Introduction

Water is vital to life but in some areas of the world, its availability is dwindling. This originates from a growing population, climate change, and emerging new demands driven by economic and population growth. Water scarcity has an even greater impact in regions where water from one source needs to be shared among multiple countries. A famous example of a transboundary water source is the Nile River in northeast Africa (Melesse et al. 2014).

As the longest river in the world, the Nile serves as a central source of water for a growing population of more than 487 million people (NBI 2017). As a consequence, pressure is put on the limited water resource (NBI 2017; Oestigaard 2012). Water scarcity, accompanied by water control along the Nile, has always been a fundamental concern in the history of the Nile Basin (Oestigaard 2012). Egypt and Sudan have signed the 1959 agreement which assigns 55.5 billion cubic metres (BCM) of Nile water to Egypt and 18.5 BCM to Sudan (Oestigaard 2012; Cascao 2009). But since then, the population in both countries are increasing dramatically and the pressure on water resources is increasing proportionally.

Additional vulnerability comes with climate change which causes increasingly unpredictable droughts as well as floods (Oestigaard 2012). After all, this situation poses a threat to water and in addition, food security since agriculture depends on a secure water supply. The agricultural sector in the Nile Basin, where water is needed for irrigation, largely depends on the Nile’s water supply (NBI 2017). However, if there is not enough water available, cultivation of land and therefore food production is only limited or not possible at all. The challenge is to manage the water resource to sustain a sufficient water supply and meet the increasing food demand for all eleven riparian countries of the Nile (Oestigaard 2012).

Considering the recent construction of the Grand Ethiopian Renaissance Dam (GERD) careful water management among the riparian states is especially important now (Ali et al. 2014). The GERD is located at the Blue Nile in Ethiopia and it will be the largest concrete dam (Webuild 2020) and largest hydroelectric power plant (NBI 2017) in Africa. Impacts on the Nile’s hydrology, in particular in the downstream countries Egypt and Sudan, are expected during the filling period of the dam and in its long-term operation (Mordos et al. 2020). This endangers the major source of freshwater in both downstream countries (NBI 2017). Regardless of the results of the current negotiations between the three countries, the establishment of the GERD will regulate the flow coming into Sudan from the Blue Nile over the year. This could encourage an uncontrolled extension of the irrigated agriculture in Sudan which could significantly affect the water flowing to Egypt if it exceeds the water share of Sudan from the Nile [18.5 BCM (Oestigaard 2012; Cascao 2009)].

This study aims to evaluate how much water volume Sudan’s irrigated agriculture sector requires.

The potential for irrigated agriculture in Sudan is evaluated to estimate the future increase in water pressure in the region. For this, the potential for irrigated agriculture in Sudan is evaluated first. This is achieved by implementing a land suitability analysis for irrigated agriculture in Sudan under consideration of several parameters which represent the prevailing conditions and define limitations for irrigated agriculture. The parameters are chosen as determined from literature research and case-related reflections. A land suitability analysis is implemented by building a model in ArcMap Model Builder. The result of the model are maps that show the suitable and not suitable areas for irrigated agriculture in Sudan.

The calculated water demand for irrigation demonstrates whether Sudan is able to expand its irrigated agriculture sector or whether it is limited by the current share of water. The amount of water calculated can be used as a basis for new negotiations on the water volumes shared between the riparian countries of Ethiopia, Sudan, and Egypt.

This study shows that it is possible to implement an LSA using ArcMap and including the AHP process. It demonstrates that the guidelines for an LSA introduced by FAO in 1993 (FAO 1993) can be followed with ArcMap and the AHP process can be integrated by making use of extensions for ArcGIS. However, further LSAs with different approaches are needed to reassure the results.

Materials and methods

Study area

The study area is the whole country of Sudan. Sudan is the third-largest country in Africa with an area of about 1.88 million km2 (FAO 2015). Sudan’s population was around 42.8 million in 2019 and is constantly growing with an annual growth rate of around 2 percent (2.4% over the 2017–2019 period) (The World Bank 2019). The population density is 20 inhabitants/km2 and 70% of the population lives in rural areas (FAO 2015). Most of the population lives along the Nile, which flows from the south of the country to the north, with the White Nile and the Blue Nile merging in the capital city of Sudan, Khartoum (The World Bank 2021b; FAO 2015). Specific features of the country, which were also used for the land suitability analysis, are presented in chapter 2.3.

Land suitability analysis

A land suitability analysis (LSA) or land suitability evaluation (LSE) is part of a land-use planning and decision-making process (Malczewski 2003; FAO 1993). It is a method where homogeneous areas are identified and related to suitability for specific uses (Hopkins 2007). It serves as a tool to identify the most suitable places for locating future land uses (Collins et al. 2001).

The FAO published a Framework for land suitability evaluation in 1976 (FAO 1976) which still builds the basis for any kind of land suitability evaluation. It is a collection of concepts, principles, and procedures which serve as an orientation to develop evaluation systems (Verheye et al. 2009). After the FAO approach to LSA, land should be assessed on its suitability based on some basic criteria, the principles of the FAO framework: the requirements of specific land use; a comparative analysis of inputs vs. benefits; the physical, economic, and social context and potential environmental impacts (George 2005; Verheye et al. 2009). Sys et al. (1991) also name the two major aspects that LSE deals with: physical resources and socio-economic resources. Literature review shows that most LSA applications focus on the assessment of the physical potential of land only (Verheye et al. 2009). However, the main physical land-use resources to be taken into account are climate, topography, wetness, physical soil characteristics, fertility characteristics, salinity, and alkalinity (Sys et al. 1991).

