transition points between water and land around the 30 se- (2005) and Annor et al. (2009) independently developed equations that
lected small reservoirs within the study area were selected. This took relate the area of small reservoirs to their volume in the UER of Ghana.
the form of recording the GPS coordinates of the sample points (76 in Their studies used bathymetric measurements of some small reservoirs
total), which were then superimposed onto the mosaicked band 6 sa- to relate the surface area obtained from satellite images to the volume.
tellite images to extract their corresponding re ectance values. A As the region is ‘geomorphologically’ (Annor et al., 2009) identical,fl
scatter plot of the distribution of the re ectance was done and a mean they were able to develop precise relations between the surface areafl
was then computed to obtain a precise re ectance threshold value (Thr) and volume with goodness of fit above 95% in both studies. Thefl
for slicing the image, (see Fig. 3(c)). equations are (3) and (4) from Leibe et al. (2005) and Annor et al.
(2009) respectively.
2.5. Reservoirs delineation Volume = 0.00857*Area
1.4346 (m³) (3)
Volume = 0.00857*Area1.44 (m³) (4)
The binary image created was processed to exclude water pixels
from all other pixels leaving a water only layer. Small reservoirs were Where:
separated from other categories of water bodies. Considering the 30 m
spatial resolution of the Landsat images used, the probability of in- Volume is the volume of the reservoir measured in cubic meter and
accurately identifying smaller features is higher (Leibe et al., 2005). Area is the surface area of the reservoir measured in squared meter
Leibe et al. (2005) defined small reservoirs to have a surface area be-
tween 1 and 100 ha based on the 30 m moderate resolution of Landsat Thus, both studies produced very similar area to volume relations
ETM images. Likewise, due to the 30 m moderate resolution of the for the region. As Leibe et al. (2005) used Landsat images as well as
Landsat 8 images used, this study regarded small reservoirs to have similarity in the geomorphology between the Upper East Region and
surface area between 1 and 100 ha inclusive. As such all waterbodies the sub-basin, the relation established by Leibe et al. (2005), was used
outside this range were deleted from the layer. Though natural re- to compute the volume of the small reservoirs delineated in this study.
servoirs (ponds) are used for fishing activities in this area, they are not
known to be used for irrigational purposes (Mul et al., 2018) and were 2.8. Proximity of reservoir to communities
therefore considered to be of limited importance to the overall purpose
of this water accessibility study. Moreover, majority of these ponds As one of the primary objectives, this study assessed which com-
completely dries out during the dry season and ceased to be of benefit munities within the study area have access to small reservoirs. To do
in terms of water accessibility. As such, all reservoirs within the this assessment, a proximity analysis was performed in ESRI ArcGIS
floodplain of the major river systems in the sub-basin were deleted from using the Generate Near Table tool. Based on Dijkstra's algorithm to
the layer. This was achieved by masking off all waterbodies within the find the shortest path between an input feature P, and a near feature S,
floodplains of the major rivers by superimposing a flood inundation this tool computes the shortest distance L, angle (θ⁰), coordinates
layer (developed using Height Above Nearest Drainage (HAND) con- (XP , YP)and (XsYs) and other proximity information between the two
tours (Nobre et al., 2016)) of the White Volta River on the small re- closest features. In cases where there are multiple points (as in this
servoirs layer generated. This step was necessary to ensure that almost study) in both the input feature and the near feature, the tool computes
all natural reservoirs (ponds) has been removed and the small reservoirs the distances between all the points and selects the shortest distances
layer obtained contains only artificially built infrastructures. and other proximity information between each closest input feature.
111
B. Ghansah et al. Remote Sensing Applications: Society and Environment 9 (2018) 107–115
This tool then writes the results to a new table (ArcGIS Desktop, 2017a, Table 2
2017b). Composition and distribution of the small reservoirs.
