Artificial Intelligence Integrated GIS for Land Suitability Assessment of Wheat Crop Growth in Arid Zones to Sustain Food Security

Abstract

Developing countries all over the world face numerous difficulties with regard to food security. The purpose of this research is to develop a new approach for evaluating wheat’s suitability for cultivation. To this end, geographical information systems (GIS) and fuzzy inference systems (FIS) are used as the most appropriate artificial intelligence (AI) tools. Outcomes of investigations carried out in the western Nile Delta, Egypt. The fuzzy inference system used was Mamdani type. The membership functions used in this work are sigmoidal, Gaussian, and zmf membership. The inputs in this research are chemical, physical, and fertility soil indices. To predict the final soil suitability using FIS, it is required to implement 81 IF-THEN rules that were written by some experts. The obtained results show the effectiveness of FIS in predicting the wheat crop’s suitability compared to conventional methods. The research region is split into four classes: around 241.3 km² is highly suitable for wheat growth, and 224 km² is defined as having moderate suitability. The third soil suitability class (low), which comprises 252.73 km², is larger than the unsuitable class, which comprises 40 km². The method given here can be easily applied again in an arid region. Decision-makers may benefit from the research’s quantitative findings.

1. Introduction

The demand for agricultural products and food is increasing, and there is a growing need to expand and improve the use of agricultural land to meet this demand, which puts extra strain on natural resources [1,2,3]. Recently, many developed and developing nations have recognized the significance of the issues and changed their agricultural policies to preserve and properly utilize their agricultural areas [4]. Additionally, the impacts of climate change and potential food security adjustments need to be given more consideration in dry and semi-arid regions [5]. Thus, ensuring the sustainability of agricultural production is the primary objective of new policies [6,7]. Comparatively to other land uses, agricultural land use is more demanding in terms of soil performance and quality. Due to the different crop nutrient needs and the physico-chemical characteristics of soils, not all soils can be used for agriculture, and not all crops can be successfully produced under certain soil conditions [8]. Researchers believe that mapping land suitability is crucial for this reason [1,9,10]. After rice and corn, wheat is the field crop with the largest cultivation area (217 million hectares), the highest production (776 million tons), and the highest commerce volume (189 million tons) [11]. To fulfill the demands of the growing global population, the yearly production of 642 million tons of wheat needs to be raised to 840 million tons by the middle of the 21st century [12]. The world’s largest wheat collection at the N.I. Vavilov Institute of Plant Industry (VIR) evaluated winter wheat adaptation to climate change for 50 years of genepool research [13], as the control of wheat diseases using bioagents is not well studied under field conditions [14]. In this regard, land suitability analyses are necessary to guarantee an efficient and sustainable supply of wheat from agricultural lands, which are scarce natural resources [15,16,17]. Parent materials, soil texture, organic matter, slope, and depth are intrinsic soil features that affect land suitability. Similarly, elements that may be influenced by human management, such as drainage, irrigation, soil and water quality, soil fertility, and crop management [1], can also affect land sustainability. In certain wheat-growing regions, excessive nitrogen levels also caused a drop in agronomic productivity [18].

A number of high-quality decision-making tools that execute complex treatments involving numerous variables have been made possible by the new models and analysis techniques [19,20]. Transferring fertile lands to future generations for their food and fiber products requires an accurate assessment of the suitability of the land for multiple uses [21]. The physical and chemical characteristics of the soil are analyzed in terms of their appropriateness for the chosen irrigation systems [22,23]. However, all of these studies fail to take into account how soil qualities vary over time, which makes their evaluations insufficiently accurate [24]. The geometric mean algorism (GMA), which is the n^th root of a sequence of values, has been extensively utilized to evaluate soil quality and crop suitability [25,26,27]. GMA was used to determine the chemical, physical, and fertility indicators, demonstrating its meticulous performance in assessing the suitability for wheat and rice crops [16,28].

The characteristics of soil change over time issue can be solved by using an effective optimization technique such as fuzzy inference system (FIS), which, can be considered one of the most effective artificial intelligence techniques based on expert systems [29] dealing with uncertainty in soil indicators. However, these uncertainties affect the final decision. When planning for the production of a particular crop, the combination of fuzzy algorithms and GIS techniques is an appropriate strategy for assessing the suitability of the land and minimizing the harmful environmental effects of agricultural activities in many previous studies [30,31,32,33,34,35].

In [31], the FAO framework for land evaluation is used to determine suitability. The Analytical Hierarchy Process (AHP) was used for MCE to judge the parameters and compute the priority index for each parameter. To create a land suitability map with unique characteristics, the various thematic layers were overlaid using ArcGIS version 9.2 software. The estimation of agricultural land suitability depends on the inference rules relating land features to suitability classes. The fuzzy inference is built with specified evaluation criteria, such as value ranges for fuzzy linguistic terms and weights of land variables in fuzzy logic modeling for agricultural land evaluation [32]. A GIS-integrated fuzzy rule-based inference system for land suitability evaluation in agricultural watersheds, incorporating both land potential and surface water potential. The terrain, soil physico-chemical properties, soil moisture stress, and feasibility of supplementary irrigation from surface water resources are all taken into account when determining suitability. To make modeling easier, a large number of attributes are used [33]. GIS is regarded as a valuable tool that the agricultural sectors in developing countries must employ because of its interactive and obvious capacity for developing wise decisions that result in effective agriculture management systems [31]. As [34,35] mentioned, the fuzzy inference system model is an effective and robust system for land assessment, with higher accuracy than conventional methods because of its high accuracy. The proposed model was found to be reasonably useful for assessing land suitability.

