1,720,965 research outputs found
Affidabilità di approcci predittivi di regionalizzazione delle proprietà geologico-tecniche dei depositi superficiali di versante
Full text linkLe frane indotte dagli eventi meteorologici intensi nelle terre al di sopra del substrato consistente (depositi superficiali di versante - DS) costituiscono un fattore di pericolo per gran parte del territorio collinare e montano. I caratteri geologico-tecnici dei DS svolgono un ruolo rilevante per la comprensione dei fenomeni di franosità superficiale. Il tempo ed i costi necessari per raccogliere dati geologico-tecnici sui DS riducono la densità areale con cui questi vengono acquisiti, determinando limitazioni di affidabilità per le valutazioni predittive regionalizzate di pericolosità da frana superficiale.
In questo quadro le performance degli approcci di regionalizzazione delle caratteristiche puntuali dei DS basate su fattori spazialmente continui, quali litologia, struttura e caratteristiche geologico-tecniche del substrato geologico, uso del suolo, morfometria dell’area indagata, hanno un ruolo di fondamentale importanza.
Vengono qui esposti i primi risultati di affidabilità di approcci di spazializzazione delle caratteristiche geologico-tecniche dei DS, in particolare la profondità (pDS), basate sulla stratificazione di criteri geologici e la segmentazione dello spazio morfometrico.
Lo spazio morfometrico viene descritto attraverso variabili quali pendenza, curvature e flow accumulation, derivate da un modello digitale del terreno con cella di 10 m. Impiegando un insieme di osservazioni puntuali (training dataset di circa 170 pDS tra profili e trivellate) raccolte in una prima area di studio (ADS1: regione a litologia costante del substrato corrispondente alla Formazione del Macigno, media valle del fiume Serchio), le variabili morfometriche sono state classificate mediante approcci hard e fuzzy di statistica multivariata (cluster analysis, neural network). È stata così ottenuta per l’ADS1 una rappresentazione continua in classi di pDS.
Le “firme morfometriche” (insieme delle relazioni pDS vs variabili morfometriche) individuate nell’ADS1 sono state utilizzate, con approccio supervisionato, per ottenere una rappresentazione in classi di pDS di una nuova area di studio (ADS2: Alpi Apuane occidentali) geograficamente distinta dalla precedente, ma localizzata sulla stessa formazione di substrato. Un nuovo dataset puntuale delle proprietà geologico-tecniche dei DS acquisito nella ADS2 (test dataset) ha consentito di valutare l’accuratezza predittiva delle pDS stimate tramite estrapolazione delle firme morfometriche di ADS1. Lo stesso dataset è stato utilizzato come training dataset per spazializzare le classi di pDS anche nella ADS2 tramite cluster analysis
A method for engineering-geological mapping: application to the Arezzo and Lucca provinces (Tuscany, Italy)
No full textLithological and geomechanical characters of outcropping rocks are relevant inputs for those applications involving geological issues. For instance, such information are used to implement spatial planning actions/rules influencing land use and transport infrastructures. The same data may also be used when mapping landslide susceptibility/hazard and preparing for landslide risk management. Many geomechanical classification systems for rock masses have been developed for engineering-geological applications (DEERE, 1963; BIENIAWSKI, 1973; BARTON et alii, 1974; HOEK, 1994). Nevertheless, these are site-specific systems or they are applied for specific engineering works (e.g. tunnels), so they are not fully adequate for continuous representations of engineering-geological properties over wide (map scale) areas. In this work, we describe the implementation of a GIS integrating lithological-geomechanical data collected in the field and laboratory with existing geological database to obtain an engineering-geological map at the scale of 1:10,000 for the provinces of Arezzo and Lucca (Tuscany, Italy). The study area is representative of different structural and lithologic settings within the Northern Apennines
Cluster analysis applied to engineering geological mapping
Full text linkCluster analysis of morphometric variable is reported in this paper to support characterization of rock masses and deposits. The first technique is related to fast mechanical characterization of bedrock and the second one on the mapping of the depth of superficial deposits. In order to extrapolate site-specific information to the whole study area two techniques are applied to morphometric space: supervised and unsupervised classifications through the algorithms maximum likelihood and ISODATA, respectively. The analysis of morphometric space with these techniques has provided significant results in order to discriminate bedrocks with different mechanical characteristics and the depth of superficial deposits
