This research aims at characterizing susceptibility conditions to gully erosion by means of GIS and multivariate statistical analysis. The study area is a 9.5 km 2 river catchment in central-northern Sicily, where agriculture activities are limited by intense erosion. By means of field surveys and interpretation of aerial images, we prepared a digital map of the spatial distribution of 260 gullies in the study area. In addition, from available thematic maps, a 5 m cell size digital elevation model and field checks, we derived 27 environmental attributes that describe the variability of lithology, land use, topography and road position. These attributes were selected for their potential influence on erosion processes, while the dependent variable was given by presence or absence of gullies within two different types of mapping units: 5 m grid cells and slope units (average size = 2.66 ha). The functional relationships between gully occurrence and the controlling factors were obtained from forward stepwise logistic regression to calculate the probability to host a gully for each mapping unit. In order to train and test the predictive models, three calibration and three validation subsets, of both grid cells and slope units, were randomly selected. Results of validation, based on ROC (receiving operating characteristic) curves, attest for acceptable to excellent accuracies of the models, showing better predictive skill and more stable performance of the susceptibility model based on grid cells.

No abstract

A statistical approach was employed to model the spatial distribution of rainfall-triggered landslides in two areas in Sicily (Italy) that occurred during the winter of 2004-2005. The investigated areas are located within the Belice River basin and extend for 38.5 and 10.3 km2, respectively. A landslide inventory was established for both areas using two Google Earth images taken on October 25th 2004 and on March 18th 2005, to map slope failures activated or reactivated during this interval. Geographic Information Systems (GIS) were used to prepare 5 m grids of the dependent variables (absence/presence of landslide) and independent variables (lithology and 13 DEM-derivatives). Multivariate Adaptive Regression Splines (MARS) were applied to model landslide susceptibility whereas receiver operating characteristic (ROC) curves and the area under the ROC curve (AUC) were used to evaluate model performance. To evaluate the robustness of the whole procedure, we prepared 10 different samples of positive (landslide presence) and negative (landslide absence) cases for each area. Absences were selected through two different methods: (i) extraction from randomly distributed circles with a diameter corresponding to the mean width of the landslide source areas; and (ii) selection as randomly distributed individual grid cells. A comparison was also made between the predictive performances of models including and not including the lithology parameter.The models trained and tested on the same area demonstrated excellent to outstanding fit (AUC > 0.8). On the other hand, predictive skill decreases when measured outside the calibration area, although most of the landslides occur where susceptibility is high and the overall model performance is acceptable (AUC > 0.7). The results also showed that the accuracy of the landslide susceptibility models is higher when lithology is included in the statistical analysis. Models whose absences were selected using random circles showed a significantly better performance when learning and validation samples were extracted from the same area; whereas, conversely, no significant difference was observed when testing the models outside the training area

The aim of this study is to analyze the susceptibility conditions to gully erosion phenomena in the Magazzolo River basin and to test a method that allows for driving the factors selection. The study area is one of the largest (225 km2) watershed of southern Sicily and it is mostly characterized by gentle slopes carved into clayey and evaporitic sediments, except for the northern sector where carbonatic rocks give rise to steep slopes. In order to obtain a quantitative evaluation of gully erosion susceptibility, statistical relationships between the spatial distributions of gullies affecting the area and a set of twelve environmental variables were analyzed. Stereoscopic analysis of aerial photographs dated 2000, and field surveys carried out in 2006, allowed us to map about a thousand landforms produced by linear water erosion processes, classifiable as ephemeral and permanent gullies. The linear density of the gullies, computed on each of the factors classes, was assumed as the function expressing the susceptibility level of the latter. A 40-m digital elevation model (DEM) prepared from 1:10,000-scale topographic maps was used to compute the values of nine topographic attributes (primary: slope, aspect, plan curvature, profile curvature, general curvature, tangential curvature; secondary: stream power index; topographic wetness index; LS-USLE factor); from available thematic maps and field checks three other physical attributes (lithology, soil texture, land use) were derived. For each of these variables, a 40-m grid layer was generated, reclassifying the topographic variables according to their standard deviation values. In order to evaluate the controlling role of the selected predictive variables, one-variable susceptibility models, based on the spatial relationships between each single factor and gullies, were produced and submitted to a validation procedure. The latter was carried out by evaluating the predictive performance of models trained on one half of the landform archive and tested on the other. Large differences of accuracy were verified by computing geometric indexes of the validation curves (prediction and success rate curves; ROC curves) drawn for each one-variable model; in particular, soil texture, general curvature and aspect demonstrated a weak or a null influence on the spatial distribution of gullies within the studied area, while, on the contrary, tangential curvature, stream power index and plan curvature showed high predictive skills. Hence, predictive models were produced on a multi-variable basis, by variously combining the one-variable models. The validation of the multi-variables models, which generally indicated quite satisfactory results, were used as a sensitivity analysis tool to evaluate differences in the prediction results produced by changing the set of combined physical attributes. The sensitivity analysis pointed out that by increasing the number of combined environmental variables, an improvement of the susceptibility assessment is produced; this is true with the exceptio...

Abstract. The main assumption on which landslide susceptibility assessment by means of stochastic modelling lies is that the past is the key to the future. As a consequence, a stochastic model able to classify past known landslide events should be able to predict a future unknown scenario as well. However, storm-triggered multiple debris flow events in the Mediterranean region could pose some limits on the operative validity of such an expectation, as they are typically resultant of a randomness in time recurrence and magnitude and a great spatial variability, even at the scale of small catchments. This is the case for the 2007 and 2009 storm events, which recently hit north-eastern Sicily with different intensities, resulting in largely different disaster scenarios.The study area is the small catchment of the Itala torrent (10 km 2 ), which drains from the southern Peloritani Mountains eastward to the Ionian Sea, in the territory of the Messina province (Sicily, Italy). Landslides have been mapped by integrating remote and field surveys, producing two event inventories which include 73 debris flows, activated in 2007, and 616 debris flows, triggered by the 2009 storm. Logistic regression was applied in order to obtain susceptibility models which utilize a set of predictors derived from a 2 m cell digital elevation model and a 1 : 50 000 scale geologic map. The research topic was explored by performing two types of validation procedures: self-validation, based on the random partition of each event inventory, and chronovalidation, based on the time partition of the landslide in- Both of the two predictions resulted in largely acceptable performances in terms of fitting, skill and reliability. However, a loss of performance and differences in the selected predictors arose between the self-validated and the chronovalidated models. These are interpreted as effects of the nonlinearity in the domain of the trigger intensity of the relationships between predictors and slope response, as well as in terms of the different spatial paths of the two triggering storms at the catchment scale.