Many studies on the application of fuzzy systems and GIS in determining optimal cultivation areas in various parts of the world have been conducted, and a number of these studies have used fuzzy systems to evaluate agricultural land suitability [7,24,36,37]. The Nile Delta is one of the most densely populated regions in the world, and it is well known for its agricultural activities [38]. Hence, in order to boost the production of agricultural crops, alternative methods of land evaluation are required.

In this article, the main objective is to evaluate and predict the suitability of the land for growing wheat by means of developing a novel approach based on the fuzzy inference system (FIS), an effective artificial intelligence technique, and GIS in a semi-arid and arid region (West Nile Delta, Egypt). The crop suitability discussed in this article will take into consideration the seventeen soil characteristics. Decision-makers and stakeholders may benefit from the study’s findings as they work to further boost agricultural output in the area and pursue sustainable development goals.

2. Materials and Methods

2.1. Study Area

The Nile Delta in northwest, Egypt is where we decided to conduct our research. the area is 797.00 km², and it is located between the coordinates 30°15′0″ 30°40′0″ E and 31°7′15″ 31°30′45″ N, as indicated in Figure 1. The location is categorized as having a Mediterranean climate [39]. During the dry season, August frequently records a relatively high average maximum temperature of 30 °C. The typical low temperature in January is 13 °C. With a typical annual rainfall of roughly 17.23 mm/year from November to February. Precipitation is frequently light and foggy. Due to the comparatively high temperatures in June and September, evaporation rates are at their maximum. The lowest rates of evaporation are noticed in January and December because of the low temperatures. The most widely produced field crops in the research area are rice, wheat, maize, and alfalfa. The three tree fruits that are most frequently planted in orchards are citrus (orange), guava, and mango [40]. Typic Torrifluvents, Typic Torripsamments, Typic Haplosalids, Vertic Natrargids, and Vertic Torrifluvents are some proposed classifications for these soils [40].

2.2. Collecting Samples and Laboratory Testing

Based on geomorphological field mapping of the study region [41] using the global positioning system (GPS), a total of 15 soil profiles were georeferenced, as shown in Figure 2. According to FAO [42] and USDA [43], morphological descriptions and classifications of soil profiles were conducted, respectively. Soil profiles were excavated down to the water table, or at a depth of 150 cm. Soil profiles vary in depth from 80 to 150 cm. Chemical analyses, such as salinity (EC), soil response (pH), percentage of calcium carbonate (CaCO₃), and percentage of exchangeable sodium (ESP), are conducted. Additionally, the physical properties of the soil, such as particle size distribution, hydraulic conductivity (HC), and water holding capacity (WHC) [44], and fertility, which is measured by the percentage of soil organic matter (SOM%) and the amount of accessible nitrogen (N), phosphorous (P), potassium (K), and zinc (Zn) in the soil, were also carried out [45,46,47,48,49]. The analysis was conducted in the accredited soil, water, and plant laboratory at Tanta University’s Faculty of Agriculture in accordance with ISO/IEC 17025:2017 requirements.

2.3. Soil Properties Mapping with Inverse Distance Weight (IDW)

The term “inverse” denotes that, in comparison to sample points that are far away, sample points that are close have larger weights and have more of an impact on the calculation of missing or unknown points [50]. IDW interpolation is an exact method with a linear combination of data. Because IDW is uncomplicated, simple, and easy to use [51,52,53,54] as well as superior and precise to kriging [55,56,57,58], it is frequently employed in soil research, utilizing the ArcGIS 10.7 IDW tool, which is frequently used to interpolate maps of certain soil attributes [59,60]. Equation (1) illustrates how the local impact reduces as one moves away from the measurement site.

$${\mathrm{z}}_{\mathrm{p}}=\frac{{{\displaystyle \sum }}_{\mathrm{I}=1}^{\mathrm{n}}\left(\frac{{\mathrm{z}}_{\mathrm{i}}}{{\mathrm{d}}_{\mathrm{i}}}\right)}{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}\left(\frac{1}{{\mathrm{d}}_{\mathrm{i}}}\right)}$$

(1)

where ${\mathrm{z}}_{\mathrm{p}}$ is the value anticipated at point P, z_i is the z value at the observed location, i and d_i is the spacing between those two points.

2.4. Crop Suitability Assessment Using FIS

Land suitability analysis is a land evaluation method that assesses the degree to which land is suitable for a certain use [16]. The current study is a quantitative evaluation of land to identify its suitability for wheat cultivation based on its soil properties in the study area. According to [61,62], Table S1, which shows the selection of influencing elements, was based on the wheat crop’s growth requirements. Seventeen parameters have been used in this paper to investigate land suitability for wheat. These are salinity, pH, ESP, CaCO₃, drainage, texture, depth, topography, surface stoniness, hard pan, hydraulic conductivity, water holding capacity, organic matter, N, P, K, and Zn. In determining land suitability, three thematic indicators were used: soil chemical, physical, and fertility suitability indices. The GMA was used to calculate the three suitability indices [16,25,26,27,28] from Equation (2) as follows:

$${\mathrm{Index}}_{\mathrm{x}}=\sqrt[\mathrm{n}]{{\mathrm{S}}_1\times {\mathrm{S}}_2\times {\mathrm{S}}_3\times \dots \times {\mathrm{S}}_{\mathrm{n}}}$$

(2)

where x is the suitability index, S is the score of the parameter, and n is the number of parameters.

