Assessment of Landslide Susceptibility using the AHP and GIS Techniques for the Zurbatiya Region, East of Iraq

Abstract


Introduction
The landslide susceptibility index map is a significant matter because of its geospatial analysis, which submits a useful tool for planning for disaster mitigation and hazard.According to Zillman (1999), landslides are ranked as the third most significant natural hazard in terms of the extent of the danger they pose and their global ramifications and their capability of inflicting significant harm to human life and critical infrastructure.The occurrence of landslides, which involves the movement of the earth materials downward of the slope, is a multifaceted operation that is impacted by several factors, may be caused by natural factors, such as an increase in the forces that cause failures, moisture, the freezing and thawing of snow, the force of Earth's gravity, as well as rock structures that play a role in failures, such as discontinuities.The failure may occur suddenly, gradually, over time, or at great distances.In addition to the artificial or human activity which is the major causes that occur due to road construction on mountain, revegetation, (Faraj et al., 2023).
Multi-criteria decision analysis (MCDA) is a commonly employed approach for evaluating landslide susceptibility (Ahmed, 2015), this technique involves the use of qualitative and quantitative parameters to establish weighted evaluations, resulting in the identification of optimal solutions to a given problem (Saaty, 1980).The analytical hierarchy process (AHP) is a widely used technique for determining weights in MCDA studies (Kayastha et al., 2013;Bera et al., 2019).The analytic hierarchy process employs comparisons using pairs to determine the degree to which one alternative surpasses another based on specific criteria (Saaty, 1980).There are previous studies related to the assessment of landslides using the AHP method, but there is only one study in Iraq (Shams and Abdullah, 2018).
The aim of this research is assessment of landslide susceptibility for the Zurbatiya region by generating a landslide susceptibility index map, using geographic information system (GIS) and remote sensing techniques.The AHP method is employed in this process.Furthermore, 14 stations in landslideprone areas have been chosen, then congruent the landslide susceptibility index map with the field observations.The choice of this area was based on two causes, the first cause is its geo-tourism and geological features, which will likely attract engineering projects in the future, such as the construction of new roads, expansion of existing roads, and development of residential or tourist buildings.Additionally, the region has experienced significant tourism activity, and as such, it is crucial to investigate the risks posed by unstable slopes.The second cause is this study is considered the first of its kind in this region, as there are no previous studies of this kind.Here lies the importance of this research.The study area is small and only covers an area of 111 km 2 , the main nearby cities that exist in the study area are Badrah, Jassan, and Zurbatiya, which are surrounded by several villages (Fig. 1).

Topographically
Generally, the surface of the research area is characterized by flat and undulated surface, and it can be divided into two parts: The first part includes the highlands in the eastern part of the area, while the second part represents the flat and less undulated features within the Mesopotamian fore land (Hamza, 1997).The highest point in the project area is (850) m a.s.l and the lowest point in the project area is 14 m a.s.l.It is characterized by the presence of ephemeral and perennial valleys, with only two main streams dissecting through the study area, they are characterized by shallow and braided channels, which become narrow and deep after leaving the alluvial fans.All of them lead into a regional shallow depression called Hore Al-Shuwaicha, forming small inland deltas, these streams are Al-Hashima and Shushirin (Fig. 1) (Hamza, 1997).

Tectonically
According to Fouad (2015), the study area is a part of the Outer Platform within the low folded zone, it has been affected by the late regional intensive tectonic deformation that caused the uplifting of Himreen Structure.It lies within both, the central-eastern parts of the Mesopotamian Zone and the southeastern part of the Low-Folded Zone.These two zones represent the outer and central units of the Unstable Shelf of the Nubio-Arabian Platform (Buday and Jassim, 1987;Al-Kadhimi et al., 1996), only the southern part of the south Himreen structure represents the Low Folded Zone within the location, which is characterized by long and narrow anticlines and synclines of NW -SE trend, change to the N -S trend northwards.Their axial length varies from 20 to 33 km, and their width is from 1 to 6 Km, accordingly, they exhibit a high aspect (length/width) ratio and fall within linear folds' categories such as Himreen, Koolic1, Koolic2 Anticlines, and Al-Faraee Syncline (Mahmoud et al., 2018).The faults within the study area are: • Al-Kachaa thrust Fault: It is extended in NW -SE direction, with a length of 25 km, cut by ShuShirin valley.This Fault thrust the Upper Member of the Fatha Formation over the Injana and Al-Muqdadiya Formations.The vertical displacement is about 50m, while the horizontal displacement ranges from a few hundred meters to one kilometer (Mahmoud et al., 2018).• ShuShirin strike-slip Fault: -It has an E-W trend, right-lateral strike-slip.The displacement is (0.5) Km.It coincides with Shushirin Valley and is approximately extended for 6 Km (Mahmoud et al., 2018).

