论文信息

Sarah Debbagh / Ammar Lachehab

Research Square

论文解读

AI 辅助整理
方法
Landsat 8与SRTM数据在GEE和ArcGIS中按30m网格叠加,用AH
发现
47.4%土地适宜农业,ROC的AUC达0.935,山前高适宜带应保护而非集约化。
机会点
可将该AHP分区框架与时间序列遥感、气候情景结合,开展动态适宜性评估与城市扩张预警。

论文摘要

摘要:农业土地适宜性评价是地中海半干旱地区可持续土地利用规划的关键工具,这些地区面临气候变率、土地退化和城市扩张压力。本研究通过整合五个空间标准——土壤水分指数(SMI)、归一化差异植被指数(NDVI)、地表温度(LST)、土地利用/土地覆盖(LULC)以及由坡度、高程和距水体距离构成的综合农业-地貌适宜性因子——开发并验证了基于GIS-遥感的阿尔及利亚北部布利达省农业潜力区划(APZ)模型。多源数据集——2025年生长季获取的Landsat 8 OLI/TIRS影像、SRTM DEM和OpenStreetMap水系数据——在Google Earth Engine平台上统一至30 m网格,并在ArcGIS 10.8中进行整合。通过层次分析法(AHP)特征向量法确定标准权重,其中SMI(0.4502)和NDVI(0.2502)起主导作用;五因子矩阵和四因子矩阵的一致性比率分别为6.25%和3.34%,证实了所获取判断的逻辑一致性。加权线性组合生成了覆盖148,046 ha的五级APZ图,其中47.40%被归类为适宜,22.53%为中等适宜,12.58%为高度适宜,11.43%为勉强适宜,6.05%为不适宜。ROC验证的曲线下面积为0.935,表明具有优异的判别性能。结果勾勒出米提贾平原的适宜农业基质、布利达阿特拉斯山前的应被解读为以保护为导向而非集约化指令的高度适宜带,以及城市化山麓走廊——其4,015 ha的残余适宜土地构成有限储备,需要明确保护以抵御城市蔓延。该框架为半干旱地中海景观的可持续农业规划提供了透明、可重复的筛选工具。

全文

Research Square

GIS-Based Agricultural Potential Zoning Using
Remote Sensing Indicators and the Analytic
Hierarchy Process: A Case Study of Blida Wilaya,
Northern Algeria
Sarah Debbagh 
University of Sciences and Technology Houari Boumediene
Ammar Lachehab 
University of Sciences and Technology Houari Boumediene
Research Article
Keywords: Agricultural land suitability, GIS, Remote sensing, Analytic Hierarchy Process, Landsat 8,
AUC/ROC validation, Blida, Algeria
Posted Date: November 13th, 2026
DOI: https://doi.org/10.21203/rs.3.rs-11473908/v1
License:   This work is licensed under a Creative Commons Attribution 4.0 International License.  
Read Full License
Additional Declarations:
No competing interests reported.
Tables 1 to 3 are available in the Supplementary Files section.

1

GIS-Based Agricultural Potential Zoning Using Remote Sensing Indicators and the Analytic Hierarchy Process: 1
A Case Study of Blida Wilaya, Northern Algeria 2
3
Affiliations 4
DEBBAGH Sarah 1*. LACHEHAB Ammar 2 5
1 DEBBAGH Sarah, Laboratory of Geography and Planning, Faculty of Ea rth Sciences, Departement of 6
Geography and Territorial Planning, University of Sciences and Technology Hou ari Boumediene (USTHB), Algiers, 7
Algeria, saradebbagh16@gmail.com 8
2 LACHEHAB Ammar, Laboratory of Geography and Planning, Faculty of Earth Science s, Departement of 9
Geography and Territorial Planning, University of Sciences and Technology Hou ari Boumediene (USTHB), Algiers, 10
Algeria, ammarlachehab@yahoo.fr 11
12
Acknowledgments 13
The authors sincerely thank the reviewers for their insightful comments and constructive suggestions, which 14
significantly improved the quality and clarity of this paper. 15
Funding source 16
The authors declare that no funds, grants, or other financial support were received for the design, execution, and 17
publication of this study. 18
Competing interest 19
All authors declare no competing interests. 20
Author contributions 21
Sarah Debbagh conceived and designed the study, developed the research methodology, prepared all figures and 22
tables, and wrote the original draft of the manuscript and supervised the overall project. Ammar Lachehab, as supervisor, 23
refined the research question and the methodological framework, verified the weight derivations and consistency ratios, 24
contributed regional and field knowledge of the Blida-Mitidja-Atlas system, prov ided laboratory and institutional 25
resources, and critically reviewed and edited the manuscript. Both authors read a nd approved the final manuscript and 26
agree to be accountable for all aspects of the work. 27
Abstract 28
Agricultural land suitability assessment is a critical instrument for sustainable land-use planning in 29
Mediterranean semi-arid regions exposed to climate variability, land degradation, and expanding urban pressure. This 30
study develops and validates a GIS–remote sensing–based agricultural potential zoning ( APZ) model for Blida Wilaya, 31
northern Algeria, by integrating five spatial criteria: the Soil Moisture Index (SMI), the Normalized Difference Vegetation 32
Index (NDVI), Land Surface Temperature (LST), Land Use/Land Cover (LULC), and a composite Agro-Physiographic 33
Suitability factor derived from slope, elevation, and distance from water bodies. Multi-source datasets—Landsat 8 34
OLI/TIRS imagery acquired during the 2025 growing season, SRTM DEM, an d OpenStreetMap hydrography—were 35
harmonized to a 30 m grid on the Google Earth Engine platform and in tegrated within ArcGIS 10.8. Criterion weights 36
were derived through the Analytic Hierarchy Process (AHP) eigenvector method, with SMI (0.4502) and NDVI (0.2502) 37
exerting dominant influence; consistency ratios of 6.25% and 3.34% for the five-factor and three-factor matrices, 38
respectively, confirmed the logical coherence of the elicited judgments. Weighted linear combination yielded a five-class 39
APZ map covering 148,046 ha, of which 47.40% was classified as suitable, 22.53% as moderately suitable, 12.58% as 40
highly suitable, 11.43% as marginally suitable, and 6.05% as not suitable. ROC valid ation produced an area under the 41
curve of 0.935, indicating excellent discriminatory performance. The results delineate a suitable agricultural matrix across 42
the Mitidja plain, a highly suitable belt along the Blidean Atlas front that should be interpreted as conservation-oriented 43