The result of an LSA is presented as a suitability map, which shows the spatial pattern of requirements, preferences, or predictors of some activity (Hopkins 2007). From the interactions of those factors and requirements, a ranking for the land in the study area can be developed. For the ranking of land suitability, the FAO suggests a classification from ‘suitable’ to ‘not suitable’ as shown in Table 1 (FAO 1993).

Table 1 Structure of the FAO land suitability classification (FAO 1993)
Full size table

Features of the study area used for land suitability analysis

The following parameters were considered for the land suitability analysis for Sudan. The parameters were selected according to recommendations in the literature (Qu et al.; 2013; Hopkins; 2007; FAO 1976, 1993; Sys et al. 1991) and other LSAs conducted, potentially with the use of GIS software (AL-Taani et al. 2021; Fekadu and Negese 2020; Aldababseh et al. 2018; Elmobarak and Salih 2013; Qu et al. 2013; Feizizadeh and Blaschke 2012; Chandio and Matori 2011; Paiboonsak et al. 2011; Jafari and Zaredar 2010; Baniya 2008; Hopkins 2007; Malczewski 2003).

The slope of the earth’s surface in Sudan ranges from 0 to about 26 degrees, see Fig. 1. The country is generally flat except for the Jebel Marra Mountains in the west, the Red Sea Hills in the northeast, and the Nuba Mountains in the south. The three ecological zones in Sudan from north to south are the desert, the semi-desert, and the low rainfall savannah. Sudan has a tropical sub-continental climate, extending from desert climate in the north through a belt of summer-rain climate to semi-dry climate (FAO 2015). This can be seen in Fig. 2 for the mean annual temperature and in Fig. 3 for the average annual precipitation. The mean annual temperature in Sudan ranges from 14 to 30 °C. It is hottest in the eastern central part and colder temperatures appear in the northwestern region as well as in the mountainous regions. The precipitation ranges from 1 to 1300 mm with a clear increase of rainfall from north to south. Overall, the climate is highly variable and prone to erratic rainy seasons. This results in droughts and floods, either localised due to rainfall and runoff or widespread, caused by an overflow of the Nile and its tributaries (FAO 2015).

Fig. 1
figure 1

Surface slope

Full size image
Fig. 2
figure 2

Average annual temperature

Full size image
Fig. 3
figure 3

Average annual precipitation

Full size image

On the other hand, from Fig. 4, it can be seen that there is generally more vegetation in the south than in the north. The north of the country is mostly covered with sand. This refers to Sudan’s soils which can be divided geographically into sandy and sand dunes respectively in the northern and west-central areas, clay in the central region, red ironstone in the south, and alluvial soils which are less extensive and widely separated (FAO 2015; Mahgoub 2014). However, half of the country is bare rocks and soil, such as wind-blown sands free of vegetation in hyper-arid regions (FAO 2015). The selected soil properties for this study, organic carbon, depth, pH value, and texture, are presented in Figs. 5, 6, 7, and 8. Most of the soil in Sudan has a depth of 75–100 cm. In the mountainous regions, the soil is less deep with ranges between 0 and 25 cm. The percentage of organic carbon in the soil in Sudan ranges from 0 to about 2.4%. It is lower in the northwest and higher in the southeast region. The pH value of soil ranges from 0 to 7. In the northwest values around 0 to 6 are found and in the southeast, values mainly range from 6.7 to 7. The soil texture in Sudan is mainly sand, but clay appears in the southern region between White Nile in the west to Atbara in the east. The clays in central Sudan are agriculturally the most important soils. Many of the agricultural schemes such as the Gezira, Rahad, New Halfa, and others make use of them (Mahgoub 2014).

Fig. 4
figure 4

Landcover

Full size image
Fig. 5
figure 5

Soil depth

Full size image
Fig. 6
figure 6

Soil organic carbon

Full size image
Fig. 7
figure 7

Soil pH value

Full size image
Fig. 8
figure 8

Soil texture

Full size image

Sudan has limited internal renewable water resources (IRWR) (FAO 2015). The two sources of surface water are rainfall and water from rivers (Mahgoub 2014). But rainfall appears erratically and is concentrated in a short season. Surface water from rivers mainly comprises the Nile River system, so-called Nilotic water, and a few non-Nilotic streams. The Nile River flows about 6700 km from south to north through eastern Africa (NBI 2017; Melesse et al. 2014; Mikhailova 2001; Sutcliffe and Parks 1999). It is the longest river in the world and comprises two major tributaries, the White Nile and the Blue Nile (NBI 2017; Melesse et al. 2014; Mikhailova 2001). In Khartoum, the capital of Sudan, the White Nile and the Blue Nile merge and are simply called the Nile from there on. Below Khartoum, 325 km north, the Nile is joined by its last tributary, the Atbara (Sutcliffe and Parks 1999). In northern Egypt, the Nile reaches the Mediterranean Sea through the Nile Delta which has an area of about 22,000 km2 (Fishar 2018; Hamza 2009; Mikhailova 2001). The Nile river basin spans 10 countries (Oestigaard 2012; Sutcliffe and Parks 1999), or 11 countries since South Sudan’s independence on July 9, 2011 (Oestigaard 2012). It covers an area of 3.18 million km2 (NBI 2017; FAO 1997; Sutcliffe and Parks 1999), which represents 10% of the African continent (NBI 2017; Oestigaard 2012; FAO 1997). The Nile system within Sudan consists of the Blue Nile and Atbara rivers, the White Nile downstream of Malakal, and a small part of the Bahr El Ghazal basin, which is mainly located in South Sudan. The Nile River flows are characterised by high variability. Generally, the Blue Nile contributes about 60% of the water that makes up the Main Nile but has an extreme range in discharge between the peak and the low periods (FAO 1997). However, the White Nile systems discharge is more uniform (FAO 1997). Just before the Blue Nile and the White Nile merge, the White Nile at Mogren has a flow rate of 26.0 BCM (1911–1995) and the Blue Nile at Khartoum of 48.3 BCM (1900–1995). At Dongola, after all, tributary inflows, the mean annual flow of the Main Nile is 84.1 BCM (1890–1995) (FAO 2021g).

