Sinkhole susceptibility, Lazio Region, central Italy

Figures & data

Figure 1. Map of the conditioning factors that play an important role in the process of sinkhole formation: lithology type, depth of the substratum, presence of mineral and thermo-mineral springs, soil gas emissions, active faults, location of the earthquake epicentres and their magnitude, and the presence of karst dolines.

Source: Data have been provided by the Area Difesa Suolo e Bonifiche, Regione Lazio.

Figure 2. Workflow of the procedures and calculations applied to obtain the base grid layers (2 × 2 km) used within the LR.

Table 1. Classification of the observations in the model.

		Estimated
		Sinkhole absence/presence
Observed		0	1	Corrected percentage
Sinkhole absence/presence	0	868	150	85.3
	1	118	170	59.0
Total percentage				79.5

Table 2. Variables in the regression equation.

Variables	B	Standard error	Significance	Exp(B)
Variables in the equation
Geology	0.392	0.059	0.000	1.479
Substratum	0.421	0.106	0.000	1.524
Mineral springs	0.535	0.152	0.000	1.707
Thermo-mineral springs	0.831	0.114	0.000	2.296
Gas emanations	−0.575	0.179	0.001	0.562
Dolines	0.527	0.206	0.010	1.694
Earthquakes M > 2	0.807	0.124	0.000	2.241
Constant	−7.413	0.689	0.000	0.001
Variables not in the equation
Springs	0.097	−	0.756	−
Earthquakes M < 2	0.492	−	0.483	−
Faults	0.012	−	0.914	−
Water table elevation	0.932	−	0.334	−

Figure 3. Receiver operating characteristics (ROC) curve that represents the goodness of the regression model constructed by using test data (about 70% of the total data).