High-Frequency Tide Gauge Optimization for Detecting Short-Period Volcanic Tsunamis: A Case Study of the 2018 Anak Krakatau Event

Tsunamis generated by volcanic mass-failure processes pose a distinct monitoring challenge because their energy is concentrated at shorter periods than typical tectonic events. Conventional tide gauge sampling rates are often too coarse to capture these signals. This study quantifies how sampling interval controls waveform fidelity and automatic detection performance using the 2018 Anak Krakatau tsunami as a benchmark. Clean model tsunami records from three landslide-volume scenarios and three representative stations spanning contrasting bathymetric settings were resampled to practical intervals (1–60 s) and evaluated using time-domain errors, spectral distortion, and detection latency with three commonly used algorithms. To emulate realistic coastal conditions, mixed records were constructed by combining tsunami signals with short-period harbor and vessel-related disturbances at multiple intensity scales, and real-time noise handling was tested using sliding-window filters. Results show that coarse sampling (15–60 s) produces severe distortion consistent with undersampling and aliasing, while a 3–5 s interval provides the best overall compromise between waveform fidelity and detection speed. A lightweight Moving-Average filter (optimal window 32 s under low SNR) improves signal clarity while preserving dominant tsunami structure. These findings support site-calibrated high-frequency monitoring with station-tuned filtering for short-period nonseismic tsunamis.