The scores of parameters ranged from 0.2 to 1, namely from the worst condition to the best condition based on [61,62], as shown in Table S1.

Each index was split into four groups, with Class 1 denoting a high suitability, Class 2 a moderate suitability, Class 3 a low suitability, and Class 4 a very low suitability. The range of values for each index was divided by the four intervals to obtain the width of each interval. By adding this number to each index’s lowest value and continuing in the same way until the index’s top range was reached, the upper limit of the first interval was generated [25,63].

To achieve the goal of crop suitability (CS) assessment, an efficient, advanced, time-saving, and more accurate method is required. All these features are available in the widely used artificial intelligence technologies of recent times. In this work, the fuzzy inference system (FIS) will be used as one of the most important applications of artificial intelligence, which has proven its effectiveness and efficiency in various fields of agricultural research [7,32,35,36,37,64]. This application depends on several steps, as shown in Figure 3.

2.4.1. Fuzzification

Fuzzification is the process of converting sharp attribute values into the common range of 0 to 1 using the Membership Function (MF) [65,66,67]. The MF in the FIS is a function that assesses a variable’s membership in a given class and determines its degree of truth (membership grade/possibility). A variable’s MF must be defined in terms of a precise quantitative value. Where appropriate, membership is chosen for the function used [65,66,67]. In this paper, sigmoidal, Gaussian, and zmf membership functions are chosen to express the different input degrees depending on their ability to specify this problem. Zmf membership (Equation (3)) is assumed to describe a very low variable; Gauss MF (Equation (4)) is assumed for low and moderate variables; and sigmoidal (Equation (5)) describes a very high variable. The symmetric Z-shaped membership function depends on two parameters, a and b, as given by:

$$\mathrm{f}\left(\mathrm{x};\mathrm{a}, \mathrm{b}\right)=\{\begin{array}{c}1, \mathrm{x}\le \mathrm{a} \\ 1-2{\left(\frac{\mathrm{x}-\mathrm{a}}{\mathrm{b}-\mathrm{a}}\right)}^2, \mathrm{x}\le \frac{\mathrm{a}+\mathrm{b}}{2}\\ 2{\left(\frac{\mathrm{x}-\mathrm{b}}{\mathrm{b}-\mathrm{a}}\right)}^2, \frac{\mathrm{a}+\mathrm{b}}{2}\le \mathrm{x}\le \mathrm{b}\\ 0, \mathrm{x}>\mathrm{b}\end{array}$$

(3)

where f is the Zmf of variable x; a and c are the Zmf membership parameters, which depict the shape of the zmf function.

However, the symmetric Gaussian function depends on two parameters σ and c, as given by:

$$\mathrm{f}\left(\mathrm{x};\mathsf{\sigma }, \mathrm{c}\right)={\mathrm{e}}^{\frac{-{\left(\mathrm{x}-\mathrm{c}\right)}^{\mathrm{m}}}{2{\mathrm{e}}^{\mathrm{x}}}}$$

(4)

where f is the Gaussian MF of a variable x; σ is the standard deviation and c is the mean value, which depict the shape of the Gaussian function. Here c represents center, σ represents width and m represents fuzzification factor and it is equal 2.

Moreover, the sigmoidal function, sigmf (x [a, c]), as given in the following equation by f (x; a, c) is a mapping on a vector x, and depends on two parameters a and c as following:

$$\mathrm{f}\left(\mathrm{x};\mathrm{a}, \mathrm{c}\right)=\frac{1}{1+{\mathrm{e}}^{-\mathrm{a}\left(\mathrm{x}-\mathrm{c}\right)}}$$

(5)

2.4.2. Input Attribute Descriptions for the MF for Site Suitability

Chemical Suitability Ranking

The chemical suitability index (CSI) is an indispensable factor in selecting a suitable site for crop agriculture as it defines soil degradation. The chemical suitability index can be calculated from Equation (6) as follows:

$$\mathrm{CSI}=\sqrt[4]{\mathrm{EC}\times \mathrm{pH}\times \mathrm{ESP}\times {\mathrm{CaCO}}_3}$$

(6)

where CSI = chemical suitability index; EC = soil salinity; pH = soil reaction; ESP = soil exchangeable sodium percentage; and CaCO₃ = proportion of soil calcium carbonate. Four categories—very low, low, moderate, and high—are used in the present research to categorize the suitability of chemical substances, as shown in Table S2. Using Equations (7)–(10) for various chemical suitability index values, it is possible to calculate the chemical suitability index MF of each class. The chemical suitability index is expressed in percentage terms in these equations. If ${\mathsf{\mu }}_{\mathrm{CSI}}\left(\mathrm{VL}\right)$, ${\mathsf{\mu }}_{\mathrm{CSI}}\left(\mathrm{L}\right)$, ${\mathsf{\mu }}_{\mathrm{CSI}}\left(\mathrm{M}\right)$ and ${\mathsf{\mu }}_{\mathrm{CSI}}\left(\mathrm{H}\right)$ are the MFs for the classes very low, low, moderate, and high, then,

$${\mathsf{\mu }}_{\mathrm{CSI}}\left(\mathrm{VL}\right)=\mathrm{f}\left(\mathrm{CSI};{\mathrm{a}}_{\mathrm{CSIVL}},{ \mathrm{b}}_{\mathrm{CSIVL}}\right)$$