Stratigraphically
In the study area, the exposed rock formations are; Jeribe Formation (Lower Miocene), which consists of 70 m of massive dolomitic limestone (Jassim and Goff, 2006).It is exposed in the core of Koolic anticline in the eastern north part of the study area.Fatha Formation (Middle Miocene), the upper member was exposed only and contains layers of Gypsum, red claystone, limestone, and marl in a cyclic manner.The maximum exposed thickness is 257m.Injana Formation (Late Miocene), consists of an alternation of claystone and sandstone, the thick sandstone predominates over the claystone and becomes frequent.
The upper part is characterized by very thick claystone (up to 30 m) and thin sandstone beds.The total thickness is 350 m.Mukdadiya Formation (Pliocene), consists of rhythmic clastic cycles.The alternation of sandstone and claystone is lenticular as a mode of deposition, with many lateral changes to each other.The total thickness is 110 m.Bai Hassan Formation (Pliocene-Pleistocene), consist of conglomerates, they are fairly compacted to the size of pebbles ranges from 1-30 cm, cemented by calcareous sand, the total thickness reaches up to 25 m.Additional to some of the Quaternary sediments of the Pleistocene-Holocene age, such as the Alluvial fan and Sheet Runoff sediments (Mahmoud and et al., 2018).These rock formations and quaternary sediments show in Fig. (1). 2023, 56 (2D), 201-213 204

Data and Method
The objective of the present study is achieved through, the following: • Stage 1: Data preparation; • Stage 2: Data correlation analysis; • Stage 3: Landslide susceptibility modeling; These stages are described respectively that explain the adopted methodological framework as in the next flowchart.

Data Preparation
The first step in any landslide assessment involves gathering all available information and data pertaining to the region of interest (VanWesten et al., 2008).Therefore, establishing a relationship between causal parameters and landslides poses a significant and challenging issue.Various data layers can be utilized in modeling landslide susceptibility, with the number of layers ranging from a few to several parameters.Researchers' parametric preferences for preparing landslide susceptibility maps were evaluated by Hasekiogullari ( 2010), based on a detailed study of 114 scientific studies indexed in the Science Citation Index (SCI) and published in different scientific journals between 2000 and 2010 (Fig. 2).Pourghasemi (2014) reviewed 220 scientific papers published between 2005 and 2012 in various ISI journals to evaluate the parametric preferences for preparing landslide susceptibility maps too (Fig. 3).Thematic layers such as slope angles, geology, slope aspect, topographic elevation, plan curvature, distance from valley, and distance from fault are available and have the most impact on landslides in the study region.These factors are integrated to prepare a susceptibility map based on the study area's integrated database.