Sample point features of 720 communities (240 communities from
Area (ha) Volume (m3) Total number
each region) obtained from the national database and from the field
survey was used as the input feature. The small reservoirs layer deli- Northern 298.540 4963761.825 51
neated in this study was then used as the near feature. It was observed UE 1077.864 14693963.71 171
from the fieldwork that almost all the small reservoirs surveyed were UW 208.861 2885254.95 32
Total 1585.265 22542980.49 254
sited within 2 km of the community/communities they serve. It was
therefore assumed that any community/communities without a re-
servoir within a 2 km radius is deemed not to have access to a small whiles this study obtained 171 small reservoirs in the UER. The dif-
reservoir. As such, a 2 km distance was specified for the proximity ference in the number of small reservoirs in this region indicate an
analysis. increase of 17 small reservoirs in-between the dates of acquisition of the
satellites images used in these studies (1999/2000 and 2016).
3. Results
3.3. Accuracy assessments
3.1. Waterbodies extraction
Out of the 100 coordinates that were projected onto the Google
The thresholding technique was successful in extracting the small Earth, 94 of them fell on reservoirs. Of the 30 coordinates points which
reservoirs from their adjacent uplands. By acquiring training samples were known earlier through the ground survey, all the coordinates fell
from the field to define the spectral threshold value between the land exactly on the designated reservoirs. As such, an accuracy of 94% was
and waterbodies, precision of the thresholding technique was obtained. obtained for the positional accuracy assessment. This provided a re-
A reflectance threshold value of 0.055 was obtained in this study and liability of the classification methods used (Fig. 5).
was thus used to delineate the water from the adjacent surrounding. All The lateral accuracy assessment indicated that the classification
reflectance values between −0.100 and 0.055 inclusive were deemed procedure used in this study marginally underestimated and over-
to be water and values between 0.056 and 0.609 (Fig. 3(b), Section estimated the surface areas. The maximum difference was 1.909 and
2.4.1) were deemed to be adjacent uplands of the waterbodies. the minimum was 0.165 ha, respectively. A t-test of the observed sur-
face areas and delineated surface areas returned a value of 0.317, in-
3.2. Small reservoirs in the white volta sub-basin dicating the similarity between the two sets of values, (Table 3). Con-
sidering the differences in field observation dates and image acquisition
The results obtained from this study indicated that there are an dates, the estimation was deemed precise.
estimated 254 small reservoirs in the White Volta sub-basin in Ghana.
These are man-made dams with surface areas ranging between 1 and 3.4. Communities access to small reservoirs
54 ha (Fig. 4) and with storage capacity between 4.780 × 103 and
1.398 × 106 m3 (Table 2). The proximity analysis estimated that only 255 out of the 720 com-
In a comparison of the results obtained in this study and Leibe et al. munities has access to a small reservoir (Fig. 6). The analysis also showed
(2005), Leibe et al. (2005) obtained 154 small reservoirs in the UER that one reservoir might serve more than one community Table 4
Fig. 4. Distribution of small reservoirs by
surface area.
112
B. Ghansah et al. Remote Sensing Applications: Society and Environment 9 (2018) 107–115
Fig. 5. Delineated small reservoirs superimposed on a Google Earth image.
4. Discussion
Table 3
Measured and delineated small reservoirs.
The results point out some general determinants of distribution of
Name of Measured Area Delineated Area Estimated volume small reservoirs in the sub-basin namely: clusters of communities and
reservoir (ha) (ha) (m3) high human and livestock population centers. Fig. 6 indicates that areas
around Bolgatanga, Navrongo, Bawku and Tamale, which are the ca-
1 Binduri 16.231 16.958 272296.3
2 Bogorogo 5.132 3.878 32794.24 pitals towns of the Bolgatanga municipality, Kassena-Nankana East
3 Botingli 22.124 23.482 434347.5 municipality, Bawku municipality and the Tamale metropolitan re-
4 Chiana 3.152 2.144 14013.8 spectively, have high clusters of small reservoirs. These municipalities
5 Dane 12.810 13.952 205814.7 and metropolitan areas are the most populous administrative areas
6 Dimbasinia 8.093 9.024 110153
7 Djarwia 8.044 9.303 115071.4 within the sub-basin (Ghana Statistical Service (GSS), 2012) and their
8 Garu 10.697 11.682 159529.5 population has been increasing since the 1960s (Ghana Statistical
9 Guropisi 2.791 4.515 40789.92 Service (GSS (2013)). They also have high number of peri-urban set-
10 Heng 9.172 9.007 109855.4 tlements (Ghana Statistical Service (GSS (2013)) and thus are high
11 Kagbiri 18.633 19.001 320563.1 population density areas. MOFA Ministry of Food and Agriculture
12 Kajelo 7.594 6.040 61923.63
13 Kandiga 15.562 16.381 259103.7 (2004) indicated that large and commercial livestock keepers are pre-
