Using interpretative structure model and analytical network process for optimum site selection of airport locations in Delta Egypt

1 Introduction

Airports are the main pillars of the economy of any country because they are very important and prominent. It is an important infrastructure project and contributes to communications [1]. It plays an important role as a catalyst for local growth and in effective plans in this field and can greatly affect and enhance the local logistics services on the social and economic levels. Infrastructure projects in Egypt are currently progressing rapidly in all fields and in all regions, especially the Nile Delta. The establishment of new international airports has now become an important destination for decision makers in Egypt.

Egypt is located in the north of the African continent, surrounded by the Mediterranean Sea in its north, with a coastline of 995 km. Egypt is considered the main hub of communication and constantly welcomes visitors and investments between European and African countries [2]. Due to tourism, production, service industries, and other fields, those who do not have access to airport services will benefit from developing a new airport in the chosen research area (Nile Delta, Egypt). This article focuses on determining the optimum locations for constructing new international airports, which are essential to be chosen properly. The residents of the study area face difficulty in accessing the nearby airport services; the residents of the selected study area will also benefit from the construction of a new international airport [3]. These changes can simultaneously enhance regional well-being and prosperity through a stable transportation system and a suitable location to address the problem of geographic integration of airport location selection using multi-criteria decision-making (MCDM) deepening on remote sensing and geographic information system (GIS) technology.

GIS technology allows for the storage, organization, and analysis of all forms of geographic data required to establish an international airport. The article combines two models, which are the interpretative structure model (ISM) and analytical network process (ANP) together ISM–ANP methods. The results are compared with traditional methods in site selection such as analytic hierarchy process (AHP) and weighted overly tool (WOA). ISM is used to determine the relationships between the criteria and standards, but ANP is a way to derive the weights of each standard, but the WOA method implements several stages in one modular tool.

One of the best-paired comparisons for determining weights for choice criteria is the AHP approach, which is used to derive ratio measures and is employed in many different sectors of decision-making procedures. Then, the resulted weights were used for ArcGIS input layers (standard maps). The techniques of the ANP, the AHP, and the overweight method are some examples of MCDM used to calculate the weights of criteria [4,5,6].

Beshr et al. used fuzzy-AHP to detect the optimum site selection for isolation hospitals for coronavirus patients in the Nile Delta, Egypt [7]. Mallick et al. [8] used GIS and ANP for wind farm site selection in East Azerbaijan Province, Iran. Exposure of the Anambra state to floods was modeled using the Interval value fuzzy rough number–decision making trial and evaluation laboratory–ANP model in a GIS environment. The location of the general bread factory in Turkey was chosen in an unknown area and their results showed the efficiency of the method and its application to the bread sector [8]. Using the AHP method and application of GIS to landslides in the watershed of Abha, Saudi Arabia, has been investigated. The results showed the feasibility of the strategy to understand how to preserve watersheds in terms of landslides, according to El Jazouli et al. [9].

AHP by Balusa and Gorai [10] helped to create a digital map with important details regarding current and upcoming landslides to enable them to occur. The results showed that the methodology succeeded in selecting the mining method using the GIS-based AHP model and was compared by Balusa and Gorai [10]. Parizi et al. [11] used the two methods of receiver operating characteristic curve (ROC) and AHP to find the best location for the construction of the airport in Libya. According to Erkan and Elsharida [12], a GIS-based approach to the selection of rainwater harvesting sites. Sayl et al. [13] chose the optimal location for groundwater wells by integrating GIS and water geophysical data. Integration of maps of the groundwater potential area using GIS and WOA of northeastern Iraq [14,15]. Multi-criteria GIS resolution analysis by Feizizadeh and Blaschke [16] for landslide sensitivity mapping: comparison of three methods for the Urmia Lake Basin, Iran. Chen [17] has used ISM and artificial neural networks techniques in developing the user expected command location predicting approach.

To help decision-makers identify the most suitable areas for new construction projects, numerous researchers have employed a wide range of techniques to identify the best, or even optimal, locations for particular projects [18,20,21,22,23,24,25,26].