(7)

$${\mathsf{\mu }}_{\mathrm{CSI}}\left(\mathrm{L}\right)=\mathrm{f}\left(\mathrm{CSI};{\mathsf{\sigma }}_{\mathrm{CSIL}},{ \mathrm{c}}_{\mathrm{CSIL}}\right)$$

(8)

$${\mathsf{\mu }}_{\mathrm{CSI}}\left(\mathrm{M}\right)=\mathrm{f}\left(\mathrm{CSI};{\mathsf{\sigma }}_{\mathrm{CSIM}},{ \mathrm{c}}_{\mathrm{CSIM}}\right)$$

(9)

$${\mathsf{\mu }}_{\mathrm{CSI}}\left(\mathrm{H}\right)=\mathrm{f}\left(\mathrm{CSI};{\mathrm{a}}_{\mathrm{CSIH}},{ \mathrm{c}}_{\mathrm{CSIH}}\right)$$

(10)

where CSI is the chemical suitability index, and the subscripts (CSIVL, CSIL, CSIM, and CSIH) denote the classes of chemical suitability index for which the MF parameters a, b, and c fall into the very low, low, moderate, and high categories, respectively.

Physical Suitability Ranking

As it defines soil degradation, the physical suitability index (PSI) is also an important component in selecting a suitable location for crop production. As stated in the following Equation (11), the physical suitability index can be derived.

$$\mathrm{PSI}=\sqrt[8]{\mathrm{R}\times \mathrm{T}\times \mathrm{D}\times \mathrm{FF}\times \mathrm{SS}\times \mathrm{HP}\times \mathrm{HC}\times \mathrm{WHC}}$$

(11)

where PSI = physical index; R = drainage; T = texture; D = depth; FF = topography; SS = % surface stoniness; HP = hard pan; HC = hydraulic conductivity (cm/h); and WHC = water holding capacity (%). In this study, physical appropriateness is broken down into four categories: very low, low, moderate, and high, as shown in Table S3. Equations (12)–(15) for various chemical suitability index indices can be used to calculate the MFs for each class of physical suitability index. In these equations, the physical suitability index is expressed as a percentage. If ${\mathit{\mu }}_{\mathit{PSI}}\left(\mathit{VL}\right)$, ${\mathit{\mu }}_{\mathit{PSI}}\left(\mathit{L}\right)$, ${\mathit{\mu }}_{\mathit{PSI}}\left(\mathit{M}\right)$, and ${\mathit{\mu }}_{\mathit{PSI}}\left(\mathit{H}\right)$ are the MF for the classes very low, low, moderate, and high, then,

$${\mu }_{\mathit{PSI}}\left(\mathit{VL}\right)=f\left(\mathit{PSI};a_{\mathit{PSIVL}},{ b}_{\mathit{PSIVL}}\right)$$

(12)

$${\mu }_{\mathit{PSI}}\left(L\right)=f\left(\mathit{PSI};{\sigma }_{\mathit{PSIL}},{ c}_{\mathit{PSIL}}\right)$$

(13)

$${\mu }_{\mathit{PSI}}\left(M\right)=f\left(\mathit{PSI};{\sigma }_{\mathit{PSIM}},{ c}_{\mathit{PSIM}}\right)$$

(14)

$${\mu }_{\mathit{PSI}}\left(H\right)=f\left(\mathit{PSI};a_{\mathit{PSIH}},{ c}_{\mathit{PSIH}}\right)$$

(15)

PSI is the physical suitability index, and the subscripts (PSIVL, PSIL, PSIM, and PSIH) indicate that the MF parameters a, b, and c belong to the very low, low, moderate, and high classes of the physical suitability index, respectively.

Fertility Suitability Ranking

Because soil fertility loss and nutrient depletion are the main causes of low production, it is obligatory to reduce loss and improve usage efficiency in order to achieve sustainable development. The fertility suitability index (FSI) was calculated using the following Equation (16):

$$\mathrm{FSI}=\sqrt[5]{\mathrm{N}\times \mathrm{P}\times \mathrm{K}\times \mathrm{OM}\times \mathrm{Zn}}$$

(16)

where FSI = fertility suitability index; N, P, and K = available nitrogen, phosphorus, potassium, respectively; OM = organic matter (%); and Zn = available zinc. The appropriateness of fertility is divided into four groups in the present investigation: very low, low, moderate, and high, as shown in Table S4. The MF for each class of fertility appropriateness index can be calculated using Equations (17)–(20) for different fertility suitability indexes. The fertility appropriateness index is represented as a percentage in these equations. If the MF for the classes very low, low, moderate, and high is ${\mathit{\mu }}_{\mathit{FSI}}\left(\mathit{VL}\right)$, ${\mathit{\mu }}_{\mathit{FSI}}\left(\mathit{L}\right)$, ${\mathit{\mu }}_{\mathit{FSI}}\left(\mathit{M}\right)$, and ${\mathit{\mu }}_{\mathit{FSI}}\left(\mathit{H}\right)$, respectively, then

$${\mu }_{\mathit{FSI}}\left(\mathit{VL}\right)=f\left(\mathit{FSI};a_{\mathit{FSIVL}},{ b}_{\mathit{FSIVL}}\right)$$