14 Karaga 13.512 14.220 211509.8 dominantly located in urban- and peri-urban areas. Blench (2006) re-
15 Kayoro 3.241 2.513 17599.36 ported that small reservoirs had been traditionally constructed to trap
16 Kwalugu 16.063 15.689 243546.4 water to serve the growing needs of both humans and livestock, though
17 Mengwe 6.223 4.831 44946.96
18 Nafkuliga 22.023 22.426 406601.9 irrigational purposes has taken over this lead services. Thus, the de-
19 Pulluma 30.941 31.490 661697.2 mand for water for urban and peri-urban agriculture, livestock wa-
20 Samboligo 2.626 1.638 9524.354 tering, as well as domestic use, has been the main driver for the con-
21 Sandema 6.186 4.996 47165.51 struction of small reservoirs within the sub-basin. This accounts for the
22 Sankpala 2.678 1.082 5253.918 presence of high number of small reservoirs within areas where the
23 Shia 12.982 13.722 200964.7
24 Tumu 7.391 6.430 67739.23 population density is high and where there is cluster of communities.
25 voggo 40.243 42.152 1005414 Also, the increase in the number of small reservoirs in the UER between
26 Walembelle 12.937 13.489 196087.4 the years 1999/2000 and 2016 (Section 3.2) and population statistics
27 Wempisi 13.936 15.171 232093.9 from the GSS indicates that increasing human population correlates
28 Wiaga 9.825 9.366 116191
29 Winkogo 5.142 6.560 69712.57 with increasing number of small reservoirs, which thus collaborates
30 Yusin dam 4.435 5.229 50352.87 with the above discussion.
The results also show that though the UER is the smallest (in land
size) among the three Northern Regions, it is home to most of the re-
servoirs within the sub-basin. The region is home to 67% of the total
number of reservoirs in the three regions. The high number of re-
servoirs in this region can be attributed to the low amount of rainfall
113
B. Ghansah et al. Remote Sensing Applications: Society and Environment 9 (2018) 107–115
Fig. 6. Communities with and without ac-
cess to small reservoir.
Table 4 agricultural activities making the use of reservoirs not highly patronize
Statistics of the accessibility of small reservoirs by communities. as compared to the UER.
Communities with Communities without Total number of
access to small access to small communities 5. Conclusion
reservoirs reservoirs
This work mapped the distribution and assessed the accessibility of
Number % Number % artificially built small reservoirs in communities within the White Volta
Northern 56 21.96 184 39.57 240 Basin in Ghana. Data used for the study included six Landsat 8 OLI
UE 162 63.53 78 16.77 240 images covering the study area, ground information obtained through
UW 37 14.51 203 43.66 240 field surveys of the area and ancillary data obtained from national
Total 255 100 465 100 720 departments and online sources. Histogram thresholding technique was
applied on the images to delineate the waterbodies from their adjacent
uplands. Accuracy assessments were done through field observed data
received in the region as compared to the other two regions (Obuobie and Google Earth images. The method used produced a positional ac-
et al., 2017; Obuobie et al., 2018), encouraging the building of more curacy of 94% and a t-test value of 0.317 was obtained for lateral ac-
reservoirs to sustain life in that region. Fig. 4 reveals that, the north- curacy. The study estimated that there are approximately 254 small
eastern part of the region has larger reservoirs compared to the western reservoirs of surface areas between 1 and 53 ha in the sub-basin. Of the
side. This is mainly due to water demand for the large-scale cultivation total number of reservoirs, 67% are located in the UER whiles the re-
of onions in this part of the region. The occurrence and suitability of the maining 20% and 13% are located in the NR and UWR respectively. The
soil (Lixisols) in the northeastern part of the region for the cultivation of combined surface area and volume of the small reservoirs were esti-
onions (Sinnadurai and Abu, 1977), which is a highly economic com- mated to be 1585.265 ha and 21.154 × 106 m3 respectively. The study
modity, is higher than the other parts of the region and even the other estimated that only 35% of communities within the sub-basin have
two regions. access to small reservoirs. Conclusions were drawn to the finding that
The NR is the biggest region (in land size) in the sub-basin. The cluster of communities and human and livestock population density are
UWR is the second largest. However, the NR and UWR has only 20% the general determinants of the distribution of small reservoirs in the
and 13% respectively of the total number of the reservoirs in the sub- sub-basin. However, rainfall distribution and suitability of soil for large-
basin. This is explained by the reasons that rainfall amount received in scale crop production may also be some important factors dictating the
these regions is higher than the UER (Obuobie et al., 2017; Obuobie distribution of small reservoirs in the White Volta sub-basin in Ghana.
