Next Article in Journal
A Mathematical Method for Estimating the Critical Slope Angle of Sheet Erosion
Next Article in Special Issue
Analysis of Debris Flow Damage Using High-Resolution Topographical Data
Previous Article in Journal
Evaluation of Groundwater Infiltration in Sewer Networks Using Fluorescence Spectroscopy
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Water
Volume 15
Issue 19
10.3390/w15193340
water-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Analysis of the Role of Precipitation and Land Use on the Size of the Source Area of Shallow Landslides

by
Alessia Giarola
,
Massimiliano Bordoni
*,
Francesco Zucca
and
Claudia Meisina
Department of Earth and Environmental Sciences, University of Pavia, Via Adolfo Ferrata 1, 27100 Pavia, PV, Italy
*
Author to whom correspondence should be addressed.
Water 2023, 15(19), 3340; https://doi.org/10.3390/w15193340
Submission received: 28 July 2023 / Revised: 10 September 2023 / Accepted: 20 September 2023 / Published: 23 September 2023
(This article belongs to the Special Issue Water Role in Landslide Hazards Formation: Occurrence, Prevention and Mitigation)
Browse Figures
Versions Notes

Abstract

:
Rainfall-induced shallow landslides cause damage to human activities and infrastructureseach year, and although the size of a landslide correlates to the damage that it causes, very few studies have previously investigated the factors that influence the size of a shallow landslide. The aim of this work was to assess the role of precipitation and land use in controlling the size of the source area of rainfall-induced shallow landslides. After ruling out the impact of the slope angle and bedrock lithology in the two selected catchments, the role of land use was investigated: the statistical tests showed that woodlands and vineyards had statistically different source area size distributions, possibly due to the different hydrological behaviours between the two. A correlation was additionally found between the size of a source area and the amount of cumulated rainfall in the three days prior to each analysed event. In both cases, an increase in source area size was linked to an increase in pore pressure. This study aimed to fill the existing gap in the research to help implement policies aimed at reducing the size, and consequently the impact, of shallow landslides.

1. Introduction

Rainfall-induced shallow landslides cause damage to human activities and infrastructures each year [1,2] as they mobilise the uppermost soil layer until 2 m from ground level [3] and are usually longer than they are wide [4]. They are also characterised by high velocities and occur on steep slopes after significantly abundant rainfall events, which can also be concentrated in time [2,5].
While factors such as geological and geomorphological features of hillslopes, land use, soil properties and rainfall amounts have been linked to the triggering of shallow landslides [6,7,8,9], their role in determining the size of a landslide is lesser known [4,10]. The size of a shallow landslide is, however, directly related to the hazard that it poses [4,11,12], with larger landslides causing greater loss per single event [13]: a positive correlation between the size of a landslide and the volume of mobilised sediments has been proven in the past [14,15,16,17], with higher volumes of mobilised sediments in shallow landslides source areas resulting in longer runout paths [2,18,19].
Despite landslide size and specifically source area size playing a key role in determining the impact of a shallow landslide, research on the factors controlling its dimension is lacking: multiple authors [8,20,21,22] have in fact discovered that the landslides seem to adhere to size–frequency distributions thatfollow a power function [8]; however, these works are not usually specific to shallow failures [23], nor do they investigate the possible controlling factors that determine landslide size. Analyses of large inventories, which include different types of landslides, have previously linked the size of landslides to features of the slope, such as the slope angle and the lithology [23,24,25,26,27], but the details of such correlations remain elusive [26].
Various studies found that landslide size is related to both the cohesive strength of slope material and hillslope-scale relief [24,27,28]. Jeandet et al. [26] tried to define a mechanical model based on Mohr–Coulomb stability analysis and, based on their results, concluded that features such as slope, height and concavity of the slope seem to play a key role in controlling the size of unstable areas. Pelletier et al. [20] have investigated the impact of soil moisture on the frequency–size distribution of rainfall-induced landslides, characterised by different depths on the sliding surfaces, and discovered that soil moisture indirectly conditions the frequency–size distribution of landslides by impacting cohesion, pore pressure and internal friction.
A rule thatexplains the distribution of the size of shallow landslides is however yet to be discovered [23] since few works have analysed the impact of some slope and material properties on the different sizes of shallow landslides.
Bellugi et al. [23] proposed an evaluation of the size of unstable areas of a shallow landslide according to the impact of some geological and geomorphological parameters, such as slope elevation, soil depth and root strength in soil. They discovered that the interaction of multiple factors is needed to reproduce real-size distributions of past events, observing additionally that a root strength decrease lowers the minimum size required to trigger a landslide, increasing the probability of small failures occurring. For larger failures, Bellugi et al. [29] discovered that their distribution cannot be explained by topographical features alone and that an explanation of their pattern could be found in the strength properties of the slope materials.
Rickli and Graf [30] previously investigated the size of shallow landslides in woodlands compared to open fields in six alpine catchments located in different geological settings and triggered by different rainfall amounts: the analyses on landslide sizes were inconsistent between different catchments, with greater landslides occurring in one or the other land cover depending on the test site. They additionally investigated the range of slope angles where the landslides occurred and linked the higher stability angle threshold of forested areas to the stabilising role of roots.
However, according to Tanyas et al. [31], inventories thatfocus on the source areas and omit runout are preferred when studying the factors affecting the size distribution of landslides because the coalescence of more landslides can alter the measured landslide size and introduce biases. Besides these effects, no studies were carried out on analysing the role of different slope properties, specifically related to land cover, and of triggering factors, namely rainfall attributes, on the size of shallow landslide source areas. Starting from this consideration, this work aims to fill those gaps by investigating the role of different land uses and of different triggering rainfall features on the size of the source areas of rainfall-induced shallow landslides. While the soil depth has previously been linked to the volume of mobilised sediments [16,32,33], it was not possible to directly take this factor into account in this study; however, the role of the slope angle was investigated and this factor has previously been found to influence the available soil depth [34].
The analyses were carried out by taking into account shallow landslide inventories collected in two representative catchments of northern Italian Apennines susceptible to the formation of rainfall-induced shallow landslides in the last 14 years. The activities have been carried out partially in the LIFE DRIVE (Drought Resilience Improvement in Vineyard Ecosystems) project, where the relationship between land use and shallow landslides was investigated.
Other than providing information regarding the possible degree of hazard posed by developing landslides, studying the size distribution of shallow landslides could additionally help to correctly identify objects compatible with shallow failures in the automatic or semi-automatic detection and mapping of shallow landslides because, while promising, these techniques, which exploit both satellite imagery and aerial photography for identification purposes, still face some challenges when it comes to discerning landslide bodies [35,36,37].

2. Materials and Methods

2.1. Study Areas

Two study areas have been selected in this work: the lower portion of the Scuropasso-Versa catchments (SVC) and the Ardivestra catchment (AVC), both located in the Oltrepò Pavese, a hilly area in the southwestern portion of the Italian region of Lombardy representative of the typical geological and geomorphological features of northern Italian Apennines [38] (Figure 1). The study areas have been outlined based on the geolithological and geotechnical features of the soil deposits to achieve homogeneous zones in order to exclude the impact of the soil as a conditioning factor of the size of the source area. The SVC was chosen because, in April 2009, a single rainfall event produced 367 shallow landslides, scattered throughout different land uses, while the AVC was chosen because 350 landslides have developed in the area over the past decade (2009 to 2020) during multiple rainfall events, most of which were located in arable lands.
In these areas, shallow landslides were triggered as a consequence of sudden and abundant precipitations (between 30 and 160 mm over a few hours or a few days; Table 1).
These phenomena can be classified as translational and roto-translational earth slides that evolve into flows [39]. The ratio between the length and width of such landslides is 1.0 to 7.1: the absolute length spans between 10 and 500 m, while the width is between 10 and 70 m.
A comprehensive multitemporal inventory of past landslide phenomena, which includes location, size, landslide classification and date of occurrence, is available for this area. The outer perimeters of the shallow landslides were outlined manually through the analysis of field surveys and of aerial and satellite images [40,41]. The landslides thatdeveloped on 27–28 April 2009, which make up all landslides in the SVC, were mapped manually through the use of high-resolution colour aerial photographs taken immediately after this event, with an image resolution of 15 cm (photo scale of 1:12,000), and were validated by repeated field surveys [42].
The inventories thatwere exploited for statistical analysis in the AVC were instead outlined manually from the observation of Google Earth images [43] with an image resolution of 0.5 m, and were also validated through field surveys.
Starting from the available shallow landslides inventories in the study areas, the source areas were extracted through a semi-automatic method, implemented by [44,45]. It calculates the source area as 25% of the total landslide body, which is in line with the observed shallow landslides in the Oltrepò Pavese area [46].