(17)

$${\mu }_{\mathit{FSI}}\left(L\right)=f\left(\mathit{FSI};{\sigma }_{\mathit{FSIL}},{ c}_{\mathit{FSIL}}\right)$$

(18)

$${\mu }_{\mathit{FSI}}\left(M\right)=f\left(\mathit{FSI};{\sigma }_{\mathit{FSIM}},{ c}_{\mathit{FSIM}}\right)$$

(19)

$${\mu }_{\mathit{FSI}}\left(H\right)=f\left(\mathit{FSI};a_{\mathit{FSIH}},{ c}_{\mathit{FSIH}}\right)$$

(20)

The subscripts (FSIVL, FSIL, FSIM, and FSIH) indicate that the MF parameters a, b, and c belong to the very low, low, moderate, and high classes of the fertility suitability index.

Final Crop Suitability Ranking

The final crop suitability index (FCSI) was described according to Equation (21):

$$\mathrm{FCSI}=\sqrt[3]{\mathrm{CSI}\times \mathrm{PSI}\times \mathrm{FSI}}$$

(21)

where FCSI = final crop suitability index; CSI = chemical suitability index; PSI = physical suitability index; and FSI = fertility index. In the current study, the suitability of the final crop is classified as unsuitable, low, moderate, and high, as shown in Table S5. Equations (22)–(25) for different final crop suitability indexes can be used to calculate the MF for each class of final crop suitability appropriateness index. In these equations, the final crop suitability and appropriateness index are represented as a percentage. If the MF is ${\mathsf{\mu }}_{\mathrm{FCSI}}$(US), ${\mathsf{\mu }}_{\mathrm{FCSI}}$(L), ${\mathsf{\mu }}_{\mathrm{FCSI}}$(M), and ${\mathsf{\mu }}_{\mathrm{FCSI}}$(H) for the classes unsuitable, low, moderate, and high, respectively, then

$${\mathsf{\mu }}_{\mathrm{FCSI}}\left(\mathrm{US}\right)=\mathrm{f}\left(\mathrm{FCSI};{\mathrm{a}}_{\mathrm{FCSIUS}},{ \mathrm{b}}_{\mathrm{CSIUS}}\right)$$

(22)

$${\mathsf{\mu }}_{\mathrm{FCSI}}\left(\mathrm{L}\right)=\mathrm{f}\left(\mathrm{FCSI};{\mathsf{\sigma }}_{\mathrm{FCSIL}},{ \mathrm{c}}_{\mathrm{FCSIL}}\right)$$

(23)

$${\mathsf{\mu }}_{\mathrm{FCSI}}\left(\mathrm{M}\right)=\mathrm{f}\left(\mathrm{FCSI};{\mathsf{\sigma }}_{\mathrm{FCSIM}},{ \mathrm{c}}_{\mathrm{FCSIM}}\right)$$

(24)

$${\mathsf{\mu }}_{\mathrm{FCSI}}\left(\mathrm{H}\right)=\mathrm{f}\left(\mathrm{FCSI};{\mathrm{a}}_{\mathrm{FCSIH}},{ \mathrm{c}}_{\mathrm{FCSIH}}\right)$$

(25)

The subscripts (FCSIUS, FCSIL, FCSIM, and FCSIH) indicate that the MF parameters a, b, and c belong to the unsuitable, low, moderate, and high classes of the final crop suitability index.

2.4.3. MF Parameters for the Input Variables

The triangle curve’s structure is defined by the MF parameters, which have a minimum value of 0 and a maximum value of 1. This method depends on the availability of data and the desired application, and the MF parameters are often defined by expert knowledge and/or developed using measured data. Numerous researchers have employed optimization methods to determine the ideal set of MF parameters for a certain FIS application [68,69,70]. However, because there is not enough data to compare the results of the FIS output to the observed data, the MF parameters in this paper are determined based on a review of the literature and the knowledge of some experts in the field. The MF parameters of fuzzy sets translate the input variable into a number of overlapping fuzzy regions, as opposed to conventional sets with sharp borders [33,65,71]. The advantage of fuzzy sets is that the elements in the set become partial because the fuzzy regions overlap, and the transition from one region to another is gradual. The MFs of the different are as shown in (Figure 4).

As shown in Figure 4a, many MF classes overlap, and an individual element can belong to multiple classes. As shown in Figure 4, the MFs of the input variables overlap. As such, a value of 8% in the chemical suitability index falls in both the low and medium classes with varying degrees of belongingness (Figure 4a). For the low, medium, and high classes, the membership values for the 8% slope would be 0.2, 0.8, and 0, respectively. It should be noted that a given element’s various classes must all have membership values that add up to one. In Figure 4b, the MFs of physical and fertility are depicted similarly in Figure 4c.

Table 1. Fuzzy membership function parameters.

	Very Low		Low		Moderate		High
	a	b	c	$\mathrm{\Sigma }$	c	$\mathrm{\sigma }$	a	c
CSI	0.2	0.38	0.38	0.066	0.53	0.061	0.65	0.57
PSI	0.53	0.63	0.64	0.054	0.77	0.05	0.78	0.88
FSI	0.167	0.29	0.32	0.078	0.51	0.073	0.52	0.59

lists the membership function limits to input the three variables according to their types and as decided by wheat growers.

