**4. Discussion**

The assessment of soil sensitivity to water erosion using a multi-criteria approach calls on several criteria which, when put together, makes it possible to establish a balance sheet and a diagnosis of the physical degradation of soils in a given area. This study employed five criteria (slope, land use, lithology, drainage density, and precipitation) whose weighting was ensured by the AHP method. This method envisages a good understanding of the criteria involved in the erosion process, based on the pair-wise comparison of criteria at the same hierarchical level [28]. Land use was the most determining criterion in the process. Results obtained from the weighting of the criteria corroborate those of Refs. [29–31], in which plant cover protects the soil against ablation and reduces risk of erosion. Vegetation, therefore, acts as a protective screen against aggressive climatic conditions by intercepting the energy released by raindrops [32]. The topographic factor through the rigidity of the slope accentuates the erosive force of sheets' flowing water. But its behavior with respect to infiltration depends on the type of soils [33].

*Modeling of Soil Sensitivity to Erosion Using the Analytic Hierarchical Process: A Study… DOI: http://dx.doi.org/10.5772/intechopen.111742*

**Figure 5.** *Overlay of the different information planes (model & field data).*

However, the model produced on the basis of these five criteria after validation gave satisfaction at a rate of 73.87%. The non-satisfaction of the model (23.13%) stems from the input data used in this study. They are mostly medium resolutions. One of the parameters whose imprecision impacted the final result was land use. Indeed, the satellite image used to establish the land cover map has a spatial resolution of 30 m x 30 m. This spatial resolution certainly did not allow more detailed observation of the earth's surface. The consequence of this imprecision on the result of this study is that the soils which could have been identified as less sensitive to water erosion were not, and *vice versa*. Forests with small dimensions may not have been taken into account in the analyses. The topographic data used to produce the slope and drainage density maps were generated with a scale of 1/50,000, which does not provide excellent precision. These inaccuracies in the model's input data can therefore influence the final result obtained in this study.

Moreover, the AHP method used for the weighting of the criteria, although relatively efficient, presents difficulties. One of the difficulties of this method concerns the subjectivity in the choice of the scale of scores ranging from 1 to 9 with their reciprocal correspondence [34]. The choice of the score corresponding to a criterion is arbitrary and may influence the calculation of the weight of the criterion considered. However, this method (AHP) is appropriate for this study because it performs well when the number of actions is reduced [35]. Nevertheless, despite these imprecisions on the input data, the quality of the model proves to be satisfactory for simulating the sensitivity of soils to erosion, which is necessary for any spatial planning study. It could be improved through the acquisition of high spatial resolution data.
