Despite the large number of recent advances and developments in landslide susceptibility mapping ... more Despite the large number of recent advances and developments in landslide susceptibility mapping (LSM) there is still a lack of studies focusing on specific aspects of LSM model sensitivity. For example, the influence of factors such as the survey scale of the landslide conditioning variables (LCVs), the resolution of the mapping unit (MUR) and the optimal number and ranking of LCVs have never been investigated analytically, especially on large data sets. In this paper we attempt this experimentation concentrating on the impact of model tuning choice on the final result , rather than on the comparison of methodologies. To this end, we adopt a simple implementation of the random forest (RF), a machine learning technique, to produce an ensemble of landslide susceptibility maps for a set of different model settings, input data types and scales. Random forest is a combination of Bayesian trees that relates a set of predic-tors to the actual landslide occurrence. Being it a nonpara-metric model, it is possible to incorporate a range of numerical or categorical data layers and there is no need to select unimodal training data as for example in linear discriminant analysis. Many widely acknowledged landslide predisposing factors are taken into account as mainly related to the lithol-ogy, the land use, the geomorphology, the structural and an-thropogenic constraints. In addition, for each factor we also include in the predictors set a measure of the standard deviation (for numerical variables) or the variety (for categorical ones) over the map unit. As in other systems, the use of RF enables one to estimate the relative importance of the single input parameters and to select the optimal configuration of the classification model. The model is initially applied using the complete set of input variables, then an iterative process is implemented and progressively smaller subsets of the parameter space are considered. The impact of scale and accuracy of input variables , as well as the effect of the random component of the RF model on the susceptibility results, are also examined. The model is tested in the Arno River basin (central Italy). We find that the dimension of parameter space, the mapping unit (scale) and the training process strongly influence the classification accuracy and the prediction process. This, in turn, implies that a careful sensitivity analysis making use of traditional and new tools should always be performed before producing final susceptibility maps at all levels and scales.
Un catasto fiumi per il controllo e la mitigazione delle pericolosità di collasso arginale dell&#... more Un catasto fiumi per il controllo e la mitigazione delle pericolosità di collasso arginale dell'Arno e dei suoi principali affluenti
Despite the many recent attempts at devising a general physically-based model to predict the dyna... more Despite the many recent attempts at devising a general physically-based model to predict the dynamics of rainfall induced landslide in real-time, there remain several drawbacks related to the difficulty in estimating the required parameters which show a high spatial variability. For these reasons approaches based on empirical thresholds of rainfall intensity and duration are still widely used and in many
Climate change driven landslide forecasting implementation in early warning activities is strictl... more Climate change driven landslide forecasting implementation in early warning activities is strictly dependent on the quality and timing of rainfall measurements and on an accurate empirical or deterministic relationship linking water infiltration to slope stability. The scientific literature offers a wide range of choices based on distributed physically based models or on empirical rainfall thresholds but a common weakness of all those approaches is that they are, for quite obvious reasons, very sensitive to errors in the rainfall estimation. This is especially true for temperate or cold climate regions in which a not negligible part of precipitation may be withhold in a snow cover until a temperature change that triggers a sudden water input in the ground. Northern Apennines of Italy are no exception to this rule as demonstrated by the very recent event of Christmas 2009 in which several hundred landslides have been triggered by a rapid snow melt in Tuscany and Emila-Romagna regions...
Atti dei Convegni Lincei: XI Giornata mondiale dell'Acqua, Acqua ed Energia, Mar 2011
Il presente studio comprende il corso del fiume Arno nel Valdarno Medio e nel tratto più a valle ... more Il presente studio comprende il corso del fiume Arno nel Valdarno Medio e nel tratto più a valle del Valdarno Superiore per una lunghezza totale di 65 km. In particolare il tratto preso in considerazione si estende all’interno della Provincia di Firenze dal Borro Vacchereccia al confine con la Provincia di Arezzo (Valdarno Superiore) fino alla confluenza con l’Ombrone in corrispondenza del limite della Provincia di Prato, estendendosi in sponda sinistra fino al circondario Empolese-Valdelsa in località Camaioni (Valdarno Medio). Il lavoro è nato dalla constatazione che gli sbarramenti artificiali e trasversali all’asse del corso d’acqua (traverse o più comunemente pescaie) hanno in generale uno stato di conservazione alquanto critico e che se lasciate all’incuria del tempo ulteriori danni potrebbero inficiare la funzionalità di tali opere ed indurre l’Arno a modificare, se pur localmente, il proprio profilo di equilibrio con conseguenze pratiche per le ormai innumerevoli infrastrutture prospicienti al corso del fiume (scalzamento delle fondazioni dei ponti, occlusione o rottura del sistema fognario, instabilità di manufatti, ecc.). Per questo motivo è quanto mai prioritario progettare un recupero di tali opere idrauliche e si è pensato di legarlo allo sfruttamento del salto idraulico. Da queste valutazioni è scaturita una collaborazione fra Provincia di Firenze e Dipartimento di Scienze della Terra dell’Università di Firenze.
Presentazione di Renzo Crescioli.
La sicurezza idraulica nel territorio fiorentino rappresenta... more Presentazione di Renzo Crescioli.
