Anthropogenic activities and urbanization frequently spread over areas underlain by unstable karst ground or man-made cavities, leading to potential risk situations for buildings, infrastructure and population. The roof of the cavities may rapidly propagate upwards, eventually resulting in progressive ground settlement and/or sudden collapse. These processes are frequently accelerated or triggered by various human activities (e.g., overloading, water recharge, water table decline). In this paper, the susceptibility to instability of artificial cavity networks under an urban area in Apulia Region, Southern Italy, was assessed and evaluated. Here, there is dense network of underground quarries excavated for the extraction of massive calcarenite for building and ornamental stone that are currently affected by severe instability problems. Identifying the zones of the cavities most prone to instability is of great practical importance. Susceptibility models have been developed analyzing the statistical relationships between the cavity sections with evidence of instability and a number of predisposing factors mainly related to the 3D geometry of the voids, roof thickness and the presence of overloads. The models produced by the discriminant analysis and the logistic regression approaches were independently and quantitatively evaluated through the construction of ROC curves. Although the two models are characterized by different input data treatment, as the logistic regression allows using continuous and discrete variables or any combination of both types even if they do not have a normal distribution, the ROC curves highlighted, for both methods, a satisfactory congruence between the model results and the observed data. Consequently, from the model performance analysis, the two models were evaluated comparable in terms of result reliability. The methodology should be considered as a general first-level analysis for near surface underground cavities aimed at identifying critical cave sectors that can be the focus of more detailed analyses, especially those associated with sensitive zone related to the presence of buildings, roads and frequent presence of human activity.
Susceptibility mapping of instability related to shallow mining cavities in a built-up environment
PELLICANI, ROBERTA;SPILOTRO, Giuseppe;
2017-01-01
Abstract
Anthropogenic activities and urbanization frequently spread over areas underlain by unstable karst ground or man-made cavities, leading to potential risk situations for buildings, infrastructure and population. The roof of the cavities may rapidly propagate upwards, eventually resulting in progressive ground settlement and/or sudden collapse. These processes are frequently accelerated or triggered by various human activities (e.g., overloading, water recharge, water table decline). In this paper, the susceptibility to instability of artificial cavity networks under an urban area in Apulia Region, Southern Italy, was assessed and evaluated. Here, there is dense network of underground quarries excavated for the extraction of massive calcarenite for building and ornamental stone that are currently affected by severe instability problems. Identifying the zones of the cavities most prone to instability is of great practical importance. Susceptibility models have been developed analyzing the statistical relationships between the cavity sections with evidence of instability and a number of predisposing factors mainly related to the 3D geometry of the voids, roof thickness and the presence of overloads. The models produced by the discriminant analysis and the logistic regression approaches were independently and quantitatively evaluated through the construction of ROC curves. Although the two models are characterized by different input data treatment, as the logistic regression allows using continuous and discrete variables or any combination of both types even if they do not have a normal distribution, the ROC curves highlighted, for both methods, a satisfactory congruence between the model results and the observed data. Consequently, from the model performance analysis, the two models were evaluated comparable in terms of result reliability. The methodology should be considered as a general first-level analysis for near surface underground cavities aimed at identifying critical cave sectors that can be the focus of more detailed analyses, especially those associated with sensitive zone related to the presence of buildings, roads and frequent presence of human activity.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