2

rather than as a mandate for intensification, and an urbanized piedmont corridor whose 4,015 ha of residual suitable land 44
constitutes a finite reserve requiring explicit protection against urban sprawl. The framewo rk provides a transparent, 45
reproducible screening tool for sustainable agricultural planning in semi-arid Me diterranean landscapes. 46
Keywords: Agricultural land suitability; GIS; Remote sensing; Analytic Hierarchy Process; Landsat 8; AU C/ROC 47
validation; Blida; Algeria 48
Introduction 49
Ensuring global food security and sustainable agricultural development has bec ome increasingly challenging 50
due to the compounded effects of climate variability, land-use change, and po pulation growth. The Mediterranean basin 51
and North African regions, including Algeria, are particularly vulnerable to these pressures, experiencing heightened land 52
degradation, water scarcity, and fluctuating agricultural productivity (Zhu et al., 2022; Faraslis et al., 2023; Abdelrahman 53
& Arafat, 2020). In this context, robust agricultural land suitability assessment is essential for optim izing land use, 54
enhancing crop yields, and supporting adaptive management strategies under chan ging environmental conditions 55
(Choudhary et al., 2023; Anusah et al., 2023). 56
Traditional approaches to land evaluation have often relied on field surveys and exp ert judgment, which can be 57
time-consuming and limited in spatial coverage. The integration of Geographic In formation Systems (GIS) and Remote 58
Sensing (RS) technologies has revolutionized spatial analysis by enabling the efficient collectio n, processing, and 59
interpretation of large-scale environmental data (Mostafiz et al., 2021; Saha et al., 2019). These tools facilitate the 60
extraction of critical biophysical indicators such as the Soil Moisture Index (SMI), No rmalized Difference Vegetation 61
Index (NDVI), Land Use/Land Cover (LULC), Land Surface Temperature (LST), and agro-physiographic parameters 62
each playing a pivotal role in determining agricultural potential under variable climatic r egimes (Saha et al., 2019; 63
Donmez et al., 2024). 64
Multi-Criteria Decision Analysis (MCDA) frameworks have further enhanced land su itability modeling by 65
systematically integrating diverse criteria into a unified decision-making process. Amo ng MCDA techniques, the 66
Analytical Hierarchy Process (AHP), particularly when employing the eigenvector method fo r weight derivation, stands 67
out for its transparency, consistency checking, and ability to incorporate both quantitative data and expert knowledge 68
(Saaty, 1980; Malczewski, 2004; Choudhary et al., 2023, Wijesinghe, 2024). The AHP’s pairwise comparison matrix 69
enables rigorous weighting of factors such as slope, elevation, distance from water bodies subcomponents that collectively 70
define agro-physiographic suitability (Faraslis et al., 2023). 71
Model validation remains a critical step in spatial suitability analysis. The Area Under the R eceiver Operating 72
Characteristic Curve (AUC/ROC) approach is widely recognized for quantifying model predictive accuracy by comparing 73
mapped suitability against observed outcomes or independent reference data. High AUC values (>0.80) reported in recent 74
applications underscore the reliability of GIS-AHP frameworks for guiding land management decisions in complex 75
environments (Shinde et al., 2024; Küçüker & Giraldo, 2022). 76
Despite methodological advances globally, there remains a notable research gap regardin g comprehensive 77
agricultural zoning using integrated GIS- RS-AHP approaches in North African contexts particularly within Algeria’s 78
Blida Wilaya. Existing studies often focus on broader Mediterranean or Middle Eastern settings or address adjacent 79
applications such as groundwater or erosion risk mapping rather than dir ect agricultural suitability under current climate 80
pressures (Chemirik et al., 2024; Mihi & Benaradj, 2022). 81
This study aims to address this gap by developing a high-resolution agricultural zoning model by using GIS-RS-82
derived indicators weighted via the AHP eigenvector method. By incorporating SMI, NDVI, LULC, LST, and an 83
integrated Agro-physiographic parameter and validating results with AUC/ROC analysis the research seeks to provide 84
actionable insights for sustainable land use planning in semi-arid Mediterranean landscapes. 85
Materials and Methods 86
Study Area 87
Blida Wilaya (wilaya code 09) is situated in the Tell region of northern Algeria, appr oximately 50 km south-west of 88
Algiers and centered near 36°28′ N, 2°49′ E (Fig. 1). It extends over 1,480 km2 and is bounded by the wilayas of Tipaza 89
to the west, Algiers to the north, Boumerdès to the east, and Bouira, Médéa and Aïn Defla to the south and south-west. 90

3

Administratively, the wilaya comprises 25 municipalities and hosts more than one million inhabitants (1,002,937 at the 91
2008 General Population and Housing Census), corresponding to a density of nearly 6 78 inhabitants/km 2 among the 92
highest in Algeria (ONS, 2008). 93
The territory is organized into three contrasting physiographic units that structure its agricultural geography (Fig. 1). The 94
northern sector is occupied by the western part of the Mitidja plain, one of Alg eria’s most fertile alluvial basins (ca. 65 95
km in total length), whose deep Quaternary soils and extensive groundwater r eservoir support intensive citrus, orchard, 96
vegetable and cereal production (Bellout et al., 2020). Elevations below 150 m cover 39.14% of the wilaya, and gentle 97
slopes (<5%) account for 37.24% of its surface, both concentrated in this u nit. To the south, the plain rises abruptly into 98
the Blidean Atlas, the easternmost calcareous massif of the Tell Atlas, whose summits e xceed 1,500 m and of which 99
roughly 2.9% of the wilaya lies above 1,200 m. This mountainous hinter land hosts the Chréa National Park (established 100
1985; ca. 26,000 ha), part of the UNESCO Chréa Biosphere Reserve, r enowned for its Atlas cedar (Cedrus atlantica), 101
holm oak and Aleppo pine stands and for the Barbary macaque (Macaca Sylvanus) populations (UNESCO-MAB, 2025). 102
Between the two lies a piedmont corridor of plains and foothills along which the largest urban centres (Blida, Ouled 103
Yaïch, Beni Mered, Boufarik, Larbaa and Meftah) have coalesced, exposing s ome of the wilaya’s best alluvial soils to 104
rapid peri-urban expansion. 105
The wilaya experiences a Mediterranean climate (Köppen Csa) of semi-arid to sub-h umid character, with hot, dry 106
summers and mild, wet winters. Mean annual precipitation is approximately 640 m m at plain stations and increases to 107
more than 800 mm in the Atlas massif, as documented in the Chiffa basin (Madani et al., 2024), whereas the mean annual 108
temperature is close to 17°C at Blida station. Rainfall is strongly concentrated between November and March, an d the 109
pronounced summer water deficit constitutes the principal climatic constraint on rainf ed agriculture, underpinning the 110
importance of soil moisture as a suitability criterion. The hydrographic network drains the mountain front and crosses the 111
plain: Oued Chiffa, incised in the eponymous gorges, joins Oued Mazafran alon g the western boundary before reaching 112
the sea; Oued Bou Roumi is impounded by the Bou Roumi dam for ir rigation and water supply; Oued Djer drains the 113
south-western margin towards the Chelif basin; and the eastern plain is drained toward s the Algiers region by Oued El 114
Harrach. This juxtaposition of a highly productive irrigated plain, a conserved forested mountain belt and an expanding 115
urban corridor, under seasonal water stress, renders Blida Wilaya a representative case for agricultural potential mapping 116
in semi-arid Mediterranean landscapes. 117

