Advertisement

Regional Environmental Change

, Volume 18, Issue 2, pp 437–449 | Cite as

Impacts of past and future land changes on landslides in southern Italy

  • Stefano Luigi Gariano
  • Olga Petrucci
  • Guido Rianna
  • Monia Santini
  • Fausto Guzzetti
Original Article

Abstract

Land use and land cover (LULC), as well as their geographical and temporal variations, affect landslide occurrence and the related risk, in ways that are difficult to determine. Here, we propose a method for the regional analysis of variations in landslide frequency and distribution in response to observed and projected LULC changes. The method is quantitative and reproducible. We test it in Calabria, southern Italy, where a catalogue of 7037 landslides occurred between 1921 and 2010 is available. First, we defined empirical relationships linking the observed LULC variations to landslide occurrence. We found that, in the period 1921–1965, the majority of the landslides occurred in forests and grassland areas, while, in the period 1966–2010, the landslide density became higher in grassland areas, lower in arable lands and remained about constant in the forests. We consider this an evidence of the positive effect of agricultural practices and land management in reducing landslide occurrence. We noticed that both the landslide occurrence and the distribution of the vulnerable elements changed in the observation period. Then, we calculated the projected variations (to 2050) in landslide occurrence related to 32 scenarios of LULC changes. Our projections reveal a modest increase in landslide occurrence in all scenarios, in the range from 0.9 to 3.2%, with an average increase of 2%. Considering all scenarios, we expect an increase in the number of landslides due to LULC variations in 291 municipalities in Calabria (71%), with 4 municipalities where the increase is expected to exceed 50%. We maintain that the observed and the projected variations in the occurrence of landslides in Calabria are related to changes in natural (i.e. the number and distribution of the triggering events) and environmental and societal (i.e. the number and the distribution of the exposed elements) components.

Keywords

Land use Land cover Landslide risk Environmental changes Future projections Regional scale Calabria 

Introduction

Land use and land cover (LULC), as well as their geographical and temporal variations, are known to influence landslide occurrence (Glade 2003; Sidle and Dhakal 2002). They act as landslide predisposing factors, contributing to control landslide hazard, and affect the distribution of the vulnerable elements so contributing to determine landslide risk (Glade 2003; Beguería 2006; Promper et al. 2014). Changes in LULC, including urbanization in hazardous areas, abandonment of rural and mountain areas, agricultural and forest practices, irrigation, deforestation and afforestation, affect the distribution, abundance and activity of landslides. The impact of the changes can be exacerbated or reduced by concurrent variations in the precipitation regimes, foreseen under the effects of the changing climate. For these reasons, LULC changes should be taken into account when analysing past and future changes in landslide risk. Moreover, several feedbacks among climatic, environmental, demographic and economic changes increase the complexity of the question.

Here, we exploit a catalogue of historical landslides that occurred in Calabria, southern Italy, in the 90-year period 1921–2010, and measures and projections of LULC changes in the same region in the periods 1956–2000 and 2000–2050 to empirically study variations in the frequency and geographical distribution of slope failures in relation to known and foreseen changes in LULC. We use the term “land cover” to describe the “physical and biological cover of the earth’s surface” and the term “land use” to describe the “functional dimension or socio-economic purpose” of a territory (EU 2007/2/EC).

Background

The analysis of the effects of climate and environmental changes on landslides is performed adopting modelling or empirical approaches (Gariano and Guzzetti 2016). Modelling approaches investigate variations in the slope stability conditions driven by rainfall and/or pore pressure variations obtained from future rainfall projections, downscaled from global climate models and used as input to physically based, statistical or regional slope stability models. Empirical approaches exploit landslide records to determine spatial and temporal variations in the activity or the frequency of landslides. In this perspective, a long and sufficiently populated catalogue of historical landslides, representing the chronology of all (or almost all) landslide activations recorded in the study area in the investigated period, should be available. The completeness and quality of the landslide records varies with the analysed epoch (Petrucci and Pasqua 2008), depending on the abundance and type of information sources, and the skill of the investigators (Guzzetti et al. 2005).

Several studies adopting an empirical approach have investigated the role of LULC changes on landslides (chiefly in the twentieth century). In particular, a clear role of deforestation in increasing slope instability in several parts of New Zealand, from the late Holocene to the 1990s, was detected by Glade (2003). Remote sensing approaches were exploited to examine variations in landslide occurrences in Mexico, between 1989 and 1999, through multiple satellite images (Alcàntara-Ayala et al. 2006) and in the Spanish Pyrenees, in the twentieth century, through a sequence of aerial photographs (Beguería 2006). Both concluded that land degradation, abandonment of agricultural areas and deforestation can be assumed as precursors of slope instability.

The growth of ploughing driven by EU-sponsored wheat cultivation on steep slopes was recognized partly responsible of the increase in landslide activity in Daunia, southern Italy, in the period 1976–2006 (Wasowski et al. 2010). On the same area, taking into account also reductions in the monthly rainfall and rainfall intensity, an increasing trend in landslide occurrence in urban areas and a decrease in forested areas, from 1928 to 2007, was identified (Lonigro et al. 2015).

Differently, Van Beek (2002) and Van Beek and Van Asch (2004) adopted a modelling, physically based approach and found a marginal decrease in the temporal activity and in the spatial frequency of landslides in a small catchment in Spain under three land use scenarios.

Attempts were also made to study the influence of LULC, and their changes, on landslide susceptibility (e.g. Jaedicke et al. 2014; Nadim et al. 2006; Reichenbach et al. 2014; Persichillo et al. 2017; Pisano et al. 2017; Van Den Eeckhaut et al. 2012) and landslide risk (Papathoma-Köhle and Glade 2013; Promper et al. 2014). For susceptibility analysis, the majority of the approaches usually recognize land use as one of the main proxies regulating landslide occurrence. Reichenbach et al. (2014), using two different land use maps to construct susceptibility models through multivariate statistics, found changes in landslide susceptibility in a 1.7-km2 area in southern Italy from 1954 to 2009, which they attributed to an increase in bare soils and a reduction in forested areas. Promper et al. (2014) used a landslide inventory of almost 700 landslides occurred in Lower Austria, analysed past variations in land cover using aerial photographs and modelled four future scenarios of land cover changes covering the period 1962–2100. Their analysis revealed changes in land cover and in the distribution of vulnerable elements, with the development of new landslide risk hotspots forced by land cover variations.

