Computers and Electronics in Agriculture 157 (2019) 110–125

Contents lists available at ScienceDirect

Computers and Electronics in Agriculture

journal homepage: www.elsevier.com/locate/compag

Original papers

Effect of climate change on land suitability for surface irrigation and T

irrigation potential of the shallow groundwater in Ghana
⁎
Abeyou W. Worqlula, , Yihun T. Dileb, Jaehak Jeonga, Zenebe Adimassuc, Nicole Lefored,
Thomas Gerika, Raghavan Srinivasanb, Neville Clarkee
a
Texas A&M AgriLife Research, Temple, TX, USA
b
Texas A&M University, College Station, TX, USA
c
The International Water Management Institute, Accra, Ghana
d
International Water Management Institute (IWMI), Pretoria, South Africa
e
The Norman Borlaug Institute for International Agriculture, Texas A&M AgriLife Research, College Station, TX, USA

A R T I C LE I N FO A B S T R A C T

Keywords: Estimating the potential land resources suitable for irrigation and evaluating the possible impact of climate
Land suitability change on land suitability is essential for planning a sustainable agricultural system. This study applied a GIS-
Groundwater based Multi-Criteria Evaluation (MCE) technique to evaluate the suitability of land for irrigation in Ghana for a
Storie index baseline period (1990 to 2010) and future time horizons 2050s (2041 to 2060) and 2070s (2061 to 2080). Key
Multi-criteria
factors considered to evaluate the suitability of the land for irrigation include biophysical features (such as
Ghana
climate, land use, soil, and slope) and socioeconomic factors (such as proximity to roads and population density).
Climate change
These factors were weighted using a pairwise comparison matrix then reclassified and overlaid on a 30 m grid to
estimate the irrigation potential of the country. Groundwater data from the British Geological Survey (BGS) were
superimposed onto the land suitability map layer to evaluate the irrigation potential and the accessibility of
shallow groundwater with simple water lifting technologies. Downscaled and bias-corrected future climate data
from HadGEM2-ES under Representative Concentration Pathways (RCP) 4.5 emission scenario were used to
represent the future climate horizon. Due to climate change, on average, rainfall will increase by 15 mm and
20 mm from the baseline period in the 2050s and 2070s, respectively. The average temperature shows a con-
sistent increase in the majority of Ghana and a higher rate of increase is expected in the 2070s. Consequently, the
rising temperature will increase the potential evapotranspiration by 6.0% and 7.6% in the 2050s and 2070s,
respectively. The suitability analysis indicates that approximately 9% of the country is suitable for surface ir-
rigation under the baseline period. A large portion of the potential land is located in the southwestern part of the
country. The potential suitable land has an average groundwater access of 12 m from the surface with an average
borehole potential yield of 2.5 L/second, which makes it favorable for utilization of simple water lifting tech-
nologies. Due to climate change, 9.5% of the suitable land will become unfavorable for irrigation in 2050s, and it
is expected to reach 17% in 2070s.

1. Introduction vulnerability (Antwi-Agyei et al., 2012; Laube et al., 2012). Less than
2% of the total cultivatable area in Ghana is estimated to be under
Agriculture is critical to the economy of Ghana, which employs irrigated production (Gumma and Pavelic, 2013; Mendes et al., 2014;
more than 50% of the population on a formal or informal basis and Namara et al., 2011a, b). Moreover, the area under small-scale irriga-
contributing 25% of the gross domestic production (GDP) and export tion by individual farmers in Ghana has been estimated to be about five
earnings (Forkuor et al., 2013; Heintz, 2005). Agriculture has grown times that under formal private or public communal irrigation schemes
consistently by more than 5% annually for the last 25 years (Mendes (Dittoh et al., 2013). The direction of recent government policy direc-
et al., 2014). Despite this consistent growth, the country remains a tion is to increase irrigated agriculture to enhance productivity (MOFA,
major net importer of agricultural food products (Ashitey and Rondon, 2014) and ensure food security. Hence, assessing the suitability of areas
2012) due to heavy reliance on rainfed production and drought for irrigation will be critically important for planning and

⁎
Corresponding author. Tel.: +1 254 774 6020.
E-mail address: [email protected] (A.W. Worqlul).

https://doi.org/10.1016/j.compag.2018.12.040
Received 3 August 2018; Received in revised form 18 December 2018; Accepted 21 December 2018
0168-1699/ Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).
A.W. Worqlul et al. Computers and Electronics in Agriculture 157 (2019) 110–125

Fig. 1. Spatial distribution of rainfall gauging stations in Ghana with the three major river basins, and a 30 m-resolution digital elevation model as a background.

implementation. Some studies suggest that irrigation development, in- mechanisms to put in place to ensure resilience to reduce the vulner-
cluding agricultural reformulation in terms of policy, technology, and ability facing the nation in the future.
management practices, could contribute to greater food security and This study, therefore, uses a GIS-based Multi-Criteria Evaluation
increase overall economic growth (Giordano and de Fraiture, 2014; (MCE) technique to identify the spatial distribution of the land that is
Giordano et al., 2012; Worqlul et al., 2018a). This paper, therefore, potentially suitable for shallow groundwater irrigation under both
explores the potential for implementing irrigated agriculture in Ghana, current and future climate conditions. The MCE technique identifies the
particularly using shallow groundwater. Ghana appears to have abun- potential suitable land by considering multiple factors affecting the
dant water resources to increase irrigation rainwater harvesting, sur- suitability of a certain land area for irrigation. In MCE, these factors
face storage and groundwater (Ofosu, 2011). However, rainfall is un- affecting are mapped, weighted to determine their relative impact, and
evenly distributed across the country, both spatially and seasonally then combined to develop a single indexed output. MCE has been ap-
(Adimassu et al., 2016; Rademacher-Schulz et al., 2014). Mendes et al. plied to land suitability analysis for irrigation (Assefa et al., 2018;
(2014) indicated that if managed well, the country’s surface runoff and Mustafa et al., 2011; Pereira and Duckstein, 1993; Prakash, 2003;
groundwater could meet most domestic and irrigation requirements. Zhang et al., 2015), and also in other field of studies including site
However, irrigation using surface water is limited due to the capital selection for convenience store location (Kuo et al., 2002), urban land
investment required for channeling the surface water to the potential use planning (Mosadeghi et al., 2015), hospital site selection (Vahidnia
irrigable areas over long distances (Obuobie et al., 2013). Widespread et al., 2009), thermal power plant site selection (Choudhary and
groundwater availability as a buffer against drought and climate Shankar, 2012), solid waste disposal site identification (Şener et al.,
change variability could potentially transform the agricultural sector in 2010; Sumathi et al., 2008; Wang et al., 2009), mapping of erosion-
Ghana (Obuobie et al., 2012). However, there are only limited studies prone areas (Assefa et al., 2015; Pal, 2016), and financial decision-
that provide a robust estimate of the potential suitable land for irriga- making (Steuer and Na, 2003). This study will be the first of its kind to
tion using groundwater in Ghana. Moreover, climate change has be- provide spatially explicit potential suitable land for irrigated agri-
come one of the biggest challenges of this century. The threat of climate culture for the current and future time horizons. The findings may
change is multidimensional; it affects agriculture and food security provide valuable information to researchers, decision makers, practi-
(Nelson et al., 2009; Rosenzweig and Parry, 1994; Schmidhuber and tioners, and investors to prioritize irrigated agriculture scaling in
Tubiello, 2007), human health, livestock production and marine eco- Ghana. The paper may also provide insight into understanding the
system (Hoegh-Guldberg and Bruno, 2010; McMichael et al., 2007; Patz impact of climate change on irrigated agriculture, and thereby devise
et al., 2005) and the environment (Galbraith et al., 2002; Green et al., policies to mitigate its impacts.
2011). Climate change will have a significant effect on the agriculture
production of sub-Saharan countries due to higher seasonal rainfall
2. Materials and methods
variability and lower water availability. A temperature increase and
rainfall reduction and higher variability will make many agricultural
2.1. Study area
areas less productive or even completely unsuitable for agriculture
(Mendelsohn and Dinar, 1999). Understanding the effect of climate
This study was carried out for the entire Republic of Ghana, located
change on land use suitability will help the effort of developing
along the Gulf of Guinea and the Atlantic Ocean in West Africa.