La sicurezza idraulica nel territorio fiorentino rappresenta elemento principe delle politiche di Difesa del Suolo portate avanti dalla Provincia di Firenze. Al presente, su notevoli porzioni di questo territorio ci troviamo a convivere quotidianamente con un rischio molto elevato, risultato di un’alta pericolosità idraulica, sommatoria di fattori fisici di questo territorio, e soprattutto di una dissennata attività insediativa che ha “saturato” estese porzioni di ambiti fluviali, in cui il fiume poteva esondare liberamente. Sono fatti ormai noti, come è anche noto lo sforzo che stiamo portando avanti come Provincia nell’ambito di accordi interistituzionali per attuare una serie di interventi sul territorio che consentiranno di diminuire la pericolosità idraulica mediante la realizzazione di opere di laminazione. Il nostro impegno è altresì concreto nel cercare di garantire uno sviluppo sostenibile investendo su fonti energetiche rinnovabili, limitando l’aumento di GAS serra atmosferici da molti indicato come una delle principali cause di un generalizzato aumento della intensità degli eventi pluviometrici che interessano anche le nostre latitudini.
Quello del risanamento e della mitigazione del rischio idrogeologico sono percorsi lunghi che terranno impegnate le amministrazioni per i prossimi anni; ci sono tuttavia una serie di azioni sia di ordine conoscitivo che di ordine manutentivo da mettere in atto al fine di garantire il non aggravio della pericolosità idraulica. Questo è lo scopo del progetto Plantario con cui è stata effettuata una capillare ricostruzione dello stato di consistenza delle sponde e delle arginature presenti lungo l’Arno e i suoi principali affluenti, in ottica di poter programmare le attività di ordine strutturale - manutentivo. Ritengo che i risultati ottenuti siano significativi e pongano le basi per una corretta gestione di queste opere idrauliche.
Despite the large number of recent advances and developments in landslide susceptibility mapping ... more Despite the large number of recent advances and developments in landslide susceptibility mapping (LSM) there is still a lack of studies focusing on specific aspects of LSM model sensitivity. For example, the influence of factors of paramount importance such as the survey scale of the landslide conditioning variables (LCVs), the resolution of the mapping unit (MUR) and the optimal number and ranking of LCVs have never been investigated analytically, especially on large datasets. In this paper we attempt this experimentation concentrating on the impact of model tuning choice on the final result, rather than on the comparison of methodologies. To this end, we adopt a simple implementation of the random forest (RF) classification family to produce an ensamble of landslide susceptibility maps for a set of different model settings, input data types and scales. RF classification and regression methods offer a very flexible environment for testing model parameters and mapping hypotheses, al...
ABSTRACT The dynamic evolution of a river and the adjacent morphological environment are particul... more ABSTRACT The dynamic evolution of a river and the adjacent morphological environment are particularly important especially if there are communities that concentrate in these areas their socio-economic activities. So a proper hydraulic risk management is an increasingly felt necessity, but when working at small scales no established fast methodology exists to map the position and the height of the various elements with centimetric accuracy. In the current work an operative methodology likely to obtain this purpose is proposed on the basis of data obtained from a real test area. It is along the Arno river (Italy) which could be considered on the whole a representative case study of other realities in the world. Various issues have been deepened. Firstly RTK-GPS measurements and information about all the natural and artificial elements, connected to hydraulic risk and fluvial dynamics, were collected. All these elements were mapped with high accuracy, in particular a local geoid model, related only to the study area, was developed to obtain orthometric heights affected with errors ≤ 0.05 m. Consequently a GIS geodatabase was built to visualize the spatial distribution of the mapped elements and to store the most important technical data. Such geodatabase provides an overview of the territories connected with the fluvial dynamics of the main rivers near the city of Firenze. This is confirmed by some applications, realized to verify the capability of the instrument. First of all the real hydraulic risk in the study area has been checked out. So the comparison between the measured dike height and the hydraulic modeling conducted by the Arno River Basin Authority has identified areas at risk of overflowing for various return periods (T30, T100, T200 , T500). Subsequently a deeper analysis of hydraulic hazards has been carried out in the urban area of Firenze. A model of surface-water flows concentrated on the historic center has provided a comprehensive response of this area to the sudden appearance of surface-water flows due mainly to the overflowing of the Arno, but also to the excess rainwater and eventually the superficial fluids from other sources. The modeling has been carried out uniquely on the basis of a geomorphological analysis, processing new detailed LiDAR images in GIS environment. For the first time the urban water flows have been identified with extreme precision during three possible phases: during the normal flowing conditions of the Arno, in the event of river obstruction at bridges and in case of undifferentiated run-off out from the riverbed. The three simulations provide the likely scenarios in the urban area of Firenze which can however be integrated with other information for the definition of specific issues. In this work in order to better define the level of safety of the city all the critical elements mapped during the field inspections and the dikes at risk of overflowing previously determined have been incorporated in three models identifyng some critical urban areas.