Data and method

Study area

Calabria extends for 15,080 km2 in southern Italy and comprises 409 municipalities ranging in size from 2.4 km2 to 292.0 km2 (Fig. 1). The average municipality area is 38.4 km2; 224 municipalities have an area < 30 km2. According to the last official Italian census (2011, www.istat.it), 1,959,050 persons live in the region, with a negative trend compared to the previous census (2001). Five municipalities (Reggio di Calabria, Catanzaro, Cosenza, Lamezia Terme and Crotone) have more than 50,000 inhabitants, and 73 municipalities have less than 1000 residents, with an average of 4790 people per municipality. Landscape is hilly or mountainous (five mountain ranges characterizes the region), with lowland areas limited chiefly to the coastal plains. Annual rainfall averages 1150 mm, with more precipitation falling in the mountain areas (> 2000 mm) than in the coastal areas (< 500 mm), and the eastern side of the region dryer than the western side. About 70% of the annual precipitation falls from October to March and only 10% in the summer.
Fig. 1

a Number of landslide events (LEs) in each municipality in Calabria in the 90-year period 1921–2010. b Number (orange bars) and cumulated number (black line) of LE per year in the period 1921–2010. Dashed line shows average number of LE in the 90-year period (colour figure online)

The analysis of rainfall observations reveal the presence of statistical significant decrease in annual and winter amount of rainfall, over two overlapping periods (1920–2000 and 1960–2000; Ferrari and Terranova 2004) and over 1916–2006 (Brunetti et al. 2012) while an opposite tendency is retrieved for summer rainfall (Caloiero et al. 2011; Brunetti et al. 2012).

Catalogue of landslide events in Calabria

We exploited a catalogue of information on the occurrence of 7600 landslides in Calabria from January 1921 to December 2010 (Fig. 1) (Gariano et al. 2015). The catalogue was compiled using multiple sources: local and national newspapers, websites, official reports from administrations and post-event field surveys. For a detailed description of the sources and the catalogue, see Petrucci and Versace (2005, 2007), Petrucci et al. (2009), and Palmieri et al. (2011). All landslides included in the catalogue have produced a reported damage (i.e. the damage was described in a chronicle or technical document) to population or public or private properties; thus, we can refer to them as “damaging landslides”.

We defined a landslide event (LE) as the occurrence of one or more landslides in a given municipality and in a given date (day, month, year) (Gariano et al. 2015), so identifying 7037 LEs in the 90-year period 1921–2010. Each record in the LE catalogue lists the following: (a) a unique LE identification number, (b) the date (and when available the time) of occurrence of the landslide(s), (c) the geographical location of the landslide(s), (d) a description of the landslide(s), (e) an indication of whether “single” or “multiple” (two or more) landslides were reported and (f) information on the size of the landslide(s) (Gariano et al. 2015). For 1618 LEs (23%, mostly occurred after 1970), the exact (street name or coordinates, with an uncertainty of about 1 km2) or approximate (toponym, district, or locality, with an uncertainty of about 10 km2) geographical coordinates of the site(s) where the landslide(s) occurred is known. For the remaining 5419 LEs (77%), only the municipality where the landslide(s) occurred is known. For the latter LE, we attributed to the landslide(s) the geographical coordinates of the centroid of the municipality where the landslide(s) was (were) reported.

Figure 1a shows the number of LE in each of the 409 municipalities in Calabria in the observation period. The average number of LE per municipality is 17; 147 municipalities (36%) have experienced less than 10 LEs and 4 municipalities more than 100 LEs. These include the largest and most populated cities of Catanzaro (142 LEs), Cosenza (120) and Reggio di Calabria (117) and the village of Scilla (104) located in an area highly prone to landslides (Iovine et al. 2014; Terranova et al. 2016). Figure 1b portrays the temporal distribution of the LE, with an average of 78 LEs per year in the observation period. For 17 years (19%), more than 100 LEs were recorded in the region, including 2009 (492 LEs) and 2010 (499 LEs). Figure 1b also shows the cumulated number of LE, a proxy measure for the completeness of the catalogue (Guzzetti 2000; Gariano et al. 2015; Wood et al. 2015). The gradient of the cumulated curve rises after 1950 and increases sharply after 2008.

Observed changes in land use and land cover from 1956 to 2000

To investigate the spatial and temporal evolution of LULC changes in Calabria, we exploited two existing maps. The first is the “land use map” (LUM) prepared in 1956 at 1:200,000 scale (minimum unit map 64 ha) by the Italian National Research Council (CNR) and the Italian Touring Club (TCI). The LUM is the first land use cartographic document produced at the Italian national level and, remarkably, it allows defining the land use in the period just before the recent phase of urbanization and industrialization that has characterized Italy. We obtained the LUM in printed format; thus, we digitized it in a GIS environment for our analysis. The second map is the “CORINE land cover” (CLC) map at 1:100,000 scale (minimum mapping unit 25 ha), released in 2000 by the European Environment Agency, and available in digital format. The two maps do not show the same information. The LUM, prepared using cadastral data and aerial photos, shows 20 classes of land use and classifies the territory according to its “current and future planned functional dimension or socio-economic purpose”. The CLC map, prepared using Landsat7 satellite images, digital elevation models, and aerial photos, shows 40 classes of land cover and classifies the “physical and biological cover of the earth’s surface” (EU 2007/2/EC).

Combining and aggregating heterogeneous data for LULC change analysis may cause multiple errors, including (i) spatial aggregation errors, (ii) classification errors and (iii) thematic aggregation errors (Moody and Woodcock 1994; Petit and Lambin 2002; Falcucci et al. 2007). To avoid classification and thematic errors, after digitizing the LUM, we reclassified both maps using a common legend, in 15 classes (Table 1), loosely based on the third level of the CLC classification. Then, to avoid spatial aggregation errors and to increase comparability between the maps, we applied the common legend to the two maps, adopting the coarser available resolution.
Table 1

Description and codes for the land use land cover (LULC) types used in this study after aggregating the TCI-CNR land use map and the CORINE land cover classes, as well as classes of projections of LULC change

Land use map TCI-CNR

CORINE land cover level 2

CORINE land cover level 3

Code

Urban settlements

1.1. Urban fabric

1.1.1. Continuous urban fabric

0

1.1.2. Discontinuous urban fabric

1.2.1. Industrial or commercial units

1.2. Industrial, commercial, and transport units

1.2.2. Road and rail networks and associated land

1.2.3. Port areas

1.2.4. Airports

1.3. Mine, dump, and construction sites

1.3.1. Mineral extraction sites

1.3.2. Dump sites

1.4. Artificial, non-agricultural vegetated areas

1.4.1. Green urban areas

1.4.2. Sport and leisure facilities

Arable land

2.1. Arable land

2.1.1. Non-irrigated arable land

1

Arable land planted with trees

Arable land

Irrigated land planted with trees

2.1.2. Permanently irrigated land

Garden

Rice fields

2.1.3. Rice fields

Vineyards

2.2. Permanent crops

2.2.1. Vineyards

2

Citrus trees

2.2.2. Fruit trees and berry plantations

Fruit trees

Olive grooves

2.2.3. Olive groves

Pasture areas

2.3. Pastures

2.3.1. Pastures

3

 