4

118
Fig. 1: Geographic Location of Wilaya of Blida. 119
Data Collection 120
A comprehensive geospatial database was assembled from multi-source remote sensing imagery and open-access 121
geographic datasets (Fig. 2). The primary optical dataset consisted of Landsat 8 Oper ational Land Imager (OLI) and 122
Thermal Infrared Sensor (TIRS) scenes acquired during the 2025 growing season, selected for minimal cloud 123
contamination (<5%) and a spatial resolution of 30 m (USGS, 2025). Topographic parameters (slope and elevation) were 124
derived from the Shuttle Radar Topography Mission Digital Elevation Model (SRTM DEM) at 30 m resolution. 125
Hydrographic features (rivers, wadis, dams, and lakes) were extracted from th e OpenStreetMap (OSM) dataset and 126
clipped to the administrative boundary of the study area. All raster layers were repro jected to the Universal Transverse 127
Mercator (UTM) Zone 31N coordinate reference system and resampled to a uniform 30 × 30 m pixel grid to ensure spatial 128
interoperability across the analytical workflow. 129

5

130
Fig. 2: General flowchart of study methodology. 131
Preprocessing of satellite imagery (including atmospheric correction, radiometr ic calibration, and geometric 132
verification) was performed within the Google Earth Engine (GEE) cloud-based platform. Sub sequent raster 133
reclassification, vector overlay operations, and cartographic composition were executed in ArcGIS 10.8 (ESRI, 2016). 134
Preparation of Thematic Layers 135
Five spatial criteria governing agricultural potential were extracted and prepared as individual raster layers prior 136
to their integration within the multi-criteria decision framework. 137

  • Normalized Difference Vegetation Index (NDVI): Vegetation vigor and photosynthetic activity were 138
    quantified using the classical NDVI formulation (Cetin et al., 2022): 139
    𝑁𝐷𝑉𝐼 = −
     +
    140
    where  and  denote the surface reflectance values in the near-infrared (Band 5) and red (Band 141
  1. spectral channels, respectively. 142
  • Land Surface Temperature (LST): LST retrieval followed the single-channel radiative transfer algorithm 143
    adapted for Landsat 8 TIRS bands (Jesus & Santana, 2017, Machireddy, 20 25). The procedure involved three 144
    sequential steps. First, the Proportion Vegetation () was estimated from the NDVI: 145
     = ( 𝑁𝐷𝑉𝐼 −𝑁𝐷𝑉𝐼𝑁𝐷𝑉𝐼 −𝑁𝐷𝑉𝐼 )2 146
    Second, the land surface emissivity () was computed as: 147
     = 0.004× +0.986 148
    Third, the at-sensor spectral radiance of thermal Band 10 was calculated as: 149
    ! = "# ×$%&# +' # 150
    where "# is the band-specific multiplicative rescaling factor and '# the additive rescaling factor, both 151
    extracted from the image metadata. The brightness temperature (BT) was subsequen tly derived via the inverse 152
    Planck function: 153

6

() = *2
ln(*1! +1) −273.15 154
where *1 and *2 are band-specific thermal conversion constants. The final LST was obtained throu gh 155
emissivity correction: 156
!5) = ()
1+( 6× () )× ln
157
where 6 is the effective wavelength of Band 10 (10.9 µm) and  = ℎ8/: = 1.438×10−2m·K, with 158
ℎrepresenting Planck's constant, 8 the speed of light, and : the Boltzmann constant. 159

  • Soil Moisture Index (SMI). The SMI was derived from the normalized LST approach, exploiting the inverse 160
    relationship between surface thermal emission and root-zone water content (Mach ireddy, 2025): 161
    5"𝐼 = !5) −!5)!5) −!5) 162
    where !5) and !5) correspond to the extreme thermal values within the scene. This index 163
    provides a spatially continuous proxy of soil hydrological status directly relevant to crop water availability. 164
  • Land Use/Land Cover (LULC). A supervised classification was performed using the Random Forest algorithm 165
    within the GEE environment. The classifier was trained on spectrally homogeneous reference polygons and 166
    yielded seven land-cover categories: water bodies, built-up areas, bare areas, shrubland, forest, grassland, and 167
    cropland. Post classification accuracy assessment and majority filter smoothing were ap plied to reduce salt and 168
    pepper noise. 169
  • Agro-Physiographic Suitability. A composite terrain index was constructed by integrating three sub criteria 170
    (slope, elevation, and distance from water bodies ) through a dedicated three-factor AHP (Fig 2). Slope and 171
    elevation were extracted from the SRTM DEM using standard GIS terrain analysis algor ithms. Distance from 172
    water bodies was computed via Euclidean distance allocation from the vectorized hydro graphic network. The 173
    resulting continuous surface was reclassified into five ordinal suitability zones. 174
    Agricultural Potential Mapping Method 175
    Analytic Hierarchy Process 176
    The integration of the five thematic layers into a unified agricultural potential map was ac hieved through the 177
    Analytic Hierarchy Process (AHP), a structured multi-criteria decision analysis (MCDA) technique originally formalized 178
    by Saaty (1980). The AHP decomposes a complex spatial decision problem into a hier archical structure of criteria and 179
    sub-criteria, enabling the systematic quantification of relative factor importance through pairwise comparisons. 180
    Pairwise Comparison Matrix Construction 181
    Two levels of pairwise comparison matrices were constructed. At the sub-criterion level, a 3 × 3 reciprocal matrix 182
    (Fig. 3) was assembled for the Agro-Physiographic factor (slope, elevation, and distance from water bodies) using Saaty's 183
    (1980) fundamental 1-9 scale of relative importance (Tab . 1). At the criterion level, a 5 × 5 matrix (Fig. 4) was built for 184
    the five principal factors (SMI, NDVI, Agro-Physiographic Suitability, LULC, an d LST). Judgments were grounded in 185
    the agronomic and biophysical literature concerning Mediterranean semi-arid landscapes (Faraslis et al., 2023; Zhu et al., 186
    2022). Soil moisture was assigned to have the highest dominance owing to its im mediate control on plant-water 187
    availability, while LST received the lowest comparative weight given its more indirect thermal regulation role. The 188
    matrices satisfy the conditions of diagonal unity (< = 1) and reciprocity (<= = 1/<=). 189
    Tab. 1: Saaty's Fundamental Scale of Relative Importance (Saaty, 1980) 190
    Score Verbal Judgement Explanation
    1 Equal importance Two criteria contribute equally to the objective

7

2 Weak or slight
3 Moderate importance Experience and judgement slightly favour one over another
4 Moderate plus
5 Strong importance Experience and judgement strongly favour one over another
6 Strong plus
7 Very strong importance A criterion is favoured very strongly; its dominance is demonstrated in
practice
8 Very, very strong
9 Extreme importance The evidence favouring one over another is of the highest possible order
191
192
Fig. 3: Agro-Physiographic Suitability AHP Pairwise Comparison Matrix (Note: S=Slope, E=Elevation, D-193
WB=Distance from Waterbody). 194