et al., 2018), and hence they are less water-stressed than the UER. The Areas for potential siting of small reservoirs includes the southwestern
NR in particular, is the confluence of the two main rivers in the part of the UWR and the Western half of NR where there are cluster of
northern part of Ghana, the White Volta River and the Black Volta communities but depleted of reservoirs.
River. Majority of the tributaries of these main rivers (Fig. 1(c), Section However, the 30 m moderate resolution of the Landsat 8 images
2.1) are also found within this region (McCartney et al., 2012). Farmers used were not deemed high enough to map reservoirs with surface areas
therefore tie their water needs more to these rivers systems than to less than 1 ha. Higher resolution satellite images are therefore needed
small reservoirs. This reduces their dependence on small reservoirs for to map smaller waterbodies. Integrated with crowdsourcing data, high
114
B. Ghansah et al. Remote Sensing Applications: Society and Environment 9 (2018) 107–115
resolution images can also help to map out more communities espe- Ji, L., Zhang, L., Wylie, B., 2009. Analysis of dynamic thresholds for the normalized
cially those in remote areas. There is also a need to develop a stan- difference water index. Photogramm. Eng. Remote Sens. 75 (11), 1307–1314.
Kadyampakeni, D.M., Mul, M.L., Obuobie, E., Appoh, R., Owusu, A., Ghansah, B., Boakye-
dardized hierarchical classification system for communities based on Acheampong, E., Barron, J., 2017. Agro-climatic and hydrological characterization of
population in order to differentiate between cities, towns, villages and selected watersheds in northern Ghana. Colombo, Sri Lanka: International Water
other settlements. This is because this study classified a settlement of Management Institute (IWMI), p. 40. (IWMI Working Paper 173). 〈https://dx.doi.
about 10 building and a town of thousands of buildings as a community org/10.5337/2017.209〉.Kranjac-Berisavljevic, G., Bayorbor, T.B., Abdulai, A.S., Obeng, F., Blench, R.M., Turton,
each in the proximity analysis, due to their distinct names and C.N., Boyd, C., Drake, E., 1999. Rethinking Natural Resource Degradation in Semi-
boundary. arid Sub-SaharanAfrica: The Case of Semi-arid Ghana. University for Development
Studies, Tamale and ODI, London.
Leemhuis, C., Jung, G., Kasei, R., Liebe, J., 2009. The Volta Basin water allocation system:
Acknowledgements assessing the impact of small-scale reservoir development on the water resources of
the Volta basin, WestAfrica. Adv. Geosci. 21, 57–62.
The authors would like to appreciate Dr. Emmanuel Obuobie of the Liebe, J., Van De Giesen, N., Andreini, M., 2005. Estimation of small reservoir storage
capacities in a semi-arid environment: a case study in the Upper East Region of
Water Research Institute of the Council for Scientific and Industrial Ghana. Phys. Chem. Earth, Parts A/B/C. 30 (6), 448–454.
Research Ghana for reviewing this work. Liebe, J.R., Van De Giesen, N., Andreini, M.S., Steenhuis, T.S., Walter, M.T., 2009.
Suitability and limitations of ENVISAT ASAR for monitoring small reservoirs in a
semiarid area. IEEE Trans. Geosci. Remote Sens. 47 (5), 1536–1547.
References McCartney, M., Forkuor, G., Sood, A., Amisigo, B., Hattermann, F., Muthuwatta, L., 2012.
The Water Resource Implications of Changing Climate in the Volta River Basin.