2.4. Heterogeneous agricultural areas

2.4.1. Annual crops associated with permanent crops

4

2.4.2 Complex cultivation patterns

1

2.4.3. Land principally occupied by agriculture, with significant areas of natural vegetation

5

2.4.4. Agro-forestry areas

6

Copse forest

3.1. Forests

3.1.1. Broad-leaved forest

7

Tall forest

3.1.2. Coniferous forest

Chestnut

3.1.3. Mixed forest

Mixed forest

Grassland (also planted)

3.2. Scrub and/or herbaceous vegetation associations

3.2.1. Natural grasslands

8

Irrigated grassland (also planted)

3.2.2. Moors and heathland

3.2.3. Sclerophyllous vegetation

9

3.2.4. Transitional woodland shrub

8

Bare areas

3.3. Open spaces with little or

3.3.1. Beaches, dunes, sands

10

no vegetation

3.3.2. Bare rocks

3.3.3. Sparsely vegetated areas

11

3.3.4. Burnt areas

3.3.5. Glaciers and perpetual snow

12

4.1. Inland wetlands

4.1.1. Inland marshes

13

4.1.2. Peat bogs

4.2.1. Salt marshes

4.2. Maritime wetlands

4.2.2. Salines

4.2.3. Intertidal flats

Water bodies

5.1. Inland waters

 

500

5.2. Marine waters

Comparison of the two maps in a GIS revealed several changes in the LULC between 1956 and 2000 (Table 2). The largest difference was observed in the “arable land” class that lost more than 3000 km2 (− 21.6%), followed by “natural grassland” (− 1320 km2; − 8.8%). Conversely, the “heterogeneous agricultural area” and “forested areas” classes increased by 2500 km2 (16.4%) and 1650 km2 (11.0%), respectively. Our results are in agreement with results obtained for the whole of Italy by Falcucci et al. (2007) for the same period.
Table 2

Areas included in each class for the land use map (LUM 1956) and the CORINE land cover map (CLC 2000), as well as related differences are also reported

Class

LUM 1956

CLC 2000

Difference

(km2)

(%)

(km2)

(%)

(km2)

(%)

0

211.26

1.4

459.43

3.0

248.17

1.6

1

5675.38

37.6

2414.36

16.0

− 3261.02

− 21.6

2

2497.76

16.6

2434.60

16.1

− 63.17

− 0.4

3

12.91

0.1

72.81

0.5

59.90

0.4

4-5-6

0.00

0.0

2464.13

16.3

2464.13

16.3

7

3967.98

26.3

5626.32

37.3

1658.34

11.0

8-9

2620.14

17.4

1300.21

8.6

− 1319.93

− 8.8

10-11-12

67.30

0.4

262.83

1.7

195.52

1.3

13

0.00

0.0

0.00

0.0

0.00

0.0

500

25.45

0.2

43.51

0.3

18.07

0.1

Projected changes in land use and land cover from 2000 to 2050

An ensemble of 32 simulations of the likely LULC changes between 2000 and 2050 was obtained through the land use change model LUC@CMCC, at 500 × 500 m spatial resolution and for 14 LULC classes aggregated from the third level of the CLC classification (Table 1), as proposed by Santini and Valentini (2011). The LUC@CMCC model, an extension of the CLUE-S model (Verburg et al. 2002), has a statistical-empirical basis, applying a stepwise logistic regression between multiple explanatory factors (EFs) and the presence of a given land use type. The LUC@CMCC model can consider multiple static and dynamic EFs that condition the presence of a given land use type. Static factors are not expected to change in the period covered by the investigation (i.e. multiple decades), whereas dynamic factors change at the yearly to decadal scales. In this perspective, we assumed year 2050 as the future reference for our work. Model evaluation is conducted during model calibration, in fact part (70%) of the sample, made of land use locations and related local value of EFs, was used to calibrate the model and the remaining part (30%) for validating it. Table 3 reports the average rank of influence of 14 EFs on LULC classes. The highest influence is associated to topographic settings; concerning climate variables, in general, temperature is significant for all the land uses, while precipitation seems important for areas with high vegetation, even if less than temperature. Besides EFs, other spatial and non-spatial rules drive the LUC@CMCC model (Santini and Valentini 2011). Spatial rules include the possibility (or the lack of possibility) to change LULC in a given area, e.g. because it is a protected natural area. Non-spatial rules include the spatial and temporal influence among land uses: a new urban area likely develops near an already urbanized area. Moreover, some transitions are quicker than other, e.g. from forest to agriculture (after tree cutting) rather than from a bare land to a dense forest, and other changes are less credible, e.g. an urban area that turns into a forest in few decades.
Table 3

Explanatory factors adopted in the simulation chain for the determination of land use land cover (LULC) scenarios and their average rank across LULC classes

Explanatory factors

Type

Average rank

Soil carbon content

Static

13

Soil clay content

Static

10

Soil silt content

Static

11

Soil sand content

Static

12

Soil pH

Static

3

Soil density

Static

2

Labour force in agriculture

Dynamic, rescaled according to population density

7

Labour force in commerce

Dynamic, rescaled according to population density

4

Labour force in industry

Dynamic, rescaled according to population density

9

Labour force in institutions

Dynamic, rescaled according to population density

8

Slope

Static

6

Topographic index

Static

1

Annual precipitation amount

Dynamic, based on climate simulations

14

Mean annual temperature

Dynamic, based on climate simulations

5

The 32 simulations (ensemble members) were obtained combining all the alternatives listed below:
  • Two degrees of demographic increase, moderate (m) and strong (s), given by the Italian National Institute of Statistics (ISTAT), driving scenarios on labour force employed in different economic sectors

  • The existence (p) or not existence (np) of areas preserved from LULC change (i.e. protected areas)

  • The consideration (e) or not (ne) of the spatial influence among land uses

  • The consideration (t) or not (nt) of the temporal influence among land uses

  • Climate projections obtained from the regional climate models PROTHEUS (Artale et al. 2009) (E) or LMDZ-Med/NEMO-Med8 (http://www.lmd.jussieu.fr/~li/circe/) (I), forced by IPCC-A1B emission scenario. On average over Calabria, the E and I models project an increase of 1% and decrease of 15% of annual precipitation and an increase of 1 and 2.3 °C of temperature, respectively, allowing to consider moderate to strong changes in climate.

A unique alphanumeric string resulting from the combination of the above bold characters, according to the chosen alternatives, characterized each simulation.

Land use land cover changes and past and projected landslide occurrences

To evaluate the relationships between landslide occurrence and LULC changes, we first analysed the past relationship, using LUM and CLC map and the LE catalogue. Since our catalogue of LE is at the municipality scale, we attributed a prevailing LULC class to each of the 409 municipalities in Calabria. For the purpose, we attributed to each municipality the LULC class that covered more than 50% of the area of the municipality with a terrain gradient > 8.5° computed from a 20-m × 20-m digital elevation model. The 8.5° threshold was selected heuristically to exclude flat areas where landslides cannot occur. Moreover, we compared the areas selected with the above defined threshold with the non-susceptible landslide areas in Italy defined by Marchesini et al. 2014. We found no overlaps between the two raster maps, thus corroborating our heuristic choice. Furthermore, the 8.5° threshold allows eliminating areas with low and very low slope factor, as defined by Nadim et al. (2006) in the characterization of landslide hazard for a global landslide susceptibility map.