Therefore, this article aims to achieve the following goals:

1. The possibility of using the ISM–ANP method as a new technique in selecting the airports location in the Nile Delta, Egypt;

2. Comparing the results of site selection using the proposed technique with traditional site selection techniques; and

3. Evaluating the ISM–ANP method by applying an accuracy test using the ROC and the area under the curve (AUC).

2 Study area (Nile Delta, Egypt) and airport services

The study area for this study is Nile Delta, Egypt, which is one of the nation’s densely populated regions. The Delta region is situated between the longitudes 30°20′ 0′′E and 32° 0′ 0′′ and the two latitudes 31°30′ 0′′ N and 30°30′ 0′′ N. The Mediterranean Sea, the Suez Canal, Cairo, and Alexandria form this region’s northern, eastern, southern, and western borders, respectively. According to Figure 1, the Nile Delta lies in the Arab Republic of Egypt’s northern region. There are five governorates in Egypt’s Nile Delta: Dakahlia, Kafr El-Sheikh, Damietta, Gharbia, and Menoufia. These governorates are situated between the two branches of the Nile River (Damietta and Rashid). Some governorates, like Dakahlia and Menoufia, may have centers that extend beyond these two branches (Figure 1). Together, these five provinces account for 16735.61 km² or 1.22% of Egypt’s total land area. Table 1 displays the population, area, and density of Nile Delta governorates. The study area has a total population of 22,223,211, they account for 22.1% of Egypt’s total population (Source: Central Agency for Public Mobilization and Statistics, Egypt 2022). Agriculture, fishing, and the furniture business are the main employers in this community. One of the most significant agricultural and industrial areas is the study area the closest to the Mediterranean coast is considered the main location, so the area must be supported by an air transport network to serve the citizens of Delta and to promote international trade. The total number of Egyptian airports is 19, including 12 international airports with a total capacity of 29,020,100 people/hour and 7 airports National with a total capacity of 1,980 people/h (Source: Egyptian Ministry of Civil Aviation, 2022) (Figure 2).

Figure 1

Study area map (Nile Delta, Egypt).

Figure 2

Types and distribution of the existing airports in Egypt.

3 Methodologies and applied analysis models

In this section, the new approach method is presented and described as follows.

4 Research methodology and procedure steps

Figure 3 depicts a rough summary and methodology of the research, explaining the techniques used to determine the optimum locations for constructing new international airports in the study area and the stages involved in each method.

Figure 3

Flow chart of the research methodology.

5 Selection of criteria for international airport site selection

The selected applied criteria were chosen depending on the Federal Aviation Administration, the International Civil Aviation Organization, the literature review of prior studies, expert opinion through actual field visits to existing airports and asking their specialists, and the availability of the data including maps, documents, etc.

The five primary categories of choice criteria to select new international airport locations are environmental concerns, topographical characteristics, climatic aspects, infrastructure facilities, and operating requirements. There are sub-criteria for each of the primary criteria, and each sub-criterion is also categorized. All of these criteria and sub-criteria have been collected into a comprehensive set of elements (parameters) that are summarized in Table 2. Thus, there are a total of 16 applicable criteria in this study, as shown in Table 2.

6 Data collection for the research

The input data for this study came from a wide range of sources: Landsat images, topographic maps, and official, and governmental data required for research. All of the images were created by digitizing, scanning, and geocoding the pertinent data after they were gathered at different resolutions and sizes. In addition, GIS methods such as intersection, union, buffering, interpolation, map algebra, and overlay were used to construct the topographic maps.

The studied area (Nile Delta, Egypt) was categorized into five soil types (sand, loam, clay, clay loam, and sand loam) based on geological governmental data and soil stratification. Soil type, elevation above sea level, and other criteria outlined in Tables 2 and 3 were considered to ensure alignment with international standards. Upon examining the study area, the geological composition was found to have remained largely unchanged for over 20 years, depending on the official reports.

Because the input data utilized by different organizations are typically created and compiled for applications, they have distinct formats, scales, and projection systems. All of those data (United States Geological Survey) were georeferenced using the Transverse Mercator projection system within the GIS environment. Syntaxinthe final needed layers were then obtained through a series of GIS steps (including extract, proximity, buffer, overlay, convert, and clip) and, ultimately, rasterizing those vector maps (shape files).

For infrastructure data, the Esri firm created the ArcGIS Editor OSM tool, which makes the possibility to download directly to the program, starting with version 10. It is straightforward to download shape files of infrastructure for the necessary study zone using this program. Shape files can be downloaded directly from the website as well. Preparing the layers is done as follows.

The meteorological information, which included temperature, wind, humidity, air pressure, precipitation, and clearness index, was obtained from the NASA Imagery Satellite at (power.larc.nasa.gov):

1. A point grid was created for every study area (22 points).

2. The necessary data were obtained as an Excel file from the NASA satellite.

3. The required data’s average was determined for each point in the grid and then entered.

4. Apply the command from the Arc toolbox (3D analyst tools – Raster interpolation – Kriging) while pointing the GIS program at a grid point.