111
A.W. Worqlul et al. Computers and Electronics in Agriculture 157 (2019) 110–125

Geographically, the country is located between 3o33′ W to 1 o30′ E and groundwater for irrigation. The BGS data consists of borehole potential
4o32′ to 11o10′ N (Fig. 1). The elevation ranges from −43 to yield and depth to groundwater table.
882 m amsl (above mean sea level). The country has a tropical humid
climate with high seasonal and annual rainfall variability (Opoku- 2.2.1. Biophysical factors
Ankomah and Cordery, 1994; Swaine, 1996). Generally, the climate in The biophysical factors determine if irrigation is possible in a par-
Ghana is tropical with distinct wet and dry seasons. The annual average ticular location or not; for example, the climate and the landscape must
rainfall from 1990 to 2010 varied between 650 and 1900 mm. The be convenient for practicing irrigation. Therefore, in this study, climate,
southwestern coast of the country receives the highest rainfall above land use, soil, and slope were considered the most important de-
1400 mm/year, while the northern part of the country is the driest re- termining factors for irrigation suitability. A detailed description of the
gion receiving less than 800 mm/year. biophysical factors used to evaluate the suitability of land areas for
irrigation in Ghana is discussed below:
2.2. Methodology
2.2.1.1. Climate factor for the baseline and future time horizons. The
The land suitability study was accomplished using a multi-criteria climate of the study area was represented by long-term rainfall and
evaluation technique for baseline period using historical climate data potential evapotranspiration for baseline and future time horizons. For
(1990 to 2010) and projected time horizons 2050s (2041 to 2060) and the baseline period (1990 to 2010), rainfall data from 64 weather
2070s (2061 to 2080). In MCE, the major factors affecting the land stations (Fig. 1) distributed across Ghana were obtained from the Ghana
suitability for agriculture were mapped on a GIS environment to de- Meteorological Agency (GMA). The daily rainfall data aggregated to
velop a single-indexed output (Ceballos-Silva and Lopez-Blanco, 2003; annual average were compared with station elevation to understand the
Chen et al., 2010; Feizizadeh and Blaschke, 2013; Pohekar and relationship between rainfall and elevation. Overall, the estimated
Ramachandran, 2004; Worqlul et al., 2015). The major factors de- elevation-rainfall relationship indicated a weak linear association
termining the land suitability for irrigation were identified from the with a coefficient of determination of 0.1. Because of this poor
literature and expert consultation (Akıncı et al., 2013; Chen et al., 2010; association, elevation was not used as an auxiliary variable for the
Mendas and Delali, 2012; Worqlul et al., 2015). The biophysical factors rainfall interpolation (Goovaerts, 2000). Instead, the Inverse Distance
included were climate (baseline, 2050s, and 2070s), soil, land use, and Weighting (IDW) interpolation method was used to interpolate the
topography. The socio-economic factors included were market access station rainfall into areal rainfall. The IDW method is well-established
represented by proximity to a road and population density. Table 1 and has demonstrated outstanding performances in estimating spatial
presents the major factors considered to identify the potential land rainfall in different regions across the world (Basistha et al., 2008; Chen
suitable for irrigation in Ghana, as well as the data sources with their and Liu, 2012; Tomczak, 1998). Fig. 3a shows the IDW interpolated
respective spatial resolutions. The general framework of the land suit- annual rainfall of Ghana of the baseline period.
ability analysis is presented in Fig. 2. The potential evapotranspiration data for the baseline period was
The data were collected in vector and raster formats. The factors obtained from the improved MODIS global evapotranspiration (ET)
were weighted using a pairwise comparison matrix (Saaty, 1977), and product (Mu et al., 2011). The improved MODIS ET data were devel-
the relative importance of the factors was computed by normalizing the oped based on MODIS and global metrological data (Mu et al., 2011).
eigenvector of the factors by their cumulative sum. The overall weights The improved MODIS global ET data were validated in a diverse range
of the factors were distributed to the different levels of suitability of ecosystems across five continents (Mu et al., 2011). The validation
classes by an equal interval ranging technique. In this technique, the data includes 46 eddy-flux towers and showed a favorable agreement
class break is determined by the weight of the factor divided by the (Mu et al., 2013; Mustafa et al., 2011). The data has been widely ap-
number of classes (Alsahli and Al-Harbi, 2017; Nandi and Shakoor, plied for a global and regional analysis (Kim et al., 2012; Ramoelo et al.,
2010). The reclassified and weighted factors were combined with a 2014; Ruhoff et al., 2013) and carbon budgets (Yang et al., 2014). The
weighted overlay analysis to calculate the preliminary land suitability MODIS ET dataset covers for the period from 2000 to 2010 at 8 day
map. A constraint map, which contains land use types that limit the interval with a spatial resolution of 1 km. The long-term average annual
suitability of land for surface irrigation (such as water bodies, wetlands, potential evapotranspiration was estimated by aggregating the 8 day
urban areas, forest, and protected areas) was developed to exclude the MODIS ET (Fig. 3b).
permanently non-suitable areas. The most optimal suitable land for ir- Climate data of the future time horizons was represented with the
rigation was identified by applying a user-defined threshold number, downscaled and bias-corrected Global Circulation Models (GCMs) from
which was greater than or equal to 80% of the suitability index. That Coupled Model Inter-comparison Project phase 5 (CMIP5) (Fick
optimal suitable land was overlaid to a GIS layer of groundwater spatial Stephen and Hijmans Robert, 2017). The projected climate data used
data collected from the British Geological Survey (BGS, MacDonald was the Representative Concentration Pathway (RCP) 4.5, which is
et al., 2012) to evaluate the potential and accessibility of the considered “moderately optimistic” pathways (Hijmans et al., 2005).

Table 1
Factors included in the land suitability analysis including their source and spatial resolution.
Data Source Spatial resolution (m)

Land use Global Land Cover Datasets (GlobeLand30) 30

Soil Africa Soil Information Service (AfSIS), 2015 250
Digital Elevation Model (DEM) Enhanced Shuttle Land Elevation Data from United States Geological Survey (USGS), 2000 released in 2015 30
Population density Global Gridded Population Database, 2015 1000
Road network Digital Chart of the World (DCW), 2006 –
MODIS potential evaporation (mm) MOD16 Global Terrestrial Evapotranspiration Data Set (2000 – 2010) 1000
Rainfall (mm/year) Ghana Meteorological Agency (GMA) from 1990 to 2010 –
Groundwater depth (m) British Geological Survey, 2012 5000
Potential borehole yield (l/s) British Geological Survey, 2012 5000
Groundwater storage (mm) British Geological Survey, 2012 5000
Future time horizon rainfall Coupled Model Inter-comparison Project phase 5 (CMIP5) from 2041 to 2060 (2050s). 1000
Future time horizon temperature Coupled Model Inter-comparison Project phase 5 (CMIP5) from 2061 to 2080 (2070s). 1000

112
A.W. Worqlul et al. Computers and Electronics in Agriculture 157 (2019) 110–125

Baseline and projected time Road Population Soil

horizons climate data DEM Land use
network density

MODIS
evaporation Rainfall

Slope
Rainfall deficit Euclidian
Baseline/2050s/2070s distance
period

Reclassified
factors

Depth to shallow Potential

groundwater borehole yield Weighting by
Overlaying weighted pairwise comparison
factors
Constraints
Preliminary suitability

Irrigation area suitability map for the Potential and

baseline period, 2050s and 2070s accessibility
of the
groundwater
Groundwater depth (m) for irrigation
Potential borehole yield (l/s)

Fig. 2. Schematic diagram of potential irrigable area mapping for the baseline period and future time horizons (2050s and 2070s) and groundwater resource
assessment.

