## 1. Introduction

Landslides are gravitational processes on slopes triggered by different factors such as land use, climatic, seismic and anthropic factors that cause a strong impact on the landscape shape. Different criteria are utilized to classify the landslide type: dynamics of the movement, depth of the slip plane, type of the movement, speed of displacement, direction of movement on slopes, landslide shape, etc. The Varnes classification [1] by landslide type of movement and material is frequently used (**Figure 1**). The landslide depth is also utilized very often. In Romania, the landslide classification is mainly used according to the depth of surface rupture:

Landslide susceptibility assessment is focused on landslide space prediction, while the landslide hazard means the assessment of space and time of occurrence. Hazard assessment is not easy, long‐term spatiotemporal data being necessary. The equation utilized to determine the landslide risk is R = H × E × V [2–4] (where H = the probability of occurrence (%); E = the value of the elements at risk; V = the vulnerability of elements at risk).

The approaches made to assess the landslide susceptibility are heuristic, statistic and deterministic. There are various regional approaches according to the study area. Qualitative approaches are utilized for large areas (scales > 1:50,000); statistical analyses are suitable at scales lower than 1:50,000. Deterministic approaches (the assessment of Fs) are expensive (geotechnical data and data concerning the landslide morphometry are necessary) but most precise.

InSAR, SAR, VHR optical images, LIDAR, Total Station, and GPS are used for landslides inventory and landslides displacements measurements.

The approaches utilized to reduce the landslide risk are (1) the assessment of landslide‐prone areas; (2) landslides control works; (3) landslides restraint works and (4) early warning systems design. The equipment for landslide monitoring is varied according to the landslide type of movement and displaced material on slopes, the speed of the movement, etc. (**Figure 4**).

The geomorphic terms describing a landslide are different from those used in engineering geology. The geomorphologist prefers using terms like scarp, steps, monticles and landslide valleys, while engineering geologist would rather resort to terms like surface of rupture, geotechnical parameters, landslide geometry and safety factor.

## 2. Landslide susceptibility, hazard and risk assessment

### 2.1. Landslide susceptibility

Landslide susceptibility means assessing the probability of landslides occurrence in a specific place. The methods are different according to the size of the studied area. Different methods are utilized in Romania to determine landslide susceptibility [5–10]. Interdisciplinary approaches have been elaborated in various studies [5,6,11–16]. After 2000, GIS software was frequently utilized to elaborate the landslide susceptibility maps.

#### 2.1.1. Heuristic approaches

Heuristic approaches are usually made in the case of large areas on the scales > 1:50,000. Geomorphic mapping is a direct, qualitative method. This approach is a qualitative one involving the researchers’ experience; therefore, it is subjective. Very often, the overlay of parametric maps (the landslide condition maps as per geology, slope declivity, slope aspect and the land use) within a GIS environment is used (**Figure 5**). Sometimes the results are very close to those obtained by statistical approaches. The advantage of this method is that it is easy and rapid to assess. The disadvantage is subjectiveness (different researchers may obtain different results).

#### 2.1.2. Statistical approaches

Statistical approaches [17–22] are made on the scales of < 1:50,000. Bivariate and multivariate analyses are common methods (frequently used) (**Figure 6**). A method utilized by us to elaborate landslide susceptibility maps was the bivariate analysis and the index of entropy. Three maps of landslide susceptibility were elaborated in The Buzau Subcarpathians (Sibiciu basin, Panatau basin and Saratel basin) using the bivariate analysis and the index of entropy [20]. The conditions of occurrence‐parametric maps (geology, slope declivity, slope aspect, curvature and the land use) were compared with the landslide inventory map. The index of entropy was used to assign weights to the parametric maps. The bivariate statistical analysis, using the index of entropy [19,23,24], was utilized to assess the probability of landslide occurrence. The weight value, obtained from the level of entropy, represents the approximation‐to‐normal distribution of the probability [19,24]. The index of entropy expresses which parameters of a natural environment are most relevant for mass movement occurrence [24]. The analysis, based on the entropy index, characterizing the level of chaos in the environment, offers the possibility to state the unfavorable combinations for slope instability in the studied area.

The equation used to calculate the weight value for the parameter as a whole (W_{j}) is as follows [24]:

where I_{j} is the information coefficient and p_{ij} is the density of probability [24].

The equation used to calculate the values of landslide susceptibility maps was:

(2) |

where landuse_recl2,…,aspect_recl2 are the values in a particular cell for the secondarily reclassified parametric map and Wj is the weight value of the parametric map [19].

The result of this summation is a continuous interval and represents various landslide susceptibility levels. The natural breaks classification method was used to divide the interval into five classes [19,20,24] (**Figure 6**).

Several methods are used to evaluate the performance of the susceptibility analysis. Very often, the pilot area is divided into two sectors. The first sector is used to determine the landslide susceptibility. The other sector is used for the validation. Another method uses random selected areas, one to determine the landslide susceptibility and another to validate the results [25].

Probabilistic, neural networks, fuzzy logic [4,22,26–29] are also employed. When using the bivariate analysis, the landslide inventory map is compared with the other parametric maps (lithology, slope declivity, slope aspect, land use sometimes also the curvature). Discriminant analysis [30] and logistic regression are the most utilized techniques [27]. The softwares used within a GIS system are ArcInfo, ArcView, ArcGIS, SPANS, IDRISI, GRASS and ILWIS and statistical packages, such as Statgraph or SPSS. The MATLAB system is also used when Neural Networks (Artificial Neural Network) method is resorted to.