2.1.1. The Lower Portion of Scuropasso-Versa Catchments (SVC)

The SVC test site is characterised by poorly cemented sandstones and conglomerates overlying marls and evaporitic deposits. In the SVC, soils are mainly low plastic clayey or clayey–sandy silts, characterised by unit weight in the order of 16–19 kN/m3 and saturated hydraulic conductivity in the order of 10−5–10−7 m/s [40] (Table 2). In this area, soil thickness typically ranges between a few centimetres and about 2.5 m from ground level. Altitudes in the area range between 252 and 353 m.a.s.l., while slope angles range between 0 and 60 degrees.
All the analysed landslides in the area were triggered during the rainfall event of 27–28 April 2009, causing one fatality and significant damage to roads and vineyards [34,42] (Figure 2). The event lasted 62 h and the measured cumulated rain amounted to 160 mm [34] and triggered landslides whose source areas range between a minimum of 1 m2 and a maximum of 2210 m2, with a median value of 157 m2 (Figure 3, Table 1). These landslides occurred across different land uses characterizing this study area, such as woodlands, shrublands, vineyards and croplands [42].

2.1.2. Study Areas: ArdivestraCatchment (AVC)

The AVC test site has a bedrock characterised by an alternation of sandstones, marls and clayey melanges. The soils that originated from these bedrocks are predominantly silty clays, reaching up to 4 m of depth [39]. Instead, the soils in the AVC are mainly high plastic silty clays, with unit weight and saturated hydraulic conductivity in the range of 18–20 kN/m3 and 10−5–10−8 m/s, respectively [39,40] (Table 3). These soils have a thickness higher than 1 m from ground level, which can reach values of over 2.5 m in correspondence of hillslopes characterised by the presence of deep-seated slow-moving landslides. Altitudes in this area range between 190 and 802 m.a.s.l., while the slope angles range between 0 and 68 degrees.
The source areas of the shallow landslides analysed in this testsite range between a minimum of 1 m2 and a maximum of 3572 m2, with a median value of 207 m2 (Table 1). These failures were triggered between the years 2009 and 2020, during the course of 7 different rainfall events lasting between 14 and 72 h and characterised by a cumulated rainfall of between 45.2 and 101.8 mm (Table 3; Figure 4 and Figure 5). Despite the presence of different land uses in the area, shallow slope failures in the AVC almost exclusively affected croplands, cultivated mostly with alfalfa, wheat and barley.

2.2. Statistical Techniques

Some statistical tests were adopted to study the effects of different parameters on source area sizes in both study areas; specifically, past landslides were compared through the use of statistical tests with geomorphological attributes, land use distributions and rainfall features.
Geomorphological features were retrieved from a digital elevation model (DEM) created with LiDAR data that were acquired between 2008 and 2010 by the Italian Ministry for Environment, Land, and Sea, at a spatial resolution of 1 m × 1 m. Land use maps were derived from the 2007 DUSAF database of the Region of Lombardy. Rainfall features were collected at hourly resolution from the rain gauge of Fortunago, which is located in the AVC study area and belongs to the ARPA Lombardy meteorological network.
Figure 6 shows a flow chart of the adopted methodology of statistical analysis: the starting point is the observation that there is high variability in the size of the source areas. What follows is the identification of homogeneous study areas to study the impact of land use (SVC) or the impact of rainfall (AVC) on source area sizes. The impact of the slope angle and the lithology were additionally analysed in both the test sites due to the significant role played by these parameters on the proneness of a hillslope to the formation of shallow landslides [25,47], as was the interaction between slope angle, lithology and, depending on the catchment, land use or rainfall.
The significance level (p-value) of all the statistical tests was set to 0.01. Firstly, the Shapiro–Wilk’s and Levene’s tests were applied to test the normality and the homogeneity of the variance of the distribution. Once defining the shape of the distributions of the different parameters considered in the analyses, some statistical tests were adopted to verify similarities or differences between the distributions of source area values between different classes of the analysed influencing parameters [48]. Thus, if both assumptions were verified, the one-way ANOVA test was adopted to evaluate if the source area was significantly different within different land use classes (SVC) or rainfall events (AVC) and then Tukey’s honestly significant difference test was applied for grouping classes thatdid not statistically differ when in source area size. If the assumptions of the Shapiro–Wilk’s and Levene’s tests were not verified, the Kruskal–Wallis test was applied instead, since it is considered to be comparable to the one-way ANOVA, and if the results of the test were significant, Dunn’s test was applied with the goal of grouping classes thatwere not statistically different in terms of distribution of the source areas.
Furthermore, correlation functions between the median value of source area inventories and a triggering rainfall parameter were defined for the AVC test in order to identify which rainfall parameters had the strongest impact on the size of source areas for different events. The significance of each correlation function was evaluated by an F-statistical test, while the root mean square error (RMSE) was quantified to verify the reliability of each correlation. The correlation functions were adopted to compare, alternatively, the median value of source areas inventories with one triggering rainfall parameter, such as (a) the event cumulated rainfall amount (event-rainfall), defining an event as a time span during which consecutive intervals of nil hourly rainfall do not last more than 3 h between the months of May and September or more than 6 h between the months of October and April [49]; (b) the event mean intensity; (c) the 3-day antecedent cumulated rainfall amount; (d) the 5-day antecedent cumulated rainfall amount; (e) the 7-day antecedent cumulated rainfall amount; (f) the 14-day antecedent cumulated rainfall amount; (g) the 30-day antecedent cumulated rainfall amount; (h) the 60-day antecedent cumulated rainfall amount. These parameters were chosen to represent the effects of the triggering event or of antecedent rainfalls of different lengths on the dimension of a particular slope instability.
Other tests were applied to assess the possible effects of the interactions between two factors, namely land use/lithology and land use/slope angle for SVC area and triggering rainfall event/lithology and rainfall event/slope angle for AVC, on the size of source areas. Among the one-way tests, the Shapiro–Wilk’s and Levene’s tests were chosen to test the normality and the homogeneity of the variance of the distribution. If both assumptions were verified, a two-way ANOVA test was adopted, and Tukey’s honestly significant difference test was applied for grouping classes that did not differ statistically when it comes to the size of the source area. Otherwise, if the assumptions of the Shapiro–Wilk’s and Levene’s tests were not verified, the two factors were combined into a single factor and the Kruskal–Wallis test was applied. If the results of the test were significant, Dunn’s test was then applied with the goal of grouping classes that were not statistically different in terms of the distribution of the source areas. To apply this test considering slope angle as a factor, the parameter was reclassified into classes according to the following scheme: (a) slope angle between 0 and 5°; (b) slope angle between 5 and 10°; (c) slope angle between 10 and 15°; (d) slope angle between 15 and 20°; (e) slope angle between 20 and 25°; (f) slope angle between 25 and 30°; (g) slope angle over 30°.