ABSTRACT The purpose of this study is to apply a regression method, based on a specific version o... more ABSTRACT The purpose of this study is to apply a regression method, based on a specific version of the random forest algorithm, to produce a series of susceptibility maps of the Arno river basin (Central Italy) and to analyze the contribution that each selected preparatory variable has on the final outcome according to varying scales and parameter sets. Random forest is a combination of tree (usually binary) bayesian predictors that permits to relate a set of contributing factors with the actual landslides occurrence. Being it a nonparametric model, it is possible to incorporate a range of numeric or categorical data layers and there is no need to select unimodal training data. The study area is divided into three distinct macro-areas, homogeneous from a geological and lithological point of view. Several classical and widely acknowledged landslide predisposing factors have been taken into account as mainly related to: the lithology, the land use the land surface geometry (derived from of DTM). In addition, for each factor we also included in the parameter set the standard deviation (for numerical variables) or the variety (for categorical ones). The use of random forest enables to estimate the relative importance of the single input parameters and to select the optimal configuration of the regression model. The model was initially applied using the complete set of input parameters at disposal, automatically assigning them a rank by relevance and calculating the ROC curve (with relative AUC value) using an independent testing dataset. Subsequently reduced versions of the random forest model were applied taking into account a progressively lower number of parameters. Step by step the least relevant parameters were discarded and the AUC values of every run was used to assess the effectiveness of the regression model. This procedure has been applied for each area in order to check which parameters need to be taken into account to best evaluate the landslide susceptibility in the study area. Considering the best set of parameters for each macro-area and the impact of scale and accuracy of input variables, the consequences on susceptibility applications are discussed.
ABSTRACT In this work we propose a snow accumulation-melting model (SAMM) to forecast the snowpac... more ABSTRACT In this work we propose a snow accumulation-melting model (SAMM) to forecast the snowpack height and we compare the results with a simple temperature index model and an improved version of the latter.For this purpose we used rainfall, temperature and snowpack thickness 5-years data series from 7 weather stations in the Northern Apennines (Emilia Romagna Region, Italy). SAMM is based on two modules modelling the snow accumulation and the snowmelt processes. Each module is composed by two equations: a mass conservation equation is solved to model snowpack thickness and an empirical equation is used for the snow density. The processes linked to the accumulation/depletion of the snowpack (e.g. compression of the snowpack due to newly fallen snow and effects of rainfall) are modelled identifying limiting and inhibitory factors according to a kinetic approach. The model depends on 13 empirical parameters, whose optimal values were defined with an optimization algorithm (simplex flexible) using calibration measures of snowpack thickness. From an operational point of view, SAMM uses as input data only temperature and rainfall measurements, bringing the additional advantage of a relatively easy implementation. In order to verify the improvement of SAMM with respect to a temperature-index model, the latter was applied considering, for the amount of snow melt, the following equation: M = fm(T-T0), where M is hourly melt, fm is the melting factor and T0 is a threshold temperature. In this case the calculation of the depth of the snowpack requires the use of 3 parameters: fm, T0 and ?0 (the mean density of the snowpack). We also performed a simulation by replacing the SAMM melting module with the above equation and leaving unchanged the accumulation module: in this way we obtained a model with 9 parameters. The simulations results suggest that any further extension of the simple temperature index model brings some improvements with a consequent decrease of the mean error between model and experimental data of the snowpack thickness.
In this paper the first results of the PLANTARIO Project are presented. The project consists in t... more In this paper the first results of the PLANTARIO Project are presented. The project consists in the creation of a GIS-related database containing all the natural, urban, hydrological and morphological elements close to the main rivers within the Florence urban and suburban area. The greatest part of the data were newly acquired from aerial photographs or by direct surveys. An extensive GPS mapping survey was carried out granting a very accurate spatial localization of the elements (less than 5cm 3D error). This project provides local public administrations with an helpful tool for managing hydrological risk, hydraulic policy, urban planning and river restoration. We discuss also two applications resulted from the use of the database: a study to identify the weirs and the contiguous buildings suitable to be converted to the generation of hydroelectric energy and a preliminary study on the Arno dikes stability.
Despite the many recent attempts at devising a general physically-based model to predict the dyna... more Despite the many recent attempts at devising a general physically-based model to predict the dynamics of rainfall induced landslide in real-time, there remain several drawbacks related to the difficulty in estimating the required parameters which show a high spatial variability. For these reasons approaches based on empirical thresholds of rainfall intensity and duration are still widely used and in many
Natural Hazards and Earth System Sciences Discussions, 2013
ABSTRACT Despite the large number of recent advances and developments in landslide susceptibility... more ABSTRACT Despite the large number of recent advances and developments in landslide susceptibility mapping there is still a lack of studies focusing on specific aspects of LSM model sensitivity. For example, the influence of factors of paramount importance such as the survey scale of the landslide conditioning variables (LCVs), the resolution of the mapping unit (MUR) and the optimal number and ranking of LCVs have never been investigated analytically, especially on large datasets. In this paper we attempt this experimentation concentrating on the impact of model tuning choice on the final result, rather than on the comparison of methodologies. To this end, we adopt a simple implementation of the random forest (RF) classification family to produce an ensemble of landslide susceptibility maps for a set of different model settings, input data types and scales. Random forest is a combination of tree (usually binary) bayesian predictors that permits to relate a set of contributing factors with the actual landslides occurrence. Being it a nonparametric model, it is possible to incorporate a range of numeric or categorical data layers and there is no need to select unimodal training data. Many classical and widely acknowledged landslide predisposing factors have been taken into account as mainly related to: the lithology, the land use, the land surface geometry (derived from of DTM), the structural and anthropogenic constrains. In addition, for each factor we also included in the parameter set the standard deviation (for numerical variables) or the variety (for categorical ones). The use of random forest enables to estimate the relative importance of the single input parameters and to select the optimal configuration of the regression model. The model was initially applied using the complete set of input parameters then, with progressively smaller subsamples of the parameter space. Considering the best set of parameters we also studies the impact of scale and accuracy of input variables and the of RF model random component on the susceptibility results. We apply the model statistics to a test area in central Italy, the basin of the Arno river (ca. 9000 km2), we present the obtained results and discuss them. Results confirm that the choice of parameter set, mapping unit resolution and training sampling method highly influences the overall accuracy of classification and prediction results. This, in turn, implies that a careful sensitivity analysis making use of traditional and new tools should always be performed before producing final susceptibility maps at all levels and scales.