Andreini, M., Schuetz, T., Senzanje, A., Rodriguez, L., Andah, W., Cecchi, P., Boelee, E., Colombo, Sri Lanka. International Water Management Institute (IWMI) (IWMI
Giesen, N., Kemp-Benedikt, E., Liebe, J.R., 2009. Small Multi-purpose Reservoir Research Report 146).
Ensemble Planning. McCully, P., Pottinger, L., 2009. Spreading the water wealth: making water infrastructure
Annor, F.O., Van De Giesen, N., Liebe, J., Van de Zaag, P., Tilmant, A., Odai, S.N., 2009. work for the poor. Ecol. L Curr. 36, 177.
Delineation of small reservoirs using radar imagery in a semi-arid environment: a McFeeters, S.K., 1996. The use of the Normalized Difference Water Index (NDWI) in the
case study in the upper east region of Ghana. Phys. Chem. Earth Parts A/B/C 34 (4), delineation of open water features. Int. J. Remote Sens. 17 (7), 1425–1432.
309–315. Mishra, K., Prasad, P.R.C., 2015. Automatic extraction of water bodies from landsat
ArcGIS Desktop, 2017a. Generate Near Table: An overview of the proximity toolset. 2017 imagery using perceptron model. J. Comput. Environ. Sci. 2015, 1–9.
Esri. Online 〈http://pro.arcgis.com/en/pro-app/tool-reference/analysis/generate- MOFA (Ministry of Food and Agriculture), 2004. Livestock Development in Ghana
near-table.htm〉. Policies and Strategies. Ministry of Food and Agriculture, Ghana, Accra.
ArcGIS Desktop, 2017b. What is positional accuracy assessment: Assessing positional Mul M., Pettinotti L., Amonoo N.A., Bekoe-Obeng E., Obuobie E., 2018. Dependence of
Accuracy, ArcMap 10.5. Available online at 〈http://desktop.arcgis.com/en/arcmap/ riparian communities on ecosystem services in Northern Ghana. International Water
latest/extensions/data-reviewer/what-is-positional-accuracy-assessment.htm〉. Management Institute (IWMI). IWMI Working Paper.
Bhardwaj, A., Singh, M.K., Joshi, P.K., Singh, S., Sam, L., Gupta, R.D., Kumar, R., 2015. A Mul, M., Obuobie, E., Appoh, R., Kankam-Yeboah, K., Bekoe-Obeng, E., Amisigo, B.,
lake detection algorithm (LDA) using Landsat 8 data: a comparative approach in Logah, F.Y., Ghansah, B., McCartney, M., 2015. Water Resources Assessment of the
glacial environment. Int. J. Appl. Earth Obs. Geoinf. 38, 150–163. Volta River Basin 166 International Water Management Institute (IWMI), Colombo,
Blench, R.M., 2006. background Conditions in Upper East Region, Northern Ghana, 2005. Sri Lanka (IWMI).
International Fund for Agricultural Development (IFAD), Rome. Nath, R.K., Deb, S.K., 2010. ‘Water-body area extraction from high resolution satellite
Deka, J., Tripathi, O.P., Khan, M.L., 2011. A multi-temporal remote sensing approach for images-an introduction, review, and Comparison’. Int. J. Image Process. (IJIP) 3 (6),
monitoring changes in spatial extent of freshwater Lake of Deepor Beel Ramsar Site, a 353–372. 〈http://www.cscjournals.org/csc〉 (3)(6)(, pp. 353-372).
major wetland of Assam. J. Wetl. Ecol. 2011 (5), 40–47. Nobre, A.D., Cuartas, L.A., Momo, M.R., Severo, D.L., Pinheiro, A., Nobre, C.A., 2016.
Landsat, U.S.G.S., 2016. 8 (L8) Data Users Handbook Version 2.0. EROS, Sioux Falls, South HAND contour: a new proxy predictor of inundation extent. Hydrol. Process. 30 (2),
Dakota. 320–333.
Douxchamps, S., Ayantunde, A., Barron, J., 2012. Evolution of Agricultural Water Obuobie, E., Logah, F., Asante-Sasu, C., Osei-Owusu, M., 2018. Climate Trends and
Management in Rainfed Crop-Livestock Systems of the Volta Basin. Colombo, Sri FutureClimate Change Projections for the White Volta, Black Volta and Oti River
Lanka: CGIAR Challenge Program for Water and Food (CPWF). 74p. (CPWF R4D basins.