Engineering geology characterization of slope deposits and physically-based assessment of shallow landslide susceptibility (Alpi Apuane, Italy)
Full text linkIn this work we present the results of engineering geology characterization of slope deposits and assessment of shallow landslide susceptibility by means of a probabilistic physically-based model for the Western sector of the Alpi Apuane (Northern Apennines, Italy). The Alpi Apuane are a Tertiary metamorphic complex which is undergoing fast tectonic uplift, exhumation and erosion in respect to neighboring regions (the coastal Versilia Plain and Garfagnana Valley). For these reasons, the morphology of the Alpi Apuane is characterized by high relief energy, as highlighted by elevation differences up to around 2,000 m and deep river valleys with steep slopes. Moreover, the study area records annual precipitations among the highest in Italy (up to around 2,500 mm/y) and, especially in the last decades, frequent intense rainfall events (i.e.: 1996, 1998, 2000, 2011, 2013, 2014). In this framework landslides are widespread, especially shallow landslides involving unconsolidated slope deposits overlying bedrock. In order to assess shallow landslide susceptibility, we used a hydrological model coupled to a limit-equilibrium infinite-slope stability model. Reliability of results by physically-based models depends on accuracy of map distribution of input data which, however, is usually almost unknown. Hence, fieldwork and laboratory tasks were carried out to map engineering geology characters of slope deposits. For a set of hundreds of field sampling points, we acquired: depth to the bedrock, geotechnical horizons, unit weight, as well as soil samples for lab analysis. The distribution of points were chosen by observing that engineering geology properties of slope deposits depend on both bedrock lithology and morphometric conditions. Then, for a subset of the sampling points, we performed hydraulic conductivity measurements. Geotechnical determinations allowed us to estimate the friction angle ranges for different slope deposit types. In order to obtain the map distribution of engineering geology parameters, we implemented a spatial analysis by clustering morphometric variables stratified as a function of bedrock lithological units. Multitemporal visual interpretation of orthophotos (2003-2016) allowed us to obtain the database for a new shallow landslide inventory, which later underwent field accuracy assessment. By integrating the inventory to geology, we identified those bedrock lithological units where the infinite-slope assumption for shallow landslide modeling could be reasonably applied. In order to take into account and evaluate the effects of input parameters uncertainty, we implemented the slope stability-hydrological model by means of a Monte Carlo simulation. Assuming that the cohesion of slope deposits changes in space and time depending upon seasonal variation of land cover and precipitations, we calibrated the model by means of a back-analysis aimed at estimating the cohesion intervals which allow for optimization of the final predictive performance within the shallow landslide regions. This task was performed by using both prediction-rate curve and ROC diagrams. Finally, the results of susceptibility assessment, as well as maps/diagrams useful to describe the variability/uncertainty of results are critically discussed
Modelling-mapping slope deposits depth and uncertainty assessment by means of machine learning approaches
Full text linkShallow landslides triggered by heavy rainfall are a common natural phenomenon in mountain areas. Climate changes and increasing urban pressure make this phenomenon a widespread source of natural hazard. For this reason, the interest of scientific community concerning the development of robust shallow landslide susceptibility/ hazard assessment methods for wide areas (regional scale) has steadily increased in the last decades. Many methods are available to achieve this goal, however, researchers are generally focused on statistics (data driven) or physically-based methods. For both approaches, the depth of Slope Deposits (SD: the surficial soil involved by landsliding which covers the bedrock) is an important parameter in order to perform accurate analysis. Furthermore, the SD depth is required for many physically-based models available in the literature. Nevertheless, this information is generally unknown at map scale, which affects uncertainty and reliability of susceptibility/hazard assessments.
In this context, this work is focused on obtaining predictive SD depth maps for wide areas by means of geostatistics methods suitable to consider variability and uncertainty of the input/output data.