The downscaled climate data includes rainfall and minimum and spatial resolution of 30 m.
maximum temperature at a spatial resolution of 1 km. This study used the 2010 land use data to identify the potential land
The projected daily rainfall data were aggregated to determine the suitable for irrigation in Ghana. The land use data indicated that the
annual average rainfall. Projected temperature data were used to pro- largest proportion of the country is coved with grassland (36%), fol-
ject the baseline of MODIS potential evapotranspiration into the future lowed by forest (32%), and shrubland (18%) (Fig. 3c). Agricultural land
time horizons. Temperature is one of the major factors affecting the rate covers approximately 9% of the country, the major crops grown in
of evapotranspiration and it has a direct linear relationship (Hargreaves Ghana includes cassava, peanut, maize, and legumes. Waterbodies oc-
and Samani, 1985; Trajkovic, 2005; Xu and Singh, 2001). Due to this, cupy approximately 3.3% of the country. The remaining land use
there are various popular temperature-based potential evapotranspira- groups were agricultural forest, urban areas, and wetlands.
tion methods (Xu and Singh, 2001). The change in temperature be-
tween the baseline period and future time horizons was used to project
2.2.1.3. Soil factor. Soil is one of the major factors that determine the
the baseline MODIS evapotranspiration into the future. The relationship
suitability of land for irrigation. Soil controls the partitioning of
developed between change in temperature and evapotranspiration for a
rainfall/irrigation into infiltration and runoff, limits evaporation, and
synthetic scenario using the Penman-Monteith (Penman, 1948) was
provides nutrients to the plant (Delwart et al., 2008; Kim and van Zyl,
used to determine evapotranspiration in the future time horizons. The
2009). For this study, soil data such as soil texture, organic carbon
method is briefly discussed under Appendix-1.
content, drainage class, and soil depth were combined into a single soil
suitability index using the modified Storie Index (SI, O'Geen, 2008;
Storie, 1978). The soil data of Ghana were obtained from the Africa Soil
2.2.1.2. Land use factor. Land use data of Ghana were collected from
Information System (AfSIS, Vågen et al., 2010). The AfSIS soil data has
GlobeLand30 (GL). The data were developed by the National Geomatics
a spatial resolution of 250 m and comes with six layers (0–5 cm,
Center of China (NGCC) and donated by the Chinese government to the
5–15 cm, 15–30 cm, 30–60 cm, 60–100 cm, and 100–200 cm); each
United Nations as a contribution towards the global sustainable
layer includes soil texture, organic carbon content and drainage class.
development and combating climate change (Jun et al., 2014). The
The soil texture was classified using the USDA soil classification,
GL data covers the Earth between the latitude of 80° N and 80° S and
resulting to eight soil classes. Majority of the country is covered by
developed by integrating satellite images from Landsat Thematic
sandy clay loam (49%) and silt loam (36%) (Fig. 3d).
Mapper (TM), Enhanced TM plus (ETM+) and Chinese
Environmental and Disaster (HJ-1) (Brovelli et al., 2015). The data is
available for 2000 and 2010 and it comprises ten land use groups with a 2.2.1.4. Slope factor. Slope plays a major role in determining the

113
A.W. Worqlul et al. Computers and Electronics in Agriculture 157 (2019) 110–125

Fig. 3. Factors used to assess the suitability of a land for surface irrigation in Ghana. (a) long-term annual average rainfall (mm) for the period 1990 to 2010, (b) long-
term average potential evapotranspiration (mm) for the period 2000 to 2010, (c) land use, d) soil texture.

suitability of land for irrigation by affecting land preparation and integrating national population data and housing censuses at the ad-
irrigation efficiency, especially in surface irrigation. The slope was ministrative level. The population density map of Ghana indicated the
calculated using a digital elevation model (DEM) which has a spatial highest population density (> 10,000 people/km2) in the Accra me-
resolution of 30 m. The average slope of the country is approximately tropolitan areas of the Greater Accra Region, and in the Kumasi me-
6.5%, and approximately 22% of the country has a slope of less than tropolitan areas in the Ashanti Region (Fig. 4b).
2%, which is considered as highly suitable for surface irrigation Road network data collected from the Digital Chart of the World for
(Fig. 4a). the year 2006 were used to determine the proximity to a paved road.
The distance of a certain land area from the paved road was calculated
at 30 m resolution using Euclidean distance (Fig. 4c). The analysis in-
2.2.2. Socioeconomic factors dicated that the average distance of a certain area from a paved road in
The success of irrigation adaptation will depend on access to both Ghana is approximately 17 km, while the farthest point is approxi-
agricultural inputs and market for selling agricultural products mately 88 km away.
(Worqlul et al., 2017; You et al., 2011). This study determined market
access using population density and proximity to a paved road network.
The population density of 2015 was obtained from the Global Gridded 2.2.3. Groundwater potential
Population Database available at the Center for International Earth The groundwater data were obtained from British Geological Survey
Science Network (CIESIN) and International Center for Tropical Agri- (BGS) and includes borehole potential yield (l/s) and depth to
culture (CIAT), Version 4 (GPWv4) (CIESIN, 2016; Doxsey-Whitfield groundwater (m), with a spatial resolution of 5 km (MacDonald et al.,
et al., 2015). The global population density was developed by 2012). The BGS potential borehole yield was compared with observed

114
A.W. Worqlul et al. Computers and Electronics in Agriculture 157 (2019) 110–125

Fig. 4. Factors used to assess the suitability of land for surface irrigation in Ghana. (a) slope (%), (b) population density (per km2), (c) distance to major paved roads
(km) and d) potential borehole yield (l/s).

groundwater yield in the central part of Ethiopia obtained from the factor weights were randomly assigned or not. For a consistent pairwise
Ethiopian Agricultural Transformation Agency (Worqlul et al., 2017). comparison, Saaty (1980) suggested that a CR value should not be
The result indicated that the BGS groundwater potential yield matched higher than 0.1, otherwise, the comparison matrix should be revised.
well with the observations from ATA. The groundwater data from the
BGS were used to identify the potential of the groundwater to irrigate
2.2.5. Classification of factors and preliminary suitability
the potential suitable land using simple water lifting technologies. The
The data collected for the suitability analysis were obtained in two
technologies considered included human, animal, diesel or solar-pow-
different formats. The continuous geographic phenomena such as soil,
ered pumps with a maximum operating depth of 25 m and pumping
land use, population density, elevation, and potential evapotranspira-
capacity of 0.6 l/s (Assefa et al., 2018). Fig. 4d presents the borehole
tion were in a grid format, while other discrete geographic phenomena
potential yield.
such as rainfall and road network were obtained in vector format. The
approach in Worqlul et al. (2017) was followed to reclassify the factors.
2.2.4. Weighting of factors The land use map of Ghana has eight classes (Fig. 3c), and these were
The factors identified for estimating the suitability of the land were reclassified into four levels of suitability according to FAO’s re-
weighted using a pairwise comparison technique (Saaty, 1977), which commendation (FAO, 1976; FAO, 1985; FAO, 1989; Walker, 1989).
helped to prepare the importance comparison matrix. In the compar- Table 2 presents the FAO framework of land suitability classification.
ison matrix, the highest value corresponds to the absolute importance The agricultural land was classified as highly suitable (class S1);
and the lowest value is the reciprocal indicating absolute triviality in- grassland, which requires land clearing and leveling was classified as
serted at the transpose position of the matrix (Saaty and Vargas, 1991; moderately suitable (S2); shrubland, which requires higher initial in-
Worqlul et al., 2015; Worqlul et al., 2017). The relative importance vestment for land preparation was classified as marginally suitable (S3);
(weights of the factors) were computed by normalizing the eigenvector and forest, water, urban and wetland were classified as not suitable (S4)
of the factors by the cumulative sum. The consistency of the pair-wise since they provide other ecosystem services and/or are inconvenient for
comparison was evaluated by employing consistency ration (CR, Franek practicing irrigation.
and Kresta, 2014). The CR will evaluate the likelihood of whether the The soil properties indicating the water holding capacity of the soil

115
A.W. Worqlul et al. Computers and Electronics in Agriculture 157 (2019) 110–125

Table 2
Framework of land suitability classification of FAO (FAO, 1976; Worqlul et al., 2015){Worqlul, 2015 #80;FAO, 1976 #137}.
Class Land Description

Class S1: Highly suitable Land without significant limitations. This land is the best possible and does not reduce productivity or require increased inputs.
Class S2: Moderately suitable Land that is clearly suitable but has limitations that either reduce productivity or require an increase of inputs to sustain productivity
compared with those needed on S1 land.
Class S3: Marginally suitable Land with limitations so severe that benefits are reduced and/or the inputs required to sustain production need to be increased so that this
cost is only marginally justified.
Class S4 (N1): Currently not suitable Land that cannot support the particular land use on a sustained basis or land on which benefits do not justify inputs.