To manage the highs and lows of the Nile River there are several storage facilities along its river course. In Sudan, there are five main dams that have an estimated total active storage capacity of 19.17 BCM. The water of the dams is used for generating hydropower and supplying irrigation water to irrigation schemes (Ahmed and Ribbe 2011).

Another water resource in Sudan is groundwater resources which are found in seven transboundary aquifers. The largest groundwater aquifer is the Nubian sandstone aquifer system (FAO 2015). As overgrazing appears around water points, groundwater yield has been reduced, aquifer levels have lowered in Dafur and Kassala states, and intrusion of seawater appeared in the Red Sea coastal zone (FAO 2015). After all, groundwater resources play a minor role in the water supply for the country due to the high costs of pum** and a weak infrastructure (Mahgoub 2014).

In total, Sudan’s annual water withdrawal is estimated at 26.1 BCM for 2015 (FAO 2015), which means that the country uses most of its renewable water resources (NBI 2017).

The agricultural sector in Sudan is the principal source of income and livelihood for 60 to 80% of the population (Mahgoub 2014). The main sub-sectors within the agricultural sector are livestock, crop production, and forestry, and fishery. The main patterns for crop production in Sudan are irrigated agriculture, semi-mechanized rainfed agriculture, and traditional rainfed agriculture. The area under irrigation is about 1.68 million hectares, semi-mechanized rainfed agriculture is practiced on 6.7 million hectares of land and traditional rainfed agriculture covers about 9 million hectares (FAO 2020). This makes up a total area for crop production of 17.4 million hectares (FAO 2020), which is 9.2% of the total area of Sudan. If permanent meadows and pasture, as well as the cultivated land, are taken into account, the agricultural land makes up 60% of the total area of the country. The area for rainfed agriculture is the largest but it varies greatly from year to year as it is depending on the variability of rainfall. The rainfed farming system consists of mainly small farms, with low input levels and poor yields. The total area for irrigated agriculture is the smallest of the three patterns. Nevertheless, irrigated agriculture has a key role as Sudan has the largest irrigated area in Sub-Saharan Africa and the second largest in the whole of Africa. The country’s agricultural production depends on irrigation because of droughts, rainfall variability, and uncertainty. With irrigated agriculture, more than twice the yield can be obtained as with rainfed agriculture (The World Bank 2021a; FAO 2002) which makes this sector Sudan’s central option to boost the economy and increase the living standard for its population (FAO 2015). The large national irrigation schemes, which use the river flow from the Nile and its tributaries, are the Gezira, Suki, New Halfa, and Rahad schemes, see Fig. 9 (FAO 2020).

Fig. 9
figure 9

Location of irrigation schemes after Ali et al. (2014)

Full size image

In the eastern region of Sudan, the Gash Delta and the Tokar Delta (Cole 2021) are two large irrigation schemes that make use of seasonal floods (FAO 2020; Mahgoub 2014). The crops cultivated overall in Sudan are diverse. The crop portfolio includes cereals such as sorghum, wheat, rice, or maize, oilseeds like groundnuts or sesame, industrial crops like cotton or sugarcane, pulses such as broad beans, and fodder crops, and horticultural crops (FAO 2020).

Sudan’s agriculture depends on weather conditions, especially rainfall which is the most important driver of national food crop production (FAO 2020; Elgali and Mustafa 2017). Precipitation is crucial to the rainfed sector as well as to the irrigation sector as rainfall supplements irrigation water and also supports crop establishment and development (FAO 2020). However, in the last forty years, summer rainfall has been decreasing up to 20 percent, while temperatures have recorded an increasing trend (Elgali and Mustafa 2017). These occurrences result in less precipitation. Combined with higher evaporation rates, due to higher temperatures, available surface water resources are reduced further. As a consequence, the change in climate leads to a decrease in the supply of cereals (Elgali and Mustafa 2017). In the irrigated agriculture sector, the crop yields remain low compared to world standards. Sequential extreme weather events due to climate change are one side responsible for increased incidences of pests, diseases, and weeds. The crops are weakened by such events and become more susceptible to diseases. On the other side, poor maintenance and silting up of canals, the shortage of efficient modern pumps, and the adoption of traditional agricultural practices affect the agricultural yield. These prevent maximum efficient use of a constant water resource and therefore, the full potential of more intensive farming cannot be exploited (FAO 2020).

There are mainly two categories of irrigated agriculture in Sudan: traditional irrigation and modern schemes (NBI 2008; UNEP 2007). Irrigation systems in modern irrigation include surface, sprinkler, and localised irrigation systems like pumps (FAO 2015; NBI 2008; UNEP 2007). Also, furrow irrigation, long furrow irrigation, gated pipes, drip irrigation, as well as centre-pivot systems are used (NBI 2008). The irrigation schemes are gravity irrigation schemes, which means the water is transported from the river to the fields and through the irrigation canals by gravity (Mahgoub 2014). Poor maintenance of the infrastructure has, however, led to canal sedimentation and weed growth inside the canals, clogging them and reducing irrigation efficiency. Overall, this leads to lowered crop** returns (Mahgoub 2014; UNEP 2007).

