The significance of intra-mountain valleys to infrastructure and human settlements and the need to mitigate the geo-hazard affecting these assets are fundamental to the economy of Italian alpine regions. Therefore, there is a real need to recognize and assess possible geohazards affecting them. This study proposes the use of GIS-based analyses to construct a sinkhole susceptibility model based on conditioning factors such as land use, geomorphology, thickness of shallow deposits, distance to drainage network and distance to faults. Thirtytwo models, applied to a test site (Enemonzo municipality, NE Italy), were produced using a method based on the Likelihood Ratio (λ) function, nine with only one variable and 23 applying different combinations. The sinkhole susceptibility model with the best forecast performance, with an Area Under the Prediction Rate Curve (AUPRC) of 0.88, was that combining the following parameters: Nearest Sinkhole Distance (NSD), land use and thickness of the surficial deposits. The introduction of NSD as a continuous variable in the computation represents an important upgrade in the prediction capability of the model. Additionally, the model was refined using a kernel density estimation that produced a significant improvement
Integration of multi-criteria and nearest neighbour analysis with kernel density functions for improving sinkhole susceptibility models: the case study of Enemonzo (NE Italy)
CALLIGARIS, CHIARA;DEVOTO, STEFANO;ZINI, Luca;
2017-01-01
Abstract
The significance of intra-mountain valleys to infrastructure and human settlements and the need to mitigate the geo-hazard affecting these assets are fundamental to the economy of Italian alpine regions. Therefore, there is a real need to recognize and assess possible geohazards affecting them. This study proposes the use of GIS-based analyses to construct a sinkhole susceptibility model based on conditioning factors such as land use, geomorphology, thickness of shallow deposits, distance to drainage network and distance to faults. Thirtytwo models, applied to a test site (Enemonzo municipality, NE Italy), were produced using a method based on the Likelihood Ratio (λ) function, nine with only one variable and 23 applying different combinations. The sinkhole susceptibility model with the best forecast performance, with an Area Under the Prediction Rate Curve (AUPRC) of 0.88, was that combining the following parameters: Nearest Sinkhole Distance (NSD), land use and thickness of the surficial deposits. The introduction of NSD as a continuous variable in the computation represents an important upgrade in the prediction capability of the model. Additionally, the model was refined using a kernel density estimation that produced a significant improvementFile | Dimensione | Formato | |
---|---|---|---|
Calligaris et al., IJS - Enemonzo.pdf
accesso aperto
Tipologia:
Documento in Versione Editoriale
Licenza:
Creative commons
Dimensione
5.34 MB
Formato
Adobe PDF
|
5.34 MB | Adobe PDF | Visualizza/Apri |
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.