5. Creation of a raster layer and land cover layer through:

   1. The source from which the layer of land cover for Egypt was downloaded.

   2. Using the study zone’s mask layer and the extract-clip from Arc Toolbox function.

   3. Using Egyptian datum to geocode the layer of land cover.

7 Reclassifying and redistributing the criteria and sub-criteria

Every criterion layer has a different unit from the others. Therefore, in order to implement a weighted overlay process, they must typically be in the same units. As a result, standardization is necessary to make the dimension units uniform, but in doing so, the scores typically lose both their dimensions and their measurement unit [24]. To create the appropriateness index map, all input layers were transformed into raster layers and categorized before being added to the weighted overlay (Figures 4–7). Assigning rating values from 1 to 9 (from the least to the most) based on the data from the literature research, the opinions of experts, and specifics on the safe distances and buffering zones to an airport site define the reclassifying task suitableness, as shown in Table 4. Note that, in Figures 4–7, the number 9 indicates the highest effect. The lower the number indicates the gradation of the effect from the highest, which is 9, to the lowest, which is 1.

Figure 4

Reclassification of the topographical map.

Figure 5

Reclassification of the infrastructure map.

Figure 6

Reclassification of operational and environment condition map.

Figure 7

Reclassification of the climatic factor map.

8 Implementation and outcomes of ISM

Finding the problem-related variables is the first step in this method of structuring the problem indicators. The contextual links between the indicators are then established utilizing specialists’ experience, theoretical knowledge, and practical expertise.

The provision of a multi-level structural model comes last. Fifteen university professors who are specialists in crisis management, urban planning, and seismic engineering collaborated in this study and answered a questionnaire on this method. The following were the steps in the implementation: formation of the SSIM.

Step 1: The experts received the pertinent questionnaire in the form of a square matrix with the same rows and columns as the urban physical indicators. The final structural self-interaction matrix is obtained after compiling all the finished opinions and matrices according to the outcomes of paired comparisons. Therefore, a relationship between the two components is considered if it is supported by the majority of the expert group (N/2 + 1 votes); otherwise, there is no relationship between the two components. This matrix is shown in Table 5.

Step 2: Creation of the IRM

The SSIM and the symbol substitution method are used to create the IRM.

Step 3: Creation of the FRM

The equation is then used to determine the consistency of the initial reachability matrix and produce the final reachability matrix (1).

Step 4: Grouping and ordering indicators according to the importance

This stage uses the final reachability matrix to determine two sets of reachability set and antecedent set. The level partitioning (LP) of the indicators used in the current study is shown in Table 6.

Step 5: Model assessment and the creation of the interpretive structure model digraph

Using the relationships taken from the final reachability matrix and the preceding stage, a group of seven experts assessed the model. Subsequently, the ultimate model is created based on the outcomes of the fourth phase and the model assessment findings. In the ISM model, an indicator will be at a lower level if it has an impact on others. On the other hand, an indicator will be of higher quality if it is strongly influenced by other indications. The ISM model displays an effect from bottom to top. There are five levels in the final model this study created (Figure 8). This model shows the most significant relationships. The directions of the arrows indicate how each factor affects the others.

Figure 8

The diamgraph of ISM-based model.

9 Implementation of ANP

Determining the importance of each urban physical indicator using this methodology is done.

Physical toughness: This model’s calculations were carried out using Super Decision and MATLAB software. The model’s phases for implementation are as follows:

Step 1: The network diagram and creation of the ANP model

Three clusters of urban physical indicators are identified in this study.

Inside each cluster, there are multiple linked signs. The ISM model produces clusters (Figure 9).

Figure 9

The procedure used to calculate the weight vectors.

Step 2: Forming the unweighted and weighted super matrices

The unweighted super matrix was created in this stage using the weights gained in the preceding step. The weighted supermatrix has dimensions (16 rows × 16 columns) and was created by standardizing the columns of the unweighted supermatrix table.

The unweighted super matrix was created in this stage using the weights gained in the preceding step. The weighted super matrix was created by standardizing the columns of the unweighted supermatrix. Limit Super matrix generation is specified in Step 4 (Figures 10 and 11).

Figure 10

The weight of the applied indicators.

Figure 11

Output map of applying ANP method.

10 AHP application

Based on the opinions of the experts, the analytic hierarchy technique was used in this study to evaluate the factors that influence and characterize urban physical resilience. The same set of experts from the ANP model offered their input and advised importance ratings in paired comparisons for this reason (Table 7) [6]. According to Table 7, professionals use paired comparison matrices to express their viewpoints.

The weight of each attribute is established by creating paired comparison matrices and figuring out the consistency ratio. Based on the judgments of the experts, Table 8 displays the relative relevance of several features in relation to one another (the weight of the main criteria) (Figures 12–14) (Table 9).

Figure 12

Output map of applying the AHP method.

Figure 13

Weighted overlay tool model for the study’s requirements.