In traditional irrigation, crops are irrigated by surface and flood irrigation (NBI 2008). They are practised in three different ways. The first method makes use of the highly fertile, so-called ‘gerf’ lands, which are exposed after the withdrawal of annual floods. This technique uses the residual moisture in the soil profile that is available when the floodwater recedes. The second type of traditional irrigation technique is the use of small-scale irrigation pumps. This evolved from previously used hand-operated levers and animal-driven waterwheels. The third type is spate irrigation. It relies on the capture and redirection of seasonal runoff to flood-wide areas of arable land (UNEP 2007).

This study is carried out to evaluate how much water volume Sudan’s irrigated agriculture sector requires. For this, the potential for irrigated agriculture in Sudan is evaluated. The water demand can then be calculated if the irrigated agriculture sector is expanded to a maximum. Several steps must be taken to receive the resulting water volume. For the final calculation of the water demand, a land suitability analysis (LSA) where areas suitable for irrigated agriculture are defined, is implemented first. A model is designed with the Geographic Information System (GIS) software ArcMap to perform the LSA. Datasets that represent the prevailing conditions in Sudan are combined with restrictions for crop cultivation under irrigated agriculture. After all, the LSA is carried out separately for Faba bean, sorghum, sugarcane, and an idealistic plant x. The output of the LSA is crop-specific suitability maps. From these maps, the sizes of suitable areas for irrigated agriculture for each crop can be calculated. Additionally, for the final calculation, the irrigation water demand for the cultivation of the selected plants is evaluated. Therefore, a crop** pattern is developed where the crops from the LSA are combined into a schedule for cultivation on one field throughout the year. Subsequently, the irrigation water demand for the resulting growing period for each crop is calculated with the software CLIMWAT, CROPWAT and AquaCrop provided by the Food and Agriculture Organization of the United Nations (FAO). The software programmes have been applied numerous times in comparable environments, for example in Ahmed (2020), Starr et al. (2020) and Elsheikh (2016). Finally, the respective results are combined and the yearly water demand for irrigated agriculture in Sudan is calculated.

GIS-based land suitability analysis

Since the FAO framework was published in 1976 technology has advanced. This created the opportunity to facilitate the implementation of the framework’s principles. The main advance came with the development of affordable GIS software. Computerized databases and modelling programmes included in GIS software facilitate the computational aspects of land evaluation such as matching requirements with land qualities (George 2005), and the efficiency and the accuracy of LSE improved greatly (Qu et al. 2013). Where overlay techniques had been done by hand to create suitability maps, they can now be implemented with GIS software (Malczewski 2003). Today, GIS-based land suitability analysis has become the mainstream within this research field (Qu et al. 2013; Malczewski 2003; Collins et al. 2001). In recent years, many LSA have been conducted with the use of GIS (Al-Taani et al. 2021; Fekadu and Negese 2020; Qu et al. 2013; Feizizadeh and Blaschke 2012; Chandio and Matori 2011; Paiboonsak et al. 2011; Baniya 2008).

For this study, a GIS-based land suitability analysis was implemented with the software ArcMap by ESRI. A model was developed in ArcMap Model Builder which allows for the building of a geoprocessing workflow to automate and document spatial analysis and data processing. The model is represented as a diagram that chains together sequences of processes and tools, see Fig. 10. Within the model, the output of one process or tool is used as the input to another process (ESRI 2021). The developed model performs a crop-specific land suitability analysis. The result is a map that shows areas suitable for the cultivation of a crop with a distinction between the suitability classes S1, S2, and S3 and also not suitable areas with the suitability class S4. The suitability class N was redefined as class S4 for the model in this study to use only numbers as suitability classes, meaning 1, 2, 3, and 4, in the model and thus in the software. A separate model is built for each crop to consider the respective crop requirements. However, the overall model chain is the same for all crops.

Fig. 10
figure 10

Model chain of the developed land suitability model in ArcMap

Full size image

The model was run for the crops Faba beans (Vicia Faba), sorghum (Sorghum bicolor), and sugarcane (Saccharum officinarum). These plants were chosen for the land suitability analysis as they play a major role in Sudan’s agriculture and/or for Sudan’s population. Also, the cultivation area of the crops differs so that altogether a large area of Sudan is covered. The following facts which apply to Sudan support this argumentation:

  • Sorghum: Main export crop (FAO 2015); most important food source (Bayer 2021; Elgali and Mustafa 2017); principle crop of the irrigated sector (FAO 2020); cultivation in eastern and southern Sudan (Bayer 2021; FAO 2020)

  • Sugarcane: Main export crop (FAO 2015); Sudan main producer among Arab and African countries (FAO 2020; Ismail 2006); principal crop of the irrigated sector (FAO 2020) grown mainly in central and eastern Sudan (Ismail 2006)

  • Faba beans: Major part of the daily diet (STP 2021; Gasim et al. 2015); good suitability for crop rotation (STP 2021; Gasim et al. 2015); cultivated in northern and central Sudan (STP 2021; Gasim et al. 2015)

The selected input parameters for the model are Sudan’s average annual precipitation, the soil organic carbon content, soil depth, the surface slope, the soil pH value, the average annual temperature, the landcover, and the soil texture, as seen in Fig. 10 at the very top. These parameters have been chosen as input parameters according to recommendations in the FAO Framework for Land Evaluation (1976) and also Sys et al. ‘Principles in land evaluation and crop production calculations’ (1991). They are also integrated as the agricultural potential is to be assessed and thereby soil data, as well as climatic data, play an important role. For example organic carbon as well as pH value indicate the health status of the soil (FAO et al. 2009). Precipitation and its distribution over the year for example determines the length of a growing season (Sys et al. 1991), is crucial for the growth and yields of crops, and irrigation is needed if too little rainfall appears for crop growth (Aldababseh et al. 2018).

