Landslide detection in mountainous forest areas using polarimetry and interferometric coherence

Earth, Planets and Space

Table 4 Accuracies of landslide detection by the supervised classification for various conditions

#	Parameters used in machine learning	Dataset		Accuracy (κ)			Recall	Precision
#	Parameters used in machine learning	Training	Test	LIA < 30	LIA > 30	Total	Recall	Precision
1	PolSAR	Site F	Site F	0.643	0.376	0.474	0.378	0.918
2	PolSAR	H	H	0.663	0.577	0.606	0.512	0.946
3	PolSAR + DEM	F	F	0.671	0.411	0.508	0.419	0.894
4	PolSAR + DEM	H	H	0.660	0.561	0.594	0.494	0.958
5	PolSAR + InSAR	F	F	0.708	0.552	0.605	0.502	0.962
6	PolSAR + InSAR	H	H	0.716	0.634	0.661	0.577	0.941
7	PolSAR + InSAR + DEM	F	F	0.738	0.608	0.653	0.566	0.935
8	PolSAR + InSAR + DEM	H	H	0.733	0.656	0.682	0.616	0.913
9	PolSAR + InSAR + DEM	F	H	0.480	0.670	0.591	0.811	0.599
10	PolSAR + InSAR + DEM	H	F	0.240	0.279	0.266	0.188	0.958
11	Dual-PolSAR + InSAR + DEM	F	F	0.678	0.518	0.573	0.468	0.959
12	Dual-PolSAR + InSAR + DEM	H	H	0.565	0.483	0.510	0.405	0.964
13	Single-PolSAR + InSAR + DEM	F	F	0.659	0.435	0.513	0.399	0.990
14	Single-PolSAR + InSAR + DEM	H	H	0.520	0.189	0.305	0.226	0.915