### 2.2. Hazard assessment

Hazard assessment is more difficult and it means defining the place and the time of landslides occurrence. To calculate the landslide hazard, different landslide inventories made at different time of occurrence are necessary [31]. Being difficult to elaborate, few landslide hazard maps are elaborated. The general steps to elaborate a landslide hazard map are presented (**Figure 7**):

It is assumed that slope failures will occur in the future under the same triggering factors and conditions that caused them in the past. The probability of landslide occurrence is obtained using some equations. Two probability models are commonly used to compute the occurrence of random point‐events in time: (1) the Poisson model and (2) the binomial model. Multiple‐temporal inventories are utilized [27].

### 2.3. Risk assessment

The risk is defined as the product between the probability of occurrence, vulnerability and the elements at risk. Assessing landslide risk is complex, undertaking the equation as follows [2–4]:

where R is the landslide risk; H is the hazard; the probability of spatiotemporal occurrence (%); V is the vulnerability of elements at risk (%) and E is the value of elements at risk (expressed in terms of measure of unit of the element taken into account).

The elements at risk [32] are the population, properties, economic activities etc. Vulnerability [32] is the degree of loss to a given element or set of elements at the risk. Vulnerability is a measure of the exposure to a potentially damaging landslide.

Total landslide risk consists of all the specific landslide risk levels. According to the potential damage to structure, infrastructure, population, we distinguish: very high total landslide risk; high total landslide risk; moderate total landslide risk; low total landslide risk; very low total landslide risk.

A complete risk assessment involves the quantification of a number of different types of losses (FEMA, 2004) (direct and indirect) [35]. In the absence of quantitative data, qualitative assessments are estimated (geomorphic risk assessment). We consider that these are not conformable with the risk definition.

We would mention the map elaborated in Ref. [3], where the elements at risk are roads and railways. The value of the damage divided to the cost for building a new structure was used to estimate vulnerability. The risk was estimated in EUR/pixel/year. Another attempt to elaborate a risk map was applied for uranium waste tailing, urbanized territory and road network for a period of 50 years. The information considered to estimate the vulnerability referred to road network, urbanized territory, radioactive waste tailing. Vulnerability was based on construction type, the land use, road network (average density of vehicles on the road) [33].

## 3. Landslides control works

**Surface drainage control works** are designed to control the influence of rainfall water and springs. Drainage collection and drainage channels (pipes or U‐ditches) are built to collect the surface water flow.

**Subsurface drainage works** are designed to remove the ground water and to prevent the infiltrations.

**Shallow subsurface drainage control works**. *Interceptor under drains and interceptor trench drains* are used to drain the shallow ground water up to 3 m deep. Control works are often used in a combination with surface drainage. In the case of horizontal gravity drains, the shallow groundwater can be removed also using horizontal gravity (30–50 m long). The pipes, of the PVC type or steel‐made pipes, are drilled at an angle of 10–15° from the horizontal surface.

**Deep subsurface drainage control works.** *Drainage wells* are built in areas with high quantities of water. With a diameter of up to 3.5 m and a depth down up to 25 m, drainage wells are made of steel or reinforced concrete (the bottom and the upper part). Radially horizontal gravity drains (deep horizontal gravity drains) are built at different levels to drain the water (**Figure 8**).

*Drainage tunnels* are built below the slip surface and are connected with drainage wells to drain the water out of the landslide mass. Drainage tunnels diameters are between 1.8 m and 2.5 m.

**Soil removal** works are used at the upper part of the landslide (small‐ and medium‐sized landslides). **Buttress Fill Works** are placed usually at the lower part of the landslide to counterweight the landslide mass. **River structures**, like check dams, bank protection, are built to prevent the action of river erosion on slopes [34].

## 4. Landslides restraint works

*Pile works* are used to stabilize the landslide and stop the movements. Piles are built of steel and filled with concrete.

*Large‐diameter cast‐in‐place pile* has a large diameter (1.5–6.5 m) and a high resistance to bending stresses. *Anchor works* (**Figure 9**) are made at the lower part of the landslide. The tensile force of anchor is used through landslide body and bedrock. *Retaining wall works* are generally designed to restraint the movements of small landslides or when a secondary landslide occurs at the toe of large landslides. Crib walls are also very often utilized.

## 5. Concluding remarks

Landslides are geomorphic processes affecting the shape of the landscape; therefore, the assessment of landslide susceptibility is of permanent interest for landslide researchers (engineering geologist, civil engineers, geomorphologists, etc.). Earth observation techniques such as airborne images (ASTER, SRTM, InSAR, LIDAR), remote sensing (higher spatial resolution‐Landsat, Spot, IRS‐IC, IKONOS and QuickBird) and higher spectral resolution as well as digital collection in the field (Fieldlog, Penmap, GSM, MapLT, Pocket‐GIS and ArcPAD) are very useful to determine landslide features and dynamics. Landslide susceptibility assessment is focused on landslide space prediction, while landslide hazard assesses space and time of occurrence. Hazard assessment is not easy; it requires long‐term spatiotemporal data. To estimate the landslide risk, detailed studies concerning the hazard, vulnerability and the elements at risk assessment are required. Vulnerability shows the degree of exposure of elements at risk to a potential damage phenomenon. The assessment of direct and indirect loss is very important to determine the landslide risk. Deterministic approaches are expensive but most reliable. Landslide susceptibility, hazard and risk assessment are important to delineate the landslide‐prone areas, to estimate the probability of occurrence, the vulnerability of elements at risk and finally to estimate the risk value. Landslide control works and landslide restraint works are complex, involving research results from different domains like geomorphology, engineering geology, civil engineering. Without being exhaustive, the chapter presents the main methods to elaborate the susceptibility maps, hazard and risk. A brief overview of landslide control works and restraint works is presented too.