Figure 14

Resulted map of applying the WOA method.

Five categories emerged from the use of the three methods: optimum suitability, excellent suitability, very good suitability, good suitability, and average to good suitability. By comparing the results obtained with the three methods, it was found that there was some convergence in the analysis of the results obtained.

11 Suitability assessment of the suggested resulted airport sites

After the application process in the three methods, three candidate sites were selected to satisfy all site requirements and are located at different locations within the highest suitability index regions (suitability index 7–9). Each of the suggested site will have assigned letter – A, B, C – and each cover about 50 km² (5,000 ha); max (Figure 15) in order to determine the optimum suitable site (Figures 16 and 17).

Figure 15

Optimum location map, using the ANP method.

Figure 16

Optimum location map, using the AHP method.

Figure 17

Optimum location map, using the WOA method.

12 Analysis and results

The ROC curve is a graphical representation of the performance of a binary classifier mode. Plotting the true positive rate vs the false positive rate at various categorization criteria is how it is made. The AUC/ROC is a measure of the classifier’s ability to discriminate. The curve is important because it provides a single number that summarizes the performance of the classifier across all possible thresholds. It is also insensitive to unbalanced class distribution. This curve is a useful measure for evaluating binary classification models. By calculating the area under the ROC curve, a single number can be obtained that summarizes the classifier’s ability to distinguish between classes, which makes it easy to evaluate the performance of classification models.

AUC is a synthetic measure created for ROC curves that defines the probability that an event categorized as positive by the test will occur in reality given all potential positive test results. The quantitative and qualitative relationships between the American University of Cairo and the prediction rate are expressed as follows: unsatisfactory (0.5–0.6), satisfactory (0.6–0.7), good (0.7–0.8), very good (0.8–0.9), and excellent (0.9–1). The ISM–ANP model is validated using AUC/ROC. The research region was gathered and used as a validation dataset. These locations were imported into the GIS Environment and validated using the ROC tool (Figures 18–21).

Figure 18

Steps to determine the AUC.

Figure 19

Test quality for AHP using the AUC/ROC curve.

Figure 20

Test quality for ANP using the AUC/ROC curve.

Figure 21

Test quality for WOA using the AUC/ROC curve.

According to the verification results, the ISM–ANP approach obtained outstanding validation accuracy with an AUC value of 0.990, while the AHP method achieved 0.936, as did the WOA method. This indicates that the ISM–ANP model is the most successful at evaluating airport site selection.

13 Conclusion

This article presents a combination of two techniques, ANP and ISM models, integrating GIS to detect the optimum site selection of new international airports in Nile Delta, Egypt. The suggested technique is compared with traditional methods using an accuracy test. Using suggested models integrating with GIS technology and multiple environmental and scientific standards followed in developed countries represents an efficient and elaborate technique in the selection of appropriate sites for airports in Egypt and makes the vision clear for decision-makers to choose the best places to establish new international airports or any other national project. Therefore, the following conclusions can be drawn.

Three new locations for the construction of international airports were found and selected throughout three governorates in the research region (Delta Nile, Egypt) based on the used mathematical models: Kafr El Sheikh, El Dakahlia, and El Menoufia. We were able to identify the most important zones for additional investigation using these models. Using ArcGIS 10.5, 16 input criteria (layers) were added to an overlay analysis process with GIS to solve the issue of where to locate the airport in the suggested area, a technological advancement that has the capacity to handle massive amounts of data from several sources. The following parameters were considered in this analysis: distance from residential areas (noise and pollution), land cover type, precipitation, temperature, wind speed, atmospheric pressure, relative humidity, altitude above sea level, land slope percentage, soil properties, ease of access to roads, water resources, power lines, communication stations, oil and gas lines, and urban centers.

Using ANP methodologies, the weights of the standards were established based on the geographical features of the study area (Nile Delta, Egypt), relevant laws and regulations, a review of the literature on previous research, and expert opinion. The finished map contained five different types of suitability indications, ranging from modest to very efficient.

The ISM–ANP approach was compared to the usual methods (AHP), and the three generated maps were compared. Three sites have been proposed for airport sites out of numerous locations with the greatest suitability index according to the approach map used.

The quality of result test was carried out for the chosen method using ROC_AUC with the usual methods for evaluating the reliability of the method employed, and the ranking of the suggested resulted sites. The obtained findings demonstrate the approach adopted. Despite variances in decision weights within the domain, it entirely outperformed the other potential sites.

Field investigations and satellite imagery analysis show that the indicated sites are consistent with the form’s results. The results of this study show that the performance is accurate. The model used to locate Egypt’s airport is quite precise. As a result, it can be customized as a decision-support tool for decision-makers and planners.