All input parameters are in a raster format with a resolution of 30 min, which is a cell size of approx. 1 km2 (WorldClim 2020b). The spatial extent of the data is the country Sudan.

Requirements for crop growth are needed to define restrictions for irrigated agriculture. The crop requirements are obtained from Galdos et al. (2009), Paiboonsak et al. (2011), and Sys et al. (1993). Table 2 shows the classification for selected land suitability criteria for sorghum. Accordingly, the data was collected and listed for sugarcane and Faba beans. They are incorporated into the model with the ‘Reclassify’ tool to assign a suitability class to the corresponding values of the input rasters. For values of the suitability class S1, the number one is assigned, for class S2 the number two, for S3 three, and N the number four.

Table 2 Classification of selected land suitability criteria for sorghum
Full size table

Within the LSA limiting factors are defined, which means that the attributes of the factor appear too unfitted to preclude any possibilities of successful use of irrigated agriculture (FAO 1976), and therefore are considered not suitable. In the model, the not suitable values from the factors landcover, temperature, slope, soil pH, and soil texture are considered limiting. Within the model, the limiting values are extracted from the respective input raster by the tool ‘Extract by Attributes’. The resulting rasters are merged into one map which shows all not suitable areas according to the limiting factors. In the last step of the model chain, the map from the limiting factors is added to the map from the weighted overlay process so that the final suitability map is created.

With the weighted overlay procedure, all parameters or rasters are overlaid according to a measurement scale and also weighting, which depends on the defined importance of a factor (ESRI 2016). In the LSA model, the overlay procedure is implemented with the tool ‘Weighted Overlay’. The measurement scale is chosen as ‘1–4 by 1’ following the suitability classification. The weight of each raster must be defined in a separate process. For this study, the weight of each raster is defined by performing an Analytical Hierarchy Process (AHP) which is an approach derived from Multi-Criteria Decision Analysis (MCDA). The Analytical Hierarchy Process (AHP) was developed by Saaty in 1970 (Collins et al. 2001; Saaty 1977) and is a procedure for weighing and comparing (Baniya 2008). It is a multi-criteria decision-making approach in which factors are arranged in a hierarchic structure (Saaty 1990). It is one of the most popular methods to obtain criteria weights in multi-criteria decision-making (Sulieman et al. 2015; Chandio and Matori 2011; Chen et al. 2010; Jafari and Zaredar 2010). Mathematical relations of the AHP are explained in detail in Saaty (2001) and Saaty (1977). In a GIS-based AHP process, the weights associated with criterion map layers are calculated with the help of a preference matrix. In this matrix, all identified criteria are compared against each other with preference factors (Chen et al. 2010). By pairwise comparison, based on mathematical representation, relative weights are calculated (Collins et al. 2001) to obtain an ascending hierarchical order (Saaty 2005). To automate the complex calculation of the weights in this study the extension extAhp20 (ESRI 2017) for ArcGIS software was used.

The preference matrix for the input layers is shown in Table 3. The scale of the preference factors ranges from one to nine per Malczewski (1999). The specific preference factors which compare each parameter against each other (pairwise comparison) are obtained from literature research. The literature consulted is: Al-Taani et al. (2021), Fekadu and Negese (2020), Nigussie et al. (2019), Feizizadeh and Blaschke (2012), Chandio and Matori (2011), Chen et al. (2010). Where no values were found it is assumed that two parameters have equal importance and a value of one is assigned. This is the case for the pairwise comparison of temperature and landcover, landcover and soil organic carbon, landcover and soil pH, and landcover and precipitation.

Table 3 Pairwise comparison matrix for analytical hierarchy process with calculated final weights and consistency ratio CR
Full size table

The result of the AHP gives the relative weights of the layers as shown in the rightest column in Table 3. On top of the relative weights, the extAhp20 extension gives the Consistency Ratio (CR) of the matrix, which is a part of the AHP to measure how consistent the matrix is (Saaty 1990). If the CR is smaller than or the same as 0.1 (10%) the matrix is considered to be consistent enough and the weight values can be utilised, if it is larger than 0.1 the judgments in the matrix require revision (Chen et al. 2010; Baniya 2008; Saaty 1990). The matrix created for this study has CR equal to 0.095781 which is smaller than 0.1 and therefore the weights are valid and can be used for the ‘Weighted Overlay’ tool.

For model validation, values from literature about existing irrigation schemes as well as satellite images of the Nile region in Sudan obtained from Google Maps are compared to the results of the model. As there is a lack of studies on land suitability analysis to evaluate the potential for irrigated agriculture in Sudan it is not possible to validate the model by comparing its results to studies by other authors. However, with values from literature about existing irrigation schemes the aggregation of suitable and not suitable areas can be reviewed. It can be checked on one side whether any gradation of suitability from south to north or west to east can be supported and on the other side if the suggested suitable areas match the current areas under cultivation for the individual crops. With satellite images existing irrigation schemes can be identified to see if they lay in areas suggested as suitable land for irrigated agriculture by the model. If they overlap, the model results can be considered verified. For this study, the validation by satellite images is focussed on the Nile region in Sudan.