In municipalities (33.8% of the total) where no single LULC class covered more than 50% of the areas with terrain gradient > 8.5°, the class of LULC covering the largest proportion of slopes is used.

We split the LE catalogue in two non-overlapping 45-year temporal subsets: the period 1921–1965 (P1) listing 2265 LEs and 1966–2010 (P2) listing 4474 LEs. The LEs in the older period (P1) were attributed the LULC classes shown in the LUM (1956) map, and the LEs in the recent period (P2) were attributed the LULC classes shown in the CLC (2000) map.

Next, we grouped the municipalities with the same attributed LULC in each period and calculated the total number of LE in each LULC class listed in Table 4. Since the number of LE in the two periods was different (2265 in P1, and 4474 in P2), first we calculated the value of the number of LE in each LULC classes normalized by the total number of LE in the period (fourth and eighth columns in Table 4). Then, we calculated the percentage variations, between P1 and P2, in the normalized number of LE occurred in each class. In this way, we obtained the rates of change (percent variation) of the normalized values of LE in each LULC class between the two periods.
Table 4

Municipalities and land slides for the land use land cover classes, as defined in Table 1

LULC class

P1

P2

#Mun

#LE

N#LE

DR

#LE

#LE

#LE

#LE

0

0

0

0

 

0

0

0

 

1

57

290

12.8

0.89

8

124

2.6

0.54

2

114

562

24.8

1.08

78

794

16.6

1.12

3

0

0

0

 

3

36

0.8

1.85

4

0

0

0

 

46

606

12.7

 

5

0

0

0

 

31

398

8.3

1.02

6

0

0

0

 

1

2

0.1

 

7

198

1235

54.5

1.18

232

2631

55.1

0.99

8

40

178

7.9

0.48

9

107

2.2

1.21

9

0

0

0

 

1

74

1.6

 

10

0

0

0

 

0

0

0

 

11

0

0

0

 

0

0

0

 

12

0

0

0

 

0

0

0

 

13

0

0

0

 

0

0

0

 

500

0

0

0

 

0

0

0

 

For P1 (1921–1965) and P2 (1966–2010) periods, the following are reported, per each LULC class: the number of municipalities with LE (#Mun); the number of landslide events (#LE); the number of landslide events normalized by the total number of LE in the period (N#LE); and a density ratio (DR, see text for explanation)

Moreover, in order to standardize the observed variations in the relation between LE and LULC, we calculated a density ratio (DR)
$$ DR=\frac{\#{LE}_{LULC}/{A}_{LULC}}{\#{LE}_{tot}/{A}_{tot}} $$
where #LE LULC is the number of LE occurred in municipalities with a given prevailing LULC class; A LULC is the total area of the municipalities with the same prevailing LULC class; #LE tot is the total number of LE occurred in the whole study area; and A tot is the area of the region. As for the frequency ratio, often used in landslide susceptibility analysis (e.g. Lee and Talib 2005; Regmi et al. 2010; Persichillo et al. 2017; Pisano et al. 2017), high values of DR indicate a greater propensity to landsliding of the LULC. Values > 1 indicate that the density of LE in the LULC class is higher than the density in the entire region, while the opposite occurs for values < 1.

Furthermore, in order to define future LE variations, we calculated the projected changes in the number of LE in each municipality, as a function of the rates of change in landslide occurrence related to LULC changes in “past” and “future” and of the projected LULC classes defined in “Projected LULC changes from 2000 to 2050” section. Finally, adopting the 32 projections of LULC change and the rates of change in landslide occurrence related to the LULC change, for each municipality, we obtained 32 values for the future variation in the number of LE.

Results and discussion

Past changes in the spatial and temporal distributions of landslide events

Figure 2 shows the spatial distribution of LE in P1 (1921–1965) and P2 (1966–2010) periods. In both periods, the municipality with the largest number of LE is Catanzaro, with 30 LEs in P1 and 112 LEs in P2. In the second period, the municipalities of Cosenza and Reggio di Calabria experienced 96 and 92 LEs, respectively. Apart from a diverse total number of LE in the two periods (2265 in P1 and 4474 in P2), related also to the increasing information availability during the 90-year period, some differences in the distributions of LE in the two periods, can be recognized. During P1, the municipalities with the largest number of LE were located mostly in the inner, mountainous part of the region (Fig. 2). Conversely, in P2, municipalities with large number of LE (> 20) were located in hilly terrain along the eastern and western coasts. This result is linked to the fact that our catalogue is made by damaging landslides, i.e. landslides that affected vulnerable elements causing reported damages. Historically, in Calabria, after land reclamation occurred in the second part of the P1 period, the population migrated towards coastal areas, increasing urban pressure where communication infrastructures could be easily built (Petrucci and Polemio 2007). This means that not only landslide hazard changed in the twentieth century in Calabria, but also the distribution of the vulnerable elements and, consequently, landslide risk. Considering the entire period 1921–2010, several municipalities (95) that experienced < 10 LE are located in the inner mountainous part of the region (including the three municipalities with the lowest number of LE; Fig. 1a), whereas some municipalities with numerous (> 20) LE are located along the external parts of the region, particularly along the E coast (Figs. 1a and 2). Such findings stress the significance of proper jointly evaluation of landslide hazard and vulnerability. In the same direction, Ferrara et al. (2015), overlapping European landslide susceptibility map reporting municipal boundaries with several indicators at municipal level, stated that landslide risk is strictly associated with socio-economic factors and agro-environmental factors, in particular to a socio-environmental spiral based on land abandonment.
Fig. 2

Maps showing the number of landslide events (LEs) in each municipality in Calabria in a 1921–1965 (P1) and b 1966–2010 (P2) periods

Relationship between landslide events and land use and land cover in the past

Figure 3 shows the prevailing LULC class assigned to each municipality in Calabria in P1 and P2 periods. For 173 municipalities, the prevailing LULC changed between the two periods, while it remained the same for the remaining 263. The average variation, in absolute value, in the number of LE occurred between P1 and P2 in the 173 municipalities that experienced LULC changes is equal to 8.5, while it is equal to 6.3 for municipalities with the same LULC. Moreover, considering the normalized values of LE, the differences are higher: 0.24 average variation in the normalized number of LE for the first ones and only 0.03 for the second ones.
Fig. 3

Prevailing land use land cover class in each municipality in Calabria in a 1921–1965 (P1) and b 1966–2010 (P2) periods. Key: 0 (red), urban settlements; 1 (pink), arable land; 2 (yellow), permanent crops; 3 (brown), pastures; 4-5-6 (light green), heterogeneous agricultural areas; 7 (dark green) forests; 8-9 (dark brown), natural grassland; 10-11-12 (grey), areas with little or no vegetation; 13 (light blue), wetlands; 500 (blue), water bodies (colour figure online)

Table 4 lists the number of LE occurred during P1 and P2 in each LULC class. In P1, most of the LEs (1235 out of 2265, 54.5%) occurred in forested areas, and the remaining LE in municipalities were characterized by arable land (290 LEs, 12.8%), permanent crops (562 LEs, 24.8%) and grassland (187 LEs, 7.9%). In P2, the distribution of the LE is more heterogeneous. Also, in this period, most of the LE occurred in forested areas (2631 LEs out of 4772, 55.1%), followed by arable land (794 LEs, 16.6%), annual crops associated with permanent crops (606 LEs, 12.7%) and heterogeneous agricultural areas (398 LEs, 8.3%). The remaining 343 LEs (7.2%) are distributed in 5 classes.