ABSTRACT In this paper we made a comparison between various methods to enter soil thickness as a ... more ABSTRACT In this paper we made a comparison between various methods to enter soil thickness as a spatial variable in a deterministic basin scale slope stability simulator. We used a slope stability model that couples a simplified solution of Richards infiltration equation and an infinite slope model with soil suction effect. Soil thickness was entered in the stability modelling using spatially variable maps obtained with four state-of-art methods: linear correlation with elevation; linear correlation with slope gradient; exponential correlation with slope gradient; a more complex geomorphologically indexed model (GIST model). Soil thickness maps and the derivate Factor of Safety (FS) maps were validated. Results confirmed that FS is very sensitive to soil thickness and showed that the same slope stability model can be highly sensitive or highly specific depending on the input soil thickness data. The uncertainty in the FS calculation can be reduced by applying more precise soil thickness input data: mean error of soil thickness maps is closely related to the sensitivity or specificity of the FS computation, while the overall performance of the stability simulation depends on mean absolute error and skewness of the frequency distribution of the errors of soil thickness maps. Despite the fact that slope-based methods are the most used in literature to derive soil thickness, in our application they returned poor results. Conversely, the use of the GIST model improved the performance of the stability model.
Climate change driven landslide forecasting implementation in early warning activities is strictl... more Climate change driven landslide forecasting implementation in early warning activities is strictly dependent on the quality and timing of rainfall measurements and on an accurate empirical or deterministic relationship linking water infiltration to slope stability. The scientific literature offers a wide range of choices based on distributed physically based models or on empirical rainfall thresholds but a common weakness of
ABSTRACT In Central Italy a significant number of landslides occurrence have been triggered by ra... more ABSTRACT In Central Italy a significant number of landslides occurrence have been triggered by rapid snow melt: recent seasonal events in the Northern Apennines, the study area, demonstrate that it is necessary to consider this phenomenon and to integrate snow precipitation within existing statistical models for landslide prediction. The proposed snow melt modeling (SMM) is divided in two modules depending on whether or not a threshold temperature is exceeded: the first one for the accumulation of solid rainfall in the snowpack and the latter for the snow melting. For the modeling identification we employ empirical data of depth of snow cover using an optimization algorithm to deduce the optimal values of the model parameters. This work is developed to increase the predictive capacity of the statistical models for landslide prediction based on rainfall thresholds. In the study area an improvement was achieved: several landslides, caused by snow melting, were correctly detected.
Despite the large number of recent advances and developments in landslide susceptibility mapping ... more Despite the large number of recent advances and developments in landslide susceptibility mapping (LSM) there is still a lack of studies focusing on specific aspects of LSM model sensitivity. For example, the influence of factors such as the survey scale of the landslide conditioning variables (LCVs), the resolution of the mapping unit (MUR) and the optimal number and ranking of LCVs have never been investigated analytically, especially on large data sets. In this paper we attempt this experimentation concentrating on the impact of model tuning choice on the final result , rather than on the comparison of methodologies. To this end, we adopt a simple implementation of the random forest (RF), a machine learning technique, to produce an ensemble of landslide susceptibility maps for a set of different model settings, input data types and scales. Random forest is a combination of Bayesian trees that relates a set of predic-tors to the actual landslide occurrence. Being it a nonpara-metric model, it is possible to incorporate a range of numerical or categorical data layers and there is no need to select unimodal training data as for example in linear discriminant analysis. Many widely acknowledged landslide predisposing factors are taken into account as mainly related to the lithol-ogy, the land use, the geomorphology, the structural and an-thropogenic constraints. In addition, for each factor we also include in the predictors set a measure of the standard deviation (for numerical variables) or the variety (for categorical ones) over the map unit. As in other systems, the use of RF enables one to estimate the relative importance of the single input parameters and to select the optimal configuration of the classification model. The model is initially applied using the complete set of input variables, then an iterative process is implemented and progressively smaller subsets of the parameter space are considered. The impact of scale and accuracy of input variables , as well as the effect of the random component of the RF model on the susceptibility results, are also examined. The model is tested in the Arno River basin (central Italy). We find that the dimension of parameter space, the mapping unit (scale) and the training process strongly influence the classification accuracy and the prediction process. This, in turn, implies that a careful sensitivity analysis making use of traditional and new tools should always be performed before producing final susceptibility maps at all levels and scales.