2.4.4. Creation of the Crop Suitability Fuzzy Rule Base

The studies [72,73,74,75], in addition to the consulted experts, served as the foundation for the fuzzy rule established in this work. Because each of the three input types is divided into four categories using the fuzzy outputs of the FIS, as was already mentioned, there are 81 IF-THEN rules in the FIS. All levels of suitability—unsuitable, low, moderate, and high—are acceptable. A common illustration of the FIS rules is:

If the CSI is low, the PSI is low, and the FSI is high, check the compatibility of the soil.

In this research, all three variables are given equal weighting in the FIS rules. Each rule base regulation takes a group of input variables from the fuzzy set and then provides the suitable score for the soil in the suitability class for a particular crop. Some of the rules-based expert system for the crop suitability score will be demonstrated in Figure 5, and the rest of the 81 rules will be in the same manner. As shown in Figure 5, the rules will be as follows:

1.

If (CSI is very low) and (PSI is very low) and (FSI is moderate), then (FCSI is low), as shown in orange lines.

2.

If (CSI is low), (PSI is moderate), and (FSI is very low), then (FCSI is unsuitable), as shown in the green lines.

3.

If (CSI is high) and (PSI is very low) and (FSI is high), then (FCSI is moderate), as shown in the violet lines.

4.

If (CSI is high) and (PSI is low) and (FSI is low), then (FCSI is high), as shown in the blue lines.

2.4.5. Aggregation of Rules

For assessing the overall suitability of a site for a specific crop, the output from all the fuzzy rules is combined. The method of obtaining the desired output from the rule base is referred to as rule aggregation. This study employs the Mamdani Implication approach, a maximum-minimum aggregation method [33,65,76]. This method selects a criterion’s MF property based on its lowest value. The outputs from each rule’s set are selected for their maximum values, and the corresponding outputs from each rule are then combined using the fuzzy union operator, as shown in Equations (26) and (27) below [33,65]:

$$\mathrm{Soil} \mathrm{suitabilty} \left(\mathsf{\mu }\right)= \max \left(\min \left({\mathsf{\mu }}_{\mathrm{c}}^{\mathrm{k}}\left(\mathrm{variable} 1\right),{\mathsf{\mu }}_{\mathrm{c}}^{\mathrm{k}}\left(\mathrm{variable} 2\right),{\mathsf{\mu }}_{\mathrm{c}}^{\mathrm{k}}\left(\mathrm{variable} 3\right)\right)\right)$$

(26)

$$\mathrm{k}=1, 2, \dots , 81 \left(\mathrm{n}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r} \mathrm{o}\mathrm{f} \mathrm{r}\mathrm{u}\mathrm{l}\mathrm{e}\mathrm{s}\right) \mathrm{a}\mathrm{n}\mathrm{d} \mathrm{c}=1, 2, 3)$$

(27)

where soil suitability (μ) is the membership grade of the suitability, ${\mathsf{\mu }}_{\mathrm{c}}^{\mathrm{k}}$ is the membership level of the input variable in rule k, and variable 1. Moreover, variables 1, 2, and 3 are the input variables to the FIS. Figure 2 shows a schematic representation of the developed FIS. The inference engine considers the fuzzified input variables and evaluates all fuzzy rules; the resulting fuzzed-up output is then defuzzified to determine the suitability class.

2.4.6. Defuzzification of the FIS Output

In the Mamdani method of rule aggregation, the output of the FIS is a fuzzy variable that needs to be defuzzified in order to be used in the decision domain. There are several defuzzifications, including these:

• The centroid technique [65];
• The principle, mean maximum membership [65];
• The weighted average method [33];
• The use of inflection points [77].

The most frequently used technique is maximum membership. In this study, the maximum membership principle was applied during defuzzification. The maximum of the largest membership value of the suitability class serves as the most representative value in this method.

Based on the chemical, physical, and fertility characteristics of the soil, FIS is used in this study to determine the best locations for growing wheat crops, which aids in increasing crop yield. The Fuzzy Logic Toolbox and the MATLAB module were used to further this research, and all aspects of the study’s model were subjected to this process. Figure 6 depicts the crop suitability score’s output membership as it is presented using FIS.

3. Results and Discussion

3.1. Soil Characteristics in the Study Area

The study area’s soil characteristics are listed in

Table 2. Statistics of selected soil properties (n = 61).

	EC	pH	ESP	CaCO3	Depth	WHC	HC	AN	AP	AK	AZn	OM
Statistical Param.	dS m−1	1:2.5	%	g kg−1	cm	%	cm h−1	mg kg−1				g kg−1
Min.	0.64	8.08	3.22	7.50	80.00	5.47	0.29	7.50	6.30	9.30	0.20	2.40
Max.	19.64	8.86	24.94	90.40	150.00	50.63	14.56	81.00	22.30	457.10	1.50	12.20
Mean	5.45	8.49	11.27	39.04	128.67	36.78	4.39	48.34	14.97	277.00	1.16	8.16
St. Dev.	5.31	0.25	6.06	20.26	26.42	18.41	4.87	24.53	5.10	173.91	0.52	3.26
Skewness	1.80	−0.19	0.66	0.85	−0.80	−1.12	1.51	−0.53	−0.23	−0.81	−1.10	−0.76
Kurtosis	3.14	−1.05	0.17	2.09	−0.76	−0.70	0.74	−1.06	−1.28	−1.10	−0.55	−0.43