Fig. 5. Factors used to assess the soil suitability for irrigation using Storie Index (SI). (a) organic carbon content (permilles), and (b) soil drainage class.

such as texture, drainage, depth, and organic carbon content were zero to 13,090 persons per square kilometer was divided into eight
combined to a single soil suitability index called Storie Index (SI, classes of suitability using equal intervals ranging technique. Eight
O'Geen, 2008; Storie, 1978). Approximately half of the soil textures in classes were adopted since the population density map showed a higher
Ghana are composed of silt loam, loam, and clay loam, which has a very variability as proximity to the road.
high water holding capacity (Fig. 3d). Organic carbon has an important The suitability of the climate for irrigation was estimated by cal-
role in physical, biological and chemical properties of the soil, influ- culating the monthly rainfall deficit (rainfall minus potential evapo-
encing nutrient holding capacity, nutrient turnover, and soil stability transpiration) for both baseline and future time horizons (Worqlul
(Krull et al., 2004). The organic carbon content of the soil in Ghana et al., 2015). A positive deficit reflects no irrigation requirement, while
varies between zero and 84 permilles, with an average value of a negative value suggests a need for irrigation. The monthly rainfall
20 permilles. The southern part of the country is dominated by forest deficit was aggregated to annual amounts. The rainfall deficit was re-
cover and is rich in organic carbon, whereas the northern part of the classified into eight classes of suitability using an equal interval ranging
country, which is driest, showed a lower organic carbon content technique.
(Fig. 5a). The soil depth varied from zero to 200 cm with an average The weight of the factors estimated with a pair-wise comparison was
depth of 180 cm. The soil drainage collected from AfSIS fits into seven distributed to the different suitability classes with an equal interval
classes, the largest three classes were “somewhat poorly drained” ranging technique. Thereafter, Weighted Overlay Analysis was used to
(38%), “well drained” (29%), and “moderately well drained” (28%) assess the preliminary suitability of land for irrigation. The preliminary
(Fig. 5b). The computed SI indicates values ranging from 1 (low sui- suitability map was then constrained by land use types that limit the
table) to 8 (highest suitable), which were classified into eight suitability suitability of the land for surface irrigation such as water bodies, wet-
classes applying an equal interval ranging technique. lands, urban, forest, and protected areas to create the final land suit-
The percent slope map computed from a 30 m resolution DEM ability map.
(Fig. 4a) was classified into five levels of suitability classes. These levels
were 0 to 2%, 2 to 8%, 8 to 12%, 12 to 20% and above 30% which were
2.2.6. Climate change effect on land use suitability
denoted as highly, moderately, marginally, low and not suitable, re-
The likely effect of climate change on land use suitability is a major
spectively. The major paved road networks interpolated with Euclidean
concern globally in recent years (Olesen and Bindi, 2002). Climate
distance were reclassified into eight classes of suitability using an equal
change is expected to affect the suitability of land for agriculture at a
interval ranging technique. With the maximum distance in Ghana from
different level depending on the direction and magnitude of change. It
a paved road being approximately 88 km, road proximity was divided
will have a significant impact in sub-Saharan African countries and
into eight classes. The population density (Fig. 4b), which ranges from
specifically Ghana where the majority of the food production is relying

116
A.W. Worqlul et al. Computers and Electronics in Agriculture 157 (2019) 110–125

Table 3
Pair-wise comparison matrix and weights of factors.
Factors Soil Land use Population density Road proximity Rain deficit Slope Weight (%)

Soil 1 3 1/3 2 1/2 1/3 11.4

Land use 1/3 1 1/4 2 1/4 1/4 6.4
Population density 3 4 1 3 2 1/2 24.9
Road proximity 1/2 1/2 1/3 1 1/5 1/3 5.8
Rainfall deficit 2 4 1/2 5 1 1/2 20.1
Slope 3 4 2 3 2 1 31.4
Total 100

on a rainfed system (Cooper et al., 2008). In this study, the effect of the preliminary suitability map, and the area above the threshold is
climate change on the suitability of land due to climate change were shown in Fig. 8. For example, at a threshold of 85%, there were thou-
evaluated by substituting the rainfall deficit of baseline period by the sands of suitable areas ranging from 1 km2 to 35 km2. At the 85%
respective future time horizons climate data while keeping the other threshold, nearly 2% of the landmass was suitable for surface irrigation,
factors of the MCE similar (Fig. 2). Then the change in land use suit- and at the 80% threshold, nearly 9% (22,600 km2) of the land was
ability due to climate change in the future time horizons were com- suitable for surface irrigation.
pared with the baseline period of estimated potential land suitability. The potential suitable lands with a suitability of 80% and above
were extracted for the three major river basin systems in Ghana
3. Results and discussion (Table 4). This type of analysis can help in identifying and prioritizing
land use planning at basin level. The Volta River basin, which covers
3.1. Weighting of factors for surface irrigation suitability mapping approximately 67% of the country has the lowest potential suitable land
for irrigation. The South Western River Basin System, which is the
The pair-wise comparison matrix and overall weights of the factors second largest basin covering 24% of the country has the largest sui-
selected for the study are shown in Table 3. The slope was the most table land for irrigation (9690 km2). The Coastal River System is the
important factor for suitability of land for surface irrigation in Ghana smallest basin but has the highest percentage (37%) of suitable land for
indicating that land preparation may require a significant initial in- irrigation.
vestment that weakens the economic feasibility. The population density
was the second most important factor for determining the suitability of 3.3. Potential of shallow groundwater
an area for irrigation, highlighting the strong role of market access for
agricultural input and produce. Rainfall deficit and soil showed modest The estimated depth to groundwater in Ghana is 14 m on average
influence on land suitability, while land use and road proximity showed from the surface, with a maximum depth of 50 m. The majority of the
the lowest influences. The credibility of the pair-wise comparison was country (98%) has groundwater access under 25 m from the ground
tested with a consistency ratio to test if the matrix ratings were ran- surface. The aquifer productivity indicates that the borehole potential
domly generated. The result indicated that the comparison matrix to be yield in the country ranges between 0.1 and 20 l/s. A large proportion
trustworthy, with a consistency ratio of 5% (Ceballos-Silva and Lopez- of the country (56%) has aquifer potential yield between 1 and 5 l/s,
Blanco, 2003; Saaty, 1980). while 2% of the country has the highest aquifer potential yield between
5 and 20 l/s.
3.2. Preliminary land suitability analysis for surface irrigation The most suitable land suitable for agriculture, which has a suit-
ability value of greater than 80% was superimposed over the borehole
The factors outlined above were reclassified into different levels of potential yield and depth to groundwater map to evaluate the potential
suitability. Fig. 6 illustrates the reclassified factors affecting land suit- of shallow groundwater for irrigation and its accessibility using simple
ability for surface irrigation in Ghana. The overall weights of the factors water-lifting technologies. The potentially suitable land has shallow
were distributed to the different levels of suitability classes by an equal groundwater access ranging between 1 and 50 m from the surface, with
interval ranging technique. For example, the overall weight for land an average depth of 12 m (Fig. 9a). The borehole yield over the po-
use, which was 6.4 (Table 3) was distributed to the four classes of tential suitable land ranges between 0.0 and 13.55 l/s and the average
suitability, pixels which are classified as highly suitable (agricultural borehole yield was approximately 2.5 l/s (Fig. 9b). The combination of
land) received a value of 4/4 * (6.4), the second most suitable land use access to shallow groundwater and moderate borehole yield (approxi-
group (grassland) received 3/4 * (6.4) and the third suitable group mately a recharge of 600 mm) makes the suitable land favorable for
(shrubland) received a weight of 2/4 * (6.4), and the least suitable irrigation development using human, animal, diesel and solar powered
group received 1/4 * (6.4) weights. The most important factor, slope, water lifting technologies.
indicated that the majority of the country (74%) was classified as highly
and moderately suitable for agriculture. The reclassified rainfall deficit 3.4. Climate change and its effect on land use suitability for irrigation
indicated a near-normal distribution; where the moderate and mar-
ginally suitable land totals 71% while the rest tends towards extremely 3.4.1. Rainfall for baseline and future time horizons
dry and extremely wet. The downscaled rainfall data of the future time horizons were
After the overall weights of the factors were distributed to the dif- compared with the baseline period. Fig. 10a and b shows the rainfall
ferent levels of suitability groups, the preliminary suitability map was difference between baseline and future time horizons. The projected
computed by overlaying weighted factors using a Weighted Overlay rainfall data difference the baseline period (Fig. 10a and b) did not
Tool available in ArcGIS under Spatial Analyst toolbox. The preliminary show a consistent change throughout the country for both future time
land suitability map provided values which ranges between 26% and horizons. In 2050s, approximately 65% of the country will receive a
94%, where 26% represents the lowest suitability and 94% the most higher rainfall while a reduction in rainfall is expected in 32% of the
suitable land (Fig. 7). A constraint map was used to exclude areas that country located in the southern coast compared to the baseline period.
are not suitable for irrigation. The most suitable land was extracted In 2070s, the area receiving higher rainfall will increase to 68% of the
with a variable threshold from 80% to 94% (with a 1% increment) from country compared to the baseline period. In general, under RCP 4.5