Un catasto fiumi per il controllo e la mitigazione delle pericolosità di collasso arginale dell&#... more Un catasto fiumi per il controllo e la mitigazione delle pericolosità di collasso arginale dell'Arno e dei suoi principali affluenti
Despite the many recent attempts at devising a general physically-based model to predict the dyna... more Despite the many recent attempts at devising a general physically-based model to predict the dynamics of rainfall induced landslide in real-time, there remain several drawbacks related to the difficulty in estimating the required parameters which show a high spatial variability. For these reasons approaches based on empirical thresholds of rainfall intensity and duration are still widely used and in many
Climate change driven landslide forecasting implementation in early warning activities is strictl... more Climate change driven landslide forecasting implementation in early warning activities is strictly dependent on the quality and timing of rainfall measurements and on an accurate empirical or deterministic relationship linking water infiltration to slope stability. The scientific literature offers a wide range of choices based on distributed physically based models or on empirical rainfall thresholds but a common weakness of all those approaches is that they are, for quite obvious reasons, very sensitive to errors in the rainfall estimation. This is especially true for temperate or cold climate regions in which a not negligible part of precipitation may be withhold in a snow cover until a temperature change that triggers a sudden water input in the ground. Northern Apennines of Italy are no exception to this rule as demonstrated by the very recent event of Christmas 2009 in which several hundred landslides have been triggered by a rapid snow melt in Tuscany and Emila-Romagna regions...
Atti dei Convegni Lincei: XI Giornata mondiale dell'Acqua, Acqua ed Energia, Mar 2011
Il presente studio comprende il corso del fiume Arno nel Valdarno Medio e nel tratto più a valle ... more Il presente studio comprende il corso del fiume Arno nel Valdarno Medio e nel tratto più a valle del Valdarno Superiore per una lunghezza totale di 65 km. In particolare il tratto preso in considerazione si estende all’interno della Provincia di Firenze dal Borro Vacchereccia al confine con la Provincia di Arezzo (Valdarno Superiore) fino alla confluenza con l’Ombrone in corrispondenza del limite della Provincia di Prato, estendendosi in sponda sinistra fino al circondario Empolese-Valdelsa in località Camaioni (Valdarno Medio). Il lavoro è nato dalla constatazione che gli sbarramenti artificiali e trasversali all’asse del corso d’acqua (traverse o più comunemente pescaie) hanno in generale uno stato di conservazione alquanto critico e che se lasciate all’incuria del tempo ulteriori danni potrebbero inficiare la funzionalità di tali opere ed indurre l’Arno a modificare, se pur localmente, il proprio profilo di equilibrio con conseguenze pratiche per le ormai innumerevoli infrastrutture prospicienti al corso del fiume (scalzamento delle fondazioni dei ponti, occlusione o rottura del sistema fognario, instabilità di manufatti, ecc.). Per questo motivo è quanto mai prioritario progettare un recupero di tali opere idrauliche e si è pensato di legarlo allo sfruttamento del salto idraulico. Da queste valutazioni è scaturita una collaborazione fra Provincia di Firenze e Dipartimento di Scienze della Terra dell’Università di Firenze.
Presentazione di Renzo Crescioli.
La sicurezza idraulica nel territorio fiorentino rappresenta... more Presentazione di Renzo Crescioli.
La sicurezza idraulica nel territorio fiorentino rappresenta elemento principe delle politiche di Difesa del Suolo portate avanti dalla Provincia di Firenze. Al presente, su notevoli porzioni di questo territorio ci troviamo a convivere quotidianamente con un rischio molto elevato, risultato di un’alta pericolosità idraulica, sommatoria di fattori fisici di questo territorio, e soprattutto di una dissennata attività insediativa che ha “saturato” estese porzioni di ambiti fluviali, in cui il fiume poteva esondare liberamente. Sono fatti ormai noti, come è anche noto lo sforzo che stiamo portando avanti come Provincia nell’ambito di accordi interistituzionali per attuare una serie di interventi sul territorio che consentiranno di diminuire la pericolosità idraulica mediante la realizzazione di opere di laminazione. Il nostro impegno è altresì concreto nel cercare di garantire uno sviluppo sostenibile investendo su fonti energetiche rinnovabili, limitando l’aumento di GAS serra atmosferici da molti indicato come una delle principali cause di un generalizzato aumento della intensità degli eventi pluviometrici che interessano anche le nostre latitudini.
Quello del risanamento e della mitigazione del rischio idrogeologico sono percorsi lunghi che terranno impegnate le amministrazioni per i prossimi anni; ci sono tuttavia una serie di azioni sia di ordine conoscitivo che di ordine manutentivo da mettere in atto al fine di garantire il non aggravio della pericolosità idraulica. Questo è lo scopo del progetto Plantario con cui è stata effettuata una capillare ricostruzione dello stato di consistenza delle sponde e delle arginature presenti lungo l’Arno e i suoi principali affluenti, in ottica di poter programmare le attività di ordine strutturale - manutentivo. Ritengo che i risultati ottenuti siano significativi e pongano le basi per una corretta gestione di queste opere idrauliche.