8

195
Fig. 4: Agricultural Potential Zones AHP Pairwise Comparison Matrix (Note: SMI=Soil Moisture Index, 196
NDVI=Normalized Difference Vegetation Index, A-PS= Agro-Physiographic Suitability, LULC=Land Use Land 197
Cover, LST=Land Surface Temperature). 198
Priority Vector (Eigenvector Method) 199
Relative weights were computed following Saaty's principal right eigenvector method (Saaty, 1977, 2003). The 200
procedure comprised two operations. First, each column of the pairwise compariso n matrix was normalized by dividing 201
every element by its respective column sum, yielding the normalized matrix '̅. Second, the priority vector ? was obtained 202
by averaging the rows of '̅: 203
@ = 1A∑ < ̅===1 204
By construction, ∑ @=1 = 1, ensuring that weights represent proportional contributions to total criterion 205
importance (Escobar et al., 2004; Klein, 2019). For the five-factor matrix, the derived weights were: 206
SMI = 0.4502, NDVI = 0.2502, Agro-Physiographic Suitability = 0.1697, L ULC = 0.0872, and LST = 0.0427 207
(Tab. 3). 208
For the three-factor Agro-Physiographic sub-matrix, the weights were: Slope = 0.6333, Elevation = 0.2605, and 209
Distance from Waterbody = 0.1062 (Tab. 2). 210
Consistency Assessment 211
The internal logical coherence of pairwise judgments was verified through the Cons istency Ratio (CR) (Saaty, 212
1987). The maximum eigenvalue (6) was estimated as: 213
6 = 1A∑ ('⋅@ )@

=1
214
The Consistency Index (CI) was then calculated: 215

9

J𝐼 = 6 −AA−1 216
and normalized against the Random Index (RI) corresponding to the matrix dimension: 217
JK = J𝐼K𝐼 218
For the five-factor matrix (A = 5, K𝐼 = 1.12), the computed values were 6 = 5.28, J𝐼 = 0.070, and JK =219
6.25%. 220
For the three-factor sub-matrix ( A = 3, K𝐼 = 0.58), 6 = 3.04, J𝐼 = 0.019, and JK = 3.34%. Both ratios 221
fall well below the 10% acceptability threshold (Saaty, 1980), confirming that the elicited judgments are logically 222
consistent and that the derived weight vectors are reliable for spatial aggregation. 223
Spatial Aggregation and Reclassification 224
The final Agricultural Potential Zone map was generated through weighted linear comb ination (WLC) of the 225
five reclassified raster layers, each multiplied by its respective AHP-derived weight. The continuous suitability score was 226
subsequently segmented into five ordinal classes: Not Suitable for Agriculture, Marginally Suitable, Moderately Suitable, 227
Suitable for Agriculture, and Highly Suitable for Agriculture. The same five-class reclassification scheme was applied to 228
each individual thematic layer to ensure uniformity in the interpretation of suitability gradients across the study area. 229
Validation 230
Model validation was performed using the Area Under the Receiver Operating Characteristic Curve 231
(AUC/ROC), a widely applied method for assessing the predictive accuracy of spatial suitability models. The ROC curve 232
describes the trade-off between true positive and false positive rates across classification thresholds, while the AUC 233
quantifies the model’s ability to discriminate between suitability classes. Higher AUC values indicate stronger predictive 234
performance and greater reliability of the agricultural potential map (Shinde et al., 2024). 235
Results and Discussion 236
Suitability Analysis of Reclassified Factors 237
Slope 238
Slope governs the intensity of surface runoff, the rate of soil erosion, and the ease of agricultural mechanization, 239
all of which directly condition land workability and moisture retention capacity. Gentle slopes favor water infiltration and 240
reduce erosion hazards, thereby enhancing agricultural suitability, whereas steep slopes accelerate runoff, intensify soil 241
loss, and constrain the use of farm machinery. Within the Agro-Physiogr aphic AHP pairwise comparison matrix, slope 242
received the highest normalized weight among the three topographic sub-criteria (Wi = 63.33%), reflecting its dominant 243
control over the composite Agro-Physiographic Suitability factor. The reclassification of the slope map yielded five 244
classes (Fig. 5): (1) slopes below 5% (37.24% of the study area), assigned V ery High suitability owing to their near-flat 245
topography favorable for cultivation and irrigation; (2) slopes of 5–15% (14.54%), rated High; (3) slopes of 15–30% 246
(23.99%), rated Medium; (4) slopes of 30–50% (21.29%), rated Low due to increasing erosion susceptibility; and (5) 247
slopes exceeding 50% (2.94%), rated Very Low, corresponding to the steep, erosion-prone terrain least suited to 248
agricultural use. 249

10

250
Fig. 5: Slope Map. 251
Elevation 252
Elevation exerts an indirect but significant influence on agricultural potential throu gh its control over local 253
thermal regimes, precipitation gradients, soil depth accumulation, and accessibility to farming infrastructure. Lower 254
elevations in the study area generally coincide with deeper alluvial soils, milder temperatur es, and easier access for 255
irrigation and equipment, while higher elevations impose thermal limitations and thinner, less developed soils that reduce 256
cultivable potential. Elevation was assigned a normalized weight of 26.05% within the Agro-Physiographic AHP matrix, 257
the second most influential topographic criterion after slope. The reclassified elevatio n map comprises five classes (Fig. 258
6): (1) below 150 m (39.14%), rated Very High; (2) 150–300 m ( 13.42%), rated High; (3) 300–600 m (20.57%), rated 259
Medium; (4) 600–1,200 m (23.95%), rated Low; and (5) above 1,200 m (2.92%), rated Very Low, representing the 260
mountainous fringes of the wilaya where agricultural activity is markedly constrained. 261

11

262
Fig. 6: Elevation Map. 263
Distance from Waterbody 264
Proximity to surface water bodies (rivers, wadis, and reservoirs) determines the accessibility of irrigation water, 265
sustains near-channel soil moisture through lateral seepage, and supports loca l groundwater recharge factors that are 266
particularly critical for agricultural productivity under the semi-arid Mediterranean climate of Blida. Parcels located close 267
to watercourses benefit from a more reliable water supply and reduced drought risk, whereas distant parcels depend 268
primarily on rainfed conditions and are more exposed to water stress. Distance from Waterbody carried the lowest weight 269
among the three Agro-Physiographic sub-criteria (Wi = 10.62%), reflecting its comparatively secondary, though still 270
relevant, role relative to slope and elevation. The reclassification produced five buffer-based classes (Fig. 7): (1) less than 271
500 m (7.66%), rated Very High; (2) 500–1,000 m (6.86%), rated High; (3) 1,000–2,000 m (13.92%), rated Medium; (4) 272
2,000–5,000 m (35.18%), rated Low; and (5) beyond 5,000 m (36.39%), rated Very Low, this last class covering the 273
largest share of the study area and underscoring the limited surface-water accessibility across much of the wilaya. 274