Calculation of water demand

From the areas of the individual suitability classes and crops, the water demand for cultivation under irrigated agriculture is calculated. For this, the crops are combined in one crop** pattern to illustrate the growth of the plants on one field in a year. The crop-specific planting and harvesting dates are obtained for poor and dense savannah zone from FAO Crop Calendar for Sudan (FAO 2021c) and recommendations and information on crop** and crop rotation is taken from FAO Crop Information (FAO 2021d). For the calculation of the plant-specific irrigation water demand, the software AquaCrop by the FAO is used (FAO 2021a). For this study, the required irrigation water volume is determined with the software based on the interaction of climate, soil profile, plant and plant growth, and irrigation in a defined simulation period. Climate data is obtained from the FAO software CLIMWAT and CropWat (FAO 2021b, e) whereby the data from the climate station in Ed-Dueim in Sudan was chosen. Ed-Dueim is located 200 km south of Khartoum, on the banks of the Nile River. As an irrigation method, furrow irrigation is chosen as it is a common but ineffective irrigation technique (Waller and Yitayew 2016; FAO 2002) which contributes to simulating a worst-case scenario. Simplifying, clay is assumed to be the soil for Sudan as it is agriculturally the most important soil in Sudan (Mahgoub 2014) and suits all types of irrigation systems for irrigated agriculture (Elmobarak and Salih 2013). The simulation period in AquaCrop is defined according to the growing cycle of the crop as defined in the crop** pattern. The growing period, hence simulation period, for Faba beans was set from 1st November to 15th March (135 days), for sorghum from 1st July to 31st October (123 days), and for sugarcane from 1st June to 31st May (365 days).

Data

Climate data such as temperature and precipitation were obtained from the WorldClim (2020a) data website, a database of high spatial resolution global weather and climate data. Climate data is available on the website for the years 1970–2000 and a spatial resolution of 30 s (approx. 1 km) was used. The elevation data was also retrieved from this website with the same resolution as the climate data. With the ArcMap tool ‘slope’, the slope was calculated from the elevation data for the study area.

Data on soil properties such as percentage of organic carbon, depth, pH value, and texture were obtained from the Harmonized World Soil Database. The database is a 30 arc-second raster database with worldwide soil data stored in a standardized structure which allows the linkage of attribute data with the raster map (FAO 2021f).

The land cover data is retrieved from the Multipurpose Landcover Database for Sudan. The data is based on visual interpretation of LANDSAT TM images acquired mainly in the period 1994–1999 (FAO 2003). For this study, the land cover was aggregated to 16 land cover classes according to their land cover classification. The resulting classes are Shrubland, Woodland, Herbaceous Crops, Built-Up Areas, Natural Waterbodies, Open Shrubs, Grasslands, Consolidated Bare Areas, Tree Crops, Artificial Waterbodies, Unconsolidated Bare Areas, Mixed Class, Forest, Non Built-Up Areas, Thicket and Closed Shrubs.

The plant-specific requirements are obtained from Paiboonsak et al. (2011), Galdos et al. (2009) and Sys et al. (1993). As the main source of crop requirements, Sys et al. (1993) were used and supplemented by the other sources if values were not available. For sorghum and sugarcane, this is the case for temperature values which were obtained from Galdos et al. (2009) and precipitation values which were obtained from Paiboonsak et al. (2011).

Results

The land suitability evaluation shows similar results for sugarcane and sorghum and deviating therefrom for Faba beans. The model calculates the largest suitable area for Faba beans with 134.9 million ha, followed by sugarcane at 68.3 million ha and sorghum at 67.7 million ha, see Table 4.

Table 4 Sizes of suitable and not suitable areas defined by the land suitability analysis with ArcMap
Full size table

The suitable area for Faba beans is larger as the soil texture ‘sand’ was assumed as marginally suitable (S3) for Faba beans, in contrast to sugarcane and sorghum where sand is defined as not suitable in the crop requirements. Since soil texture is a limiting factor in the land suitability analysis, and sand is the soil texture to a large extent in the western part from north to south in Sudan, the not suitable area is also correspondingly large. For the following comparison of the suitability maps, see Figs. 11, 12, and 13. For all three crops, an accumulation of not suitable areas in northern Sudan and increasing suitable areas further south are determined, whereby this distribution is less significant for Faba beans. The highly suitable (S1) areas are located in the southeast for all three crops, and for Faba beans and sorghum, some S1-areas are also in the southwest. Moderately suitable (S2) areas do not reach as far south as S1-areas. They appear for sugarcane in the southwest, central and northeast Sudan, for Faba beans in western Sudan from the south of the country to the north, and in eastern Sudan from north southwards to a little more than half the length of the country and are continuously interspersed with not suitable areas in the northern half of the country. For sorghum, the S2-areas are located in the northern part of the country whereby smaller areas are in the west and centre and vast areas are concentrated in the northeast. For all three crops, marginally suitable (S3) areas accumulate increasingly from north to south along the Nile River and the Blue and White Nile, some S3-areas are also scattered across the country. However, the areas are significantly smaller than for all other suitability classes and are not visible on the figures due to the chosen scale. The not suitable (N) areas have the largest share of all suitability classes for sugarcane and sorghum. For Faba beans the reason for the smaller area with suitability class ‘not suitable’ and therefore larger suitable area was elaborated above. Not suitable areas appear to a greater extent northeastern Sudan, where the Red Sea Hills are located. For sugarcane and sorghum large not suitable areas are also located in the east from north to south and the south central region. Altogether the different distributions of suitable and not suitable areas result from the different crop requirements of each plant and also the weighting process in the model.

