Daniela Lagomarsino

Updating and tuning a regional-scale landslide early warning system

This work presents the last improvements of an operative regional-scale warning system developed for the management of the risk related to rainfall-induced landslides (both shallow and deep seated). The warning system is named Sistema Integrato Gestione Monitoraggio Allerta, and it is based on a set of spatially variable statistical rainfall thresholds (Martelloni et al. Landslides 9(4): 485–495, 2012b). The performance of the warning system was enhanced using a larger landslide dataset for the calibration of thresholds and readjusting the boundaries of the territorial units (TUs, the basic spatial unit of application of the warning system). Our tuning leads to define a larger number of TUs and to change some of the previous reference rain gauges. In particular, a statistical analysis highlighted that the spatial organization of missed and correctly predicted landslides does not depend on lithology, land use, and morphometric attributes; therefore, the redefinition of TUs was based on the administrative borders between municipalities. This allowed combining the TU outputs into a decisional procedure which, in a completely automated way, is able to forecast the warning levels based on objective and quantitative criteria (the number of expected landslides), in full accordance with the regional civil protection guidelines. The implementation of these updates was straightforward and could be conveniently applied to similar warning systems based on rainfall thresholds.

Updating EWS rainfall thresholds for the triggering of landslides

In this paper, the updating of rainfall thresholds for landslide early warning systems (EWSs) is presented. Rainfall thresholds are widely used in regional-scale landslide EWSs, but the efficiency of those systems can decrease during the time, so a periodically updating should be required to keep their functionality. The updating of 12 of the 25 thresholds used in the EWS of Tuscany region (central Italy) is presented, and a comparison between performances of new and previous thresholds has been made to highlight the need of their periodical update. The updating has been carried out by collecting ca. 1200 new landslide reports (from 2010 to March 2013) and their respective rainfall data, collected by 332 rain gauges. The comparison has been made by the use of several statistical indexes and showed a marked increasing in the performances of the new thresholds with respect to previous ones.

Landslide Susceptibility Mapping at National Scale: The Italian Case Study

Landslide susceptibility maps are key tools for land use planning, management and risk mitigation. The Landslide susceptibility map of Italy, scale 1:1,000,000 is being realized by using the Italian Landslide Inventory – Progetto IFFI and a set of contributing factors, such as surface parameters derived from 20 to20 m DEM, lithological map obtained from the geological map of Italy 1:500,000, and land use map (Corine Land Cover 2000). These databases have been subjected to a quality analysis with the aim of assessing the completeness, homogeneity and reliability of data, and identifying representative areas which may be used as training and test areas for the implementation of landslide susceptibility models. In order to implement the models, physiographic domains of homogeneous geology and geomorphology have been identified, and landslides have been divided into three main classes in order to take into account specific sets of conditioning factors: (a) rockfalls and rock-avalanches; (b) slow mass movements, (c) debris flows. The modelling tests performed with different techniques (Discriminant Anaysis, Logistic Regression, Bayesian Tree Random Forest) provided good results, once applied with the appropriate selection of training and validations sets and with a significant number of statistical units.

Shifts in the eruptive styles at Stromboli in 2010–2014 revealed by ground-based InSAR data

Ground-Based Interferometric Synthetic Aperture Radar (GBInSAR) is an efficient technique for capturing short, subtle episodes of conduit pressurization in open vent volcanoes like Stromboli (Italy), because it can detect very shallow magma storage, which is difficult to identify using other methods. This technique allows the user to choose the optimal radar location for measuring the most significant deformation signal, provides an exceptional geometrical resolution, and allows for continuous monitoring of the deformation. Here, we present and model ground displacements collected at Stromboli by GBInSAR from January 2010 to August 2014. During this period, the volcano experienced several episodes of intense volcanic activity, culminated in the effusive flank eruption of August 2014. Modelling of the deformation allowed us to estimate a source depth of 482 ± 46 m a.s.l. The cumulative volume change was 4.7 ± 2.6 × 105 m3. The strain energy of the source was evaluated 3–5 times higher than the surface energy needed to open the 6–7 August eruptive fissure. The analysis proposed here can help forecast shifts in the eruptive style and especially the onset of flank eruptions at Stromboli and at similar volcanic systems (e.g. Etna, Piton de La Fournaise, Kilauea).

Mapping landslide phenomena in landlocked developing countries by means of satellite remote sensing data: the case of Dilijan (Armenia) area

ABSTRACT Landslide detection and mapping are essential issues for reducing impact of such natural disasters, and for improving the future built-up expansion and planning strategies, especially in developing countries where a reasonable land-use design is an important concern for sustainable growth and environmental management. Armenia is a landlocked country and its urban development is strongly tied to the improvement of infrastructures, which must takes into account the environmental setting and the slope instability of the area, in order to identify risks and possible damages to settlements and economic activities. The use of satellite-based Earth Observation data has advanced significantly in the last decade and has turned out to be very useful for measuring and monitoring slow-moving surface deformation phenomena with millimetric precision. In this framework, this study aims at providing a remote sensing-based Landslide Inventory Map (LIM) and a Landslide Susceptibility Map (LSM) over Dilijan (Armenia) area, performed within the Secondary Cities Urban Development in Armenia project. In particular, LIM and LSM in the study area were produced by using ground deformation measurements derived from satellite Synthetic Aperture Radar (SAR) data, acquired by ALOS and ENVISAT sensors from 2003 up to 2010, and integrated with photo-interpretation of recent optical images and morphological analysis of Digital Elevation Model (DEM). Given the extensive presence of vegetation in the area of interest, satellite SAR images were processed to produce both SqueeSAR™ and Temporary Coherent Scatterers data, which are PSI (Persistent Scatterer Interferometry) data conceived as evolution of PSInSAR™ approach and particularly suited for non-urban and rural areas characterized by low density of coherent terrain benchmarks over time. Landslide mapping produced through this work identifies the most hazardous landslide-affected and landslide-prone areas around Dilijan city, and can be used for further estimating environmental risks for urban infrastructure development in the area.