Min—Minimum; Max—Maximum; SD—standard deviation; EC—electrical conductivity; pH—soil reaction; ESP—exchangeable sodium percentage; CaCO₃—calcium carbonate percentage; WHC—water holding capacity; HC—hydraulic conductivity; AN—available nitrogen; AP—available phosphorous; AK—available potassium; AZn—available zinc; and OM—organic matter.

and interpolated in Figure S1. There are differences in soil texture between clay, silt clay, silt clay loam, sand, and sandy loam. Hydraulic conductivity (HC), which ranged from 0.29 to 14.56 (cm h⁻¹), expresses water flow and pore structure in soil [78]. Hydraulic conductivity is a crucial indicator of soil pore structure and water movement [27]. Moreover, the water holding capacity (WHC) of soil varies widely from 5.47 to 50.63%. The depth of the soil was between 80 and 150 cm. Both non-saline and high-salinity soils may be found in this region, where the EC values range from 0.64 to 19.64 dS m⁻¹, with an average value of 5.45 ± 5.31 dS m⁻¹. Most salinized soils are found in drylands because of the arid climate and high evaporation rates [79]. This fits the overall pattern of the northern delta, where excessive soil salinity characterizes the majority of the soil [80,81,82,83,84]. Leaching high-salinity soils requires high-quality water [85,86]. The soil pH in the research region ranges from 8.08 to 8.86, making the environment alkaline to strongly alkaline. Areas in the northeast and southeast of the study area had the highest pH values (Figure S1b). There is a significant degree of pH similarity among the various units in the research area, as indicated by the standard deviation (SD) of pH = 0.25 [87]. Physical, chemical, and biological aspects are known to be influenced by soil pH [88,89,90]. Calcium carbonate (CaCO₃) has a range of 7.5 to 90.4 g Kg⁻¹. Shell particles could be the cause of the highest CaCO₃ value [91]. The data show that the CaCO3 content varies greatly within the study area (SD = 20.26). CaCO₃ levels were the highest in the middle and southwest of the investigated area (Figure S1d). The formation of very hard layers that are impermeable to water and crop roots, as well as phosphorus fixation fertilizer in calcareous soils, can result in areas with the highest CaCO₃ values [92]. The range of ESP values is 3.22% to 24.94%, with an average of 11.27 ± 6.06%, indicating sodicity hazards in the area [93]. The upper northwest of the research area has seen an increase in the spatial patterns of EC and ESP (Figure S1a,c). Clay and SOM content are related to the CEC, which ranges from 5.82 to 42.24 cmol kg⁻¹, with an average of 29.80 cmol kg⁻¹ [94]. The research area’s SOM concentration varied from 2.04 to 12.2 g Kg⁻¹. A slight SOM concentration was noted in the research area’s soils, despite the fact that SOM is crucial for enhancing the physical and chemical characteristics of soil [95,96]. The available N ranges from 7.5 to 81 mg N kg⁻¹, demonstrating that the nitrogen content in the study area differs from low to moderate [28]. According to [88], the research area’s available P and K content is categorized as moderate since the average values are 14.97 mg P Kg⁻¹ and 277 mg K Kg⁻¹, respectively.

Figure 7 demonstrates how a type-2 Mamdani system makes inferences. In this case, the fuzzified inputs generate firing strengths for both the upper and lower membership functions. If CSI = 0.3, which means very low, PSI = 0.6, which means low class, and FSI = 0.9, which indicates high class, then the final crop suitability index (FCSI) will be 0.569, which means moderate suitability for wheat crop cultivation. In the same manner, the wheat soil suitability for any value of CSI, PSI, and FSI of any soil can be obtained.

The resulting FIS model shows three-dimensional surfaces. For each possible combination of the two input variables, the control surface document displays the FIS output value. The influence of specific processing parameters, such as CSI, PSI, and FSI, in relation to the final crop suitability index (FCSI) for wheat cultivation can be identified from the 3D charts shown in Figure 8.

For dispersion analysis, the three indicators—chemical, physical, and fertility—have a clear effect on the final crop suitability index for wheat crops. Therefore, the three-dimensional figure was drawn between the physical and chemical in Figure 8a to show the most suitable places without explaining the effect of fertility, as well as between both fertility and physical in Figure 8b to show the most suitable places with an explanation of the effect of chemical properties. Figure 8c shows the suitability of soil when taking fertility properties and chemical properties into account, with an explanation of the physical properties effect. These graphs show the variance in crop suitability for growing the wheat crop depending on the CSI, PSI, and FSI.

The descriptive analysis defines the most suitable places for wheat cultivation in addition to determining them on the basis of soil chemical, physical, and fertility properties, which focus on the suitability of the soil for growing the crop. With regard to performance, soil fertility clearly affects the suitability of the soil, as 50% of the respondents affirmed that the higher the suitability of the soil, the higher the soil fertility.

3.2. Chemical Suitability Index (CSI)

The surrounding factors that affect the chemical balance in the soil, which is later reflected in the soil’s fertile state, include soil salinity, temperature, evapotranspiration, soil moisture, and other factors. These factors all alter the chemical properties of the soil [97,98,99,100]. The chemical index spatial distribution map (Figure 9) demonstrates the wide range of chemical quality, from low chemical suitability (CSI 3) to high chemical suitability (CSI 1). The following list includes the area’s chemical index: The chemical suitability of areas of 238.61 km² is high, 515.46 Km² is moderate, and 4.09 km² is low, as shown in

Table 3. Areas and classes of CSI.