Fig. 11
figure 11

Land suitability for sugarcane

Full size image
Fig. 12
figure 12

Land suitability for Faba beans

Full size image
Fig. 13
figure 13

Land suitability for sorghum

Full size image

The water demand for the growing period, as determined in the crop** pattern for this study, is calculated with AquaCrop for Faba beans at 972.60 mm, for sorghum at 923.8 mm, and sugarcane at 5133.6 mm. The resulting water demand for irrigated agriculture is calculated by multiplying the irrigation water demand for the growing period with the areas suitable for irrigated agriculture and adding the resulting values for each crop and suitability class S1, S2, and S3. In Table 5 the resulting water demands for different scenarios are presented. The final irrigation water demand depends on the scenario chosen for the simultaneous cultivation of the selected crops. Due to the combination of the crops into one crop** pattern each crop is cultivated only on a certain share of the agricultural field. The growing periods of the crops allow for Faba beans and sorghum to grow alternatively on the same share of the field and sugarcane on the remaining part of the field. It follows that the field share of either Faba beans or sugarcane plus the share of sugarcane equals 100%. Depending on the distribution of the shares, more or less irrigation water is therefore needed, as the irrigation water demand of the chosen crops varies. For example, sugarcane needs more water and hence the irrigation water demand is higher the larger the share of sugarcane cultivation on the field is.

Table 5 Evaluation of irrigation water demand of areas suitable for irrigated agriculture in sudan under different scenarios
Full size table

Discussion

Validation of the model

The largest areas of irrigation schemes are currently existing along the Blue Nile, followed by schemes along the White Nile, and then the Atbara and Main Nile systems (UN 2020). This validates the gradation of suitable areas in southern Sudan as suggested by the LSA model. Also, the regions where the crops are currently cultivated, sugarcane in the east and central Sudan (Ismail 2006), Faba beans in the north and central Sudan (STP 2021; Gasim et al. 2015), sorghum in the east (Bayer 2021) and south (FAO 2020) Sudan, overlap with the suitable areas for cultivation suggested by the LSA model. A comparison between the suggested suitability areas and satellite images from Google Maps also validate the model. On satellite images agricultural land under irrigation can be identified as, for example, canals that connect to the Nile River within the cultivation area or the round shape of fields that are irrigated via centre-pivot irrigation systems are recognisable. In Fig. 14 irrigation schemes and sites in the Nile region in Sudan are marked. Irrigation sites are marked with a circle as their extent is smaller than that of the irrigation schemes which are marked with polygons. The same accumulation from north to south shows as already stated in the literature. Also, if the locations of the sites are compared to the suitability areas in Figs. 11, 12, and 13, it shows that the existing sites and schemes are located in those areas where the model suggests suitable areas. Regions, where the model and the current agricultural status vary, can be attributed, for example, to settings made for computing within the model chain such as the layer weighting or definition of crop requirements. Furthermore, deviations occur due to simplifications of datasets. Adjustments can be made for the settings of the model but simplifications will remain due to the limited processing capacities of the software and hardware.

Fig. 14
figure 14

Irrigation schemes and sites in the Nile region in Sudan located with satellite images from Google Maps (2021)

Full size image

Use of ArcMap for LSA

Following the utilisation of Model Builder in ArcMap, some limitations result from the technical possibilities of the programme. As it is only possible to integrate tools as part of the model chain in the Model Builder, the integration of the AHP add-on is not possible. This means, the AHP process must be run beforehand and cannot be integrated into the automated process of the LSA. However, the general use of Model Builder for an LSA is beneficial. Setting up the model is a simple task where the largest amount of time is needed to obtain and prepare the input data and the tools also work smoothly.

The following aspects can improve the model for further LSA analyses:

  • Integration of additional physical input parameters

  • Combining separate crop models into one model

  • Mixing the final maps of each model to indicate in which areas which crop is suitable

Other possible further developments are related to the correctness of the GIS data. For example does the climate data date back to the time period from 1970 to 2000 (WorldClim 2020a) and also the data for the landcover dates back to the time period from 1994 to 1999 (FAO 2003). Nevertheless, this is assumed to be accurate enough. However, more recent data sources would represent the prevailing conditions more accurately.

Furthermore, the resolution for raster data is limited. On the one hand, this is due to limited computing capacities as the more detailed a raster dataset is, the more computing power and data storage space is needed. On the other hand, the limited resolution is due to the provision of the data source. For example, the raster data downloaded from WorldClim for average temperature, precipitation, and elevation have a resolution of 30 s. This is a cell size of 1 km2 which is 100 ha. The average area of an agricultural farm in 2016 in Germany was about 61 ha (BMEL 2017). This means that the resolution from the raster data, which later determines the accuracy of the suitability map, is too low to accurately represent the delineation of agricultural land from other uses. However, this study was carried out for the whole of Sudan, and for the overall calculation, it is not important to identify individual farms. Therefore is the accuracy of the raster data assumed to be sufficient.

A further improvement in accuracy, which is not due to the data source, can be made with the land cover that was obtained from FAO (2003). For this study, the land cover classes of the layer were manually summarised from 161 land cover classifiers to 16. However, the more classifiers thus different land cover types are used, the more accurate the data becomes. Concluding, a more detailed delineation of Sudan’s land cover can be obtained if the land cover classes are less aggregated.

After all, the limitations of the use of ArcMap and the Model Builder can be attributed to technical constraints and limited data accuracy. This coincides with the assumptions underlying a land suitability evaluation listed in the FAO Framework for Land Suitability Evaluation (1976). Also, the application of any GIS system and the integration of an LSA therein is limited by the user's knowledge and abilities. Therefore, the LSA is not objective and value-free as it reflects decisions made by the developer Malczewski (2003) and Qu et al. (2013) also names shortages in the whole LSA process, as it is not mature due to the inherent complexity of land systems and evaluation methods. As a result, it is difficult to fully and accurately reflect the true suitability of land. It follows that with different methods, different suitability results can be received. On top of that, there is no uniform way to present the results of an LSA which may also lead to misinterpretations (Qu et al. 2013).