An Operational Warning System for the Forecasting of Landslide Occurrence at Regional Scale

In this work we present a regional scale warning system (named SIGMA) for rainfall induced landslides. The system combines rain gauges measurements and rainfall forecasts and compares them with a series of statistical rainfall thresholds based on the total amount of rainfall. The system was specifically built to take into account both shallow and deep seated landslides. A decisional algorithm integrated in the warning system automatically provides a criticality level depending on which thresholds are exceeded. The model was integrated in the regional warning system of the Emilia Romagna region (Italy) for civil protection purposes. This paper accounts also for the main modifications that the model had during its development, passing form a base version relying on thresholds defined by means of a statistical analysis on rainfall data, to an advanced version in which thresholds were calibrated using a landslide database. The passage from a system to another was straightforward and for its flexibility and versatility, the proposed methodology seems particularly appropriate for emerging countries that have not yet gathered extensive and complete information on the location and time of occurrence of landslides.

A Tool for Classification and Regression Using Random Forest Methodology: Applications to Landslide Susceptibility Mapping and Soil Thickness Modeling

Classification and regression problems are a central issue in geosciences. In this paper, we present Classification and Regression Treebagger (ClaReT), a tool for classification and regression based on the random forest (RF) technique. ClaReT is developed in Matlab and has a simple graphic user interface (GUI) that simplifies the model implementation process, allows the standardization of the method, and makes the classification and regression process reproducible. This tool performs automatically the feature selection based on a quantitative criterion and allows testing a large number of explanatory variables. First, it ranks and displays the parameter importance; then, it selects the optimal configuration of explanatory variables; finally, it performs the classification or regression for an entire dataset. It can also provide an evaluation of the results in terms of misclassification error or root mean squared error. We tested the applicability of ClaReT in two case studies. In the first one, we used ClaReT in classification mode to identify the better subset of landslide conditioning variables (LCVs) and to obtain a landslide susceptibility map (LSM) of the Arno river basin (Italy). In the second case study, we used ClaReT in regression mode to produce a soil thickness map of the Terzona catchment, a small sub-basin of the Arno river basin. In both cases, we performed a validation of the results and a comparison with other state-of-the-art techniques. We found that ClaReT produced better results, with a more straightforward and easy application and could be used as a valuable tool to assess the importance of the variables involved in the modeling.

Susceptibility of intrusion-related landslides at volcanic islands: the Stromboli case study

Susceptibility of intrusion-related landslides in an active volcano was evaluated coupling the landslide susceptibility estimation by random forest (RF), and the probabilistic volcanic vent opening distribution, as proxy for magma injection, using the QVAST tool. In order to develop and test the method proposed here, the RF/QVAST approach was adopted for Stromboli volcano (Southern Italy) since it experienced moderate to huge instability events, it is geomorphologically prone to instability events, and it is affected by active intense volcanic activity that can produce slope instability. The main destabilizing factors of the volcanic flanks are the slope, the aspect, the terrain roughness, the land cover and the litho-technical features of the outcropping rocks. Estimation of volcanic susceptibility shows that the areas with high probability of new vent opening are located in the north-western unstable volcano flank (Sciara del Fuoco), in the volcano summit and the north-eastern volcano flank coherent with the possible re-activation of the eruptive fissures related to the regional tectonic setting. The areas with higher probability of intrusion-related landslides are located in the upper part of the Sciara del Fuoco, while the rest of the island show moderate to low probability of intrusion-related landslide occurrence.

Regional Scale Landslide Susceptibility Mapping in Emilia Romagna (Italy) as a Tool for Early Warning

The Emilia Romagna region (22,446 km2, Northern Italy) is widely affected by landslides. The Civil protection Agency of the Emilia Romagna Region uses a regional scale warning system (WS) for the management of the risk related to rainfall induced landslides. The WS is used to perform a temporal forecasting of landslides, as it provides an alert level for each of the eight subdivisions (called alert zones—AZ) of the regional territory.

How to Improve the Accuracy of Landslide Susceptibility Maps Using PSInSAR Data

Landslide susceptibility maps (LSM) are frequently used by local authorities for land-use management and planning activities. They are valuable tools used by decision makers for urban and infrastructural plans and for civil protection purposes. False negative and false positive errors can affect the accuracy of a LSM, decreasing the reliability of this useful product. False negative errors are usually the worst in terms of social and economic losses because they are related to a misclassification of areas at risk. In this paper we present a new methodology aimed at improve the accuracy of the LSMs using measurement points (PS, Permanent Scatterers and DS, Distributed Scatterers) retrieved through the multi-interferometric SqueeSAR technique. The proposed approach uses two different TerraSAR-X datasets acquired in ascending and descending geometry. PS/DS velocity are re-projected along the steepest slope direction. The integration between the LSM and the ground deformation velocity maps was performed by using an empirical contingency matrix, which takes into account the average Vslope module and the susceptibility degree obtained by using the Random Forests algorithm for an area located within Messina Province (Sicily, Italy). Results highlight that 33.37 km2 have been updated. The combination among SqueeSAR data and the LSM improves the reliability in predicting slow moving landslide which, especially, affect urbanized areas. The use of this procedure can be easily applied in different areas where multi-interferometric datasets are available. The proposed approach will help civil protection and decision making authorities to use reliable landslide susceptibility maps, correcting part of the errors of the original LSM.