Classes	Area (km2)	Area (%)
High chemical suitability (CSI 1)	238.61	31.47
Moderate chemical suitability (CSI 2)	515.46	67.99
Low chemical suitability (CSI 3)	4.09	0.54

. In some areas of the research area’s north-east, the CSI 3 class is distinguished by high ECe, ESP, and pH values, which may be promoting chemical degradation [101]. These findings are consistent with those of [28] investigations into Egypt’s north Nile Delta.

3.3. Physical Suitability Index (PSI)

According to the concept of soil conservation against potential degradation, the physical attributes of the soil depend on soil tillage conditions, organic additions, land use, fertilization, and irrigation [97,100,102,103,104]. Data from

Table 4. Areas and classes of PSI.

Classes	Area (km2)	Area (%)
High physical suitability (PSI 1)	465.34	61.38
Moderate physical suitability (PSI 2)	15.34	2.02
Very low physical suitability (PSI 4)	277.48	36.60

and Figure 10 show that PSI in the research area ranges from high physical index (PSI 1) to unsuitable (PSI 4). Due to deep soil profiles, flat surfaces, medium textures, and low gravel content, the soil physical index rating indicates that 61% of the study area has high physical suitability soil, while the remaining 2.02 and 36.6% of all agricultural areas are, respectively, classified as having moderate physical suitability (PSI 2) and very low (PSI 4) classes.

3.4. Fertility Suitability Index (FSI)

Numerous research domains, from the sustainability of soil management to the idea of precision farming, have considered soil fertility mapping a crucial concern [80]. According to the FSI, the study area fell into three classes: high (FSI 1), low (FSI 3), and very low (FSI 4), with FSI 1 being the largest at 456.34 km² (Figure 11 and

Table 5. Areas and classes of FSI.

Classes	Area (km2)	Area (%)
High fertility suitability (FSI 1)	465.34	61.38
Low fertility suitability (FSI 3)	160.72	21.20
Very low fertility suitability (FSI 4)	132.10	17.42

). Given that land degradation factors are active in some locations, which decrease nutrient availability on the one hand and increase carbon release to the atmosphere on the other, the low fertility situation may be produced by agricultural practices that have had various effects that lower soil fertility [97,98,99,100].

3.5. Land Suitability Based on FIS

The most important soil quality indicators are soil physical, chemical, and biological quality measurements [27]. The highest possible value for these variables boosts agricultural output and extends the sustainability of management systems [105]. The study region is divided into four classes based on Figure 12 and

Table 6. Areas of final crop suitability index (FCSI).

Class	Area (km2)	Area (%)
High Suitable (FCSI 1)	241.30	31.83
Moderate Suitable (FCSI 2)	224.04	29.55
Low Suitable (FCSI 3)	252.73	33.33
Unsuitable (FCSI 4)	40.09	5.29

, the land suitability index (LSI). Over 241.3 km² (31.83%) of the total research area are in the first class, which is distinguished by its high suitability for wheat cultivation. With 224 km² (29.55%) of the overall research area, the second class is distinguished by moderate suitability. The third suitability class of soil (low) takes up 252.73 km² (33.33%) of the entire study area, while the unsuitable class, which takes up 40 km² (5.29%) of the entire study area, is the lowest representative class. Low levels of OM%, CEC, N, P, and K had a detrimental impact on the suitability of the land in addition to its physical characteristics. For example, coarse texture affected the organization of the particles and pores, which had an impact on root growth, the rate of plant emergence, and the agricultural practices of water infiltration [78].

4. Conclusions

In this paper, the suitability of wheat crops was evaluated using fuzzy inference, one of the most useful AI techniques and a widespread one in this field. A fuzzy system also has the potential to represent and control uncertain or incomplete agricultural knowledge. The fuzzy inference system used in this research is of the Mamdani type. The most suitable lands for wheat cultivation were classified in this study based on the soil’s chemical, physical, and fertility properties. The accuracy of the results of the FIS program depends primarily on the selection of the appropriate membership, where sigmoidal, Gaussian, and zmf were selected to express the different scores of the inputs corresponding to their ability to describe this problem, and secondly, on the number of IF-THEN rules, which depend mainly on the expertise of experts in the field. The results proved the accuracy, effectiveness, and speed of the proposed program, which integrates FIS and GIS to determine the suitability of the soil for the cultivation of wheat compared to traditional methods. In addition to the physical qualities of the land, low levels of OM%, N, P, and K had a negative effect on wheat’s suitability for cultivation.

The proposed crop suitability evaluation program, with its accurate results in this study, will help:

• Decision-makers in obtaining useful information about the primary main limiting factors as noticed from the observed crop suitability;
• Determining the necessary improvements that will be required to achieve agricultural sustainability;
• Integration of the use of FIS with GIS for mapping soil capacity and crop suitability is critical for optimal land use and food security in arid regions such as Egypt;
• Generalization of the proposed technique for other crops.

In conclusion, it is very important to assess land suitability periodically to try to maintain a high crop yield for the purpose of reducing the gap between production and consumption, and it is suggested that field work and approaches to crop suitability calculation be increased in future studies. As a whole, the proposed crop suitability assessment program is reapplied whenever the soil’s physical, chemical, and fertility characteristics are available and influence other study areas.