Despite the large number of recent advances and developments in landslide susceptibility mapping ... more Despite the large number of recent advances and developments in landslide susceptibility mapping (LSM) there is still a lack of studies focusing on specific aspects of LSM model sensitivity. For example, the influence of factors of paramount importance such as the survey scale of the landslide conditioning variables (LCVs), the resolution of the mapping unit (MUR) and the optimal number and ranking of LCVs have never been investigated analytically, especially on large datasets. In this paper we attempt this experimentation concentrating on the impact of model tuning choice on the final result, rather than on the comparison of methodologies. To this end, we adopt a simple implementation of the random forest (RF) classification family to produce an ensamble of landslide susceptibility maps for a set of different model settings, input data types and scales. RF classification and regression methods offer a very flexible environment for testing model parameters and mapping hypotheses, al...
ABSTRACT The dynamic evolution of a river and the adjacent morphological environment are particul... more ABSTRACT The dynamic evolution of a river and the adjacent morphological environment are particularly important especially if there are communities that concentrate in these areas their socio-economic activities. So a proper hydraulic risk management is an increasingly felt necessity, but when working at small scales no established fast methodology exists to map the position and the height of the various elements with centimetric accuracy. In the current work an operative methodology likely to obtain this purpose is proposed on the basis of data obtained from a real test area. It is along the Arno river (Italy) which could be considered on the whole a representative case study of other realities in the world. Various issues have been deepened. Firstly RTK-GPS measurements and information about all the natural and artificial elements, connected to hydraulic risk and fluvial dynamics, were collected. All these elements were mapped with high accuracy, in particular a local geoid model, related only to the study area, was developed to obtain orthometric heights affected with errors ≤ 0.05 m. Consequently a GIS geodatabase was built to visualize the spatial distribution of the mapped elements and to store the most important technical data. Such geodatabase provides an overview of the territories connected with the fluvial dynamics of the main rivers near the city of Firenze. This is confirmed by some applications, realized to verify the capability of the instrument. First of all the real hydraulic risk in the study area has been checked out. So the comparison between the measured dike height and the hydraulic modeling conducted by the Arno River Basin Authority has identified areas at risk of overflowing for various return periods (T30, T100, T200 , T500). Subsequently a deeper analysis of hydraulic hazards has been carried out in the urban area of Firenze. A model of surface-water flows concentrated on the historic center has provided a comprehensive response of this area to the sudden appearance of surface-water flows due mainly to the overflowing of the Arno, but also to the excess rainwater and eventually the superficial fluids from other sources. The modeling has been carried out uniquely on the basis of a geomorphological analysis, processing new detailed LiDAR images in GIS environment. For the first time the urban water flows have been identified with extreme precision during three possible phases: during the normal flowing conditions of the Arno, in the event of river obstruction at bridges and in case of undifferentiated run-off out from the riverbed. The three simulations provide the likely scenarios in the urban area of Firenze which can however be integrated with other information for the definition of specific issues. In this work in order to better define the level of safety of the city all the critical elements mapped during the field inspections and the dikes at risk of overflowing previously determined have been incorporated in three models identifyng some critical urban areas.
ABSTRACT The purpose of this study is to apply a regression method, based on a specific version o... more ABSTRACT The purpose of this study is to apply a regression method, based on a specific version of the random forest algorithm, to produce a series of susceptibility maps of the Arno river basin (Central Italy) and to analyze the contribution that each selected preparatory variable has on the final outcome according to varying scales and parameter sets. Random forest is a combination of tree (usually binary) bayesian predictors that permits to relate a set of contributing factors with the actual landslides occurrence. Being it a nonparametric model, it is possible to incorporate a range of numeric or categorical data layers and there is no need to select unimodal training data. The study area is divided into three distinct macro-areas, homogeneous from a geological and lithological point of view. Several classical and widely acknowledged landslide predisposing factors have been taken into account as mainly related to: the lithology, the land use the land surface geometry (derived from of DTM). In addition, for each factor we also included in the parameter set the standard deviation (for numerical variables) or the variety (for categorical ones). The use of random forest enables to estimate the relative importance of the single input parameters and to select the optimal configuration of the regression model. The model was initially applied using the complete set of input parameters at disposal, automatically assigning them a rank by relevance and calculating the ROC curve (with relative AUC value) using an independent testing dataset. Subsequently reduced versions of the random forest model were applied taking into account a progressively lower number of parameters. Step by step the least relevant parameters were discarded and the AUC values of every run was used to assess the effectiveness of the regression model. This procedure has been applied for each area in order to check which parameters need to be taken into account to best evaluate the landslide susceptibility in the study area. Considering the best set of parameters for each macro-area and the impact of scale and accuracy of input variables, the consequences on susceptibility applications are discussed.