12

275
Fig. 7: Distance from Waterbody Map. 276
Agro-Physiographic Suitability 277
The Agro-Physiographic Suitability factor is a composite index derived from the integration of slope, elevation, 278
and distance from waterbody. Rather than describing a single biophysical proces s, this factor synthesizes the terrain's 279
inherent physical capacity to sustain agriculture, independent of dynamic vegetative or thermal conditions and was 280
subsequently reintroduced as a single input within the final five-factor AHP model, where it received a normalized weight 281
of 16.97%. Its reclassified output delineates five suitability zones across the study area (Fig. 8): (1) an Unfavorable Agro-282
Physiographic Zone (3.87%), corresponding to areas of steep slope, high elevation, and/or remoteness from water sources; 283
(2) a Marginally Favorable Zone (24.98%); (3) a Moderately Favorable Zone (20.36%); (4) a Favorable Agro-284
Physiographic Zone (12.75%); and (5) a Highly Favorable Agro-Physiograp hic Zone (38.05%), which occupies the 285
largest proportion of the wilaya and reflects the generally low-relief, water-accessible character of the Metija plain. 286

13

287
Fig. 8: Agro-Physiographic Suitability Map. 288
289
Soil Moisture Index (SMI) 290
The Soil Moisture Index reflects the water content available within the root zone and constitu tes a key 291
determinant of seed germination, plant water stress, and overall drought vulnerab ility, particularly under the semi-arid 292
rainfed conditions that characterize much of the study area. SMI captures short- term soil hydrological status directly 293
relevant to crop establishment and yield potential. Owing to its strong and immediate influence on plant-available water, 294
SMI was assigned the highest weight among the five factors in the final AHP matrix (Wi = 45.02%). The reclassified SMI 295
map is divided into five moisture classes (Fig. 9): (1) 0–0.2, denoting V ery Dry Soil (0.97%) and rated Very Low; (2) 0.2–296
0.4, Dry Soil (49.22%), rated Low and covering nearly half of the study area; (3) 0.4–0.6, Moderately Moist Soil (41.98%), 297
rated Medium; (4) 0.6–0.8, Moist Soil (7.82%), rated High; and (5) 0.8– 1, Saturated / Very Moist Soil (0.005%), rated 298
Very High but occupying a negligible share of the wilaya, largely confined to permanently wet or irrigated patches. 299

14

300
Fig. 9: Soil Moisture Index Map. 301
Normalized Difference Vegetation Index (NDVI) 302
NDVI quantifies vegetation vigor and green biomass density from the contrast between red and near-infrared 303
reflectance and serves as a proxy for current photosynthetic activity and land productivity. Zones of dense, healthy 304
vegetation cover indicate high existing biological productivity favorable to agricultur e, whereas negative or near-zero 305
values correspond to open water, bare soil, or degraded surfaces unsuitable for cropping. NDVI was assigned the second-306
highest weight in the final AHP matrix (Wi = 25.02%), underscoring its importance as a direct indicator of vegetative 307
health. The reclassification yielded five vegetation classes (Fig. 10): (1) values below 0, classified as Water (0.07%) and 308
rated Very Low; (2) 0–0.2, Bare Soil (4.77%), rated Low; (3) 0.2–0.4, Sparse Vegetation (16.20%), rated Medium; (4) 309
0.4–0.6, Moderate Vegetation (37.19%), rated High; and (5) values above 0.6 , Dense Vegetation (41.76%), rated Very 310
High and representing the largest share of the wilaya, consistent with the extensive cropland and forest cover of the region. 311

15

312
Fig. 10: Normalized Difference Vegetation Index Map. 313
Land Use / Land Cover (LULC) 314
LULC identifies the actual current use and cover of the land and is essential for d istinguishing areas already 315
under active cultivation from those structurally unavailable for agriculture (built-up areas, water bodies) or requiring land-316
cover conversion (shrubland, bare ground). Unlike the continuou s biophysical indices, LULC suitability was assigned 317
according to the land-use category itself rather than a numerical gradient. Seven classes were identified (Fig. 11): (1) 318
Water Body (0.05%), rated Very Low; (2) Build Up (9.40%), rated Very Low, reflecting urbanized land unavailable for 319
cultivation; (3) Bare areas (0.02%), rated Low; (4) Shrubland (10.17%), rated Medium; (5) Forest (45.01%), also rated 320
Medium and comprising the single largest land-cover class in the wilaya; (6) Grassland ( 0.42%), rated High; and (7) 321
Cropland (34.94%), rated Very High, representing land already under active agricultural exploitation. LULC was assigned 322
a normalized weight of 8.72% within the final AHP matrix. 323

16

324
Fig. 11: Land Use/ Land Cover Map. 325
Land Surface Temperature (LST) 326
Land Surface Temperature governs evapotranspiration rates and crop heat stress and also served as the basis for 327
deriving the Soil Moisture Index. Moderate- to-warm surface temperatures generally favor photosynthetic activity and 328
crop development, whereas extreme heat suppresses metabolic processes, increases crop wate r demand, and elevates the 329
risk of moisture deficit. LST received the lowest weight among the five factors in the final AHP matrix (Wi = 4.27%), 330
reflecting its comparatively indirect, though non-negligible, influence relative to t he vegetative and moisture-based 331
indicators. The reclassified LST map comprises five thermal classes (Fig. 12): (1) below 25°C, Moderate Surface 332
Temperature (2.64%), rated Medium; (2) 25–30°C, Warm Surface Temperature (19.30%), rated Very High and the most 333
favorable thermal range for crop growth; (3) 30–35°C, Hot Surface Temperature (30.39%), rated High; (4) 35–40°C, Very 334
Hot Surface Temperature (41.69%), rated Low, covering the largest share of th e wilaya; and (5) above 40°C, Extremely 335
Hot Surface Temperature (5.99%), rated Very Low, corresponding to the most thermally stressed surfaces least conducive 336
to agricultural productivity. 337

17

338
Fig. 12: Land Surface Temperature Map. 339
AHP-Derived Weights of the Suitability Factors and Their Classes 340
Tab. 2: Assigned weight and normalized weights of respective classes of factors influencing Agro-Physiographic 341
Suitability. 342
Factor
Factor’s
Normalized
Weights Wi
(%)
Classes Assigned weight
Influence on
Agro-
Physiographic
Suitability Map
Area
(ha)
Area
(%)
Slope 63,33%
< 5% 5 Very High 55137,23 37,24%
5% - 15% 4 High 21524,32 14,54%
15% - 30% 3 Medium 35520,19 23,99%
30% - 50% 2 Low 31517,46 21,29%

50% 1 Very Low 4347,24 2,94%
Elevation 26,05%
< 150m 5 Very High 57951,88 39,14%
150m - 300m 4 High 19866,05 13,42%
300m - 600m 3 Medium 30449,00 20,57%
600m - 1200m 2 Low 35461,12 23,95%
1200m 1 Very Low 4318,37 2,92%
Distance
from
Waterbody
10,62%
< 500m 5 Very High 11337,69 7,66%
500m - 1000m 4 High 10153,02 6,86%
1000m - 2000m 3 Medium 20603,77 13,92%

