Assessing and mapping wind erosion-prone areas in northeastern Algeria 1 using additive linear model, AHP, FAHP approaches, GIS, and medium 2 resolution multisource remotely sensed data

Wind erosion is one of the most severe environmental problems in arid, semi-arid and dry 26 sub-humid regions of the planet. This dissertation aimed to identify areas sensitive to wind erosion in northeastern Algeria (Wilaya of Tebessa) based on empirical model using analytic hierarchy process (AHP), fuzzy analytic hierarchy process (FAHP) approaches, and geomatics- based techniques. Sixteen causative factors were used incorporating meteorological, soil erodibility, physical environment, and anthropogenic impacts as main available inputs in this approach. Weighted linear combination (WLC) algorithm was used to combine all standardized raster layers. Area under curve (AUC) value equal 0.96 indicate an excellent accuracy for the proposed approach. Globally, wind erosion risk increase gradually from the north to south of 34 the whole area. Besides, it was found that areas with slight, moderate, high, and very high 35 covered 9.65 %, 25.83 %, 24.30 %., and 40.22 %, respectively of the total. Our results 36 highlighted the potential of additive linear model, and free available medium resolution multi- 37 source remote sensing data in studying natural hazards and disasters mainly under data-scarce 38 or areas of difficult access in developing countries. In addition, restoration and re-vegetation 39 activities of sensitive areas at high risk of wind erosion represent a challenge for researchers 40 and decision-makers. 41


45
Soil is essentially a non-renewable resource on the human time scales. It is dynamic is 46 prone to rapid degradation depending on land misuse and soil mismanagement (Blanco and Lal, 47 dust concentration occurs principally at the south of the Algerian saharan atlas region (Shao,81 2008). As result, about 20 Mha are threatened by wind erosion in Algeria (Bensaïd, 2006). 82 Covering a total area of 20 million ha, Algerian steppes are the most 83 widely distributed rangeland types in the North African countries (Hirche et al., 2011). In 84 Algeria, change in climatic and soil conditions make the steppe a fragile environment. 85 Unfortunately, about 600 000 ha of Algerian steppes have been degradation by wind erosion 86 (Bensaïd, 2006). 87 Wind erosion phenomenon is complex process and cannot be directly measured soils losses at 88 a field scale (Webb et al., 2009). To assess wind erosion land susceptibility, there are numerous 89 empirical wind-erosion modelling, change in its complexity, input data required, model outputs variables. Furthermore, most of these models are designed to estimate wind erosion risk at a 115 local scale, as well as, it's scaled up into a regional-scale still discutable and not fully consistent. 116 Against this background, the aim of this dissertation is: (i) the quantitative estimate of the wind 117 erosion vulnerability at a regional scale in the whole area of Tebessa (NE Algeria) using WLC 118 algorithm ,AHP, fuzzy AHP (FAHP) approaches, GIS, and remote sensing techniques (ii) to 119 prove the potential of using available medium resolution multi-source remote sensing data for 120 monitoring and assessing natural hazard, (iii) to provides scholars, policy makers, planner, and 121 expert of environment specialist an accurate maps of wind erosion sensitivity, in order to adopt 122 the best strategies in land restoration and wind erosion control programs.  (Fig. 6f). Seasonally, the high wind speeds values are recorded in spring season. In fact, the 137 most frequent surface winds blow mainly from the North and Northwest (46 %) bringing rain 138 during the wet season, and soften the climate by reducing hot weather during the summer 139 months, although other wind directions are southeast and southwest (32%), this winds called 140 sirocco are hot and dry and blows from the Sahara to the north in summer period (Seltzer, 1946).

141
In fact, the areas represent the mountain morphologies (Aurès Nemamcha Mountains in the 142 north of wilaya) act as barriers in the way of sirocco wind, which make the north of Wilaya not 143 very exposed to this wind type. Additional to the livestock farming of sheep, agriculture is the 144 main human activity with ~ 27.7% of the whole area (irrigated and rainfed crop cultivation).

145
The naturel vegetation is dominated by steppe type with 30%. Six main soil types are 146 distinguished in the study area-calcic, calcareous, basic alluvial, saline soil, and aeolian soil  The most important step for modelling data is to choose the appropriate existing factors that 154 could be adopted in multi criteria decision analysis (MCDA). In our study, it is almost 155 impossible to consider the entire database for the establishment of the sensitivity map because 156 the criteria involved in the wind erosion phenomenon are innumerable and make its modelling 157 very delicate, as well as, availability of input required data to use in the model represents also 158 a limiting factor in the multi criteria analysis process. To overcome the problem of data scarcity, 159 it is necessary to select only the relevant criteria for the analysis. To address these needs, the

216
All the mapped inputs data ware reclassified according human impact characteristics.

217
Meanwhile, areas having more anthropic pressure are considered as optimal for wind erosion 218 occurrence.     In the present study, desert steppe was considered the worst type of vegetation cover for wind 279 erosion risk, followed by steppe, and after that forest steppe. represent stable and consolidated soil (Karnieli, 1997 Where P is total annual precipitation in mm and ETP is mean annual evapotranspiration in mm, 313 where AI climatic classification is a unitless values and varies from 0 to 1 (UNEP, 1992). In 314 this present study, AI is reclassified in two classes or climates ( Application of analytical hierarchy process (AHP) 320 The methodological workflow of the study is summarized as a flowchart in Fig. 2. Thomas  refer to the relative importance of each factor with respect to another to determine its weight.