3. Results

3.1. Impact of Slope Angle on Shallow Landslide Source Areas

In the SVC site (Figure 7), it can be observed how the woodlands were located on the highest median slope angles (25.9°), as were the landslides thatoccurred in woodlands (33.11°). Vineyards were, in general, located at lower slope angles than shrublands (12.37° and 17.25°, respectively) while the shallow landslides thathave occurred in those land uses had similar distributions (25.79° and 28.48°, respectively). Croplands were located at the lowest slope angles (6.59°), while the source areas of the landslides in croplands had a median slope angle of 26.27°.
The result of Shapiro–Wilk’s test was the non-normality of the distribution of the slope angles in the source areas. The Kruskal–Wallis test also resulted in a p-value higher than 0.01, meaning no statistical differences in terms of the effects of slope angle to source areas (χ2 = 30.81, p-value = 0.05).
In Table 4, the landslides in each land use class were subdivided based on their source area size into three groups (smaller than 50 m2, between 50 and 100 m2 and larger than 100 m2) to offer further information on the minimum and maximum median slope angle for the landslides in each group.
In the AVC test site, with the exception of the event of 6–8 February 2009, which reached a maximum slope angle of 49.76 degrees, the majority of the landslides developed in slope angle ranges of around 5 to 30 degrees and the median slope angle values were similar for all the events (Figure 8). The result of Shapiro–Wilk’s test was the non-normality of the distribution of the slope angle between the events. The Kruskal–Wallis test showed that differences in slope angles between source areas developed during different events were not present (χ2 = 1.75, p-value = 0.42).
Table 5 presents the median slope angle values for each event, grouped into three landslide size classes (<50 m2, 50–100 m2, >100 m2) to assess the range of the median slope angle of the mapped landslides.

3.2. Impact of the Lithology on Source Area Size

In the SVC, the landslides have occurred in three lithologies (presented in order of decreasing frequency): (a) sandstones, (b) marls and (c) gypsum marls. The Shapiro–Wilk’s test had a p-value lower than 0.01, meaning non-normality between the distributions. The Kruskal–Wallis test was then applied: the result was a p-value higher than 0.01 (p-value = 0.2778), which means that there is no statistical difference between the distribution of the source areas in different lithologies.
In the AVC, the lithologies in which the landslides occurred were (with decreasing frequencies): (a) marls, (b) clays, (c) interlayered rocks (marls with clay), (d) sandstones, (e) gravel, sand, loam and (f) debris from past landslides. Similarly to the SVC, the Shapiro–Wilk’s test proved the non-normality of the distribution and the Kruskal–Wallis test had a p-value higher than 0.01 (p-value = 0.4355), which means that there is no statistical difference between the distribution of the source areas that have occurred in different lithologies.

3.3. Impact of Land Use on Source Area Size

The landslides thathave occurred in the SVC on 27–28 April2009 are distributed in four main land uses: woodlands (170 landslides), vineyards (103), shrublands (66) and croplands (28). The highest median source area values were recorded in woodlands and croplands, with a value of 68 m2 and 66 m2, respectively. The median values of source areas for vineyards and shrublands were lower: 43 m2 and 39 m2, respectively.
Table 6 shows an overview of the size of the source areas in each land use.
Figure 9 shows an eCDF and a histogram of the source area in different land uses.
The Shapiro–Wilk’s test had a p-value lower than 0.01, which means that the data arenot normally distributed and it is therefore necessary to adopt statistical tests meant for non-parametric datasets, such as the Kruskal–Wallis and Dunn’s tests.
The result of the Kruskal–Wallis test was a p-value lower than 0.01 (0.005, specifically), meaning that the distribution of the source areas among different land uses was statistically different.
Dunn’s test analysed the source areas located in each combination of land uses and found significant differences in the distributions of source areas in woodlands compared to vineyards (Table 7).

3.4. Impact of Rainfall Attributes on Shallow Landslides Source Areas

The landslides occurred in the AVC developed in croplands over the course of six rainfall events during the last decade (2009–2020). The highest median source area values were recorded during the event of February 2016, with a value of 159 m2; on the contrary, the event of February 2009 was characterised by the lowest median value of 45 m2. Table 8 shows an overview of the size of the source areas in each event: the November 2014 and December 2020 events had the lowest and highest maximum values, respectively (Figure 10).
The result of the Shapiro–Wilk’s test was a p-value lower than 0.01, which means that the data arenot normally distributed. The Kruskal–Wallis and Dunn’s tests were therefore applied: the result of the Kruskal–Wallis test was a p-value lower than 0.01 (0.0004, specifically), meaning that the distribution of the source areas among different events was statistically different.
Dunn’s test analysed the source areas of landslides that occurred during each event and found significant differences in the distributions of the source areas among some of the events, as shown in Table 9. The events of 6–8 February and 27–29 February have different distributions of source areas, as do those of 18–20 January 2014 compared to those of 27–29 February 2016.
All the considered rainfall attributes presented an increasing powerlaw relation with the median value of source areas of an event (Figure 11, Table 10). In particular, the best correspondence was identified with the event cumulated rainfall amount and 3-day and 5-day antecedent cumulated rainfall amounts, as testified by RMSE values in the range of 16.5–21.5 m2 and by a p-value of F-statistical test in the range of 0.02–0.04. Instead, the other correlations showed values of RMSE indexes higher than 28 m2 and a p-value of the F-statistical test higher than 0.08, testifying a poor correlation with the median values of shallow landslide source areas.

3.5. Effect of Interactions between Factors on Shallow Landslides Source Areas

The Shapiro–Wilk’s test showed a p-value lower than 0.01 for all the considered factors, which means that these data are not normally distributed. According to this, the Kruskal–Wallis tests were performed after combining two of each selected factors in order to evaluate the statistical effects of the interactions to source area distribution.
Regarding the SVC, interactions between land use and lithology and land use and slope angle are not statistically significant for determining differences in shallow landslides source areas distribution (χ2 = 22.84, p-value = 0.05 for lithology/land use; χ2 = 27.93, p-value = 0.02 for slope angle/land use). Similarly, for the AVC, interactions between rainfall events and lithology and rainfall events and slope angle are not statistically significant fordetermining differences in shallow landslides source areas distribution (χ2 = 15.04, p-value = 0.18 for lithology/rainfall; χ2 = 45.04, p-value = 0.02 for slope angle/rainfall).