Data Correlation Analysis
ArcGIS software is used to produce the layer maps that are employed to create the landslide susceptibility index map.Since this map is computed using raster data, all vector conditioning elements (such as elevation or slope angle, etc.) were transformed to a raster of 12.5×12.5m.Each parameter has been divided into several classes based on how likely it is to produce unstable slopes, each class was given a rating value between 1 and 9.This rating scale is almost identical to the AHP scale.Where: 9very high, 7-high, 5-moderate, 3-low, and 1-very low.
The slope angle directly affects landslide; thus, it is used in preparing landslide susceptibility maps, the maximum slope angle ranges up to 57˚ (Fig. 4a).Slopes were reclassified into five classes < 10˚, 10˚-18˚, 18˚-30˚, 30˚-45˚and >45˚, very gentle, gentle, moderately steep, steep and very steep slopes respectively.Landslide risk is anticipated to rise as slope steepness increases, peaking on very steep slopes.The geology map layer is important; the type of exposed rock units considerably affects the landslide risk.Using ArcGIS 10.8, diverse lithological units in the research region, as illustrated in Fig. 4b, were classified into several classes depending on a geological map prepared at a scale of 1:100.000by Mahmoud et al. (2018).According to their propensity to cause landslides, all lithological units were divided into six groups.Topographic Elevation Layer, to improve the usefulness of this landslide susceptibility mapping, a topographic elevation layer was derived from the ASTER DEM 12.5 m resolution data.The research area's topographic height, which has been classified into five classes, varied from less than 100 m to more than 350 m above sea level, as follows ).
The slope aspect can be defined as the slope direction which identifies the downslope direction of the maximum rate of change of elevation (Fernandes et al., 2004) and it is calculated in compass degrees (from 0 to 360), based on the surface tools in Arc GIS.It was reclassified into 9 classes, each representing an angular section of 45˚illustrated rate in (Fig. 4d): north-east (22.50-67.50), east (67.51-112.50), south-east (112.51-157.50), south (157.51-202.50), south-west (202.51-247.50), west (247.51-292.50)and north-west (292.51-337.50)north (337.51-22.50).Great attention has been paid to the conditions of the slope because the configuration and steepness play an important role in the occurrence of landslides.
Another geomorphic indication of topographic characteristics is sloping curvature, which is the rate at which the gradient of a slope changes in one direction.It affects surface erosion by directing or deflecting the flow of runoff down the hill, and at some locations, it is linked to the flow pattern on a slope.It displays two extreme positive and negative numbers.A surface that is convex upward at a certain point is often represented by a positive number, whereas surfaces that are concave upward are typically described by negative values.Flat regions are described by zero to lower positive and lower negative integers (the small absolute values) (Lee et al., 2003).According to the review of the studies, landslides are more likely to occur the larger the negative value.Landsides are least likely to be triggered on level surfaces.Using ArcGIS 10.8 software, the plan curvature map for the research region was generated from the DEM data (Fig. 4e).In order to apply weights, the range of curvature valuesweres split into 5 positive and negative classes.
Distance from fault: The faults increase landslide activity.It can be thought of as a major zone of weakness within the geologic units it cuts, regardless of whether it is tectonically active.According to Pourghasemi et al. (2012), faults and shear zones also preferentially bind and limit the groundwater flow, which affects slope stability.Distance from faults was classified into 5 classes of < 1000, 1000-1500, 1500-2000, 2000-3000, and >3000 meters (Fig. 4f)-3000, and >3000 meters (Fig. 4f).
Distance from valleys: The major and branching valleys are highly irregular; this shows in the drainage map which is derivative from DEM data in the ArcGIS10.8software.When determining a region's susceptibilities to landslides, the distance from waterways is also important.Different buffer lengths extending outward from all channels were developed, then classified into 5 classes to account for the effect of drainage on the nearby slopes, as follow: <150, 150-300, 300-500, 500-750 and >750 meter (Fig. 4g).

Landslide Susceptibility Modeling
The present study evaluates landslide susceptibility by AHP before the fieldwork.Generally, to effectively approach a complex and unstructured problem like that, it's recommended to use a systematic approach (Saaty and Vargas, 2001).Which involves the following steps: Decomposition: This involves breaking down the problem into its constituent components, in order to better understand the elements that contribute to the overall complexity.
Hierarchical ordering: The identified components are then arranged in a hierarchical order, with the more significant factors being given higher priority than others.
Subjective judgment: Numerical values are assigned to subjective judgments on the relative importance of each factor.This involves analyzing the factors based on their relevance to the problem at hand, as well as their impact on the overall outcome.
Priority determination: The assigned numerical values are then analyzed to determine the priorities to be assigned to each factor.This allows for a more structured and systematic approach to problemsolving, with a clear focus on the factors that have the greatest impact on the final outcome.(Saaty and Vargas, 2001).
According to the role that each parameter plays in the landslide occurrence, the AHP technique was utilized to determine the parameter weights.The most weight was given to the parameter that had the maximum effect on the determination of the goal (as a numerical value) in terms of its nature, physical properties, and subjective expert opinions, with the lowest weight given to the factor that had the least effect.This technique is utilized to systematically assign preferences depending on the numerical scale proposed by Al-Saaty (2000), given in Table 1.

Numerical value Description 1
The importance is equal 3 The importance is slight 5 The importance is moderate 7 The importance is very strong 9 The importance is extreme 2, 4, 6, 8 The intermediate values between two adjacent values By using a pairwise comparison matrix (PCM), all the parameters were compared against each other, which allowed for an independent assessment of each factor's contribution and redundancy, thereby producing a measure of the consistency of the judgments comparison as well as reducing the measurement error, called the consistency index (CI), which is defined as follows: CI= λMax -n / n-1 (1) Where: λMax: -the maximum eigenvalue of the matrix; n: -the number of participating parameters.To assess a so-called random consistency index (RI) (Saaty 2000), reciprocal matrices were created by Saaty utilizing random scales such as 1/9, 1/8..., 1...., 8, and 9 as shown in Table 2. Then it was used the consistency ratio (CR) was illustrated by the following equation based on Saaty 2000: Based on Saaty 1977 the weighting coefficients were acceptable if the CR value was less than or equal to 0.1, but if it was larger than 0.1, the subjective assessment needed to be changed.This is a methodical manner to produce weights for the heuristic weighted overlay method to prevent inconsistent weights, based on AHP and through normalizing of AHP-PCM.Each input parameter received a different weight based on the extent of expected landslide susceptibility.Details of each of the parameters are given in Table 3.

