3.1.5. Vegetation Cover: The Presence/Absence Effect of Ponds

This factor is the focal point of this study since by slightly changing this parameter the outcome changes significantly. At this point, two vegetative cover scenarios are presented: the first presents the actual setting of the Claise accounting for the presence of ponds, while the second simulates a scenario where ponds are removed to assess the difference in erosion outcomes with and without their presence. This last step allowed to quantify the impact of ponds on erosion in the Claise basin. In the second scenario, the land occupation group "ponds" was changed to their surrounding class (grasslands).

The large area of ponds, which makes up around 11% of the Claise (Table 3), displays their potential role as modifiers basin processes. Figure 7 presents the two considered vegetation covers as inputs for crossing with the potential soil erosion risk map to yield two different actual soil erosion risk maps. These maps were then compared to evaluate the effect of the presence and absence of ponds and their role on erosion.

**Figure 7.** (**a**) Current Claise vegetation cover and (**b**) alternative vegetation cover and corresponding CORINE classification; 1: fully protected, 2: not fully protected.

Class 0 refers to land cover categories that are not considered in CORINE; these categories are urban areas and water bodies, while classes 1 and 2 refer to the fully protected and not fully protected covers. From Figure 7b, 72.5% of the Claise basin corresponds to the fully protected class, while 27.5% of the study area is occupied by not fully protected cover. To assess the impact of pond presence, the second scenario of replacing ponds by their surrounding dominant cover was performed.

3.1.6. Actual Soil Erosion Maps Under Current and Alternative Scenarios

The two actual soil erosion risks maps were produced by multiplying the respective indices of the potential soil erosion risk map and the two vegetation cover scenarios using the "raster calculator" tool. Figure 8 reveals the outcome under both scenarios.

**Figure 8.** Actual soil erosion risk map of the Claise under (**a**) current vegetation cover and (**b**) the alternative pondless scenario.

The Actual soil erosion map was cross-checked against: Institut National de la Recherche Agronomique (INRA) 2000 erosion maps and the combined GIS sol–INRA–SOeS 2011 maps [80]. These were produced following the Modèle d'Évaluation Spatiale de l'Aléa d'Érosion des Sols (MESALES). The comparison between the INRA maps, and the produced actual soil erosion risk map, is presented in Table 6. By this comparison, it is concluded that there is a good agreement between these maps. In addition, since INRA maps are not completely adequate to be considered at the basin scale [80], the established erosion maps are considered for the no-erosion zones, overcoming, this way, the challenge of coarse representation.


**Table 6.** Verification of the established erosion map.

As can be seen from Table 6, a total of 134 validation points were chosen. These were divided into 86, 37, and 11 low-erosion, moderate-erosion, and high-erosion zones validation points, respectively. A large part of the moderate-erosion class was misinterpreted as the low-erosion class. This discrepancy is due to the fact that the no-erosion zones do not exist in the INRA maps, but are instead classified as low-erosion zones. Therefore, the error margin in the moderate-erosion class from Table 6 is justified by the finer scale representation of the produced maps, compared to the INRA maps. The overall

accuracy was determined to be 75%, while the computed Cohen's kappa coefficient [81] was found to be 0.7; this indicates a substantial agreement between the INRA maps, and the produced actual soil erosion risk map. The kappa coefficient was used since it tests inter-rater reliability; i.e., the coefficient represents the extent to which the generated data are correct representations of the measured data. In the case of this study, the generated data is the actual soil erosion risk map, while the measured data consists of the validation points obtained from the INRA maps.

From Figure 8, three main results can be drawn:

(1) The role of vegetation cover in changing erosion risks is solidified. This is particularly reflected by the setting of the Claise basin due to the agricultural and grass cover. By comparing the actual soil erosion risk map with the potential soil erosion risk map and statistics, a shift of erosion risk classes is observed. As mentioned in Section 3.1.4, considering the potential soil erosion risks, and ignoring the vegetation cover, low, moderate and high-risk areas take over 3%, 96.5% and 0.5% of the total basin area, accordingly. When the vegetation cover layer was taken into account the resulting actual soil erosion risk shifted to 65.66%, 21.68% and 0.18%, for low, moderate and high-risk areas. These observations solidify that vegetative cover is the most influential aspect for erosion assessment. In further detail regarding the vegetation cover layer, areas corresponding to agricultural classes are seen to have higher erosion risks than areas with different land cover types. This is in agreement with Verheijen et al. (2009) [82] observations that despite the considerable effect of soil type, topography and climatic conditions, the major influencer of soil erosion is the vegetative cover, especially cultivated areas.

(2) The remainder 12.48% of the actual soil erosion risk map is the no-erosion zone. As seen in Figure 7a, most of the no-erosion zone corresponds to the concentration area of ponds, while the remainder 1.48% represents the Claise River. The ponded area represents 88.23 km<sup>2</sup> of the Claise under no risk of erosion, making these ponds a counter-erosion zone.

(3) At a graphical scale, a complete shift from low to moderate risks, in the greatest part of the basin, is observed. Table 7 presents the statistical difference between the actual soil erosion risks with current vegetation cover and those of the pondless scenario. From Table 7, the effective role of ponds as an erosion counter-measure is revealed.


**Table 7.** Actual soil erosion risk (ASE) for the Claise basin under current and simulated vegetation cover.

Additionally, the impact of ponds on erosion at the scale of the basin is revealed. Not only did the no-erosion and low-erosion classes decrease by 11.36% and 65.14%, in the absence of ponds scenario, but also the moderate and high-erosion risks increased by 55.12% and 21.38%, respectively. These changes are due to several reasons:

(1) The most evident reason is that ponds effectively and directly nullify splash erosion in the areas they occupy.

(2) Their widespread, yet dense, positioning throughout the basin, counteracts runoff erosion in a twofold way: first, by intercepting eroded soils by overland flow, retaining this way, the transported material and preventing them from reaching the streams, and second, by slowing surface runoff and thus, abating its erosive force. Despite the fact that the low-slope topography does not particularly favor runoff erosion, let alone high velocities of overland flow, this obviously has some effect, especially in cases of intense rainfall events.

(3) Their dense aggregation in the basin attributes them the role of cascade check dams, containing sediments.

(4) Their large density and chain sequence where the retention effect is greatly amplified (factor of 2179 ponds) [38].

(5) The highly erodible setting of the basin resulting from a challenging pedology.