Significance of water demand

The projected volume of irrigation water demand also underlies assumptions and simplifications. The projected water demand is assumed for the case where all of the suggested suitable areas are under irrigated agricultural use, which is an idealistic approach. It is assumed that the source of water for irrigation is the Nile but in practice, it is debatable whether water from the Nile would be transported several hundred kilometres to be used for irrigation. To expand such a network, high investments would be necessary and running costs would also constantly arise due to maintenance works. In addition, the 1959 agreement with Egypt assigns an amount of 18.5 BCM to Sudan which has to be considered in future planning.

Furthermore, variations in the irrigation water demand can be expected as the following assumptions were made within the model:

  • One climate station and the corresponding climate data were assumed to represent Sudan. If a different station or various stations are chosen, other rainfall regimes occur. Therefore, a different irrigation water demand is simulated for the growing period which then affects the total water demand. For example, if Dongola is set as the representative station in AquaCrop, Faba beans have an irrigation water demand of 1323.6 mm for their growing cycle instead of 972.6 mm with El-Dueim as the climate station.

  • One uniform clay soil was chosen in AquaCrop for Sudan. This does not represent reality. If other soils are chosen, the water demand varies. For example, if clay loam is chosen as the soil in Sudan, Faba beans have an irrigation water demand for the growing period of 314.5 mm instead of 972.6 mm as calculated for clay.

  • The choice of crop** patterns and hence crop combination also has an effect on the irrigation water demand. Crops have different water requirements during their growing periods. Some plants need more water, some can grow even if little water is available. If more and different crops are chosen for the crop** pattern, different water demand is to be expected.

Moreover, the demand for irrigation water increases or decreases with the increase in the total area suitable for irrigated agriculture or vice versa. This is the case as the size of the areas is editable as the definition of suitable and not suitable attributes in the model chain is prone to limits of GIS systems but also the knowledge of the user. If characteristics which are now assigned to be not suitable are changed to be suitable, the overall suitable area and therefore the irrigation water demand increases.

From the scenarios for the calculation of water demand, the highest water volume is evaluated with the cultivation of sugarcane on 90% of the field and Faba beans and sorghum cultivated alternatingly on 10% of the field. Currently, sugarcane is the main export crop in Sudan (FAO 2015) as the country is the main producer among Arab and African countries (FAO 2020; Ismail 2006). Sugarcane is a principal crop of the irrigated sector (FAO 2020) as almost all sugarcane in the country is cultivated under irrigation (FAO 2015). The sugar is produced by the Kenana Sugar Company, the White Nile Sugar Company, and the Sudanese Sugar Company, located in the cities of Guneid, New Halfa, Sennar, and Asalaya (FAO 2020). The total harvested area of sugarcane in fully controlled irrigation schemes lies at 70,000 ha in 2019. The FAO also states that sugarcane is well suited to Sudan which shows as Sudan has one of the highest sugarcane yield in the world. Also, the National Agricultural Revival Programme from 2007 to 2012 aimed at improving water control through rehabilitation of the large irrigation schemes, encouraging the development of the agro-industry by establishing several sugar factories, and improving infrastructure (FAO 2015). It shows that there is not only great potential for the cultivation of sugarcane, but that sugarcane also has an economic significance, and an expansion of cultivation is already being sought.

Thus, if the 90/10 distribution scenario is adopted for growing Faba beans, sorghum and sugarcane, the irrigation water demand in this study is calculated at 33.5 BCM/year. However, in 2017 the Nile water withdrawal for irrigation in Sudan was 14 BCM (NBI 2017). This means around 19.5 BCM/year is needed additionally to satisfy the demand for irrigation water in Sudan with the scenario proposed in this study.

Conclusion and outlook

It shows that ArcMap Model Builder is a great tool to set up a model for LSA. The validation process shows that the developed model is suitable for the implementation of an LSA. A large potential for irrigated agriculture in Sudan has been identified, which means that the area available for agriculture can be increased. Especially in the southeast, where large irrigation schemes already exist, very suitable areas are classified. In northern Sudan, not suitable areas accumulate. However, assumptions and simplifications are made, mainly due to technical possibilities, so that the resulting values can deviate upwards or downwards, even considerably. To further enhance the results of this study, additional LSAs with more accurate data, additional input parameters, or even varying approaches for the same study area are necessary.

Based on a defined crop** pattern, the evaluation of irrigation water demand has shown that an additional water volume of approx. 20 BCM/year could be necessary in case of the maximum expansion of irrigated agriculture as estimated in this study. Yet, just as for the LSA, assumptions, and simplifications were made for the calculation. Factors such as defining one crop** pattern or defining one climate station and soil profile for the whole country of Sudan influence the overall accuracy of the final results. In a further development of the modelling process, these inaccuracies should be eliminated to form a more accurate representation of the prevailing conditions and finally receive a more accurate calculation of irrigation water demand. It is recommended for Sudan to collaborate with Egypt and other countries in the basin to identify the possible projects to increase the Nile water flow to meet future demands. Possibilities could include rainfall harvesting projects to make use of part of the 1600 BCM/year of rainfall falling over the Nile basin. One of several examples could be working together with South Sudan to resume the work of the Jonglei canal to increase the water budget of the Nile for the benefit of the three countries.