4. Discussion

The sizes of the source areas of shallow landslides have a significant impact when it comes to the modelling, prediction and mitigation of these phenomena [8]. In this study, several factors, which could potentially influence this feature and have not been studied in detail previously, have been analysed to assess their impact on source area dimensions, providing an important contribution to understanding which differences can be encountered within a dataset of phenomena triggered by similar rainfall events. The analysis was carried out in two test sites of Italian Apennines that are very prone to the formation of shallow landslides and are representative of different geological, geomorphological and land use features. The two test sites were selected to achieve homogeneous catchments, as far as soil features go, to focus on the impact of the other selected parameters (slope angle, land use, rainfall features) analysed in terms of influence on source area sizes, excluding a further factor that would have made for a more complex interpretation of the achieved statistical results. It is also worth noting that the main soil physical and geotechnical features (i.e., grain size, density, mechanical properties, soil thickness) could influence the sizes and volumes of the sediments mobilised by shallow slope failures [8,50]; thus, their role ininfluencing source area sizes will have to be investigatedfurther in more heterogeneous susceptible sites, characterised by higher variability in soil properties.
The first analysis was performed to assess the distribution of the slope angles in both catchments, since it has been proven to correlate to the size of a landslide [23] and subsequently to its source area. Some authors have discovered an inverse relationship between the slope angle and the size of a landslide [51], although results have been inconsistent [23].
This inverse relationship seems to be confirmed in SVC for croplands, where the lowest average slope angles are associated with the second-largest source areas. However, this is not true for all other land uses: for vineyards, shrublands and woodlands, greater slope angles are associated with larger landslides instead. Statistical tests have highlighted that, in the SVC test site, land use has an effect on source area size: woodlands and vineyards have different source area size distributions, whereas, in other land uses, the distributions were statistically similar. On the other hand, statistical tests performed on the slope angles did not show differences in distribution among different source area classes and, for this area, the results of the statistical tests therefore seem to also suggest that the interactions of slope angle with land use do not significantly influence the source area sizes.
In the AVC, statistical tests have highlighted differences in the distribution of the source areas related to different rainfall events, in particular for the events of February 2009 and February 2016 and for the events of January 2014 and February 2016. The distribution of the slope angles in both cases is however statistically similar, meaning that the slope angles cannot explain differences in source area size between these inventories. Moreover, also in the AVC, the results of the statistical tests seem to suggest that the interactions between the slope angle and the rainfall events are not significantly impacting shallow landslide source areas.
Regarding the impact of the bedrock lithology on the size of the source area, in both test sites, the distributions of the source areas did not show any statistical differences in distribution between landslides thathad developed in different lithologies, even though bedrock lithology has been proven to impact sediment supply [25]. Moreover, the interactions between land use and bedrock lithology in SVC and betweenrainfall events and bedrock lithology in AVC were also not statistically significant in influencing shallow landslide source areas.
The achieved results emphasise the need to investigate the role played by land use and rainfall on the sizes of shallow landslides source areas since the research regarding the impact of land use on the source area size is lacking [29]. Moser [52] discovered that landslides tend to be bigger in woodlands compared to open lands, whereas Rickli and Graf [29], due to both the lack of research on this topic and inconsistencies within their study areas, concluded that landslide size is similar between forest and open land. However, in SVC, similarly to Moser [52], landslides tend to be larger in woodlands compared to open land, namely croplands and shrublands. However, in accordance with Rickli and Graf [29] and despite larger landslides on average, the size distribution between woodlands and croplands or shrublands is statistically similar.
Statistical differences were instead found between the size distributions of woodlands compared to vineyards, and land use changes in SVC might give an insight as to why that is. Most of the areas thatare currently classified as “woodlands” have been cultivated as vineyards until the 1980s and then have successively been abandoned. The abandonment led to the growth of black locust trees (Robinia pseudoacacia), a fast-growing, invasive species [53], and, in previous studies, these areas of the Oltrepò Pavese have been associated with an increased tendency towards instability compared to areas thatare still cultivated as vineyards or historic woodlands [46].
Regarding the differences in size distribution, however, a possible explanation could be found in the hydrogeological behaviourof woodlands compared to vineyards. In active vineyards, water gets routed away at surface level through preferential rills, such as those formed by the wheels of tractors and other heavy machinery, which are associated with increased runoff tendencies and reduced infiltration [54,55,56,57]. The same is not true for woodlands, where greater root systems compared to vineyards might promote infiltration rather than runoff [58]. This behaviour would lead to an increase in pore pressure and an increase in the self-weight of the soil and subsequent mobilisation of the soil [59,60,61,62,63].While the cited studies refer to landslide susceptibility rather than size, soil moisture has also been found to affect landslide size [20].
It was established that a higher volume of mobilised sediments results in greater damage [14,17] and, consequently, in a greater loss in revenues. Studying how land use influences the size of a landslide could therefore help to link the quantity of mobilised and lost sediments to each land use. This could be especially relevant when it comes to vineyards, which are often areas of high cultural and economic relevance [64].
In the AVC, i.e., for shallow landslides thatwere located in croplands, a correlation was found between the median landslide size and the event cumulated rainfall amount and cumulative precipitation three and five days before the event occurred. For all recorded events, a greater rainfall quantity cumulated over the course of the three days prior to the triggering event, resulting in larger source areas. Higher rainfall amounts cause an increase in pore pressure within the soil [60,63] and pore pressure has been found to affect landslide size [20]. Reid et al. [60] additionally observed that while shallow pore pressure increases within a few hours of the precipitation, the increase inthe pore pressure at the slip surface is delayed. Despite the limited number of considered events (7), the results of this research seem to suggest that the cumulated rainfall of an event and of the immediate antecedent days directly affects the size of shallow landslides, with higher cumulated values being linked to larger landslides.
The importance of rainfall as a controlling factor of the landslide size is additionally suggested by the fact that, in the study area, it is possible to observe multiple instances of landslides occurring at the same location over the course of different events with significantly different sizes. For instance, shallow landslides that developed on 17–19 November 2019, the second event in terms of rainfall abundance, occur with much larger sizes at the same location in the immediate vicinity or in correspondence of the same sectors affected in the past by other shallow landslides, despite the unchanged geolithological and topographical settings. Examples of this behaviourare showcased in Figure 12, where shallow landslides triggered by different events and corresponding different sizes of their source areas are represented. In the AVC site, agricultural practices, carried out for preparing hillslopes for sowing grasses, remould and move the volumes of sediments mobilised by shallow landslides in a cultivated hillslope, such as those shown in Figure 12. These activities bring the hillslopes back to conditions comparable to those prior to a shallow landslide triggering in terms of slope features and sediment availability, explaining why different shallow landslides can occur very close to each other after different triggering events. In Figure 12, it is highlighted how subsequent landslides are not necessarily larger than landslides that have occurred at the same location, and a comparison between the landslides that occurred in February 2016 and in November 2019 is provided.
The achieved results highlight the importance of factors such as land use and rainfall parameters in controlling the size of shallow landslide source areas. Additionally, this work could contribute to the development of automatic or semi-automatic detection methods by providing insight into what object size is compatible with shallow failures.
However, some limitations must be taken into account before considering these results as generally valid for other geological–geomorphological and environmental settings. Regarding the land use classification, the adopted land use map gives a somewhat broad classification of the land uses thatare present, grouping within the same category woodlands of different ages and species, croplands cultivated with different plant types (in this case, mostly alfalfa, wheat and barley) and vineyards managed through different management techniques of the inter-row, which have been linked to different soil mobility rates [64]. In the future, provided that this information could be retrieved for past events, it would be interesting to take this factor into account. Regarding the analysis of the role of precipitation on source area size, a larger dataset of rainfall-induced events would provide additional strength to the proposed argumentation. Although correlated to the slope angle [34], taking into account the role of soil depth directly, which was proven to affect landslide size [23], could also help in improving the comprehension of different source area sizes. Lastly, it would be interesting to perform the same analyses in a different yet comparable geological setting to evaluate similarities and differences in the results achieved in the analysed context.

5. Conclusions

The aim of this work was to assess the role of usually neglected parameters, such as precipitation and land use, in controlling the size of the source area of rainfall-induced shallow landslides. Firstly, the impact of one factor generally influencing the slope instabilities dimension, namely the slope angle, was assessed. The relationship between slope angle and source area size was not linear and the slope angle alone could not explain different source area size distributions. For both catchments, the bedrock lithology did not seem to directly influence source area size as no statistical differences in distribution were found.
The role of land use was then investigated: woodlands and vineyards have statistically different distributions and a possible explanation could be found in the different hydrological behaviours between the two: in vineyards, the action of heavy machinery increases the runoff [54,55,56,57], whereas, in woodlands, the roots of trees promote infiltration [58]. The consequence is an increase in pore pressure in the latter [59,60,61,62,63], which could result in larger source areas [20]. In the future, taking into account factors such as the vineyard management techniques could also prove interesting.
A correlation was additionally found between the size of a source area and the amount of cumulated rainfall in the three days prior to each analysed event. It has been proven in the past that higher rainfall amounts cause an increase in pore pressure in the pores within the soil [59,60,61,62,63]. The provided hypothesis is that a higher amount of rainfall would cause greater pore pressure and consequently larger landslides.
This study aimed to fill the existing gap in the research regarding the factors controlling the size of a shallow landslide. A better understanding of these factors could help to implement policies aimed at reducing the size, and consequently the impact, of shallow landslides. It might additionally contribute to the development of algorithms for the automatic or semi-automatic detection of shallow failures. However, it is worth noting that the influence of other factors, such as soil physical and geotechnical properties, has to be analysed in terms of soil properties, especially in more heterogeneous settings compared to the areas considered in this study. A possible future development could be to adopt the same methodology on a different test site, employing a higher number of mapped landslide events and higher precision land use maps.