The spatial distribution of the LE differs in the two periods, being more heterogeneous in P2 (LE distributed in nine classes) than in P1 (only four classes). This is also related to the different distribution of prevailing LULC classes in the municipalities, partially due to the diverse scale and original detail of LULC information. During P1, the 409 municipalities have 4 prevailing LULC classes; conversely, in P2, they have 9 different classes (Table 4). Certainly, the different classification of the two maps used in the present work play a role in these distributions. Passing from P1 to P2, in wide parts of the region, there was a clear modification in LULC from forested to agricultural use. Overall, in P2, an increase in the total number of LE in municipalities characterized by forested and heterogeneous agricultural areas, and a decrease of landslides in arable land can be recognized.

Nevertheless, analysing the density ratio (DR; Table 4), the density of LE in municipalities with prevailing forested LULC class remained almost the same between the two periods (1.18 in P1 and 0.99 in P2). Contrariwise, high differences were observed in the DR related to arable land (decrease from 0.89, in P1, to 0.54, in P2) and to grassland (huge increase from 0.48, in P1, to 1.21, in P2). Thus, in the observed periods, DR increased in the municipalities having grassland as prevailing LULC class, decreased in those with prevailing arable land and did not show significant changes in those characterized by prevailing forested areas. These findings highlight the positive effect of forests, agricultural practices, and land management in reducing slope instability.

On the beneficial effects of forests and agricultures practices, conflicting findings are retrievable in literature; they are pointed out by several authors (e.g. Alcàntara-Ayala et al. 2006; Beguería 2006; García-Ruiz et al. 2010; Bruschi et al. 2013; Persichillo et al. 2017; Pisano et al. 2017) but disregarded in other test cases (e.g. Rickli and Graf 2009; Marston 2010), making new studies necessary (Papathoma-Köhle and Glade 2013).

Given that our landslide catalogue does not contain precise information on landslide dimension, we could not analyse the influence on LULC changes on the size distribution, as made, e.g. by Rickli and Graf (2009) and Guns and Vanacker (2014).

Projected variations of landslide events as a function of changes in land use land cover

Exploiting the 32 members of LULC change simulations (“Projected LULC changes from 2000 to 2050” section), and the rates of future variation in LE occurrence in each municipality (“LULC changes and past and projected landslide occurrences” section), we prepared 32 maps of projected variation in the LE occurrences in Calabria. Table 5 lists the average variations in the number of LE per each scenario. Figure 4 shows three maps of the LE variation due to the projected LULC changes, namely:
  1. (i)

    The best-case scenario (BS) (Fig. 4a, b), with the lowest increase in the number of LE and related to the “m_np_ne_nt_I” scenario (cf. “Projected LULC changes from 2000 to 2050” section)

     
  2. (ii)

    The worst-case scenario (WS) (Fig. 4e, f), with the largest increase in LE occurrence and related to the “s_p_e_t_E” scenario (cf. “Projected LULC changes from 2000 to 2050” section)

     
  3. (iii)

    The mean scenario (MS), with the average value of variation in LE occurrence among the 32 simulations. The related LULC scenario (Fig. 4c) was defined considering the modal value of LULC among the 32 simulations.

     
Table 5

Average variations in the normalized number of LE for the 32 scenarios of LULC change, described in the “Projected LULC changes from 2000 to 2050” section

Scenario

Variation (%)

m_np_e_nt_E

1.2

m_np_e_t_E

2.0

m_np_ne_nt_E

2.0

m_np_ne_t_E

1.9

m_p_e_nt_E

1.1

m_p_e_t_E

1.7

m_p_n_e_nt_E

1.8

m_p_ne_t_E

1.6

s_np_e_nt_E

2.3

s_np_e_t_E

2.8

s_np_ne_nt_E

2.5

s_np_ne_t_E

1.8

s_p_e_nt_E

2.7

s_p_e_t_E

3.2

s_p_ne_nt_E

2.8

s_p_ne_t_E

2.6

m_np_e_nt_I

1.8

m_n_p_e_t_I

2.0

m_np_ne_nt_I

0.9

m_np_ne_t_I

1.7

m_p_e_nt_I

1.8

m_p_e_t_I

1.8

m_p_ne_nt_I

1.2

m_p_ne_t_I

2.0

s_np_e_nt_I

2.0

s_np_e_t_I

2.5

s_np_ne_nt_I

1.7

s_np_ne_t_I

2.2

s_p_e_nt_I

2.0

s_p_e_t_I

2.7

s_p_ne_nt_I

1.9

s_p_ne_t_I

2.7

In italic and in bold, the best- and worst-case scenarios, respectively

Fig. 4

Prevailing land use land cover (LULC) class for municipalities in Calabria (a, c, e, key as for Fig. 3) and future variation landslide event (LE) occurrence (b, d, f), in the best-case (BS, a, b), mean (MS, c, d), and worst-case (WS, e, f) scenarios

Although the different importance of the configuration alternatives, the variations could be ascribed to all of them. More in detail, climate and socio-economic inputs are treated as EFs in the stepwise logistic regression, although precipitation is removed for some LULC classes since not important. Demography also drives the land demands (i.e. the amount of surface to be allocated for each LULC class in the simulation domain). Spatial relationships among LULC classes are treated still through a stepwise logistic regression to understand how a given LULC is explained by the neighbouring LULC, while the temporal relationships are expressed as follows: (i) general elasticity to conversion (e.g. forest cut is easier and faster than forest growth; urbanization is more plausible than urban areas removal) and (ii) specific elasticity considering what effectively happened in the past. Finally, in a protected area, LULC changes are not allowed. Spatial and temporal constraints among LULC classes and the presence of protected areas concentrate land available to be modified in a reduced portion of the whole domain, increasing the intensity of LULC changes per surface. Moreover, the degree of demographic increase, correlated to the extent of the LULC modifications, could clearly entail worst slope stability conditions. However, several variations among the three scenarios, especially related to the agricultural and forested LULC classes, are described below.

