KVP: a multiscale kurtosis approach for seismic phase picking [Dataset]

This research focuses on improving automatic event detection and seismic phase picking in large and complex data sets generated by modern instruments, particularly Distributed Acoustic Sensing (DAS) and Ocean Bottom Seismometers (OBS). The authors address a major limitation of these systems: data quality often varies strongly along the fibre or between stations because of environmental noise, poor coupling, and low signal-to-noise ratios, which makes accurate picking especially difficult for emergent arrivals. To overcome this problem, they developed a new multiband kurtosis-based algorithm called Kurtosis-Value-Picker (KVP), designed to enhance the detection of both impulsive and emergent seismic signals. The method computes characteristic functions in several frequency bands using sliding windows and identifies triggers from localized jumps in kurtosis, which makes it more sensitive to subtle arrivals than traditional approaches. The algorithm was validated with earthquake records from land and submarine DAS cables as well as OBS data, and its performance was compared with FilterPicker and PhaseNet. The results show that KVP provides accurate and robust picks across diverse recording conditions, preserves useful spectral information, and is well suited for the analysis of complex seismic data sets