Author Contributions

Conceptualisation, A.G., M.B. and C.M.; data curation, A.G. and C.M.; funding acquisition, C.M.; investigation, A.G. and M.B.; methodology, A.G., M.B., F.Z. and C.M.; project administration, C.M.; resources, M.B. and C.M.; supervision, C.M.; validation, A.G., M.B., F.Z. and C.M.; visualisation, F.Z.; writing—original draft, A.G.; writing—review and editing, M.B., F.Z. and C.M.All authors have read and agreed to the published version of the manuscript.

Funding

This work has been carried out partially in the framework of the project LIFE DRIVE (Drought Resilience Improvement in Vineyard Ecosystems), funded by European Union (grant number LIFE19 ENV/IT/000035).

Data Availability Statement

Data are available on request.

Acknowledgments

We refer to and thank the following projects, which funded the creation of the shallow landslides inventories used in this study: Study of the landslides triggered in 27–28 April 2009 event in Oltrepo Pavese and definition of guidelines for hillslopes management (funded by Pavia Province and Rotary Club Oltrepo Pavese; responsible for the project: Claudia Meisina); ANDROMEDA (funded by Fondazione Cariplo; responsible of the project: Claudia Meisina). The aerial photographs, used for the creation of the inventory of April 2009 shallow landslides, were from18 May 2009 and were taken by Rossi s.r.l. (Brescia). We thank the anonymous reviewers for their revisions and suggestions to the work.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Cevasco, A.; Pepe, G.; Brandolini, P. The influences of geological and land use settings on shallow landslides triggered by an intense rainfall event in a coastal terraced environment. Bull. Eng. Geol. Environ. 2014, 73, 859–875. [Google Scholar] [CrossRef]
  2. Di Napoli, M.; Di Martire, D.; Bausilio, G.; Calcaterra, D.; Confuorto, P.; Firpo, M.; Cevasco, A. Rainfall-induced shallow landslide detachment; transit and runout susceptibility mapping by integrating machine learning techniques and GIS-based approaches. Water 2021, 13, 488. [Google Scholar] [CrossRef]
  3. Bordoni, M.; Meisina, C.; Valentino, R.; Lu, N.; Bittelli, M.; Chersich, S. Hydrological factors affecting rainfall-induced shallow landslides: From the field monitoring to a simplified slope stability analysis. Eng. Geol. 2015, 193, 19–37. [Google Scholar] [CrossRef]
  4. Milledge, D.G.; Bellugi, D.; McKean, J.A.; Densmore, A.L.; Dietrich, W.E. A multidimensional stability model for predicting shallow landslide size and shape across landscapes. JGREarth Surf. 2014, 119, 2481–2504. [Google Scholar] [CrossRef] [PubMed]
  5. Lan, H.X.; Zhou, C.H.; Lee, C.F.; Wang, S.J.; Wu, F.Q. Rainfall-induced landslide stability analysis in response to transient pore pressure-A case study of natural terrain landslide in Hong Kong. Ser. Technol. Sci. 2003, 46, 52–68. [Google Scholar] [CrossRef]
  6. Saito, H.; Nakayama, D.; Matsuyama, H. Relationship between the initiation of a shallow landslide and rainfall intensity—Duration thresholds in Japan. Geomorphology 2010, 118, 167–175. [Google Scholar] [CrossRef]
  7. Posner, A.J.; Georgakakos, K.P. Soil moisture and precipitation thresholds for real-time landslide prediction in El Salvador. Landslides 2015, 12, 1179–1196. [Google Scholar] [CrossRef]
  8. Tebbens, S.F. Landslide scaling: A review. Earth Space Sci. 2020, 7, e2019EA000662. [Google Scholar] [CrossRef]
  9. Phillips, C.; Hales, T.; Smith, H.; Basher, L. Shallow landslides and vegetation at the catchment scale: A perspective. Ecol. Eng. 2021, 173, 106436. [Google Scholar] [CrossRef]
  10. Hungr, O.; McDougall, S.; Wise, M.; Cullen, M. Magnitude–frequency relationships of debris flows and debris avalanches in relation to slope relief. Geomorphology 2008, 96, 355–365. [Google Scholar] [CrossRef]
  11. Cardinali, M.; Reichenbach, P.; Guzzetti, F.; Ardizzone, F.; Antonini, G.; Galli, M.; Salvati, P. A geomorphological approach to the estimation of landslide hazards and risks in Umbria; Central Italy. Nat. Hazards Earth Syst. Sci. 2002, 2, 57–72. [Google Scholar] [CrossRef]
  12. Lombardo, L.; Tanyas, H.; Huser, R.; Guzzetti, F.; Castro-Camilo, D. Landslide size matters: A new data-driven; spatial prototype. Eng. Geol. 2021, 293, 106288. [Google Scholar] [CrossRef]
  13. Turner, A.K. Social and environmental impacts of landslides. Innov. Infrastruct. Solut. 2018, 3, 70. [Google Scholar] [CrossRef]
  14. Imaizumi, F.; Sidle, R.C. Linkage of sediment supply and transport processes in Miyagawa Dam catchment, Japan. J. Geophys. Res. Earth Surf. 2007, 112. [Google Scholar] [CrossRef]
  15. Scheevel, C.R. Predicting Landslide Stability; Runout; and Failure Velocity at Cook Lake Landslide, Wyoming; Colorado School of Mines ProQuest Dissertations Publishing: Golden, CO, USA, 2017. [Google Scholar]
  16. Cha, D.; Hwang, J.; Choi, B. Landslides detection and volume estimation in Jinbu area of Korea. For. Sci. Technol. 2018, 14, 61–65. [Google Scholar] [CrossRef]
  17. Domej, G.; Bourdeau, C.; Lenti, L.; Martino, S.; Pluta, K. Shape and Dimension Estimations of Landslide Rupture Zones via Correlations of Characteristic Parameters. Geosciences 2020, 10, 198. [Google Scholar] [CrossRef]
  18. Corominas, J. The angle of reach as a mobility index for small and large landslides. Can. Geotech. J. 1996, 33, 260–271. [Google Scholar] [CrossRef]
  19. Okura, Y.; Kitahara, H.; Sammori, T.; Kawanami, A. The effects of rockfall volume on runout distance. Eng. Geol. 2000, 58, 109–124. [Google Scholar] [CrossRef]
  20. Pelletier, J.D.; Malamud, B.D.; Blodgett, T.; Turcotte, D.L. Scale-invariance of soil moisture variability and its implications for the frequency-size distribution of landslides. Eng. Geol. 1997, 48, 255–268. [Google Scholar] [CrossRef]
  21. Massey, C.; Townsend, D.; Rathje, E.; Allstadt, K.E.; Lukovic, B.; Kaneko, Y.; Villeneuve, M. Landslides triggered by the 14 November 2016 M w 7.8 Kaikōura earthquake; New Zealand. Bull. Seismol. Soc. Am. 2018, 108, 1630–1648. [Google Scholar] [CrossRef]
  22. Zhang, J.; van Westen, C.J.; Tanyas, H.; Mavrouli, O.; Ge, Y.; Bajrachary, S.; Khanal, N.R. How size and trigger matter: Analyzing rainfall-and earthquake-triggered landslide inventories and their causal relation in the Koshi River basin; central Himalaya. Nat. Hazards Earth Syst. Sci. 2019, 19, 1789–1805. [Google Scholar] [CrossRef]
  23. Bellugi, D.; Milledge, D.G.; Dietrich, W.E.; Perron, J.T.; McKean, J. Predicting shallow landslide size and location across a natural landscape: Application of a spectral clustering search algorithm. JGR Earth Surf. 2015, 120, 2552–2585. [Google Scholar] [CrossRef]
  24. Frattini, P.; Crosta, G.B. The role of material properties and landscape morphology on landslide size distributions. Earth Planet. Sci. Lett. 2013, 361, 310–319. [Google Scholar] [CrossRef]