18

2000m - 5000m 2 Low 52075,77 35,18%

5000m 1 Very Low 53876,17 36,39%
343
Tab. 3: Assigned weight and normalized weights of respective classes of factors influencing Agriculture Potential 344
Zones. 345
Factor
Factor’s
Normalized
Weights
Wi (%)
Classes Assigned weight
Influence on
Agriculture
Potential
Zones
Area
(ha)
Area
(%)
Soil Moisture
Index (SMI) 45,02%
0 - 0,2 : Very Dry Soil 1 Very Low 1439,16 0,97%
0,2 - 0,4 : Dry Soil 2 Low 72870,52 49,22%
0,4 - 0,6 : Moderately Moist Soil 3 Medium 62152,11 41,98%
0,6 - 0,8 : Moist Soil 4 High 11577,38 7,82%
0,8 - 1 : Saturated / Very Moist Soil 5 Very High 7,40 0,01%
Normalized
Difference
Vegetation
Index (NDVI)
25,02%
< 0 : Water 1 Very Low 105,71 0,07%
0 - 0,2 : Bare Soil 2 Low 7066,70 4,77%
0,2 - 0,4 : Sparse Vegetation 3 Medium 23989,89 16,20%
0,4 - 0,6 : Moderate Vegetation 4 High 55059,21 37,19%
0,6 : Dense Vegeration 5 Very High 61824,93 41,76%
Agro
Physiographic
Suitability
16,97%
Unfavorable Agro-Physiographic Zone 1 Very Low 5722,14 3,87%
Marginally Favorable Zone 2 Low 36979,18 24,98%
Moderately Favorable Zone 3 Medium 30143,58 20,36%
Favorable Agro-Physiographic Zone 4 High 18872,37 12,75%
Highly Favorable Agro-Physiographic Zone 5 Very High 56329,15 38,05%
Land Use /
Land Cover
(LULC)
8,72%
Water Body 1 Very Low 71,65 0,05%
Shurbland 3 Medium 15053,95 10,17%
Build Up 1 Very Low 13912,37 9,40%
Grassland 4 High 620,76 0,42%
Forest 3 Medium 66629,92 45,01%
Cropland 5 Very High 51732,31 34,94%
Bare areas 2 Low 25,32 0,02%
Land Surface
Temperature
(LST)
4,27%
< 25°C : Moderate Surface Temperature 3 Medium 3902,65 2,64%
25°C - 30°C : Warm Surface Temperature 5 Very High 28567,48 19,30%
30°C - 35°C : Hot Surface Temperature 4 High 44987,31 30,39%
35°C - 40°C : Very Hot Surface Temperature 2 Very Low 61722,18 41,69%
40°C : Extremely Hot Surface Temperature 1 Low 8866,80 5,99%
346
Validation 347
The ROC curve was used to evaluate the predictive performance of the agricultural potential map derived using 348
the AHP method. The area under the curve (AUC) reached 0.93 (Fig. 13), indicating excellent discriminatory performance 349
of the model. This high AUC value suggests that the AHP-based approach effectively d istinguishes areas with different 350
levels of agricultural potential. Overall, the results provide strong evidence of the model’s predictive capability and 351
support the reliability of the generated agricultural potential map. 352

19

353
Fig. 13: Validation of APZ Map by ROC curve. 354
Discussion 355
Spatial structure of the mapped agricultural potential 356
The final APZ map (Fig. 14) organizes the wilaya into three legible bands. Red, not-suitable pixels cluster along 357
the piedmont corridor that links Blida, Ouled Yaïch, Beni Mered, Boufarik, Larbaa and Meftah, and within the cores of 358
the smaller towns; a dark-green, highly suitable belt runs south-west to no rth-east along the front of the Blidean Atlas, 359
through Chréa, Hammam Melouane and Bougara, and reappears on the eastern hills of Djebabra and Souhane; between 360
the two, a light-green suitable matrix covers the Mitidja plain and the broad int ramontane valleys. At wilaya scale the 361
suitable class dominates (70,179 ha; 47.40 % of 148,046 ha), followed by the moderately (33,351 ha; 22.53 %), highly 362
(18,625 ha; 12.58 %), marginally (16,928 ha; 11.43 %) and not suitable (8,963 ha; 6.05 %) classes. 363
The municipal statistics quantify this geography and expose its asymmetries (Fig. 15-16). The not-suitable class 364
is a corridor phenomenon: the five urbanized communes concentrate 2,461 ha (19.31 % of their 12,744 ha), against 4,139 365
ha (6.86 %) in the 60,304 ha of plain communes and 2,363 ha (3 .15 %) in the 74,998 ha of Atlas communes, with 366
communal maxima in Ouled Yaïch (448 ha; 32.0 %), Beni Mered (365 ha; 24.5 %) and Blida (1,183 ha; 22.3 %). Across 367
municipalities this share varies inversely with communal area (r = −0.61), confirmin g that it traces built-up fabric rather 368
than any biophysical gradient. The highly suitable class follows an opposite, agronomically unexpected gradient: it peaks 369
in Chréa (2,737 ha; 34.1 %), Bouarfa (1,587 ha; 23.3 %), Djebabra (6 92 ha; 23.9 %), Souhane (1,325 ha; 18.5 %) and 370
Chiffa (875 ha; 18.0 %), whereas plain communes range from 204 ha (4.7 %; Ben Khellil) to 1,061 ha (12.5 %; Larbaa); 371
zone totals invert the expected order (Atlas 11,823 ha, 15.76 %; plain 5,675 h a, 9.41 %; corridor 1,127 ha, 8.84 %). The 372
suitable class behaves as anticipated, forming the plain's matrix (28,130 ha; 46.65 %) and the Atlas valley floors (38,034 373
ha; 50.71 %), while the corridor retains only 4,015 ha (31.50 %) and dis plays the wilaya's most fragmented profile 374
(marginally + moderately suitable = 40.35 %). 375
A fourth, finer pattern deserves mention. In the western and south-western com munes (Oued Djer, El Affroun, 376
Mouzaia), the map substitutes orange and yellow mottling for the plain's green matrix: marginally and moderately suitable 377
land reaches 42.0 % (2,551 ha) in Oued Djer, 53.0 % (2,952 ha) in E l Affroun and 43.3 % (3,620 ha) in Mouzaia, 378
delineating the hilly transition between the Mitidja and the Chelif-margin reliefs, where rainfed cereals and scrub replace 379
irrigated orchards. Within the green matrix itself, the scattered red speckle that textures the plain corresponds to hamlets, 380

20

farms and linear infrastructure, the municipal counterpart of the 13,912 ha (9.40 %) of built-up land recorded in the LULC 381
criteria. 382
Three readings follow. The plain's suitable matrix is consistent with the intensive cropland of the Mitidja and 383
with the documented peri-urban pressure on its eastern sector (Bellout et al., 2 020). The corridor's not-suitable signal 384
expresses land availability rather than capacity: the same communes still hold 4, 015 ha of suitable land on some of the 385
wilaya's best alluvial soils, so their low classes act as an artificial-surface mask superimposed on an otherwise favorable 386
biophysical setting, a distinction land-evaluation theory requires to be kept explicit (FAO, 1976; Malczewski, 2004). The 387
Atlas concentration of the top class is the most contestable feature of the map: with SMI and NDVI carrying 70.04 % of 388
the total weight while terrain constraints enter only through the agro-physiographic f actor (16.97 %, of which slope is 389
63.33 %, i.e. ≈10.7 % effective weight), moist, densely vegetated forest slopes sco re highest even where steep gradients 390
preclude mechanized cropping; in a conservation landscape such as Chréa National Par k, high NDVI and SMI values 391
express ecosystem integrity rather than cropping capability. This inversion between plain and mountains is therefore best 392
treated as a testable hypothesis of the weighting scheme, not as an established result. 393
394
Fig. 14: Agricultural Potential Zones (APZ) Map. 395
396