341
In this way, the importance of each factor is determined by the weight that it is assigned, whose Where λmax and n are the largest eigenvalue and order of matrix, respectively. RI is the 366 Random Index depending on the order of the matrix given by Saaty (1980) as displayed in Table   367 2. CR changes from 0 and 1. If the CR index is greater than 0.10, the preference matrix should 368 be revised. In this present study, CRAPSE, CRSE, and CRFWER equals to 0.02, 0.02, and 0.03, 369 respectively, expressing a satisfactory consistency of judgments (Table 4).

370
Application of fuzzy analytical hierarchy process (FAHP) 371 In fact, Zadeh (1965) firstly introduced fuzzy logic and its basic idea is to consider the 372 spatial objects on a map as members of a set to deal with ambiguities and uncertainties that 373 usually exist in human judgment. In prior research, fuzzy logic represent a powerful approach,

401
Where l, m, and u represent the lower, mean, and upper bounds, respectively, of the fuzzy 402 number Ã as illustrated in Fig. 3  Where: Ã represent n*n-fuzzy judgment matrix containing TFN, ã =(lij,mij,uij), ã = 1 ã ⁄ , 426 for , = 1, … . , and ≠ , n is the total number of criteria. Afterward, geometric mean 427 method (Buckley, 1985) was used to define the fuzzy geometrical mean and fuzzy weights of Ideally, prediction without knowing data accuracy and precision are a little use in science.

481
For that reason, sand dunes areas was mapped and checked starting from high-resolution image 482 (Google Earth) and field data verification (Fig. 9a-f). Here, the choice of verification stations 483 were greatly depends on the following criteria including proximity and accessibility in terms of 484 roads, altitude, aspect, security, and data representativeness for the whole of study area. Here,

500
The DUA, DR, SG, LD, PD, and FP maps (Fig. 4a-f and Table 5) are used to draw the map 501 of anthropic pressure on the steppe environment parameter in Tebessa area. APSE map were 502 grouped into four classes to define areas most exposed to human impact along the study area.

503
The first class corresponds to slight sensitivity, which distributed exclusively in the southern 504 part, occupying 5.18 % of the whole area. The second class belongs to moderate sensitivity, 505 representing 10.28 % of the total area, and located in the south. The third class presents high 506 sensitivity, corresponding to 33.68 % of the entire area. The fourth class represents very high 507 sensitivity, which covered the northern part, with 50.87 % of the study area ( Fig.6a and Table   508 5).

509
Outs of soil erodibility factor of the study area is illustrated in Fig. 6b and Table 5. Generally, four factor in each class of soil erodibility parameter is summarized in Fig. 5a-d and Table 5. 515 Fig. 6c and Table 5 show that the majority of the vegetation in study area was highly vulnerable 516 to wind erosion since main vegetation consists of desert steppe and steppe with a cover of 52.21 517 % and 45.37 %, respectively. Very few areas have good vegetation density including forest 518 steppe (2.42 %), and water corps (0.01%). These areas are considered as the less vulnerable, 519 and located in the northern of the study area.

520
Map of sand movement extent show high sensitivity to wind erosion in the far south of the land 521 area, due to the presence of sandy soils in these areas ( Fig.6d and  Fig.6f and Table 5 depicts AI for 10 years (2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017)(2018). In general, wind erosion increase with 534 decreasing in water availability in a region. In short, dryness increase from the north to south 535 in the study area. The results indicate that almost the whole region of Tebessa is classified as 536 "Arid" with 63.08% ,except the north of study area, which are classified as "Semi-arid" with 537 rate of 36.92%.

538
Vulnerability maps of causal factors were obtained based on the above-described relations.

577
Wind erosion remains moderate in the north part. This part is characterized by a semi-arid 578 climate, moderate wind intensity, favourable soil structure, absence of anthropogenic activities, 579 and relatively dense vegetal cover. In essence, when the vegetation cover is less than 35%, wind 580 intensity is deemed as the main agent for wind erosion (Bensaïd, 2006). In reality, wind speed  The methodology proposed by this current study is a significant contribution to generalize the 617 wind-erosion risk assessment especially at steppe regions. Furthermore, the proposed approach 618 based on medium resolution satellite data available free is considered to be cost-efficient, 619 simple, and easily to apply for monitoring wind erosion risk. Although the method applied in