  25. Roda-Boluda, D.C.; D’Arcy, M.; McDonald, J.; Whittaker, A.C. Lithological controls on hillslope sediment supply: Insights from landslide activity and grain size distributions. Earth Planet. Sci. Lett. 2018, 43, 956–977. [Google Scholar] [CrossRef]
  26. Jeandet, L.; Steer, P.; Lague, D.; Davy, P. Coulomb mechanics and relief constraints explain landslide size distribution. Geophys. Res. Lett. 2019, 46, 4258–4266. [Google Scholar] [CrossRef]
  27. Medwedeff, W.G.; Clark, M.K.; Zekkos, D.; West, A.J. Characteristic landslide distributions: An investigation of landscape controls on landslide size. Earth Planet. Sci. Lett. 2020, 539, 116203. [Google Scholar] [CrossRef]
  28. Stark, C.P.; Guzzetti, F. Landslide rupture and the probability distribution of mobilized debris volumes. JGR Earth Surf. 2009, 114. [Google Scholar] [CrossRef]
  29. Bellugi, D.G.; Milledge, D.G.; Cuffey, K.M.; Dietrich, W.E.; Larsen, L.G. Controls on the size distributions of shallow landslides. Proc. Natl. Acad. Sci. USA 2021, 118, e2021855118. [Google Scholar] [CrossRef]
  30. Rickli, C.; Graf, F. Effects of forests on shallow landslides–case studies in Switzerland. For. Snow Landsc. Res. 2009, 82, 33–44. [Google Scholar]
  31. Tanyaş, H.; van Westen, C.J.; Allstadt, K.E.; Jibson, R.W. Factors controlling landslide frequency–area distributions. Earth Planet. Sci. Lett. 2019, 44, 900–917. [Google Scholar] [CrossRef]
  32. Larsen, I.J.; Montgomery, D.R.; Korup, O. Landslide erosion controlled by hillslope material. Nat. Geosci. 2010, 3, 247–251. [Google Scholar] [CrossRef]
  33. Zieher, T.; Perzl, F.; Rössel, M.; Rutzinger, M.; Meißl, G.; Markart, G.; Geitner, C. A multi-annual landslide inventory for the assessment of shallow landslide susceptibility–Two test cases in Vorarlberg, Austria. Geomorphology 2016, 259, 40–54. [Google Scholar] [CrossRef]
  34. Zizioli, D.; Meisina, C.; Valentino, R.; Montrasio, L. Comparison between different approaches to modeling shallow landslide susceptibility: A case history in Oltrepo Pavese; Northern Italy. Nat. Hazards Earth Syst. Sci. 2013, 13, 559–573. [Google Scholar] [CrossRef]
  35. Rau, J.Y.; Chang, K.T.; Shao, Y.C.; Lau, C.C. Semi-automatic shallow landslide detection by the integration of airborne imagery and laser scanning data. Nat. Hazards 2012, 61, 469–480. [Google Scholar] [CrossRef]
  36. Mondini, A.C.; Guzzetti, F.; Reichenbach, P.; Rossi, M.; Cardinali, M.; Ardizzone, F. Semi-automatic recognition and mapping of rainfall induced shallow landslides using optical satellite images. Remote Sens. Environ. 2011, 115, 1743–1757. [Google Scholar] [CrossRef]
  37. Notti, D.; Cignetti, M.; Godone, D.; Giordan, D. Semi-automatic mapping of shallow landslides using free Sentinel-2 images and Google Earth Engine. Nat. Hazards Earth Syst. Sci. 2023, 23, 2625–2648. [Google Scholar] [CrossRef]
  38. Meisina, C.; Zucca, F.; Fossati, D.; Ceriani, M.; Allievi, J. Ground deformation monitoring by using the permanent scatterers technique: The example of the Oltrepo Pavese (Lombardia, Italy). Eng. Geol. 2006, 88, 240–259. [Google Scholar] [CrossRef]
  39. Cruden, D.M. Cruden, DM, Varnes, DJ, 1996, Landslide Types and Processes, Transportation Research Board, US National Academy of Sciences, Special Report, 247: 36–75. Landslides Eng. Pract. 1996, 24, 20–47. [Google Scholar]
  40. Bordoni, M.; Bittelli, M.; Valentino, R.; Vivaldi, V.; Meisina, C. Observations on soil-atmosphere interactions after long-term monitoring at two sample sites subjected to shallow landslides. Bull. Eng. Geol. Environ. 2021, 80, 7467–7491. [Google Scholar] [CrossRef]
  41. Bordoni, M.; Vivaldi, V.; Lucchelli, L.; Ciabatta, L.; Brocca, L.; Galve, J.P.; Meisina, C. Development of a data-driven model for spatial and temporal shallow landslide probability of occurrence at catchment scale. Landslides 2021, 18, 1209–1229. [Google Scholar] [CrossRef]
  42. Persichillo, M.G.; Bordoni, M.; Meisina, C.; Bartelletti, C.; Barsanti, M.; Giannecchini, R.; D’Amato Avanzi, G.; Galanti, Y.; Cevasco, A.; Brandolini, P.; et al. Shallow landslides susceptibility assessment in different environments. Geomat. Nat. Hazards Risk 2016, 8, 748–771. [Google Scholar] [CrossRef]
  43. Bordoni, M.; Galanti, Y.; Bartelletti, C.; Persichillo, M.G.; Barsanti, M.; Giannecchini, R.; Meisina, C. The influence of the inventory on the determination of the rainfall-induced shallow landslides susceptibility using generalized additive models. Catena 2020, 193, 104630. [Google Scholar] [CrossRef]
  44. Galve, J.P.; Cevasco, A.; Brandolini, P.; Soldati, M. Assessment of shallow landslide risk mitigation measures based on land use planning through probabilistic modelling. Landslides 2015, 12, 101–114. [Google Scholar] [CrossRef]
  45. Galve, J.P.; Cevasco, A.; Brandolini, P.; Piacentini, D.; Azañón, J.M.; Notti, D.; Soldati, M. Cost-based analysis of mitigation measures for shallow-landslide risk reduction strategies. Eng. Geol. 2016, 213, 142–157. [Google Scholar] [CrossRef]
  46. Persichillo, M.G.; Bordoni, M.; Meisina, C. The role of land use changes in the distribution of shallow landslides. Sci. Total Environ. 2017, 574, 924–937. [Google Scholar] [CrossRef]
  47. Catani, F.; Lagomarsino, D.; Segoni, S.; Tofani, V. Exploring model sensitivity issues across different scales in landslide susceptibility. Nat. Hazards Earth Syst. Sci. 2013, 13, 2815–2831. [Google Scholar] [CrossRef]
  48. Luino, F.; De Graff, J.; Biddoccu, M.; Faccini, F.; Freppaz, M.; Roccati, A.; Ungaro, F.; D’Amico, M.; Turconi, L. The Role of Soil Type in Triggering Shallow Landslides in the Alps (Lombardy, Northern Italy). Land 2022, 11, 1125. [Google Scholar] [CrossRef]
  49. Bordoni, M.; Corradini, B.; Lucchelli, L.; Valentino, R.; Bittelli, M.; Vivaldi, V.; Meisina, C. Empirical and Physically Based Thresholds for the Occurrence of Shallow Landslides in a Prone Area of Northern Italian Apennines. Water 2019, 11, 2653. [Google Scholar] [CrossRef]
  50. Ramos-Scharron, C.E.; Arima, E.Y.; Chen, Y.; Ohlson, B.; Vest, B. Sediment mobilization by rain-driven landslides in a subtropical moist, mix-use, and predominantly metamorphic setting. Geomorphology 2022, 413, 108380. [Google Scholar] [CrossRef]
  51. Gabet, E.J.; Dunne, T. Landslides on coastal sage-scrub and grassland hillslopes in a severe El Nino winter: The effects of vegetation conversion on sediment delivery. Geol. Soc. Am. Bull. 2002, 114, 983–990. [Google Scholar] [CrossRef]
  52. Moser, M. Der Einfluss der PflanzendeckebeiElementarereignissen auf Grösse und Anzahl der AnbruchsflächenimBereich des mittlerenLesachtales. Kärnten Cent. Für Das Gesamte Forstwes. 1971, 88, 164–176. [Google Scholar]
  53. Vítková, M.; Müllerová, J.; Sádlo, J.; Pergl, J.; Pyšek, P. Black locust (Robiniapseudoacacia) beloved and despised: A story of an invasive tree in Central Europe. For. Ecol. Manag. 2017, 384, 287–302. [Google Scholar] [CrossRef] [PubMed]