Resulting and Validation
Landslide susceptibility index map was created after the criterion scores were normalized and weights were determined using the following equation by Voogd (1983).The AHP-based PCM determines the weight and rating values, where it is created using a weighted linear combination and a continuous numerical scale to indicate various degrees of landslide susceptibility.

LSI. = ∑ Wjwij
(3) Where: -Wj:-the weight of the parameter j; wij:-the rating of the class I in the parameter j; These equations have been applied in GIS software on the seven elements (slope angles, geology, slope aspect, topographic elevation, distance from faults, and distance from valleys ), where the slope angle is representing the most critical factor, the geology has been considered the next very important factor because the stratigraphic sequence of rocks suggests that harder lithological units like gypsum typically occupy the uppermost layers, while the little hardness such as claystone and marl, tend to reside in lower strata, this arrangement can be attributed to the differential susceptibility of these materials to the weathering, so will experiencing greater rates of degradation over time.
Moreover, the presence of Al-Kachaa thrust and Shushirin strike-slip fault in this rough.Nearness to valleys also poses some risks, mainly because of the undercutting of the slope, which may lead to failure of the slope.Elevation also poses certain risks, mainly landslides, and higher elevations are generally more susceptible to slope failure.These elements were combined to produce the landslide susceptibility index map (LSI) (Fig. 5), which was tested by the fieldwork stage, in which 14 stations were selected based on the past landslides (plate 1), distributed within the four zones (Fig. 6).The calculated landslide susceptibility classes' coverage by area as well as their proportions of the overall area are shown in Table 4.

Discussion
Landslide susceptibility index map is classified into four zones in ArcGIS software, with a specific area and ratio for each zone, as follows: The low zone of landslide susceptibility has the highest ratio 41.2% of the whole region, which covers 46.1km 2 .
The moderate landslide susceptibility class which shared is 39.9% which cover 44.4km 2 .The percentage of high landslide susceptibility 14.8% is lower than the previous classes, which cover 16.3km 2 .
The very high landslide susceptibility class, which covers 4.2 km 2 , recorded the lowest percentage 3.9%.
Overall, it can be seen the exact match when comparing the map of LSI with what was recorded of previous failures within the 14 stations.
Station 8 within the high zone.Stations 2, 9, 10, and 12 within the moderate zone.Stations 11, 13, and 14 within the low zone.
Only a few or no landslides occurred in the southern and western parts of the research area, while the susceptibility is very high and mostly high for the middle, toward the east, and the extreme northeast, as shown in the landslide susceptibility index map, which is mainly caused by higher elevations and high slope angles resulting from the impact of the Al-Kachaa thrust and Shushirin strike-slip fault, which are the principal causes of distortions in the area, including landslides.In addition to the geology and arrangements of lithological units and the degree to which they are influenced by the weathering factor.The various types of landslides can be differentiated within the study area as follows: Toppling and Rock fall, in addition to the rolling that occurs later for the separated blocks by failure.The research area has experienced significant tourism activity because of its geological diversity and geotourism, so it will likely attract engineering and tourist projects in the future, such as the construction of new roads, expansion of existing roads, and development of residential or tourist buildings.As such, it was important to investigate the risks posed by unstable slopes.

Conclusions
The study area has not been subjected to any previous studies concerned with studying and evaluating slope stability.Each landslide susceptibility study has its own set of restrictions.These restrictions are the conditioning factors, and their applicability must be correct for a successful multicriteria susceptibility analysis.In an attempt to reduce the hazard of landslides if the zone of hazard is predicted, the AHP method and GIS were used for landslide susceptibility assessment.Seven factors have been selected as the most important and available for contributing to the occurrence of landslides in the Zurbatiya region, and they are as follows: slope angle, geology, topographic elevation, aspect, plan curvature, distance from faults, and distance from valleys.These elements were arranged according to the most influential elements in landslide susceptibility.According to the experience of the researcher, the values of these factors have been arranged.Then the landslide susceptibility index map was prepared, and it is classified into four zones (low, moderate, high, and very high susceptibility), which cover an area of (46.1 km 2 , 44.4 km 2 , 16.3 km 2 , 4.2 km 2 ), respectively.The rainfall element is not included because it is very uniform, and because the area is not inhabited, the land cover was also not used.

Fig. 4
Fig. 4. a. Slope angle map, b. geologic map, c. topographic elevation, d. slope aspect map, e. plan curvature map, f. distance from fault, g. distance from the valley

Table 2 .
Random consistency index

Table 3 .
Classes, ratings, and relative importance of landslide conditioning factors.

Table 4 .
landslide susceptibility zones, areas and ratios Plate 1. shows the 14 stations of the past landslide in the research area.