Automated Detection of Recent Mud Extrusions Using UAV Imagery and Deep Learning: A Comparative Analysis of Traditional and CNN-Based Approaches

Historical earthquake records extend knowledge of past earthquakes but are uneven and incomplete. This study compares historical and early modern earthquake locations worldwide to test how well historical data reflect real earthquake patterns. We show that historical records capture large-scale tectonic features but miss details at smaller scales. This means historical earthquakes are useful for broad hazard studies, but not for detailed local assessments.

Instruments that measure extremely small ground motions are vital for monitoring earthquakes, volcanoes, and slow Earth deformation. This study presents a virtual modelling approach that predicts sensor performance before construction. By accounting for physical limits, electronic effects, and uncertainty, the method clarifies what truly limits sensitivity and how designs can be improved, supporting the development of more accurate and robust tools for Earth science.

We use uncrewed-aerial-system-derived optical and infrared data, mineralogical and geochemical analyses of rock samples, and surface degassing measurements to analyze degassing and hydrothermal alteration at the fumaroles of the La Fossa cone, Vulcano island, Italy. We give a detailed view of associated structures and dynamics, such as local alteration gradients, diffuse active units that significantly contribute to the total activity, or effects of permeability reduction and surface sealing. 

Santa Barbara and Aragona are affected by mud volcanism with episodic hazardous paroxysm events. Two potentially hazardous paroxysm exposed surfaces of 0.12 and 0.20 km² were elaborated with DSMs and with historical information on the paroxysms that occurred in the past. This paper, in the end, could be a useful tool for civil protection authorities in order to take appropriate risk mitigation measurements for exposed people and for monitoring activities.