  54. Li, Y.X.; Tullberg, J.N.; Freebairn, D.M. Wheel traffic and tillage effects on runoff and crop yield. Soil Tillage Res. 2007, 97, 282–292. [Google Scholar] [CrossRef]
  55. Materechera, S.A. Tillage and tractor traffic effects on soil compaction in horticultural fields used for peri-urban agriculture in a semi-arid environment of the North West Province, South Africa. Soil Tillage Res. 2009, 103, 11–15. [Google Scholar] [CrossRef]
  56. Comino, J.R.; Iserloh, T.; Lassu, T.; Cerdà, A.; Keestra, S.D.; Prosdocimi, M.; Ries, J.B. Quantitative comparison of initial soil erosion processes and runoff generation in Spanish and German vineyards. Sci. Total Environ. 2016, 565, 1165–1174. [Google Scholar] [CrossRef] [PubMed]
  57. Capello, G.; Biddoccu, M.; Ferraris, S.; Cavallo, E. Effects of tractor passes on hydrological and soil erosion processes in tilled and grassed vineyards. Water 2019, 11, 2118. [Google Scholar] [CrossRef]
  58. Xie, C.; Cai, S.; Yu, B.; Yan, L.; Liang, A.; Che, S. The effects of tree root density on water infiltration in urban soil based on a Ground Penetrating Radar in Shanghai, China. Urban For. Urban Green. 2020, 50, 126648. [Google Scholar] [CrossRef]
  59. Van Asch, T.W.; Sukmantalya, I.N. The modelling of soil slip erosion in the upper Komering area, South Sumatra Province, Indonesia. Geogr. Fis. Din. Quat. 1993, 16, 81–86. [Google Scholar]
  60. Reid, M.E. A pore-pressure diffusion model for estimating landslide-inducing rainfall. J. Geol. 1994, 102, 709–717. [Google Scholar] [CrossRef]
  61. Lu, N.; Godt, J.W. Infinite slope stability under unsaturated seepage conditions. Water Resour. Res. 2008, 44, W11404. [Google Scholar] [CrossRef]
  62. Lu, N.; Godt, J.W. Hillslope Hydrology and Stability; Cambridge University Press: Cambridge, UK, 2013. [Google Scholar]
  63. Bogaard, T.A.; Greco, R. Landslide hydrology: From hydrology to pore pressure. Wiley Interdiscip. Rev. Water 2016, 3, 439–459. [Google Scholar] [CrossRef]
  64. Meisina, C.; Bordoni, M.; Vercesi, A.; Maerker, M.; Ganimede, C.; Reguzzi, M.C.; Gagnarli, E. Effects of Vineyard Inter-Row Management on Soils; Roots and Shallow Landslides Probability in the Apennines, Lombardy, Italy. Proceedings 2019, 30, 41. [Google Scholar]
Water 15 03340 g001
Figure 1. Location and lithological maps of the chosen test sites.
Figure 1. Location and lithological maps of the chosen test sites.
Water 15 03340 g001
Water 15 03340 g002
Figure 2. Land use in the SVC and location of the shallow landslides thatoccurred between 27–28 April 2009.
Figure 2. Land use in the SVC and location of the shallow landslides thatoccurred between 27–28 April 2009.
Water 15 03340 g002
Water 15 03340 g003
Figure 3. Empirical cumulative distribution function (eCDF) and histogram of the distribution of the source area in the SVC in m2.
Figure 3. Empirical cumulative distribution function (eCDF) and histogram of the distribution of the source area in the SVC in m2.
Water 15 03340 g003
Water 15 03340 g004
Figure 4. Rainfall-induced shallow landslides triggered over the course of different events.
Figure 4. Rainfall-induced shallow landslides triggered over the course of different events.
Water 15 03340 g004
Water 15 03340 g005
Figure 5. Empirical cumulative distribution function (eCDF) and histogram of the distribution of the source areas in the AVC in m2.
Figure 5. Empirical cumulative distribution function (eCDF) and histogram of the distribution of the source areas in the AVC in m2.
Water 15 03340 g005
Water 15 03340 g006
Figure 6. A flow chart of the adopted methodology.
Figure 6. A flow chart of the adopted methodology.
Water 15 03340 g006
Water 15 03340 g007
Figure 7. ECDF of the slope angles both within the source area and in general (SVC).
Figure 7. ECDF of the slope angles both within the source area and in general (SVC).
Water 15 03340 g007
Water 15 03340 g008
Figure 8. ECDF of the slope angles of the source areas for each event (AVC).
Figure 8. ECDF of the slope angles of the source areas for each event (AVC).
Water 15 03340 g008
Water 15 03340 g009
Figure 9. ECDF and histogram of the source area in different land uses in the SVC.
Figure 9. ECDF and histogram of the source area in different land uses in the SVC.
Water 15 03340 g009
Water 15 03340 g010
Figure 10. ECDF of the source area by event in the AVC.
Figure 10. ECDF of the source area by event in the AVC.
Water 15 03340 g010
Water 15 03340 g011
Figure 11. Correlations between the median of shallow landslides source areas and rainfall attributes of the triggering events of the triggering events and of the antecedent conditions before the triggering events.
Figure 11. Correlations between the median of shallow landslides source areas and rainfall attributes of the triggering events of the triggering events and of the antecedent conditions before the triggering events.
Water 15 03340 g011
Water 15 03340 g012
Figure 12. Some landslides occurred at the same location in the AVC after different triggering events. A highlight of the landslides from the February 2016 and November 2019 events is provided.
Figure 12. Some landslides occurred at the same location in the AVC after different triggering events. A highlight of the landslides from the February 2016 and November 2019 events is provided.
Water 15 03340 g012
Table 1. Main characteristics of the source areas of past shallow landslides in the catchments. The table features the number of failures (“Nr”) and minimum (“Min”), maximum (“Max”), median, mean and 1st and 3rd quartiles (“1st Qu.” and “3rd Qu.”) values for the source areas (in m2) recorded in each catchment.
Table 1. Main characteristics of the source areas of past shallow landslides in the catchments. The table features the number of failures (“Nr”) and minimum (“Min”), maximum (“Max”), median, mean and 1st and 3rd quartiles (“1st Qu.” and “3rd Qu.”) values for the source areas (in m2) recorded in each catchment.
CatchmentNrMin
(m2)		1st Qu.
(m2)		Median
(m2)		Mean
(m2)		3rd Qu.
(m2)		Max
(m2)
SVC367123561431572210
AVC316127952192332557
Table 2. Main soil features of a representative soil type of each test site.
Table 2. Main soil features of a representative soil type of each test site.
CatchmentGravel Content
(%)		Sand Content
(%)		Silt Content
(%)		Clay Content
(%)		Liquid Limit
(%)		Plasticity Index
(%)		Unit Weight
(kN/m3)		Saturated Hydraulic Conductivity
(m/s)
SVC0.2–12.37.5–13.251.1–65.621.3–29.038.5–41.814.3–17.216–1910−5–10−7
AVC0.1–2.50.7–3.239.7–46.847.5–57.565.5–73.945.6–53.618–2010−5–10−8
Table 3. A brief overview of the features of rainfall events thattriggered shallow landslides in the AVC test site. The table features the number of failures (“Nr of landslides”), the duration of each event in hours and the cumulated rainfall in millimetres.
Table 3. A brief overview of the features of rainfall events thattriggered shallow landslides in the AVC test site. The table features the number of failures (“Nr of landslides”), the duration of each event in hours and the cumulated rainfall in millimetres.
