Evaluating the Effect of Modelling Methods and Landslide Inventories Used for Statistical Susceptibility Modelling (original) (raw)
Related papers
Assessing the quality of landslide susceptibility maps for Lower Austria
2012
Landslide susceptibility maps are helpful tools to identify areas potentially prone to future landslide occurrence. As more and more national and provincial authorities demand for these maps to be computed and implemented in spatial planning strategies, several aspects of the quality of the landslide susceptibility model and the resulting classified map are of high interest. In this study of landslides in Lower Austria, we focus on the model form uncertainty to assess the quality of a flexible statistical modelling technique, the generalized additive model (GAM). The study area (15 850 km 2 ) is divided into 16 modelling domains based on lithology classes. A model representing the entire study area is constructed by combining these models. The performances of the models are assessed using repeated k-fold cross-validation with spatial and random subsampling. This reflects the variability of performance estimates arising from sampling variation. Measures of spatial transferability and thematic consistency are applied to empirically assess model quality. We also analyse and visualize the implications of spatially varying prediction uncertainties regarding the susceptibility map classes by taking into account the confidence intervals of model predictions. The 95 % confidence limits fall within the same susceptibility class in 85 % of the study area. Overall, this study contributes to advancing open communication and assessment of model quality related to statistical landslide susceptibility models.
Assessing the quality of landslide susceptibility maps – case study Lower Austria
Natural Hazards and Earth System Science, 2014
Landslide susceptibility maps are helpful tools to identify areas potentially prone to future landslide occurrence. As more and more national and provincial authorities demand for these maps to be computed and implemented in spatial planning strategies, several aspects of the quality of the landslide susceptibility model and the resulting classified map are of high interest. In this study of landslides in Lower Austria, we focus on the model form uncertainty to assess the quality of a flexible statistical modelling technique, the generalized additive model (GAM). The study area (15 850 km 2 ) is divided into 16 modelling domains based on lithology classes. A model representing the entire study area is constructed by combining these models. The performances of the models are assessed using repeated k-fold cross-validation with spatial and random subsampling. This reflects the variability of performance estimates arising from sampling variation. Measures of spatial transferability and thematic consistency are applied to empirically assess model quality. We also analyse and visualize the implications of spatially varying prediction uncertainties regarding the susceptibility map classes by taking into account the confidence intervals of model predictions. The 95 % confidence limits fall within the same susceptibility class in 85 % of the study area. Overall, this study contributes to advancing open communication and assessment of model quality related to statistical landslide susceptibility models.
2014
Landslide susceptibility maps are helpful tools to identify areas potentially prone to future landslide occurrence. As more and more national and provincial authorities demand for these maps to be computed and implemented in spatial planning strategies, several aspects of the quality of the landslide susceptibility model and the resulting classified map are of high interest. In this study of landslides in Lower Austria, we focus on the model form uncertainty to assess the quality of a flexible statistical modelling technique, the generalized additive model (GAM). The study area (15 850 km 2 ) is divided into 16 modelling domains based on lithology classes. A model representing the entire study area is constructed by combining these models. The performances of the models are assessed using repeated k-fold cross-validation with spatial and random subsampling. This reflects the variability of performance estimates arising from sampling variation. Measures of spatial transferability and thematic consistency are applied to empirically assess model quality. We also analyse and visualize the implications of spatially varying prediction uncertainties regarding the susceptibility map classes by taking into account the confidence intervals of model predictions. The 95 % confidence limits fall within the same susceptibility class in 85 % of the study area. Overall, this study contributes to advancing open communication and assessment of model quality related to statistical landslide susceptibility models.
Landslide Inventories for Reliable Susceptibility Maps in Lower Austria
2013
Landslide inventories, their accuracy and the stored information are of major importance for landslide susceptibility modelling. Working on the scale of a province (Lower Austria with about 10,000 km²) challenges arise due to data availability and its spatial representation. Furthermore, previous studies on existing landslide inventories showed that only few inventories can be used for statistical susceptibility modelling. In this study two landslide inventories and their resulting susceptibility maps are compared: the Building Ground Register (BGR) of the Geological Survey of Lower Austria and an inventory that was mapped on the basis of a high resolution LiDAR DTM. This analysis was performed to estimate minimum requirements on landslide inventories to allow for deriving reliable susceptibility maps while minimizing mapping efforts. Therefore a consistent landslide inventory once from the BGR and once from the mapping was compiled. Furthermore, a logistic regression model was fitted with randomly selected points of each landslide inventory to compare the resulting maps and validation rates. The resulting landslide susceptibility maps show significant differences regarding their visual and statistical quality. We conclude that the application of randomly selected points in the main scarp of the mapped landslides gives satisfactory results.