Considering all future scenarios, forested areas (class 7) are the most frequent class, given that 245 out of the 409 municipalities (60%) have it as modal value of prevailing LULC. This is followed by arable land (class 2), with 112 municipalities. Remarkably, 282 municipalities (68.9%) have the same LULC class in all 32 scenarios, including 268 (65.5%) municipalities with no variation from the prevailing LULC in the 2000 CLC map. For 201 out of the mentioned 282 municipalities having the same LULC class in all scenarios, the prevailing class is the forested areas.

We calculated the average variation in LE occurrence for each scenario, and in each municipality. All scenarios produced an increase in the projected LE occurrence, ranging from 0.9 to 3.2% (2% on average considering all scenarios and municipalities; Table 5). Considering all scenarios, 291 municipalities (71%) will experience an increase of LE. Considering the MS, four municipalities will experience a mean increase in LE occurrence greater than 50% (Parghelia, 55.1%; Santa Maria del Cedro, 55.1; Cirò, 51.8%; Villa San Giovanni, 50.9%; Fig. 4d). A mean projected increase in LE occurrence greater than 10% will characterize 80 municipalities (20%), while 31 will experience a normalized increase in LE greater than 20%. Those municipalities are mostly located in the northern part of the region, in hilly areas, and are mainly characterized by a future forested prevailing LULC (class 7). The municipalities of Scalea and Tropea (tourist areas located in the western coast of the region) with artificial surfaces (LULC class 0) as prevailing LULC in MS will experience a high increase in LE occurrence, between 10 and 50%. Since that LULC class is related to an increase of the elements at risk, this variation will result in an increase in both landslide hazard and risk. On the other hand, 118 municipalities (28.9%) will be characterized by a reduction in LE occurrence, and, among those, the variation will be <− 50% for 12 municipalities (characterized by forest and grassland as modal value of prevailing LULC class and mostly located in the coastal areas) (Fig. 4c, d). The municipality of Catanzaro, which registered the highest number of LE both in P1 and P2 periods, will be characterized, on average, by a decrease in LE occurrence (related to LULC changes) equal to − 8.2% by the end of 2050. This is related to the transition from forested areas, in P1, to arable land, in P2 and in MS. In this case, the LULC variation might lead to a reduction in LE occurrence.

In BS (Fig. 4a, b), 104 municipalities (25.4%), mostly located in the coastal plains, will experience a reduction in LE occurrence, and 20 of them will have a significant reduction (<− 10%). Those municipalities will be characterized by arable land and grassland as prevailing LULC class. Conversely, only 12 municipalities (characterized by forested prevailing LULC and mostly localized in the Southern and eastern part of the region) will experience a significant increase (> 50%) in LE occurrence.

In WS (Fig. 4e, f), 28 municipalities (6.8%), heterogeneously distributed along the region and having prevailing forested LULC (class 7), will experience a (normalized) increase in LE occurrence greater than 50%. Almost 20% of the municipalities (79 out of 409, prevailing LULC classes 2, 4 and 7) will experience an increase > 10%. Those municipalities are mostly localized in the hilly areas of the region. Only 19 coastal municipalities (4.6%) will have a significant reduction (variation <− 10%) in LE occurrence.

Conclusions

We evaluated the spatial and temporal variations in landslide occurrence at regional scale under the effect of LULC changes. The landslide catalogue, at municipal scale, considers only landslides that have caused reported damage to either population and/or properties, and it can be considered a representative sample, complete in time and space, of the damaging landslides occurred in the study area and a complete sample of damaging landslides occurred in a relatively large region throughout a rather long period. Selection of municipalities as reference spatial unit (e.g. in assigning a prevailing LULC class) enables effective matching between environmental and socio-economic data. However, our method could be reapplied using diverse unit of analysis.

We found that, in the period 1921–1965, the majority of LE occurred in forests and grassland areas. Conversely, during 1966–2010, the LE is distributed in a larger number of LULC classes. Moreover, considering the density of LE in each LULC class, we found that the density of LE increased in grassland, decreased in arable land and remained quite stable in forested areas, thus highlighting the positive effect of agricultural practices and land management in reducing slope instability. The distribution of damaging landslides followed the population distribution (i.e. the distribution of the vulnerable elements) in the twentieth century in Calabria (Petrucci and Polemio 2007; Gariano et al. 2015).

Looking at the future, we further found that all of the analysed scenarios of expected variation in LULC resulted in an increase in the LE occurrence, ranging from 0.9 to 3.2%, with an average value equal to 0.2% (Table 5). In particular, an increase in the number of LE due to LULC variations is expected for 291 municipalities (71% of the total), while for four municipalities, the increase is expected to exceed 50%. However, we maintain that LULC change is not the only factor controlling landslide occurrence. Obviously, the combined effect of LULC change, weather forcing, and geo-lithological conditions affect landslides. Furthermore, climatic, environmental, demographic and economic changes are also strictly correlated, with several feedbacks, to landslide occurrence and variation (Sidle and Ochiai 2006; Gariano and Guzzetti 2016). Moreover, human actions—and disturbances—affect significantly landslide risk, given that they influence both landslide hazard and the distribution of vulnerable elements. We maintain that the observed and expected variations in the occurrence of damaging landslides in Calabria are due to changes in natural components (i.e. changes in triggering and predisposing factors leading to variations in the number of the events) and in environmental and societal components (i.e. changes in the number and distribution of the exposed elements).

The proposed method is reproducible, and we expect that it will be used for landslide hazard and risk assessment studies in similar regions affected by landslide risk. We further expect that the obtained results and suggested method will be used for environmental and land use planning and for landslide risk management by public administrations in Calabria, also taking advantage of the release of new information and scenarios.

Notes

Acknowledgements

We are grateful to Luca Pisano for fruitful discussions and to Paola Giostrella for having helped us find the CNR-TCI land use map of Calabria. The manuscript greatly benefited from the constructive comments and suggestions of the editor and the two anonymous reviewers, definitely.