The study area is located in Northern Tuscany where, in the last years, we developed research projects on engineering geology characterization of SD. Hence, a large dataset of SD depth obtained by field survey (more than 1,000 oobservations) is used in this work. Many geo-environmental variables such as: geology, land use, morphometric variables, are considered in the analysis. Morphometric variables (eg. flow accumulation, slope and hillslope curvature) are derived from a digital elevation model with cell size of 10 m. Two different machine learning techniques are used to map SD depth: clustering and artificial neural networks. The supervised clustering analysis is performed with probabilistic and fuzzy algorithms. For the unsupervised clustering, the results of various maps obtained by integrating different sets of input variables are spatially combined (data fusion) in order to obtain a single map. The analysis performed with artificial neural networks has been implemented by a feed-forward multi-layer neural network. In order to exploit the field measurement dataset, also the effect of samples geographic neighbourhood were considered.
The results show the feasibility of the methods for regional scale mapping. Moreover the results are discussed and analyzed in order to identify best solutions to evaluate and represent the SD depth uncertainty
Uncertainty on shallow landslide hazard assessment: from field data to hazard mapping
Full text linkShallow landsliding that involve Hillslope Deposits (HD), the surficial soil that cover the bedrock, is an important
process of erosion, transport and deposition of sediment along hillslopes. Despite Shallow landslides generally
mobilize relatively small volume of material, they represent the most hazardous factor in mountain regions due
to their high velocity and the common absence of warning signs. Moreover, increasing urbanization and likely
climate change make shallow landslides a source of widespread risk, therefore the interest of scientific community
about this process grown in the last three decades. One of the main aims of research projects involved on this
topic, is to perform robust shallow landslides hazard assessment for wide areas (regional assessment), in order to
support sustainable spatial planning.
Currently, three main methodologies may be implemented to assess regional shallow landslides hazard: expert
evaluation, probabilistic (or data mining) methods and physical models based methods. The aim of this work is
evaluate the uncertainty of shallow landslides hazard assessment based on physical models taking into account
spatial variables such as: geotechnical and hydrogeologic parameters as well as hillslope morphometry. To achieve
this goal a wide dataset of geotechnical properties (shear strength, permeability, depth and unit weight) of HD was
gathered by integrating field survey, in situ and laboratory tests. This spatial database was collected from a study
area of about 350 km2 including different bedrock lithotypes and geomorphological features. The uncertainty
associated to each step of the hazard assessment process (e.g. field data collection, regionalization of site specific
information and numerical modelling of hillslope stability) was carefully characterized.
The most appropriate probability density function (PDF) was chosen for each numerical variable and we assessed
the uncertainty propagation on HD strength parameters obtained by empirical relations with geotechnical index
properties. Site specific information was regionalized at map scale by (hard and fuzzy) clustering analysis taking
into account spatial variables such as: geology, geomorphology and hillslope morphometric variables (longitudinal
and transverse curvature, flow accumulation and slope), the latter derived by a DEM with 10 m cell size. In order
to map shallow landslide hazard, Monte Carlo simulation was performed for some common physically based
models available in literature (eg. SINMAP, SHALSTAB, TRIGRS). Furthermore, a new approach based on the
use of Bayesian Network was proposed and validated. Different models, such as Intervals, Convex Models and
Fuzzy Sets, were adopted for the modelling of input parameters. Finally, an accuracy assessment was carried
out on the resulting maps and the propagation of uncertainty of input parameters into the final shallow landslide
hazard estimation was estimated. The outcomes of the analysis are compared and discussed in term of discrepancy
among map pixel values and related estimated error.
The novelty of the proposed method is on estimation of the confidence of the shallow landslides hazard mapping
at regional level. This allows i) to discriminate regions where hazard assessment is robust from areas where more
data are necessary to increase the confidence level and ii) to assess the reliability of the procedure used for hazard
assessment
A new shallow landslides inventory for Southern Lunigiana (Tuscany, Italy) and analysis of predisposing factors
No full textA new inventory of shallows landslides for a study area located within Southern Lunigiana valley has been obtained by means of digital visual interpretation of orthophoto maps acquired in the years 2003, 2007, 2010 and 2013. A total of 331 shallow landslides occurred during the decade 2003-2013 have been identified, resulting in an average landslide density of 1.1 landslides per square kilometers. A spatial analysis has been implemented to investigate landslide distribution in respect to predisposing factors: bedrock geology, land use and morphometry (elevation, slope steepness, flow accumulation, slope aspect). Results highlight that bedrock lithology and land use are important factors affecting shallow landslide distribution. While all the morphometric parameters analyzed are correlated to shallow landslides density, slope aspect seems to be the prominent conditioning for landslide occurrence
Testing and improving the Rock Mass Quality Index (RQI) in North-Western Tuscany (Italy)
Full text linkMany geo-mechanical classification systems for rock masses have been developed for engineering geology applications. However, most of them are site-specific and they take into account the combination of rock mass data related to different geological and physico-mechanical properties. For these reasons, they may hardly be applied when regional, continuous representation over wide areas (map scale) are necessary (i.e. spatial planning, seismic microzoning).