ABSTRACT In this work we propose a snow accumulation-melting model (SAMM) to forecast the snowpac... more ABSTRACT In this work we propose a snow accumulation-melting model (SAMM) to forecast the snowpack height and we compare the results with a simple temperature index model and an improved version of the latter.For this purpose we used rainfall, temperature and snowpack thickness 5-years data series from 7 weather stations in the Northern Apennines (Emilia Romagna Region, Italy). SAMM is based on two modules modelling the snow accumulation and the snowmelt processes. Each module is composed by two equations: a mass conservation equation is solved to model snowpack thickness and an empirical equation is used for the snow density. The processes linked to the accumulation/depletion of the snowpack (e.g. compression of the snowpack due to newly fallen snow and effects of rainfall) are modelled identifying limiting and inhibitory factors according to a kinetic approach. The model depends on 13 empirical parameters, whose optimal values were defined with an optimization algorithm (simplex flexible) using calibration measures of snowpack thickness. From an operational point of view, SAMM uses as input data only temperature and rainfall measurements, bringing the additional advantage of a relatively easy implementation. In order to verify the improvement of SAMM with respect to a temperature-index model, the latter was applied considering, for the amount of snow melt, the following equation: M = fm(T-T0), where M is hourly melt, fm is the melting factor and T0 is a threshold temperature. In this case the calculation of the depth of the snowpack requires the use of 3 parameters: fm, T0 and ?0 (the mean density of the snowpack). We also performed a simulation by replacing the SAMM melting module with the above equation and leaving unchanged the accumulation module: in this way we obtained a model with 9 parameters. The simulations results suggest that any further extension of the simple temperature index model brings some improvements with a consequent decrease of the mean error between model and experimental data of the snowpack thickness.
In this paper the first results of the PLANTARIO Project are presented. The project consists in t... more In this paper the first results of the PLANTARIO Project are presented. The project consists in the creation of a GIS-related database containing all the natural, urban, hydrological and morphological elements close to the main rivers within the Florence urban and suburban area. The greatest part of the data were newly acquired from aerial photographs or by direct surveys. An extensive GPS mapping survey was carried out granting a very accurate spatial localization of the elements (less than 5cm 3D error). This project provides local public administrations with an helpful tool for managing hydrological risk, hydraulic policy, urban planning and river restoration. We discuss also two applications resulted from the use of the database: a study to identify the weirs and the contiguous buildings suitable to be converted to the generation of hydroelectric energy and a preliminary study on the Arno dikes stability.
Despite the many recent attempts at devising a general physically-based model to predict the dyna... more Despite the many recent attempts at devising a general physically-based model to predict the dynamics of rainfall induced landslide in real-time, there remain several drawbacks related to the difficulty in estimating the required parameters which show a high spatial variability. For these reasons approaches based on empirical thresholds of rainfall intensity and duration are still widely used and in many
Natural Hazards and Earth System Sciences Discussions, 2013
ABSTRACT Despite the large number of recent advances and developments in landslide susceptibility... more ABSTRACT Despite the large number of recent advances and developments in landslide susceptibility mapping there is still a lack of studies focusing on specific aspects of LSM model sensitivity. For example, the influence of factors of paramount importance such as the survey scale of the landslide conditioning variables (LCVs), the resolution of the mapping unit (MUR) and the optimal number and ranking of LCVs have never been investigated analytically, especially on large datasets. In this paper we attempt this experimentation concentrating on the impact of model tuning choice on the final result, rather than on the comparison of methodologies. To this end, we adopt a simple implementation of the random forest (RF) classification family to produce an ensemble of landslide susceptibility maps for a set of different model settings, input data types and scales. Random forest is a combination of tree (usually binary) bayesian predictors that permits to relate a set of contributing factors with the actual landslides occurrence. Being it a nonparametric model, it is possible to incorporate a range of numeric or categorical data layers and there is no need to select unimodal training data. Many classical and widely acknowledged landslide predisposing factors have been taken into account as mainly related to: the lithology, the land use, the land surface geometry (derived from of DTM), the structural and anthropogenic constrains. In addition, for each factor we also included in the parameter set the standard deviation (for numerical variables) or the variety (for categorical ones). The use of random forest enables to estimate the relative importance of the single input parameters and to select the optimal configuration of the regression model. The model was initially applied using the complete set of input parameters then, with progressively smaller subsamples of the parameter space. Considering the best set of parameters we also studies the impact of scale and accuracy of input variables and the of RF model random component on the susceptibility results. We apply the model statistics to a test area in central Italy, the basin of the Arno river (ca. 9000 km2), we present the obtained results and discuss them. Results confirm that the choice of parameter set, mapping unit resolution and training sampling method highly influences the overall accuracy of classification and prediction results. This, in turn, implies that a careful sensitivity analysis making use of traditional and new tools should always be performed before producing final susceptibility maps at all levels and scales.
ABSTRACT In this paper we made a comparison between various methods to enter soil thickness as a ... more ABSTRACT In this paper we made a comparison between various methods to enter soil thickness as a spatial variable in a deterministic basin scale slope stability simulator. We used a slope stability model that couples a simplified solution of Richards infiltration equation and an infinite slope model with soil suction effect. Soil thickness was entered in the stability modelling using spatially variable maps obtained with four state-of-art methods: linear correlation with elevation; linear correlation with slope gradient; exponential correlation with slope gradient; a more complex geomorphologically indexed model (GIST model). Soil thickness maps and the derivate Factor of Safety (FS) maps were validated. Results confirmed that FS is very sensitive to soil thickness and showed that the same slope stability model can be highly sensitive or highly specific depending on the input soil thickness data. The uncertainty in the FS calculation can be reduced by applying more precise soil thickness input data: mean error of soil thickness maps is closely related to the sensitivity or specificity of the FS computation, while the overall performance of the stability simulation depends on mean absolute error and skewness of the frequency distribution of the errors of soil thickness maps. Despite the fact that slope-based methods are the most used in literature to derive soil thickness, in our application they returned poor results. Conversely, the use of the GIST model improved the performance of the stability model.