21

397
Fig. 15: Municipal distribution of the APZ Map. 398
399
Fig. 16: Area-weighted APZ class composition of the three geographical zones. 400
Synthesis: what the validation demonstrates and what remains to be tested 401
The AUC of 0.935 demonstrates that the APZ index separates reference positives from ne gatives strongly and 402
consistently across all four class thresholds, which supports the internal log ic of the weighted combination and the 403
usability of the map as a screening instrument for territorial planning. It does not, in its present aggregated form, establish 404
per-class accuracy, statistical precision, or transferability beyond the calibration scene. Three developments would close 405
these gaps: point-level reference data with spatially blocked cross-validation and boo tstrap confidence intervals; a 406
confusion-matrix accuracy assessment of the five classes; and one-at-a-time weight perturbation to test the stability of the 407
Atlas top-class concentration. Crop-specific evaluations and multi-temporal SMI/NDVI com posites would further 408

22

decouple transient vegetation state from enduring capability. The broader metho dological lesson aligns with 409
Mediterranean suitability studies in which moisture and vegetation indices dominate expert weigh tings (Faraslis et al., 410
2023; Zhu et al., 2022): high discrimination statistics certify that the index rep roduces the environmental gradients its 411
criteria encode, not that every gradient is agronomically meaningful at every elevatio n. For planning, the map already 412
delivers two robust messages: the corridor's 4,015 ha of remaining suitable land is a finite reserve requiring explicit 413
protection against the sprawl that accounts for the 2,461 ha of not-su itable land mapped in the same communes, and the 414
plain's suitable matrix, not the mountain top class, is the agronomic backbone of the wilaya. Until the weighting hypothesis 415
of the first section is tested, the highly suitable belt of the Blidean Atlas should be interpreted as an indicator of favorable 416
moisture-vegetation conditions deserving conservation-oriented management, rather than as a mandate for agricultural 417
intensification on steep terrain. Read together, the ROC diagnostic, the municipal statistics and the map itself thus provide 418
a coherent, explicitly bounded evidence base: strong internal discrimination, a geographically legible class structure, and 419
a clearly identified agenda of soil, temporal and field validation before the APZ map is used for crop-allocation decisions. 420
Conclusion 421
This study developed and validated an integrated GIS Remote Sensing and AHP framewo rk for agricultural 422
potential zoning in Blida Wilaya, northern Algeria, a representative semi-arid Mediterranea n landscape subject to 423
compounded climatic and urban pressures. Five spatial criteria: the Soil Moisture Index , NDVI, Land Surface 424
Temperature, Land Use/Land Cover, and a composite Agro-Physiographic Suitability factor integrating slope, elevation, 425
and distance from water bodies were harmonized at 30 m resolution, weighted through the AHP eigenvector method, and 426
aggregated by weighted linear combination into a five-class Agricultural Potential Zone map. 427
The spatial configuration of the map is geographically coherent: the suitable class forms the agronomic matrix 428
of the Mitidja plain and the intramontane valleys; the not-suitable class traces the urba nized piedmont corridor linking 429
Blida, Ouled Yaïch, Beni Mered, Boufarik, Larbaa, and Meftah; and the highly suitable class concentrates along the 430
Blidean Atlas front (Chréa, Hammam Melouane, Bougara) and the eastern h ills (Djebabra, Souhane). ROC validation 431
yielded an AUC of 0.935, confirming excellent discriminatory performance a nd supporting the map's usability as a 432
screening instrument for territorial planning. Two robust planning messages follow: the 4,015 ha of suitable residual land 433
within the urbanized corridor constitutes a finite reserve of prime alluvial soils requiring explicit protection against urban 434
sprawl, and the plain's suitable matrix - rather than the mountain top class - is the agronomic backbone of the wilaya. The 435
concentration of the highly suitable class in the Atlas should be interpreted cautiously , as it reflects a weighting scheme 436
dominated by moisture and vegetation indices and may therefore signal ecosystem in tegrity and favorable hydro-437
vegetative conditions in forested terrain rather than mechanizable cropping capability; t hese areas merit conservation-438
oriented management until confirmed by field-level assessment. 439
Despite its strong internal discrimination, the model remains bounded by the agg regated form of its validation, 440
and the per-class accuracy, statistical precision, and transferability of the APZ map beyond the calibration scene have yet 441
to be established. Future work should incorporate point-level reference data with s patially blocked cross-validation and 442
bootstrap confidence intervals, confusion-matrix accuracy assessment of the fiv e suitability classes, and one- at-a-time 443
weight perturbation to test the stability of the Atlas concentration, together with crop-s pecific evaluations and multi-444
temporal SMI/NDVI composites to decouple transient vegetation states from enduring lan d capability. Within these 445
explicitly delimited bounds, the APZ map furnishes a transparent, reproducible, and geographically legible evidence base 446
for sustainable agricultural land-use planning in Blida and, more broadly, in semi-arid Mediterranean regions of North 447
Africa. 448
449
450
451
452
453