The aim of this study is to test and improve an existing method for engineering geology mapping of rock masses based on quantitative integration of geological information, fieldwork geo-mechanical measurements, lab determinations and spatial analyses. This method has been applied to a new study area located in North-Western Tuscany (Italy) where due to a complex structural setting, different structural and lithological units of the Northern Apennines chain crop out. Fieldwork measurements were performed for the outcropping geological formations by choosing sets of sites representative of different rock mass characters (lithology, weathering, jointing), both at local and wide scale. For each surface or sub-surface site, a regular grid of measuring points was defined, where each point underwent rebound measurements (R - Schmidt hammer). Frequency of grid points and measurements were chosen in order to obtain reasonable statistical stability of average site rebound values. Following methods from the literature, the Geological Strength Index (GSI) was also estimated for each investigation site. We collected representative rock samples for lab evaluation of unit weight to be used along with R to calculate the Rock Mass Quality Index (RQI). In fact, according to the literature, unit weight is related to weathering and mechanical properties of rocks. A statistical analysis of correlation between both R – GSI and RQI – GSI was performed and the results are presented and discussed. Moreover, the spatial analysis of the whole dataset confirm that the proposed method allows one to recognize different engineering geology characters among different formations, as well as to identify different geo-mechanical units within the same formation. The spatial analysis of RQI also highlights variability among different structural domains of the study area. In conclusion, this study supports this method as suitable for cartographic engineering geology applications
Influence of digital terrain model on mapping of slope deposits depth
No full textIn this work the influence of the characteristics of different Digital Terrain Models (DTMs) on Slope Deposits depth (dSD) maps is evaluated. dSD data collected during fieldwork have been mapped through GPS. Two different DTMs have been used for the analysis: one derived from LiDAR survey (1 m pixel size), the other extracted by the topographic maps of Regione Toscana (10 m pixel size). dSD maps have been extracted by integrating cluster analysis of morphometric variables with dSD data. In order to investigate the effects of spatial resolution, the LiDAR DTM has been down-sampled to both 5 m and 10 m. The weight of GPS positioning uncertainty has been assessed by performing a statistical analysis on the morphometric clusters falling into buffer areas of sampling points. Comparisons among outputs have been performed through map accuracy estimations, as well as by spatial distribution visual check and extension assessment of dSD classes. Results show that by increasing spatial resolution of input DTMs the statistical quality of the dSD maps generally reduces. Concerning the effect of buffer size around sampling points, as a general rule, the larger the buffer, the lower the map accuracy. Only the map obtained by 1 m LiDAR DTM is insensitive to buffer size
Mapping slope deposits depth by means of cluster analysis: A comparative assessment
No full textIn this work a comparison among slope deposits (SD) maps obtained by integrating field measurements of SD depth and cluster analysis of morphometric data has been performed. Three SD depth maps have been obtained for the same area (SA1) by using different approaches. Two maps have been achieved by implementing both the supervised and unsupervised approaches and exploiting the dataset of SD depths previously collected in a region (SA2) characterized by the same bedrock lithology, although located 35 km far from the SA1. The results have been validated against a reference map based on SD depth measurements acquired during this work within the SA1 and mapped by unsupervised clustering. The outcome of the study shows the feasibility of the methodology proposed to obtain depth maps of SD. Nevertheless the very low map accuracy suggests that relationships among main morphometric variables and slope deposits depth are not constant at regional scale, although considering areas characterized by the same bedrock lithology. Hence, maps of SD depth should be based on depth data specifically acquired within the area under study. In order to improve the exploitation of SD depth datasets outside their provenance area, further research are necessary on clustering algorithms performance as well as additional morphometric and environmental variables to be employed in spatial analysis
- …