117
A.W. Worqlul et al. Computers and Electronics in Agriculture 157 (2019) 110–125

Fig. 6. Reclassified factor maps: (a) rainfall deficit, (b) land use class, (c) Storie Index class, (d) slope class, (e) distance to the road class and (f) population density
class. S1 represents highly suitable areas and as the number increases, the suitability level decreases.

118
A.W. Worqlul et al. Computers and Electronics in Agriculture 157 (2019) 110–125

Fig. 7. A preliminary suitable land for irrigation map; 94% shows the most suitable area for irrigation while 26% shows the least suitable land.

Fig. 8. Suitable irrigation area (in 1000 km2) at different suitability levels. For example, approximately 4500 km2 of the land is suitable at a suitability threshold of
85% and 12,700 km2 land is suitable at a suitability threshold of 82%.

scenario, rainfall will increase in the northern part of the country (drier in the central part of Ghana.
part of the country) under both future time horizons, while a reduction
will be expected in the southern part of the country (wettest region)
3.4.2. Temperature and potential evapotranspiration in the future time
(Fig. 10a and b). Overall, on average, the country will receive an ad-
horizons
ditional 15 mm rainfall more from the baseline period in 2050s and it
The projected future time horizon minimum and maximum tem-
will increase to 20 mm in 2070s. This indicates that under future time
perature were averaged and the difference from the baseline period was
horizons, the small change in rainfall could not be the major limiting
computed (Fig. 11a and b). The rate of change of temperature is higher
factor for land suitability analysis compared to the baseline period. A
in 2070s compared to 2050s. In 2050s, the average temperature will
similar conclusion was also drawn by Srivastava et al. (2018) conducted
increase by 2.0 °C from the baseline period. In this period, the average

119
A.W. Worqlul et al. Computers and Electronics in Agriculture 157 (2019) 110–125

Table 4 The rainfall deficit of the future time horizons was used as input to
Potential suitable land area in the major river basins of Ghana. MCE to identify the potential land suitable for irrigation in the future
River basin Basin area Potential irrigable Percent potential time horizons and the difference in land suitability from the baseline
(km2) land (km2) (%) period was computed. The projected land suitability analysis indicated
that in the 2050s, 9.5% of the baseline period suitable for irrigation will
Coastal River System 20,446 7568 37.0
be unfavorable for irrigation due to increasing rainfall deficit and in
South Western River 58,189 9687 16.6
System
2070s the unfavorable land area will further increase to 17%. The
Volta Basin System 159,793 5337 3.3 difference between future and baseline land suitability analysis is
Total 238,428 22,592 9.5 shown in Fig. 13a and b. Generally, in the 2050s, 53% of the land and in
2070s 47% of the country will not be affected by climate change when
compared with the baseline suitability. However, in both future time
temperature will increase by 1.5 to 2.2 °C in the majority of the country horizons, the central part of Ghana will be highly affected by climate
(66%). In 2070s, the average temperature will further increase by 2.6 °C change.
from the baseline period. In the majority of the land (88%) temperature Although rainfall gains are expected in the majority of the country
will increase by 2.3 to 3.1 °C. The rising temperature expected in the in the future time horizon, overall the net gain due to climate change in
future will significantly impact plant development and productivity. Ghana will be negative due to a rising temperature. Therefore, the
Rising temperatures will increase the potential evapotranspiration on government of Ghana should plan adoption measures to reduce the
average by 6.0% and 7.6% in the 2050s and 2070s, respectively. The plausible impact of rising temperature, higher evaporation, and greater
projected temperature and evapotranspiration together with population rainfall variability. The planning should include enhancing the sus-
growth will place a substantial demand on water leading to inter- tainable use of water resources, crop diversification and planting a new
sectoral competition over the limited resources. variety of temperature tolerant crops. The government should also be
prepared to control pests and diseases which may expand their geo-
graphic territory as the climate warms.
3.4.3. Climate change impact on the land use suitability The most suitable land for the projected time horizons was extracted
The rainfall deficit of the projected time horizons was compared with a variable threshold from 80% to 94% (with a 1% increment) and
with the baseline period. The result indicated that the rainfall deficit in the percentage area difference from the baseline period was calculated
2050s will increase by 14% from the baseline period and in 2070s it for each group (Fig. 14). The result indicated that in 2050s, the land use
will further increase by 20% from the baseline period. The difference in suitability decreased by more than 5% in the 60% of the land suitability
rainfall deficit between future and baseline period indicated that in groups (Fig. 14). In 2017, the land use suitability decreased across all of
2050s, rainfall deficit will increase in 92% of the country and in 2070s, the suitability groups by more than 10%.
the rainfall deficit will further increase in 98% of the country (Fig. 12a The potential suitable lands of the 2050s and 2070s with a
and b).

Fig. 9. (a) Depth to groundwater (m) and (b) Aquifer productivity (l/s) over the potential suitable land (suitability greater than 80%).

120
A.W. Worqlul et al. Computers and Electronics in Agriculture 157 (2019) 110–125

Fig. 10. Annual average rainfall change from the baseline period in the future time horizons (a) Rainfall change from the baseline period in 2050s (2041 to 2060) and
(b) Rainfall change from the baseline period in 2070s (2061 to 2080).