Assessing the quality of landslide susceptibility models - case study Lower Austria
Landslide susceptibility maps are helpful tools to identify areas potentially prone to future landslide occurrence. As more and more national and provincial authorities demand for these maps to be computed and implemented in spatial planning strategies, several aspects of the quality of the landslide susceptibility model and the resulting classified map are of high interest. In this study of landslides in Lower Austria, we focus on the model form uncertainty to assess the quality of a flexible statistical modelling technique, the generalized additive model (GAM). The study area (15 850 km2) is divided into 16 modelling domains based on lithology classes. A model representing the entire study area is constructed by combining these models. The performances of the models are assessed using repeated k-fold cross-validation with spatial and random subsampling. This reflects the variability of performance estimates arising from sampling variation. Measures of spatial transferability and thematic consistency are applied to empirically assess model quality. We also analyse and visualize the implications of spatially varying prediction uncertainties regarding the susceptibility map classes by taking into account the confidence intervals of model predictions. The 95% confidence limits fall within the same susceptibility class in 85% of the study area. Overall, this study contributes to advancing open communication and assessment of model quality related to statistical landslide susceptibility models.
2013
Landslide inventories, their accuracy and the stored information are of major importance for landslide susceptibility modelling. Working on the scale of a province (Lower Austria with about 10,000 km²) challenges arise due to data availability and its spatial representation. Furthermore, previous studies on existing landslide inventories showed that only few inventories can be used for statistical susceptibility modelling. In this study two landslide inventories and their resulting susceptibility maps are compared: the Building Ground Register (BGR) of the Geological Survey of Lower Austria and an inventory that was mapped on the basis of a high resolution LiDAR DTM. This analysis was performed to estimate minimum requirements on landslide inventories to allow for deriving reliable susceptibility maps while minimizing mapping efforts. Therefore a consistent landslide inventory once from the BGR and once from the mapping was compiled. Furthermore, a logistic regression model was fitted with randomly selected points of each landslide inventory to compare the resulting maps and validation rates. The resulting landslide susceptibility maps show significant differences regarding their visual and statistical quality. We conclude that the application of randomly selected points in the main scarp of the mapped landslides gives satisfactory results.
Landslides, 2021
The reliability of input data to be used within statistically based landslide susceptibility models usually determines the quality of the resulting maps. For very large territories, landslide susceptibility assessments are commonly built upon spatially incomplete and positionally inaccurate landslide information. The unavailability of flawless input data is contrasted by the need to identify landslide-prone terrain at such spatial scales. Instead of simply ignoring errors in the landslide data, we argue that modellers have to explicitly adopt their modelling design to avoid misleading results. This study examined different modelling strategies to reduce undesirable effects of error-prone landslide inventory data, namely systematic spatial incompleteness and positional inaccuracies. For this purpose, the Austrian territory with its abundant but heterogeneous landslide data was selected as a study site. Conventional modelling practices were compared with alternative modelling designs ...
Susceptibility Maps for Landslides Using Different Modelling Approaches
Springer eBooks, 2013
This study focuses on the comparison of different modelling approaches and is part of the research project "MoNOE" (Method development for landslide susceptibility modelling in Lower Austria). The main objective of the project is to design a method for landslide susceptibility modelling for a large study area. For other objectives of the project see Bell at al. in this volume. To reach the main objective two different statistical models, Weights of Evidence and Logistic Regression are applied and compared. By using nearly the same input data in test areas it is possible to compare the capabilities of these different methods. First results of the comparison indicate that in valleys and on south directed slopes the results of the two different modelling approaches are quite similar. The results on north directed slopes show differences. In the ongoing work the reason for these differences will be analysed. Also it will be necessary to find adequate validation methods for the two modelling approaches.
Remote Sensing
Landslide susceptibility is one of the main topics of geomorphological risk studies. Unfortunately, many of these studies applied an exclusively statistical approach with little coherence with the geomorphodynamic models, resulting in susceptibility maps that are difficult to read. Even if many different models have been developed, those based on statistical techniques applied to slope units (SUs) are among the most promising. SU segmentation divides terrain into homogenous domains and approximates the morphodynamic response of the slope to landslides. This paper presents a landslide susceptibility (LS) analysis at the catchment scale for a key area based on the comparison of two GIS-based bivariate statistical methods using the landslide index (LI) approach. A new simple and reproducible method for delineating SUs is defined with an original GIS-based terrain segmentation based on hydrography. For the first time, the morphometric slope index (MSI) was tested as a predisposing facto...