References

  1. Alcàntara-Ayala I, Esteban-Chavez O, Parrot JF (2006) Landsliding related to land-cover change: a diachronic analysis of hillslope instability distribution in the Sierra Norte, Puebla, Mexico. Catena 65(2):152–165. doi: 10.1016/j.catena.2005.11.006 CrossRefGoogle Scholar
  2. Artale VS, Calmanti AC, Dell’Aquila A, Herrmann M, Pisacane G, Ruti PM, Sannino G, Struglia MV, Giorgi F, Bi X, Pal JS, Rauscher S, The PROTHEUS Group (2009) An atmosphere ocean regional climate model for the Mediterranean area: assessment of a present climate simulation. Clim Dyn 35(5):721–740. doi: 10.1007/s00382-009-0691-8 CrossRefGoogle Scholar
  3. Beguería S (2006) Changes in land cover and shallow landslide activity: a case study in the Spanish Pyrenees. Geomorphology 74(1–4):196–206. doi: 10.1016/j.geomorph.2005.07.018 CrossRefGoogle Scholar
  4. Brunetti M, Caloiero T, Coscarelli R, Gullà G, Nanni T, Simolo C (2012) Precipitation variability and change in the Calabria region (Italy) from a high resolution daily dataset. Int J Climatol 32:57–73. doi: 10.1002/joc.2233 CrossRefGoogle Scholar
  5. Bruschi VM, Bonachea J, Remondo J, Gómez-Arozamena J, Rivas V, Barbieri M, Capocchi S, Soldati M, Cendrero A (2013) Land management versus natural factors in land instability: some examples in northern Spain. Environ Manag 52(2):398–416. doi: 10.1007/s00267-013-0108-7 CrossRefGoogle Scholar
  6. Caloiero T, Coscarelli R, Ferrari E, Mancini M (2011) Precipitation change in Southern Italy linked to global scale oscillation indexes. Nat Hazards Earth Syst Sci 11:1683–1694. doi: 10.5194/nhess-11-1683-2011 CrossRefGoogle Scholar
  7. EU Parliament Directive 2007/2/EC (2007) Establishing an Infrastructure for Spatial Information in the European Community (INSPIRE). Official Journal of the European Union, 14 March 2007. L108/1-L180/14Google Scholar
  8. Falcucci A, Maiorano L, Boitani L (2007) Changes in land-use/land cover patterns in Italy and their implications for biodiversity conservation. Landsc Ecol 22:617–631. doi: 10.1007/s10980-006-9056-4 CrossRefGoogle Scholar
  9. Ferrara C, Matteo Barone P, Salvati L (2015) Unravelling landslide risk: soil susceptibility, agro-forest systems and the socio-economic profile of rural communities in Italy. Soil Use Manage 31(2):290–298. doi: 10.1111/sum.12185 CrossRefGoogle Scholar
  10. Ferrari E, Terranova OG (2004) Non-parametric detection of trends and change point years in monthly and annual rainfalls. In: Proceedings of 1st Italian–Russian workshop—new trends in hydrology, CNR Publications, Rende (CS), pp 177–188Google Scholar
  11. García-Ruiz JM, Beguería S, Alatorre LC, Puigdefábregas J (2010) Land cover changes and shallow landsliding in the flysch sector of the Spanish Pyrenees. Geomorphology 124(3–4):250–259. doi: 10.1016/j.geomorph.2010.03.036 CrossRefGoogle Scholar
  12. Gariano SL, Guzzetti F (2016) Landslides in a changing climate. Earth Sci Rev. doi: 10.1016/j.earscirev.2016.08.011
  13. Gariano SL, Petrucci O, Guzzetti F (2015) Changes in the occurrence of rainfall-induced landslides in Calabria, Southern Italy, in the 20th century. Nat Hazard Earth Syst Sci 15:2313–2330. doi: 10.5194/nhess-15-2313-2015 CrossRefGoogle Scholar
  14. Glade T (2003) Landslide occurrence as a response to land use change: a review of evidence from New Zealand. Catena 51:297–314. doi: 10.1016/s0341-8162(02)00170-4 CrossRefGoogle Scholar
  15. Guns M, Vanacker V (2014) Shifts in landslide frequency–area distribution after forest conversion in the tropical Andes. Anthropocene 6:75-85. doi: 10.1016/j.ancene.2014.08.001
  16. Guzzetti F (2000) Landslide fatalities and the evaluation of landslide risk in Italy. Eng Geol 58(2):89–107. doi: 10.1016/S0013-7952(00)00047-8 CrossRefGoogle Scholar
  17. Guzzetti F, Salvati P, Stark CP (2005) Evaluation of risk to the population posed by natural hazards in Italy. In: Hungr O, Fell R, Couture R, Eberhardt E (eds) Landslide risk management. Taylor & Francis Group, London, pp 381–389Google Scholar
  18. Iovine G, Greco R, Gariano SL, Iaquinta P, Pellegrino AD, Terranova OG (2014) Shallow-landslide susceptibility in the Costa Viola mountain ridge (Southern Calabria, Italy) with considerations on the role of causal factors. Nat Hazards 73:111–136. doi: 10.1007/s11069-014-1129-0 CrossRefGoogle Scholar
  19. Jaedicke C, Van Den Eeckhaut M, Nadim F, Hervás J, Kalsnes B, Vangelsten BV, Smith JT, Tofani V, Ciurean R, Winter MG, Sverdrup-Thygeson K, Syre E, Smebye H (2014) Identification of landslide hazard and risk “hotspots” in Europe. Bull Eng Geol Environ 73(2):325–339. doi: 10.1007/s10064-013-0541-0 Google Scholar
  20. Lee S, Talib JA (2005) Probabilistic landslide susceptibility and factor effect analysis. Environ Geol 47:982–990. doi: 10.1007/s00254-005-1228-z CrossRefGoogle Scholar
  21. Lonigro T, Gentile F, Polemio M (2015) The influence of climate variability and land use variations on the occurrence of landslide events (Subappennino Dauno, Southern Italy). Rend Online Soc Geol It 35:192–195. doi: 10.3301/ROL.2015.98 Google Scholar
  22. Marchesini I, Ardizzone F, Alvioli M, Rossi M, Guzzetti F (2014) Non-susceptible landslide areas in Italy and in the Mediterranean region. Nat Hazards Earth Syst Sci 14:2215–2231. doi: 10.5194/nhess-14-2215-2014 CrossRefGoogle Scholar
  23. Marston RA (2010) Geomorphology and vegetation on hillslopes: interactions, dependencies and feedback loops. Geomorphology 116:206–217. doi: 10.1016/j.geomorph.2009.09.028 CrossRefGoogle Scholar
  24. Moody A, Woodcock CE (1994) Scale-dependent errors in the estimation of land-cover datasets. Photogramm Eng Remote Sens 60:585–594.Google Scholar
  25. Nadim F, Kjekstad O, Peduzzi P, Herold C, Jaedicke C (2006) Global landslide and avalanche hotspots. Landslides 3(2):159–173. doi: 10.1007/s10346-006-0036-1 CrossRefGoogle Scholar
  26. Palmieri W, Petrucci O, Versace P (2011) La difesa del suolo nell’Ottocento nel Mezzogiorno d’Italia. IV Quaderno dell’Osservatorio di Documentazione Ambientale (Dip. Difesa del Suolo, UNICAL), Rubbettino, Soveria Mannelli, Italy, pp 184 (in Italian)Google Scholar
