**1. Introduction**

Shallow landslides induced by rainfall are very common phenomena that occur in hilly and mountainous areas, causing loss of human life and environmental and economic damage [1,2]. The main triggering factors of shallow landslides are represented by intense or prolonged rainfall [3–5], while the main predisposing factors are represented by lithology, morphology [6,7], and soil conditions—such as land cover and land use [8,9]. In particular, land use is constantly evolving, and its changes affect landslide occurrence [10–14]. Indeed, changes in vegetation cover have an impact on the landscape diversity [15] and relevant effects on the hydrological processes and mechanical structure of the soil, with either positive or negative consequences for slope stability [16–19]. Moreover, agricultural practices, on one hand, contribute positively to landscape and on the other hand, can have a negative impact on the slope stability. As an example, unsustainable agricultural practices characterized by heavy mechanization, such as soil tillage, can generate an excessive pressure on the soil, making the soil more susceptible to instability and degradation phenomena [20].

Several scientific contributions, among others, show the positive effects of vegetation cover on slope stability [21–25]: from a hydrological point of view, the vegetation dissipates most of the kinetic energy of the raindrops, weakening the erosion action, with a degree of interception depending on the density of the leaves and the size of the plant [26]; from a

**Citation:** Volpe, E.; Gariano, S.L.; Ardizzone, F.; Fiorucci, F.; Salciarini, D. A Heuristic Method to Evaluate the Effect of Soil Tillage on Slope Stability: A Pilot Case in Central Italy. *Land* **2022**, *11*, 912. https://doi.org/ 10.3390/land11060912

Academic Editors: Matej Vojtek, Andrea Petroselli and RaffaelePelorosso

Received: 29 April 2022 Accepted: 12 June 2022 Published: 15 June 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

geo-mechanical point of view, the most important effect is represented by the mechanical reinforcement exerted by the roots [22,27–30], consisting of an increase in the shear strength, included in the Mohr–Coulomb failure criterion as additional soil cohesion [31,32]. In particular, the main contribution of roots to soil strength is related to the presence of small roots in the most superficial soil layers, which increase the compound matrix strength. Such an effect has been largely known as root reinforcement [33]. Another action involves large roots which intersect the shear surface and mobilize a soil–root friction force instead of the entire tensile strength [14]. Moreover, root reinforcement depends on the density of roots in soils and root diameters. As an example, grasses provide significant reinforcement to the shallower layers of soil, while woody roots of trees and shrubs provide reinforcement over a greater depth of soil through a combination of both fine, fibrous roots and coarser, woody roots [34].

Conversely, the effects of agricultural practices, and in particular of soil tillage, on slope stability conditions are poorly investigated. In this case, some studies show that the modification of soil's mechanical properties can be related to the tools used to till the land. In particular, some studies investigated how the soil characteristics vary in response to practices with the aim of assessing the efficiency of agricultural machinery, e.g., [35]. As an example, tillage with a rotary paraplow can be considered as a conservation technique leaving the soil state unchanged [36]. Other works, e.g., [37], show that the effect of tillage tools on the soil depends on the nature of the soil (fine-grained or coarse-grained soil). In the case of paddy soils, different laboratory experiments were conducted to determine the effect of tillage operations on the soil's physical, rheological and mechanical properties [38,39]. A negative impact of the tillage operation was found in soil bulk density, while a change in the rheological behavior of paddy soil according to the variation of the moisture content was observed [38]. Some authors have performed field-scale analyses aimed at evaluating the effect of different tillage techniques (conventional or conservation) on soil erosion, also in comparison with non-tilled cases [40–42]. Overall, considerable increases in soil erosion, runoff, and sediment loss are observed in the cases with conventional tillage as compared with the cases with soil saving technologies (conservation tillage or no-tillage) [41,42]. On the other hand, no significant differences in hydraulic conductivity were observed [40]. Recently, Straffelini and co-authors proposed a physical modeling approach to assess runoff and soil erosion in vineyards under different soil managements and observed that continuous tillage aggravated soil erosion as compared to reference tillage, single tillage, and nectariferous [43].

To our knowledge, no articles regarding particularly the impact of soil tillage on slope stability and shallow landslide occurrence in hilly environments are currently present in the literature. Indeed, most studies have focused on steep, often terraced landscapes [see, e.g., [20] for further references] or on erosion hazards, e.g., [41,42]. Quantitative measurements of the link between tillage operations and soil mechanical characteristics are not straightforward. However, there is a general consensus that tillage, plowing, and leveling generate a modification of the mechanical properties of the soils involved, leading to a possible increase in the propensity to slope failures of an area. In particular, soil tillage could lead to a decrease of the soil cohesion (e.g., due to soil disaggregation) and an increase of the friction angle (e.g., due to soil compaction). Quantitative data on this topic are rarely available: a table reporting quantitative changes in the mechanical parameters was found only in Albiero et al. (2014) [44].

In this paper, we analyze the decrease of the soil cohesion due to soil tillage, through a back-analysis approach and with the application of a probabilistic, physically-based model for the triggering of rainfall-induced landslides [45] in an agricultural environment. The choice of a probabilistic model is motivated to take into account the natural changes of the physical and mechanical properties of soils and rocks, which are characterized by high variability in space both in horizontal and vertical dimensions [46]. In probabilistic approaches, the safety level of the slope is given by the probability of failure (PoF), i.e., the probability associated with a value of factor of safety ≤1. Probabilistic approaches can provide a high

level of reliability when a detailed description of the study area is available in terms of slope topography and physical, mechanical, and hydraulic soil properties. For landslide prediction, probabilistic approaches, which assume input data as random variables defined through their probability density functions, are more suitable than deterministic approaches, which assume the input data without uncertainty [47].