CatchmentNrDuration (h)Cumulated Rainfall (mm)Nr of Landslides
6–8 February 2009603945.260
18–20 January 2014704252.070
15 November 2014891450.689
27–29 February 20165550101.855
17–19 November 2019237291.623
7–9 December 2020196578.819
Table 4. The table presents the number of landslides per square kilometre for each land use, the number of landslides and the minimum and maximum median slope angle values for each landslide within the class. To gain a better understanding of the distribution of the source areas, landslides are subdivided per land use and then further grouped into three landslide size classes (<50, 50–100, >100 m2).
Table 4. The table presents the number of landslides per square kilometre for each land use, the number of landslides and the minimum and maximum median slope angle values for each landslide within the class. To gain a better understanding of the distribution of the source areas, landslides are subdivided per land use and then further grouped into three landslide size classes (<50, 50–100, >100 m2).
Land UseNr of Landslides per km2Source Area Size ClassesNr of LandslidesMin. Median Slope Angle ValueMax. Median Slope Angle Value
Croplands0.816<50 m21214.2431.25
50–100 m299.1232.92
>100 m2718.6531.21
Shrublands0.292<50 m2353.3348.00
50–100 m2816.9430.64
>100 m22314.3549.26
Vineyards0.229<50 m2594.7036.39
50–100 m2178.0544.01
>100 m22710.8435.58
Woodlands0.753<50 m2626.9248.09
50–100 m23623.8146.33
>100 m27215.3254.72
Table 5. The table presents for each event the number of landslides and the minimum and maximum median slope angle values, grouped into three landslide size classes (<50, 50–100, >100 m2).
Table 5. The table presents for each event the number of landslides and the minimum and maximum median slope angle values, grouped into three landslide size classes (<50, 50–100, >100 m2).
Land UseSource Area Size ClassesNr of LandslidesMin. Median Slope Angle ValueMax. Median Slope Angle Value
February 2009<50 m2311.7523.74
50–100 m277.9719.44
>100 m2221.0822.68
January 2014<50 m2316.6422.14
50–100 m21111.5219.64
>100 m2287.9219.86
November 2014<50 m2388.2120.53
50–100 m2134.8821.75
>100 m2386.7522.86
February 2016<50 m2119.6517.49
50–100 m2109.5217.63
>100 m2349.2224.40
November 2019<50 m2116.8216.82
50–100 m2711.8223.30
>100 m2154.6621.59
December 2020<50 m2310.3820.23
50–100 m2212.6318.44
>100 m21410.6119.32
Table 6. Size of the source area in each land use for the SVC.
Table 6. Size of the source area in each land use for the SVC.
Land UseNrMinimum
(m2)		Maximum
(m2)		Mean
(m2)		Median
(m2)
Croplands28152711563
Shrublands66110152727
Vineyards103112809237
Woodlands1701221018773
Table 7. Results of Dunn’s test: p-value and Z-score for each combination of land uses. Pairs that are statistically similar according to the results of Dunn’s test are in bold characters.
Table 7. Results of Dunn’s test: p-value and Z-score for each combination of land uses. Pairs that are statistically similar according to the results of Dunn’s test are in bold characters.
ComparisonZ (-)p-Value (-)
WoodlandsShrublands1.990.24
WoodlandsCroplands0.980.98
ShrublandsCroplands−0.480.63
WoodlandsVineyards3.490.00
ShrublandsVineyards0.710.96
CroplandsVineyards1.071.00
Table 8. Overview of the size of the source areas in each event of the AVC.
Table 8. Overview of the size of the source areas in each event of the AVC.
EventNrMinimum
(m2)		Maximum
(m2)		Mean
(m2)		Median
(m2)
6–8
February 2009		60199812945
18–20
January 2014		701155616375
15
November 2014		891255727395
27–29
February 2016		5511997320159
17–19
November 2019		23331275218115
7–9
December 2020		1936387164143
Table 9. Results of Dunn’s test: p-value and Z-score for each combination of events. Pairs that are statistically similar according to the results of Dunn’s test are in bold characters.
Table 9. Results of Dunn’s test: p-value and Z-score for each combination of events. Pairs that are statistically similar according to the results of Dunn’s test are in bold characters.
ComparisonZ (-)p-Value (-)
7–9 December 20206–8February 20092.440.16
7–9 December 202027–29 February 2016−0.131.00
6–8 February 200927–29 February 2016−3.630.00
7–9 December 202018–20 January 20142.150.28
6–8 February 200918–20 January 2014−0.491.00
27–29 February 201618–20 January 20143.280.01
7–9 December 202015 November 20141.790.52
6–8February 200915 November 2014−1.141.00
27–29 February 201615 November 20142.840.06
18–20 January 201415 November 2014−0.661.00
7–9 December 202017–19 November 2019−0.050.96
6–8February 200917–19November 2019−2.680.09
27–29 February 201617–19November 20190.081.00
18–20 January 201417–19November 2019−2.380.17
15 November 201417–19November 2019−2.000.37
Table 10. Functions of correlations between the median of shallow landslides source areas and rainfall attributes of the triggering events and of the antecedent conditions before the triggering events, with the corresponding statistical reliability (RMSE index) and significance (F-statistical test).
Table 10. Functions of correlations between the median of shallow landslides source areas and rainfall attributes of the triggering events and of the antecedent conditions before the triggering events, with the corresponding statistical reliability (RMSE index) and significance (F-statistical test).
PredictorFunctionRMSE
(m2)		F
(-)		p-Value
(-)
Event cumulated rainfall amount0.43x1.2616.512.000.02
Event mean intensity70.05x0.4251.50.680.46
3-day antecedent cumulated rainfall amount0.24x1.3918.28.250.04
5-day antecedent cumulated rainfall amount0.08x1.5621.58.050.04
7-day antecedent cumulated rainfall amount0.36x1.2028.05.360.08
14-day antecedent cumulated rainfall amount0.33x1.1633.42.730.17
30-day antecedent cumulated rainfall amount1.85x0.7534.81.040.36
60-day antecedent cumulated rainfall amount2.40x0.6736.10.300.61
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Giarola, A.; Bordoni, M.; Zucca, F.; Meisina, C. Analysis of the Role of Precipitation and Land Use on the Size of the Source Area of Shallow Landslides. Water 2023, 15, 3340. https://doi.org/10.3390/w15193340

AMA Style

Giarola A, Bordoni M, Zucca F, Meisina C. Analysis of the Role of Precipitation and Land Use on the Size of the Source Area of Shallow Landslides. Water. 2023; 15(19):3340. https://doi.org/10.3390/w15193340

Chicago/Turabian Style

Giarola, Alessia, Massimiliano Bordoni, Francesco Zucca, and Claudia Meisina. 2023. "Analysis of the Role of Precipitation and Land Use on the Size of the Source Area of Shallow Landslides" Water 15, no. 19: 3340. https://doi.org/10.3390/w15193340

APA Style

Giarola, A., Bordoni, M., Zucca, F., & Meisina, C. (2023). Analysis of the Role of Precipitation and Land Use on the Size of the Source Area of Shallow Landslides. Water, 15(19), 3340. https://doi.org/10.3390/w15193340

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop