Skip to main content

Table 2 Deep learning analysis of deformation at Turkish Volcanoes based on the method of Anantrasirichai et al. (2018, 2019a). The Convolutional Neural Network outputs two classes: D + S + T contains a mixture of deformation and atmospheric artefacts while S + T contains only atmospheric artefacts. The images in the D + S + T class are then checked by an expert and assigned to the true or false positive classes. In the case of Turkey, all images were either S + T or false positives. All volcano names are given using locally approved names and spelling with the GVP equivalents given in Table 1

From: Baseline monitoring of volcanic regions with little recent activity: application of Sentinel-1 InSAR to Turkish volcanoes

Volcano Number of Images CNN Outputs Expert Checks
Deformation (D + S + T) Atmosphere (S + T) True Positive False Positive
Kula Volcanic Field 796 0 796 0 0
Karapınar Volcanic Field 340 0 340 0 0
Hasandağ 670 2 668 0 2
Göllüdağ 693 0 693 0 0
Acıgöl 696 0 696 0 0
Erciyes Dağı 339 1 338 0 1
Karacadağ 469 1 468 0 1
Nemrut Dağı 323 0 323 0 0
Tendürek Dağı 349 0 349 0 0
Ağrı 271 0 271 0 0
Total 4946 4 4946 0 4