The model is applied to the Collazzone area, a cultivated area located in central Italy, characterized by a high susceptibility to landslides. The area has been the subject of several studies, e.g., [48–50] and is periodically monitored through systematic image analysis and on-site surveys [51,52]. This allowed a preliminary quantitative assessment of the effect of the crops on the stability conditions of the area. Through a back-analysis approach and with the support of sensitivity indices, a quantitative evaluation on the effect induced by soil tillage on the mechanical properties of the soil has been provided. The contribution is divided as follows: after the introduction, an overview of the theoretical aspects of the landslide model and the method used in this study are illustrated in Section 2. The description of the study area and the database available for the mechanical soil characterization are part of this paragraph. In Section 3, after defining the geotechnical and hydrological assumptions considered for the reliability analysis, the results of the model are shown and discussed. The conclusions and future research developments represent the final section of the paper (Section 4).

#### **2. Materials and Methods**

*2.1. Study Area and Data*

The Collazzone area extends for 80 km<sup>2</sup> in the Perugia province, Umbria region, central Italy (Figure 1A). The geology of the area has been investigated several times [53,54]: it consists of the alternation of recent fluvial deposits along the valley bottoms, continental gravel sand and clay, travertine deposits, sandstone and marl in various percentages and thinly layered limestone. The digital elevation model (with a 20 m resolution) reveals that the territory is mainly hilly with elevations ranging between 145 and 634 m a.s.l.; the slopes that stand on the area have a gradient varying between 5◦ and 50◦, with the highest values in the northern part of the area (Figure 1B).

**Figure 1.** (**A**) Localization of the study area (in red) within the Umbria region (blue borders); (**B**) slope distribution in the area; ( **C**) geotechnical classification of soil types according to Fanelli et al., (2016) [55]; the green circles represent the landslides.

For the purposes of this work, defining the mechanical parameters needed in input by the model: the geotechnical classification of the Perugia province proposed by Fanelli et al., (2016) [55] was adopted, which includes an estimation of the physical and mechanical properties of the soil types constituting the near-surface cover. According to this classification, which reports the geotechnical parameters and a general description of the stratigraphy, the central part of the study area is covered mainly by clays, while the rocks (e.g., marl and travertine) are distributed in the south-eastern part of the study area (Figure 1C).

The area studied represents a high percentage of cultivated land (Figure 2) and the soil is inventoried as arable land. According to the last Agriculture Report, published on the Umbria region website (https://www.regione.umbria.it/agricoltura/statistica, last accessed on 28 April 2022), about 75% of the total agricultural area is used. According to the reference soil groups of the international soil classification [56] and to the map of the soils of the Umbria region [57], most of the area can be classified as a *calcaric cambisol.*

**Figure 2.** Maps showing ( **A**) the distribution of land use and (**B**) a picture of the cultivated and uncultivated areas in the study area.

Despite the presence of vegetation, the area presents a high landslide susceptibility, typically, the slope failures are triggered chiefly by meteorological events, including intense and prolonged rainfall and rapid snow melting [53]. Using aerial and satellite images, Fiorucci and co-authors [51] estimated the landslide mobilization rates in the area in the period 2004–2005 and speculated that the remarkably high yearly rate of landslide mobilization observed in the area in the analyzed period might be due to the agricultural and land use practices. Considering the interesting case study for the period 1941–2005, a multi-temporal landslide inventory map analyzing different sets of aerial photographs and field surveys is available for the area [53]; even now, numerous annual surveys continue to be made in the area and an ongoing mapping has also been made using remote sensing product such as Lidar, monoscopic and stereoscopic satellite images [51,52,58], and field surveys carried out after intense or prolonged rainfall, when images were not available. The last survey was carried out by CNR-IRPI on 20 December 2020, following the copious rain that affected the area in the first ten days of the month. In particular, on 8 December from 03:00 A.M. to 15:00 P.M. (local time), 50 mm of cumulative rainfall fell (Figure 3), corresponding to a mean intensity of 4 mm/h (0.07 mm/min) and a peak intensity of 9.2 mm/h (0.16 mm/min).

**Figure 3.** Bar chart showing the hourly rainfall measured by the rain gauge located in Collazzone on 8 December 2020. Data provided by the administration of the Umbria region.

The landslide information was obtained through a reconnaissance survey of the area (20 December 2020) driving and walking along main, secondary, and farm roads [59]. The investigators stopped at viewing points to check slopes where single or multiple landslides were identified and took photos of each landslide or group of landslides using a camera provided with GPS, and prepared a rapid (raw) mapping of the landslide. In the laboratory, the geolocated photographs were used to improve the location of the individual landslides and to characterize the type and the size. Landslides identified in the field and in the photographs were mapped on Google Earth. The main weakness of the reconnaissance inventory is the completeness, since from viewing points some slopes are not entirely visible, and an undetermined number of landslides may not have been identified and mapped.

During this last field survey, an event inventory was defined, including 26 shallow landslides. For the aim of this work, the landslides located in areas classified as rocks according to Fanelli et al. (2016) [55] and in uncultivated areas were excluded from the analysis; thus only 19 landslides (shown in Figure 2) were considered for modelling and analysis. As can be seen from Figures 1 and 2, these 19 landslides are located in the central portion of the area; the soils involved (Figure 1C) are clays, characterized by a cohesive resistance. Figure 4 shows an example of a shallow landslide (soil slide) mapped during the field survey.

**Figure 4.** Example of shallow landslide (soil slide) in the study area. Photo taken on 20 December 2020, by CNR-IRPI.
