The overview of the susceptibility of the region to landslides depicted in Fig. 3, defines five main steps related to (i) landslide inventory; (ii) development of a database relating to spatial measurements and conditioning factors; (i) prioritizing the different classes and subclasses for SI modeling, (iv) kinetic assessment via digital photogrammetry; (v) model evaluation and safety index assessment
Landslide inventories
four field campaigns were carried out between 2022 and 2024 for a full description of the observed sites of landslides, located essentially in the Northeast and the Southeast of the study area. The collected field data are related mainly to the geotechnical properties of the soil, the structural characteristics of the rock formations, the extent and morphology of the landslides, and the environmental conditions contributing to the instability (Figs. 2 & 4)
Digital photogrammetry
Digital photogrammetry and remotely sensed data represent a significant advance in landslide risk assessment. The approach relies on the photogrammetry workstations and sentinel-2 imagery (Table 1) to highlight accurate interpretation of surface ground changes over the past years. The cartography defines a complementary approach of terrestrial photogrammetry (drone mission) carried out during May 2023. The main objective of this mission is the primary objective of the drone mission is to identify and monitor rock falls within the region to track their movement over time. The obtained data is coupled with the satellite imagery (Sentinel 2) to outline the preferential pathways of the rocky mass movement, especially during the last extreme rainfall events of 2017 and 2023, 2024.
Table 1
Metadata of the acquired Sentinel imagery
Sentinel_2 Product ID
|
Acquisition date
|
Spatial resolution
|
Cloud cover
|
2A_MSIL1C_20161112T100242_N0204
|
2016-11-12
|
10 m
|
3%
|
S2A_MSIL1C_20171117T100301_N0206
|
2017-11-17
|
10 m
|
4%
|
S2A_MSIL1C_20181122T100321_N0207
|
2018-11-22
|
10 m
|
3%
|
S2A_MSIL2A_20191127T100341_N0500
|
2019-11-27
|
10 m
|
3%
|
S2A_MSIL2A_20201121T100331_N0500
|
2020-11-21
|
10 m
|
4%
|
S2B_MSIL1C_20211111T100149_N0301
|
2021-11-11
|
10 m
|
3%
|
S2A_MSIL2A_20221121T100321_N0400
|
2022-11-21
|
10 m
|
3%
|
S2A_MSIL1C_20231126T100331_N0509
|
2023-11-26
|
10 m
|
3%
|
S2A_MSIL1C_20240504T100031_N0510
|
2024-05-04
|
10 m
|
3%
|
Susceptibility modeling
The assessment of the susceptibility to landslides relies on the agglomeration of different geo-thematic maps describing the major factors influencing this vulnerability based on the geostructural, geological, and geomorphological features of the study area to outline, via The association of the hierarchical influence of the direct or indirectly involved factors. The resulting map can be used for subsequent studies aiming the planning of the required interventions to mitigate rockfall hazards and (or) to inhibit the potential impacts. A classified susceptibility index will be helpful to delineate the appropriate section of infrastructure management and to prioritize the required safety measures and the detailed assessment of the rockfalls. The selected parameters and criteria used in this study are related to geomorphological features (slope, elevation, aspect, lithology), hydrologic characteristics (Network density, ….), and climate conditions, and they are synthesized by Table 2.