  27. Papathoma-Köhle M, Glade T (2013) The role of vegetation cover change for landslide hazard and risk. In: Renaud F, Sudmeier-Rieux K, Estrella M (eds) The role of ecosystems in disaster risk reduction. UNU-Press, Tokyo, pp 293–320Google Scholar
  28. Persichillo MG, Bordoni M, Meisina C (2017) The role of land use changes in the distribution of shallow landslides. Sci Total Environ 574:924–937. doi: 10.1016/j.scitotenv.2016.09.125 CrossRefGoogle Scholar
  29. Petit CC, Lambin EF (2002) Impact of data integration technique on historical land-use/land-cover change: comparing historical maps with remote sensing data in the Belgian Ardennes. Landsc Ecol 17:117–132. doi: 10.1023/A:1016599627798 CrossRefGoogle Scholar
  30. Petrucci O, Pasqua AA (2008) The study of past damaging hydrogeological events for damage susceptibility zonation. Nat Hazards Earth Syst Sci 8:881–892. doi: 10.5194/nhess-8-881-2008 CrossRefGoogle Scholar
  31. Petrucci O, Polemio M (2007) Flood risk mitigation and anthropogenic modifications of a coastal plain in Southern Italy: combined effects over the past 150 years. Nat Hazards Earth Syst Sci 7:361–373CrossRefGoogle Scholar
  32. Petrucci O, Versace P (2005) Frane e alluvioni in provincia di Cosenza agli inizi del ‘900: ricerche storiche nella documentazione del Genio Civile. I Quaderno dell’Osservatorio di Documentazione Ambientale (Dip. Difesa del Suolo, UNICAL), Editoriale Bios, Rende, Italy, 172 pp (in Italian)Google Scholar
  33. Petrucci O, Versace P (2007) Frane e alluvioni in provincia di Cosenza tra il 1930 e il 1950: ricerche storiche nella documentazione del Genio Civile. II Quaderno dell’Osservatorio di Documentazione Ambientale (Dip. Difesa del Suolo, UNICAL), Pubbl. GNDCI N. 2913, Editoriale Bios, Rende, Italy, 247 pp (in Italian)Google Scholar
  34. Petrucci O, Versace P, Pasqua AA (2009) Frane e alluvioni in provincia di Cosenza fra il 1951 ed il 1960: ricerche storiche nella documentazione del Genio Civile. III Quaderno dell’Osservatorio di Documentazione Ambientale (Dip. Difesa del Suolo, UNICAL), Rubbettino, Soveria Mannelli, Italy, 175 pp (in Italian)Google Scholar
  35. Pisano L, Zumpano V, Malek Z, Rosskopf CM, Parise M (2017) Variations in the susceptibility to landslides, as a consequence of land cover changes: a look to the past, and another towards the future. Sci Total Environ. doi: 10.1016/j.scitotenv.2017.05.231
  36. Promper C, Puissant A, Malet J-P, Glade T (2014) Analysis of land cover changes in the past and the future as contribution to landslide risk scenarios. Appl Geogr 53:11–19. doi: 10.1016/j.apgeog.2014.05.020 CrossRefGoogle Scholar
  37. Regmi NR, Giardino JR, Vitek JD (2010) Modeling susceptibility to landslides using the weight of evidence approach: western Colorado, USA. Geomorphology 115(1–2):172–187. doi: 10.1016/j.geomorph.2009.10.002 CrossRefGoogle Scholar
  38. Reichenbach P, Busca C, Mondini AC, Rossi M (2014) The influence of land use change on landslide susceptibility zonation: the Briga catchment test site (Messina, Italy). Env Manag 54:1372–1384. doi: 10.1007/s00267-014-0357-0 CrossRefGoogle Scholar
  39. Rickli C, Graf F (2009) Effects of forests on shallow landslides—case studies in Switzerland. For Snow Landsc Res 82(1):33–44Google Scholar
  40. Santini M, Valentini R (2011) Predicting hot-spots of land use changes in Italy by ensemble forecasting. Reg Environ Chang 11:483–502. doi: 10.1007/s10113-010-0157-x CrossRefGoogle Scholar
  41. Sidle RC, Dhakal AS (2002) Potential effects of environmental change on landslide hazards in forest environments. In: Sidle RC (ed) Environmental change and geomorphic hazards in forests, IUFRO research series, 9. CABI Publishing, Wallingford, Oxen, pp 123–165CrossRefGoogle Scholar
  42. Sidle RC, Ochiai H (2006) Landslides: processes, prediction, and land use. Water Resour. Monogr. Ser. 18, AGU, Washington DC, 312 pp. doi: 10.1029/WM018
  43. Terranova OG, Gariano SL, Bruno C, Greco R, Pellegrino AD, Iovine G (2016) Landslide-risk scenario of the Costa Viola mountain ridge (Calabria, Southern Italy). J Maps. doi: 10.1080/17445647.2016.1195300
  44. Van Beek LPH (2002) Assessment of the influence of changes in land use and climate on landslide activity in a Mediterranean environment. Netherlands Geo-Graphical Studies NGS 294, Utrecht 360 ppGoogle Scholar
  45. Van Beek LPH, Van Asch TWJ (2004) Regional assessment of the effects of land-use change on landslide hazard by means of physically based modelling. Nat Hazards 31(1):289–304. doi: 10.1023/B:NHAZ.0000020267.39691.39 CrossRefGoogle Scholar
  46. Van Den Eeckhaut M, Hervás J, Jaedicke C, Malet J-P, Montanarella L, Nadim F (2012) Statistical modelling of Europe-wide landslide susceptibility using limited landslide inventory data. Landslides 9(3):357–369. doi: 10.1007/s10346-011-0299-z CrossRefGoogle Scholar
  47. Verburg PH, Soepboer W, Veldkamp A, Limpiada R, Espaldon V, Mastura SSA (2002) Modeling the spatial dynamics of regional land use: the CLUE-S model. Environ Manag 30:391–405. doi: 10.1007/s00267-002-2630-x CrossRefGoogle Scholar
  48. Wasowski J, Lamanna C, Casarano D (2010) Influence of land-use change and precipitation patterns on landslide activity in the Daunia Apennines, Italy. Quat J Eng Geol Hydrog 43:387–401. doi: 10.1144/1470-9236/08-101 CrossRefGoogle Scholar
  49. Wood JL, Harrison S, Reinhardt L (2015) Landslide inventories for climate impacts research in the European Alps. Geomorphology 228:398–408. doi: 10.1016/j.geomorph.2014.09.005 CrossRefGoogle Scholar

Copyright information

© Springer-Verlag GmbH Germany 2017

Authors and Affiliations

  1. 1.CNR IRPIPerugiaItaly
  2. 2.Department of Physics and GeologyUniversity of PerugiaPerugiaItaly
  3. 3.CNR IRPIRende (CS)Italy
  4. 4.Euro-Mediterranean Center on Climate Change, Division on Regional Models and Geo-Hydrological ImpactsCMCC FoundationCapuaItaly
  5. 5.Euro-Mediterranean Center on Climate Change, Division on Impacts on Agriculture, Forests and Ecosystem ServicesCMCC FoundationViterboItaly
  6. 6.Far Eastern Federal UniversityVladivostokRussia

Personalised recommendations