Fig. 11. Annual average temperature difference of the future time horizons from the baseline period (a) Average temperature change from the baseline period in
2050s (2041 to 2060) and (b) Average temperature change from the baseline period in 2070s (2061 to 2080).

suitability of 80% and above were extracted for the three major river In Ghana, the majority the potential suitable land for irrigation is
basin systems of Ghana (Table 5). The result indicated that for both located in the in the southern part of the country (Fig. 9), which is
future time horizons the suitable land in South Western River Basin expected to experience a significant rainfall deficit in the future time
System will increase by 12% and 2% for 2050s and 2070s, respectively, horizons (Fig. 12). The rising temperature, which increases evaporation
while a significant suitable land area reduction will be expected in and transpiration by plants will likely reduce recharge to the ground-
Volta and Coastal River Systems for both future time horizons (compare water system. This highlights the extent of climate change impact on
with Table 4). Ghana where most of the population is located (Fig. 4b) with a greater
demand.
3.4.4. Irrigation potential of the groundwater during the future time
horizons 4. Conclusions
The irrigation potential of the groundwater during the future time
horizons was not evaluated due to the absence of projected ground- This study is the first of its kind to provide a spatially explicit land
water data and due to the complexity of the groundwater response in suitability analysis for the baseline and future time horizons in Ghana.
the future. However, in general, climate change will have a significant The findings indicate that there is a significant area of land suitable for
effect across the different water balance components (precipitation, irrigation that could be developed with shallow groundwater under the
evapotranspiration, soil moisture storage, groundwater and surface baseline condition. A large portion of the suitable land is located in the
runoff) due to a rising temperature and rainfall variability (Abdo et al., South Western River Basin System with shallow groundwater access
2009; Adem et al., 2014; Eckhardt and Ulbrich, 2003; Worqlul et al., (< 25 m) that has a moderate borehole yield (2.0–4.6 l/s), which made
2018b). the basin highly favorable for small-scale irrigation using simple water

121
A.W. Worqlul et al. Computers and Electronics in Agriculture 157 (2019) 110–125

Fig. 12. Rainfall deficit between future time horizons and baseline period (a) 2050s and (b) 2070s. The negative values indicate a further rainfall deficit increase in
the future time horizon and positive values indicated a reduction in rainfall deficit.

Fig. 13. Relative change of land suitability between future time horizons and baseline period (a) 2050s and (b) 2070s. The negative values indicates a reduction in
suitability from the base period, a positive value indicates an increase in suitability while zero indicates no change in suitability.

60
Change in land suitability (%)

2050s 2070s
40

20

-20

-40

-60
80 82 84 86 88 90 92 94
Percent threshold (%, suitability level)
Fig. 14. Percentage change of suitable land during the future time horizon were compared to the baseline period. The negative values indicate a reduction in
suitability from the baseline period and a positive value indicates an increase in land suitability for irrigation.

122
A.W. Worqlul et al. Computers and Electronics in Agriculture 157 (2019) 110–125

Table 5
Potential suitable land for irrigation under the future time horizons in the major river basins of Ghana.
River basin Basin area (km2) Potential irrigable land in 2050 s (km2) Potential irrigable land in 2070 s (km2)

Coastal River System 20,446 6433 5892

South Western River System 58,189 10,835 9802
Volta Basin System 159,793 3178 3058
Total 238,428 20,446 18,751

lifting technologies making the area favorable for small-scale irrigation can inform program planners and investors regarding the optimal lo-
using human, animal, diesels and solar powered water lifting technol- cations for intensification of agriculture using groundwater irrigation.
ogies. Ghana’s agricultural policy, as well as the more specific irrigation
The climate change study under the HadGEM2-ES RCP 4.5 emission policy, foresees the need to expand irrigated agriculture in order to
scenario shows the northern part of Ghana will benefit from increasing achieve food security aims and expand the contribution of agriculture
rainfall for both future time horizons. Overall, in Ghana, on average to economic growth. Studies that apply methods such as MCE can be
rainfall will increase in both future time horizons by 15 mm and 20 mm relevant tools for targeting investments to achieve impact.
from the baseline period in 2050s and 2070s, respectively. However,
the average temperature will rise by 2.0 °C and 2.6 °C in 2050s and
2070, respectively and the rise in temperature will increase the po- Acknowledgements
tential evapotranspiration by 6.0% and 7.6% in the 2050s and 2070s,
respectively. Due to climate change, 9.5% and 17% of the potential land This publication was made possible through support provided by
suitable under the baseline period will be unfavorable in 2050s and Feed the Future through the U.S. Agency for International
2070s, respectively. This suggests that climate change adaptation and Development, under the terms of Contract No. AID-OAA-A-13-0005.
mitigation policies should be in place to maintain the suitable land for The opinions expressed herein are those of the author(s) and do not
irrigation. For example, there are several land and water management necessarily reflect the views of the U.S. Agency for International
interventions (e.g. mulch, conservation agriculture, etc) that shifts the Development. We would like also to acknowledge the National
non-productive evapotranspiration to productive evapotranspiration, Meteorological Agency Services of Ghana, British Geological Survey,
and thereby maintain or even improve land and water productivity in United States Geological Survey, African Soil Information Service and
the face of climate change. International Food Policy Research Institute for providing us with
This study was intended to provide valuable information to policy- quality data.
level decision-makers on the potential to invest in irrigation using The editor and the two anonymous reviewers are gratefully ac-
groundwater in Ghana. The maps and identification of suitable areas knowledged for their valuable comments on our manuscript.

Appendix. -1: Temperature effect on potential evapotranspiration

Evapotranspiration is the second largest water budget of the water balance components (Vörösmarty et al., 1998; Wilson et al., 2001). Tem-
perature-based evaporation methods are common which includes Penman–Monteith (Penman, 1948), Hargreaves (Hargreaves and Samani, 1985),
Thornthwaite (Thornthwaite, 1948), Blaney-Criddle (Brouwer and Heibloem, 1986) among others. Penman–Monteith method is one of the standard
methods for estimating potential evaporation (ETo). To understand the effect of a change in temperature on potential evapotranspiration, Penman-
Monteith was used to estimate the potential evapotranspiration by changing the temperature with synthetic scenarios of changing temperature
between −5°C to 5 °C. The hypothetical temperature change indicated a strong linear association with the potential evapotranspiration (Fig. A1). For
a one-celsius temperature increase, the potential evapotranspiration will increase by 2.9%; potential evapotranspiration increases linearly with
increasing temperature (Fig. A1). A five Celsius increase from the baseline will increase the potential evapotranspiration by 14.6%.

Fig. A1. Effect of change in temperature on potential evapotranspiration for a synthetic temperature change scenario from the baseline period.

123
A.W. Worqlul et al. Computers and Electronics in Agriculture 157 (2019) 110–125

Appendix A. Supplementary material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.compag.2018.12.040.

References managing trade-offs. Agri. Water Manage. 131, 175–182.

M. Giordano C. De Fraiture E. Weight J. van der Bliek . Water for wealth and food se-
curity: Supporting farmer-driven investments in agricultural water management.
Abdo, K., Fiseha, B., Rientjes, T., Gieske, A., Haile, A., 2009. Assessment of climate Synthesis report of the AgWater Solutions Project. IWMI; 2012.
change impacts on the hydrology of Gilgel Abay catchment in lake Tana basin, Goovaerts, P., 2000. Geostatistical approaches for incorporating elevation into the spatial
Ethiopia. Hydrol. Process. 23, 3661–3669. interpolation of rainfall. J. Hydrol. 228, 113–129.
Adem, A.A., et al., 2014. Climate Change Projections in the Upper Gilgel Abay River Green, T.R., et al., 2011. Beneath the surface of global change: impacts of climate change
Catchment. In: Blue Nile Basin Ethiopia Nile River Basin. Springer, pp. 363–388. on groundwater. J. Hydrol. 405, 532–560.
Adimassu, Z., et al., 2016. Smallholder irrigation productivity for sustainable in- Gumma, M.K., Pavelic, P., 2013. Mapping of groundwater potential zones across Ghana
tensification. Water balances for high value crops in northern Ghana. using remote sensing, geographic information systems, and spatial modeling.
Akıncı, H., Özalp, A.Y., Turgut, B., 2013. Agricultural land use suitability analysis using Environ. Monitor. Assess. 185, 3561–3579.
GIS and AHP technique. Comput. Electron. Agri. 97, 71–82. Hargreaves, G.H., Samani, Z.A., 1985. Reference crop evapotranspiration from tem-
Alsahli, M.M., Al-Harbi, M., 2017. Allocating optimum sites for air quality monitoring perature. Appl. Eng. Agri. 1, 96–99.
stations using GIS suitability analysis. Urban Climate. Heintz, J., 2005. Employment, poverty, and gender in Ghana. PERI Work. Pap. 66.
Antwi-Agyei, P., Fraser, E.D., Dougill, A.J., Stringer, L.C., Simelton, E., 2012. Mapping the Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G., Jarvis, A., 2005. Very high re-
vulnerability of crop production to drought in Ghana using rainfall, yield and so- solution interpolated climate surfaces for global land areas. Int. J. Climatol. 25,
cioeconomic data. Appl. Geography 32, 324–334. 1965–1978.
Ashitey, E., Rondon, M., 2012. Ghana Exporter Guide 2012. Global Agricultural Hoegh-Guldberg, O., Bruno, J.F., 2010. The impact of climate change on the world’s
Information Network, Accra, Ghana. marine ecosystems. Science 328, 1523–1528.
Assefa, T., Jha, M., Reyes, M., Srinivasan, R., Worqlul, A., 2018. Assessment of suitable Jun, C., Ban, Y., Li, S., 2014. China: open access to Earth land-cover map. Nature 514,
areas for home gardens for irrigation potential, water availability, and water-lifting 434.
technologies. Water 10, 495. Kim, H.W., Hwang, K., Mu, Q., Lee, S.O., Choi, M., 2012. Validation of MODIS 16 global
Assefa, T.T., et al., 2015. Identification of erosion hotspot area using GIS and MCE terrestrial evapotranspiration products in various climates and land cover types in
technique for koga watershed in the upper blue Nile Basin Ethiopia. Am. J. Environ. Asia. KSCE J. Civil Eng. 16, 229–238.
Sci. 11, 245. Kim, Y., van Zyl, J.J., 2009. A time-series approach to estimate soil moisture using po-
Basistha, A., Arya, D., Goel, N., 2008. Spatial distribution of rainfall in Indian larimetric radar data. IEEE Trans. Geosci. Remote Sens. 47, 2519–2527.
Himalayas–a case study of Uttarakhand region. Water Resource Manage. 22, Krull, E.S., Skjemstad, J.O., Baldock, J.A., 2004. Functions of soil organic matter and the
1325–1346. effect on soil properties. Cooperative Research Centre for Greenhouse Accounting.
Brouwer, C., Heibloem, M., 1986. Irrigation water management: irrigation water needs. Kuo, R.J., Chi, S.-C., Kao, S.-S., 2002. A decision support system for selecting convenience
Train. Manual 3. store location through integration of fuzzy AHP and artificial neural network.
Brovelli, M.A., Molinari, M.E., Hussein, E., Chen, J., Li, R., 2015. The first comprehensive Comput. Industry 47, 199–214.
accuracy assessment of GlobeLand30 at a national level: methodology and results. Laube, W., Schraven, B., Awo, M., 2012. Smallholder adaptation to climate change: dy-
Remote Sens. 7, 4191–4212. namics and limits in Northern Ghana. Clim. Change 111, 753–774.
Ceballos-Silva, A., Lopez-Blanco, J., 2003. Delineation of suitable areas for crops using a MacDonald, A., Bonsor, H., Dochartaigh, B.É.Ó., Taylor, R., 2012. Quantitative maps of
multi-criteria evaluation approach and land use/cover mapping: a case study in groundwater resources in Africa. Environ. Res. Lett. 7, 024009.
Central Mexico. Agri. Syst. 77, 117–136. McMichael, A.J., Powles, J.W., Butler, C.D., Uauy, R., 2007. Food, livestock production,
Chen, F.-W., Liu, C.-W., 2012. Estimation of the spatial rainfall distribution using inverse energy, climate change, and health. Lancet 370, 1253–1263.
distance weighting (IDW) in the middle of Taiwan. Paddy Water Environ. 10, Mendas, A., Delali, A., 2012. Integration of multicriteria decision analysis in GIS to de-
209–222. velop land suitability for agriculture: application to durum wheat cultivation in the
Chen, Y., Yu, J., Khan, S., 2010. Spatial sensitivity analysis of multi-criteria weights in region of Mleta in Algeria. Comput. Electron. Agri. 83, 117–126.
GIS-based land suitability evaluation. Environ. Model. Softw. 25, 1582–1591. Mendelsohn, R., Dinar, A., 1999. Climate change, agriculture, and developing countries:
Choudhary, D., Shankar, R., 2012. An STEEP-fuzzy AHP-TOPSIS framework for evalua- does adaptation matter? World Bank Res. Observ. 14, 277–293.
tion and selection of thermal power plant location: a case study from India. Energy Mendes, D.M., Paglietti, L., Jackson, D., 2014. Ghana: Irrigation market brief, FAO/IFC.
42, 510–521. Cooperation Programme Rome.
CIESIN. Gridded Population of the World, Version 4 (GPWv4): Population Density MOFA, 2014. 2014 Agric Sector Annual Progress Report. Ministry of Food and
Adjusted to Match 2015 Revision UN WPP Country Totals. In: Center for Agriculture, Accra, Ghana.
International Earth Science Information Network – CIESIN – Columbia University Mosadeghi, R., Warnken, J., Tomlinson, R., Mirfenderesk, H., 2015. Comparison of Fuzzy-
(Ed.). NASA Socioeconomic Data and Applications Center (SEDAC), Palisades, NY; AHP and AHP in a spatial multi-criteria decision making model for urban land-use
2016. planning. Comput., Environ. Urban Syst. 49, 54–65.
Cooper, P., et al., 2008. Coping better with current climatic variability in the rain-fed Mu, Q., Zhao, M., Kimball, J.S., McDowell, N.G., Running, S.W., 2013. A remotely sensed
farming systems of sub-Saharan Africa: an essential first step in adapting to future global terrestrial drought severity index. Bull. Am. Meteorol. Soc. 94, 83–98.
climate change? Agri., Ecosyst. Environ. 126, 24–35. Mu, Q., Zhao, M., Running, S.W., 2011. Improvements to a MODIS global terrestrial
Delwart, S., et al., 2008. SMOS validation and the COSMOS campaigns. IEEE Trans. evapotranspiration algorithm. Rem. Sens. Environ. 115, 1781–1800.
Geosci. Remote Sens. 46, 695–704. Mustafa, A., et al., 2011. Land suitability analysis for different crops: a multi criteria
Dittoh, S., Awuni, J.A., Akuriba, M.A., 2013. Small pumps and the poor: a field survey in decision making approach using remote sensing and GIS. Researcher 3, 61–84.
the upper east region of Ghana. Water Int. 38, 449–464. Namara, R.E., et al., 2011a. Smallholder shallow groundwater irrigation development in
Doxsey-Whitfield, E., et al., 2015. Taking advantage of the improved availability of the upper east region of Ghana. IWMI.
census data: a first look at the gridded population of the world, version 4. Pap. Appl. Namara, R.E., Horowitz, L., Nyamadi, B., Barry, B., 2011b. Irrigation development in
Geography 1, 226–234. Ghana. Past experiences, emerging opportunities, and future directions.
Eckhardt, K., Ulbrich, U., 2003. Potential impacts of climate change on groundwater Nandi, A., Shakoor, A., 2010. A GIS-based landslide susceptibility evaluation using bi-
recharge and streamflow in a central European low mountain range. J. Hydrol. 284, variate and multivariate statistical analyses. Eng. Geol. 110, 11–20.
244–252. Nelson, G.C., et al., 2009. Climate change: impact on agriculture and costs of adaptation.
FAO. A framework for land evaluation. Rome; 1976. Intl. Food Policy Res. Inst.
FAO. Guidelines: land evaluation for irrigated agriculture. Soils Bulletin 55. Via delle O'Geen, A.T., 2008. A revised storie index for use with digital soils information. UCANR
Terme di Caracalla, 00100 Rome, Italy; 1985. Publications.
FAO. Guidelines for Designing and Evaluating Surface Irrigation Systems (FAO Irrigation Obuobie, E., Diekkrueger, B., Agyekum, W., Agodzo, S., 2012. Groundwater level mon-
and Drainage Paper) Via delle Terme di Caracalla, 00100 Rome, Italy; 1989. itoring and recharge estimation in the white volta river basin of Ghana. J. Afri. Earth
Feizizadeh, B., Blaschke, T., 2013. Land suitability analysis for Tabriz county, Iran: a Sci. 71, 80–86.
multi-criteria evaluation approach using GIS. J. Environ. Plann. Manage. 56, 1–23. Obuobie, E., Ofori, D., Agodzo, S.K., Okrah, C., 2013. Groundwater potential for dry-
Fick Stephen, E., Hijmans Robert, J.W., 2017. 2: new 1-km spatial resolution climate season irrigation in north-eastern Ghana. Water Int. 38, 433–448.
surfaces for global land areas. Int. J. Climatol. Ofosu, E.A., 2011. Sustainable Irrigation Development in the White Volta sub-Basin.
Forkuor, G., Pavelic, P., Asare, E., Obuobie, E., 2013. Modelling potential areas of UNESCO-IHE PhD Thesis. CRC Press.
groundwater development for agriculture in northern Ghana using GIS/RS. Hydrol. Olesen, J.E., Bindi, M., 2002. Consequences of climate change for European agricultural
Sci. J. 58, 437–451. productivity, land use and policy. Euro. J. Agron. 16, 239–262.
Franek, J., Kresta, A., 2014. Judgment scales and consistency measure in AHP. Proc. Opoku-Ankomah, Y., Cordery, I., 1994. Atlantic sea surface temperatures and rainfall
Econom. Finance 12, 164–173. variability in Ghana. J. Clim. 7, 551–558.
Galbraith, H., et al., 2002. Global climate change and sea level rise: potential losses of Pal, S., 2016. Identification of soil erosion vulnerable areas in Chandrabhaga river basin:
intertidal habitat for shorebirds. Waterbirds 25, 173–183. a multi-criteria decision approach. Model. Earth Syst. Environ. 2, 5.
Giordano, M., de Fraiture, C., 2014. Small private irrigation: enhancing benefits and Patz, J.A., Campbell-Lendrum, D., Holloway, T., Foley, J.A., 2005. Impact of regional