Table 2
Factor
|
Source
|
Formula/
expression
|
Classification
|
Assigned ratio
|
Altitude
|
DEM (30m)
|
Direct extraction
|
165–263 m
264–339 m
340–443 m
444–580 m
|
Different terrain elevations and associated risks
|
Slope
|
DEM (30m)
|
Direct extraction
|
< 5 °
6–11 °
12–18 °
19–36 °
|
Steeper slopes increase rockfall risk
|
Aspect
|
DEM (30m)
|
Direct extraction
|
Plate (-1);
North (0°-22.5°; 337.5°-360°);
Northeast (22.5°-67.5°);
East (67.5°-112.5°);
Southeast (112.5°157.5°;
South (157.5°202.5°);
Southwest (202.5°-247.5°); West (247.5°292.5°);
Northwest (292.5°-337.5°)
|
Influence of direction on rockfall occurrence
|
Plan Curvature
|
DEM (30m)
|
Direct extraction
|
-1.23 to -0.27
-0.26 to -0.06
-0.05 to 0.1
0.11 to 0.35
0.36 to 1.38
|
Influence of terrain shape on rockfalls
|
Profil Curvature
|
DEM (30m)
|
Direct extraction
|
-1.3 to -0.4
-0.39 to -0.11
-0.1 to 0.09
0.091 to 0.34
0.35 to 1.2
|
Influence on debris deposition and rockfalls
|
Lithology
|
Matmata map (1/100000)
|
Direct extraction
|
Clays and Marls
Limestones and Dolomites
Sandy clays
|
Influence of rock types on rockfall occurrence
|
Hydrographic network
|
DEM (30m)
|
Direct extraction
|
Four classes of stream
|
Influence of watercourses on rockfalls
|
Hydrographic density
|
DEM (30m
|
Direct extraction
|
< 1,98
1,99 − 3,97
3,98 − 5,95
5,96 − 7,93
7, 94 − 9,92
|
Influence of stream density on rockfalls
|
Rainfall distribution
|
Annual
precipitation
(CRDA Gabès)
|
Measured Data
|
50–80
81–110
111–140
141–200
> 200
|
Influence of rainfall on rockfall occurrence
|
The Analytical Hierarchy Process (AHP) is a semi-qualitative method that involves matrix-based pairwise comparisons to assess the contribution of various factors to landslides. As a multi-objective, multi-criteria decision-making approach, AHP enables users to derive a scale of preference from a set of alternatives (Pourghamesi et al. 2012). This method aids decision-makers in identifying the best solution aligned with their goals and understanding of the problem (Table 3). The equation used for landslide susceptibility mapping with AHP is as follows (Eq. 1):
LSI = Σni=1 (Ri x Wi) (Eq. 1)
where: Ri represents the rating classes for each layer, and Wi denotes the weights for each landslide conditioning factor.
To determine the landslide susceptibility map, the effects of each parameter relative to each other are assessed through pairwise comparisons. The final map is constructed using the following equation (Eq. 2):
LSAHP = (elevation x WAHP) + (slope x WAHP) + (slope x WAHP) + (curvature plan x WAHP) + (curvature profile x WAHP) + (rainfall x WAHP) + (hydrologic network x WAHP) + (stream density x WAHP) + (lithology x WAHP) (Eq. 2)
Here, WAHP represents the weight for each landslide conditioning factor. The pixel values obtained are then classified into four classes (low, moderate, high, and very high) based on natural breaks to determine the class intervals in the landslide susceptibility index map.
In the AHP method, the consistency ratio (CR) serves as an index of inconsistency, indicating the likelihood that the matrix judgments were randomly generated (Saaty 1980, 1994). The CR is calculated using the formula (Eq. 3):
CR = CI/RI (Eq. 3)
where (RI) is the average consistency index for a given order of the matrix, as provided by Saaty (1980), and (CI) is the consistency index, which can be expressed as (Eq. 4):
CI= (λmax -n)/(n-1) (Eq. 4 )
Here, λmax is the largest eigenvalue of the matrix and n is the order of the matrix. The CR, ranging from 0 to 1, reflects the matrix's consistency. A (CR) of 0.1 or less indicates a reasonable level of consistency (Vargas 2001), while a(CR) above 0.1 suggests the need for revising the judgments in the matrix due to inconsistencies. Using the AHP method, spatial factor weights were determined and applied in a weighted linear sum procedure (Voogd 1983) to calculate landslide susceptibility. In this study, a CR of 0.099 indicates a reasonable level of consistency in the pairwise comparisons, sufficient for recognizing factor weights. Consequently, rainfall received the highest weight, whereas the stream density had the lowest (Table 4).
Data treatment
the collected field data and the information obtained from geospatial platforms and open remote sensing sources are treated with different software namely Agisoft, Metashape Professional, and Cloud Compare.