Climate change driven landslide forecasting implementation in early warning activities is strictl... more Climate change driven landslide forecasting implementation in early warning activities is strictly dependent on the quality and timing of rainfall measurements and on an accurate empirical or deterministic relationship linking water infiltration to slope stability. The scientific literature offers a wide range of choices based on distributed physically based models or on empirical rainfall thresholds but a common weakness of
ABSTRACT In Central Italy a significant number of landslides occurrence have been triggered by ra... more ABSTRACT In Central Italy a significant number of landslides occurrence have been triggered by rapid snow melt: recent seasonal events in the Northern Apennines, the study area, demonstrate that it is necessary to consider this phenomenon and to integrate snow precipitation within existing statistical models for landslide prediction. The proposed snow melt modeling (SMM) is divided in two modules depending on whether or not a threshold temperature is exceeded: the first one for the accumulation of solid rainfall in the snowpack and the latter for the snow melting. For the modeling identification we employ empirical data of depth of snow cover using an optimization algorithm to deduce the optimal values of the model parameters. This work is developed to increase the predictive capacity of the statistical models for landslide prediction based on rainfall thresholds. In the study area an improvement was achieved: several landslides, caused by snow melting, were correctly detected.
Uploads
Papers
La sicurezza idraulica nel territorio fiorentino rappresenta elemento principe delle politiche di Difesa del Suolo portate avanti dalla Provincia di Firenze. Al presente, su notevoli porzioni di questo territorio ci troviamo a convivere quotidianamente con un rischio molto elevato, risultato di un’alta pericolosità idraulica, sommatoria di fattori fisici di questo territorio, e soprattutto di una dissennata attività insediativa che ha “saturato” estese porzioni di ambiti fluviali, in cui il fiume poteva esondare liberamente. Sono fatti ormai noti, come è anche noto lo sforzo che stiamo portando avanti come Provincia nell’ambito di accordi interistituzionali per attuare una serie di interventi sul territorio che consentiranno di diminuire la pericolosità idraulica mediante la realizzazione di opere di laminazione. Il nostro impegno è altresì concreto nel cercare di garantire uno sviluppo sostenibile investendo su fonti energetiche rinnovabili, limitando l’aumento di GAS serra atmosferici da molti indicato come una delle principali cause di un generalizzato aumento della intensità degli eventi pluviometrici che interessano anche le nostre latitudini.
Quello del risanamento e della mitigazione del rischio idrogeologico sono percorsi lunghi che terranno impegnate le amministrazioni per i prossimi anni; ci sono tuttavia una serie di azioni sia di ordine conoscitivo che di ordine manutentivo da mettere in atto al fine di garantire il non aggravio della pericolosità idraulica. Questo è lo scopo del progetto Plantario con cui è stata effettuata una capillare ricostruzione dello stato di consistenza delle sponde e delle arginature presenti lungo l’Arno e i suoi principali affluenti, in ottica di poter programmare le attività di ordine strutturale - manutentivo. Ritengo che i risultati ottenuti siano significativi e pongano le basi per una corretta gestione di queste opere idrauliche.
La sicurezza idraulica nel territorio fiorentino rappresenta elemento principe delle politiche di Difesa del Suolo portate avanti dalla Provincia di Firenze. Al presente, su notevoli porzioni di questo territorio ci troviamo a convivere quotidianamente con un rischio molto elevato, risultato di un’alta pericolosità idraulica, sommatoria di fattori fisici di questo territorio, e soprattutto di una dissennata attività insediativa che ha “saturato” estese porzioni di ambiti fluviali, in cui il fiume poteva esondare liberamente. Sono fatti ormai noti, come è anche noto lo sforzo che stiamo portando avanti come Provincia nell’ambito di accordi interistituzionali per attuare una serie di interventi sul territorio che consentiranno di diminuire la pericolosità idraulica mediante la realizzazione di opere di laminazione. Il nostro impegno è altresì concreto nel cercare di garantire uno sviluppo sostenibile investendo su fonti energetiche rinnovabili, limitando l’aumento di GAS serra atmosferici da molti indicato come una delle principali cause di un generalizzato aumento della intensità degli eventi pluviometrici che interessano anche le nostre latitudini.
Quello del risanamento e della mitigazione del rischio idrogeologico sono percorsi lunghi che terranno impegnate le amministrazioni per i prossimi anni; ci sono tuttavia una serie di azioni sia di ordine conoscitivo che di ordine manutentivo da mettere in atto al fine di garantire il non aggravio della pericolosità idraulica. Questo è lo scopo del progetto Plantario con cui è stata effettuata una capillare ricostruzione dello stato di consistenza delle sponde e delle arginature presenti lungo l’Arno e i suoi principali affluenti, in ottica di poter programmare le attività di ordine strutturale - manutentivo. Ritengo che i risultati ottenuti siano significativi e pongano le basi per una corretta gestione di queste opere idrauliche.