124
A.W. Worqlul et al. Computers and Electronics in Agriculture 157 (2019) 110–125

climate change on human health. Nature 438, 310. Thornthwaite, C.W., 1948. An approach toward a rational classification of climate.
Penman, H.L., 1948. Natural evaporation from open water, bare soil and grass. In: Geograph. Rev. 38, 55–94.
Proceedings of the Royal Society of London A: Mathematical, Physical and Tomczak, M., 1998. Spatial interpolation and its uncertainty using automated anisotropic
Engineering Sciences. The Royal Society, pp. 120–145. inverse distance weighting (IDW)-cross-validation/jackknife approach. J. Geograph.
Pereira, J.M., Duckstein, L., 1993. A multiple criteria decision-making approach to GIS- Inform. Decis. Anal. 2, 18–30.
based land suitability evaluation. Int. J. Geograph. Inform. Sci. 7, 407–424. Trajkovic, S., 2005. Temperature-based approaches for estimating reference evapo-
Pohekar, S., Ramachandran, M., 2004. Application of multi-criteria decision making to transpiration. J. Irrigat. Drain. Eng. 131, 316–323.
sustainable energy planning—a review. Renew. Sustain. Energy Rev. 8, 365–381. Vågen, T., et al., 2010. AfSIS technical specifications: soil health surveillance. World
T. Prakash. Land suitability analysis for agricultural crops: a fuzzy multicriteria decision Agroforestry Centre, Nairobi, Kenya.
making approach. ITC; 2003. Vahidnia, M.H., Alesheikh, A.A., Alimohammadi, A., 2009. Hospital site selection using
Rademacher-Schulz, C., Schraven, B., Mahama, E.S., 2014. Time matters: shifting sea- fuzzy AHP and its derivatives. J. Environ. Manage. 90, 3048–3056.
sonal migration in Northern Ghana in response to rainfall variability and food in- Vörösmarty, C.J., Federer, C.A., Schloss, A.L., 1998. Potential evaporation functions
security. Clim. Develop. 6, 46–52. compared on US watersheds: possible implications for global-scale water balance and
Ramoelo, A., et al., 2014. Validation of global evapotranspiration product (MOD16) using terrestrial ecosystem modeling. J. Hydrol. 207, 147–169.
flux tower data in the African savanna, South Africa. Rem. Sens. 6, 7406–7423. W.R. Walker. Guidelines for designing and evaluating surface irrigation systems; 1989.
Rosenzweig, C., Parry, M.L., 1994. Potential impact of climate change on world food Wang, G., Qin, L., Li, G., Chen, L., 2009. Landfill site selection using spatial information
supply. Nature 367, 133–138. technologies and AHP: a case study in Beijing, China. J. Environ. Manage. 90,
Ruhoff, A., et al., 2013. Assessment of the MODIS global evapotranspiration algorithm 2414–2421.
using eddy covariance measurements and hydrological modelling in the Rio Grande Wilson, K.B., Hanson, P.J., Mulholland, P.J., Baldocchi, D.D., Wullschleger, S.D., 2001. A
basin. Hydrol. Sci. J. 58, 1658–1676. comparison of methods for determining forest evapotranspiration and its compo-
Saaty, T.L., 1977. A scaling method for priorities in hierarchical structures. J. Math. nents: sap-flow, soil water budget, eddy covariance and catchment water balance.
Psychol. 15, 234–281. Agri. Forest Meteorol. 106, 153–168.
Saaty, T.L., 1980. The analytic hierarchy process: planning, priority setting, resources Worqlul, A.W., Collick, A.S., Rossiter, D.G., Langan, S., Steenhuis, T.S., 2015. Assessment
allocation. McGraw, New York, pp. 281. of surface water irrigation potential in the Ethiopian highlands: the Lake Tana Basin.
Saaty, T.L., Vargas, L.G., 1991. Prediction, projection, and forecasting: applications of the Catena 129, 76–85.
analytic hierarchy process in economics, finance, politics, games, and sports. Kluwer Worqlul, A.W., et al., 2018a. Multi-dimensional evaluation of simulated small-scale irri-
Academic Pub. gation intervention: a case study in dimbasinia watershed, Ghana. Sustainability 10,
Schmidhuber, J., Tubiello, F.N., 2007. Global food security under climate change. Proc. 1–23.
National Academy Sci. 104, 19703–19708. Worqlul, A.W., et al., 2017. Assessing potential land suitable for surface irrigation using
Şener, Ş., Şener, E., Nas, B., Karagüzel, R., 2010. Combining AHP with GIS for landfill site groundwater in Ethiopia. Appl. Geography 85, 1–13.
selection: a case study in the Lake Beyşehir catchment area (Konya, Turkey). Waste Worqlul, A.W., et al., 2018b. Impact of climate change on streamflow hydrology in
Manage. 30, 2037–2046. headwater catchments of the upper blue Nile basin Ethiopia. Water 10, 120.
Srivastava, A.K., Mboh, C.M., Zhao, G., Gaiser, T., Ewert, F., 2018. Climate change impact Xu, C.Y., Singh, V., 2001. Evaluation and generalization of temperature-based methods
under alternate realizations of climate scenarios on maize yield and biomass in for calculating evaporation. Hydrol. Process. 15, 305–319.
Ghana. Agri. Syst. 159, 157–174. Yang, Y., Guan, H., Shang, S., Long, D., Simmons, C.T., 2014. Toward the use of the
Steuer, R.E., Na, P., 2003. Multiple criteria decision making combined with finance: a MODIS ET product to estimate terrestrial GPP for nonforest ecosystems. IEEE Geosci.
categorized bibliographic study. Euro. J. Operat. Res. 150, 496–515. Rem. Sens. Lett. 11, 1624–1628.
Storie, R.E., 1978. Storie index soil rating. Division of Agricultural Sciences. University of You, L., et al., 2011. What is the irrigation potential for Africa? a combined biophysical
California. and socioeconomic approach. Food Policy 36, 770–782.
Sumathi, V., Natesan, U., Sarkar, C., 2008. GIS-based approach for optimized siting of Zhang, J., Su, Y., Wu, J., Liang, H., 2015. GIS based land suitability assessment for to-
municipal solid waste landfill. Waste Manage. 28, 2146–2160. bacco production using AHP and fuzzy set in Shandong province of China. Comput.
Swaine, M., 1996. Rainfall and soil fertility as factors limiting forest species distributions Electron. Agric. 114, 202–211.
in Ghana. J. Ecol. 419–428.

125