23

References 454
AbdelRahman, M. A. E., & Arafat, S. M. (2020). An approach of agricultural courses for soil conservation based on crop 455
soil suitability using geomatics. Earth Systems and Environment, 4(1), 273–285. https://doi.org/10.1007/s41748-456
020-00145-x 457
Anusha, B. N., Raghu Babu, K., Pradeep Kumar, B., Padma Sree, P., Vee raswamy, G., Swarnapriya, C., & Rajasekhar, 458
M. (2023). Integrated studies for land suitability analysis towards sustainable agricultural development in semi-459
arid regions of AP , India. Geosystems and Geoenvironment, 2(2), 1001 31. 460
https://doi.org/10.1016/j.geogeo.2022.100131 461
Bellout, A., Vaz, E., & Penfound, E. (2020). Rethinking agricultural land use in Algiers: A spatial analysis of the Eastern 462
Mitidja Plain. Habitat International, 104, 102239. https://doi.org/10.1016/j.habitatint.2020.102239 463
Binte Mostafiz, R., Noguchi, R., & Ahamed, T. (2021). Agricultural Land Suitability Assessment Using Satellite Remote 464
Sensing-Derived Soil-Vegetation Indices. Land, 10(2), 223. https://doi.org/10.3390/land10020223 465
Cetin M, Isik Pekkan Ö, Ozenen Kavlak M, et al (2022) GIS-based forest fire risk determination for Milas district, Turkey. 466
Natural Hazards. https://doi.org/10.1007/s11069-022-05601-7 467
Chemirik, C. H. K., Baahmed, D., Nedjai, R., Boudjemline, D. & Mahcer, I. (2024). Mapping Groundwater Potential 468
Zones with GIS-RS-AHP Under Climate Change: Case of Mostaganem Plateau, Northwest Algeria. Ecological 469
Engineering & Environmental Technology, 25(6), 72–89. https://doi.org/10.12912/27197050/186667 470
Choudhary, K., Boori, M. S., Shi, W., Valiev, A., & Kupriyanov, A. (20 23). Agricultural land suitability assessment for 471
sustainable development using remote sensing techniques with analytic hierarchy proce ss. Remote Sensing 472
Applications: Society and Environment, 32, 101051. https://doi.org/10.1016/j.rsase.2023.101051 473
Donmez, E., Yilmaz, M. T., & Yucel, I. (2024). Added utility of temp erature zone information in remote sensing-based 474
large scale crop mapping. Remote Sensing Applications: Society and Environment, 35, 101264. 475
https://doi.org/10.1016/j.rsase.2024.101264 476
Escobar, M. T., Aguaron, J., & Moreno-Jimenez, J. M. (2004). A note on AHP group consistency for the row geometric 477
mean priorization procedure. European Journal of Operational Research, 153( 2), 318-322. 478
https://doi.org/10.1016/S0377-2217(03)00166-0 479
ESRI (2016). ArcGIS Desktop: Release 10.8 Redlands, CA: Environmental Systems Research Institute. 480
FAO (1976). A Framework for Land Evaluation. FAO Soils Bulletin 32, Rome . 481
https://www.fao.org/3/x5310e/x5310e00.htm 482
Faraslis, I., Dalezios, N. R., Alpanakis, N., Tziatzios, G. A., Spiliotopoulos, M., Sakellariou, S., Sidiropoulos, P., Dercas, 483
N., Domínguez, A., Martínez-López, J. A., López-Urrea, R., Karam, F., Amami, H., & Nciri, R. (2023). Remotely 484
Sensed Agroclimatic Classification and Zoning in Water-Limited Mediterranean Areas towar ds Sustainable 485
Agriculture. Remote Sensing, 15(24), 5720. https://doi.org/10.3390/rs15245720 486
Jesus, j. B. D., & santana, i. D. M. (2017). Estimation of land surface temperature in caatinga area using landsat 8 data . 487
Journal of Hyperspectral Remote Sensing, 7(3), 150–157. https://doi.org/10.29150/jhrs.v7.3.p150-157 488
Klein, J. (2019). Calculating the AHP priority vector. Software Engineering Institute, Carnegie Mellon University. 489
Madani AZ, Hermassi T, Taibi S, Dakhlaoui H and Mechergui M (2024) Climate change imp acts on the Chiffa basin 490
(northern Algeria) using bias-corrected RCM data. Front. Water 6:1507961. doi: 10.3389/frwa.2024.1507961 491
Machireddy, S. R. (2025). Geospatial technology approach for soil moistur e estimation in Kurnool District, Andhra 492
Pradesh, using NDVI and LST. Sensing and Imaging, 26(1), Article 62. http s://doi.org/10.1007/s11220-025-493
00598-2 494
Malczewski, J. (2004). GIS-based land-use suitability analysis: A critical overview. Pro gress in Planning, 62(1), 3–65. 495
https://doi.org/10.1016/j.progress.2003.09.002 496
Mihi, A., & Benaradj, A. (2022). Assessing and mapping wind erosion-prone areas in Northeastern Algeria using additive 497
linear model, fuzzy logic, multicriteria, GIS, and remote sensing. Environmental Earth Sciences, 81, 47. 498
https://doi.org/10.1007/s12665-021-10154-2 499
Mumcu Küçüker, D., & Cedano Giraldo, D. (2022). Assessment of soil erosion risk using an integrated approach of GIS 500
and Analytic Hierarchy Process (AHP) in Erzurum, Türkiye. Ecological Informatics, 71, 101788. 501
https://doi.org/10.1016/j.ecoinf.2022.101788 502

24

ONS – Office National des Statistiques (2008). Recensement Général de la Population et de l’Habitat 2008 (RGPH 2008). 503
Algiers: ONS. https://www.ons.dz/spip.php?rubrique43 504
Saaty, T. L. (1977). A scaling method for priorities in hierarchical structures. Journal of Mathematical Psychology, 15(3), 505
234-281. https://doi.org/10.1016/0022-2496(77)90033-5 506
Saaty, T. L. (1980). The analytic hierarchy process: Planning, priority setting, resource allocation. McGraw-Hill. 507
Saaty, T. L. (1987). The analytic hierarchy process - what it is and how it is used. Mathematical Modelling, 9(3-5), 161-508
176. 509
Saaty, T. L. (1990). How to make a decision: The analytic hierarchy process. European Journal of Operational Research, 510
48(1), 9-26. https://doi.org/10.1016/0377-2217(90)90057-I 511
Saaty, T. L. (2003). Decision-making with the AHP: Why is the principal eigenvecto r necessary? European Journal of 512
Operational Research, 145(1), 85-91. https://doi.org/10.1016/S0377-2217(02)00227-8 513
Saaty, T.L. (1980) The Analytic Hierarchy Process: Planning, Priority Setting, Resources Allocation. Mcgraw-Hill, New 514
York. 515
Saha, A., Patil, M., Goyal, V . C., & Rathore, D. S. (2019). Assessment and Impact of Soil Moisture Index in Agricultural 516
Drought Estimation Using Remote Sensing and GIS Techniques. Proceedings, 7(1), 2. 517
https://doi.org/10.3390/ECWS-3-05802 518
Shinde, S. P., Barai, V . N., Gavit, B. K., et al. (2024). Assessment of groundwater potential zone mapping for semi-arid 519
environment areas using AHP and MIF techniques. Environmental Sciences Europe, 36, 87. 520
https://doi.org/10.1186/s12302-024-00906-9 521
NESCO-MAB (2025). Chréa Biosphere Reserve. UNESCO Man and the Biosphere Programme. 522
https://www.unesco.org/en/mab/chrea 523
USGS - United States Geological Survey (2025). Digital Elevation Model (DEM) 30m × 30m resolution. Available at: 524
https://earthexplorer.usgs.gov/. Last accessed: 20/11/2025 525
Wijesinghe, D. C. (2024). GIS-based AHP and MCDA Modeling for Cropland Suitability Analysis: A Bibliometric 526
Analysis. Gazi University Journal of Science Part A: Engineering and Innovation, 11(3), 598-621. 527
https://doi.org/10.54287/gujsa.1510527 528
Zhu, X., Xiao, G., & Wang, S. (2022). Suitability evaluation of potential arable land in the Mediterranean region. Journal 529
of Environmental Management, 313, 115011. https://doi.org/10.1016/j.jenvman.2022.115011 530

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.
tab.1.docx
tab.2.docx
tab.3.docx

论文原文与出版信息

Research Square
查看论文原文

继续探索