Working Paper Series 04).Duong, N. D., 2012. Water body extraction from multi Obuobie, E., Logah, F., Asante-Sasu, C., Osei-Owusu, M., 2017. Climate Trends and Future
spectral image by spectral pattern analysis. International Archives of the Climate Change Projections for the White Volta, Black Volta and Oti River basins.
Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume 39, p. B8. Technical report of CSIR-Water Research Institute, p. 89.
Du, Z., Li, W., Zhou, D., Tian, L., Ling, F., Wang, H., Gui, Y., Sun, B., 2014. Analysis of Obuobie, E., 2008. Estimation of groundwater recharge in the context of future climate
Landsat-8 OLI imagery for land surface water mapping. Remote Sens. Lett. 5 (7), change in the White Volta River Basin, WestAfrica. ZEF.
672–681. Ofosu, E.A., Zaag, P., van de van der Giesen, N., Odai, S.N., Amanor, R., 2014. Analysis of
Eilander, D., Annor, F.O., Iannini, L., van de Giesen, N., 2014. Remotely sensed mon- upscaling of irrigation development in the white Volta sub-Basin. JENRM 1 (1),
itoring of small reservoir dynamics: a Bayesian approach. Remote Sens. 6 (2), 36–43.
1191–1210. Ofosu, E.A., van der Zaag, P., van de Giesen, N., Odai, S.N., 2010. Productivity of irri-
Frazier, P.S., Page, K.J., 2000. Water body detection and delineation with Landsat TM gation technologies in the white volta basin. Phys. Chem. Earth 35, 706–716.
data. Photogramm. Eng. Remote Sens. 66 (12), 1461–1468. Ouma, Y.O., Tateishi, R., 2006. A water index for rapid mapping of shoreline changes of
Ghana Statistical Service (GSS, 2013. 2010 population & Housing Census. National five East African Rift Valley lakes: an empirical analysis using Landsat TM and ETM+
Analytical Report. May 2013. data. Int. J. Remote Sens. 27 (15), 3153–3181.
Ghana Statistical Service (GSS),, 2012. 2010 Population & Housing Census. Summary Sally, H., Lévite, H., Cour, J., 2011. Local water management of small reservoirs: lessons
Report of Final Results. May 2012. from two case studies in Burkina Faso. Water Altern. 4 (3), 365.
Ghansah, B., Asare, Y.M., Tchao, E.T., Forkuo, E.K., 2016. Mapping the spatial changes in Sinnadurai, S., Abu, J.F., 1977. Onion farming in Ghana. Econ. Bot. 31 (3), 312–314.
Lake Volta using multitemporal remote sensing approach. Lakes Reserv. Res. Manag. Storey, J., Choate, M., Lee, K., 2014. Landsat 8 operational land imager on-orbit geo-
21, 206–215. metric calibration and performance. Remote Sens. 2014 (6), 11127–11152.
Gyau-Boakye, P., Dapaah-Siakwan, S., 1999. Groundwater: solution to Ghana's rural United Nations, 2015. The Millennium Development Goals Report 2015. United Nations,
water supply industry? Ghana Eng (Accra, Ghana). New York, 2015.
Jawak, S.D., Kulkarni, K., Luis, A.J., 2015. A review on extraction of lakes from remotely Venot, J.P., 2011. What commons? Rethinking participation in the sub-Saharan African
sensed optical satellite data with a special focus on cryospheric lakes. Adv. Remote water sector. Hyderabad, s.n.
Sens. 4 (3), 196–213. Venot, J.P., Andreini, M., Pinkstaff, C.B., 2011. Planning and corrupting water resources
Jiang, H., et al., 2014. An automated method for extracting rivers and lakes from landsat development: the case of small reservoirs in Ghana. Water Altern. 4 (3), 399.
imagery. Remote Sens. 6, 5067–5089. Venot, J.P., Krishnan, J., 2011. Discursive framing: debates over small reservoirs in the
Jiang, L., Wang, Y., 2016. Mapping Zoige alpine wetland using random forest classifi- rural South. Water Altern. 4 (3), 316.
cation and Landsat 5 TM data. Stud. Surv. Mapp. Sci. (SSMS). 4, 24–27.
115