We attempt to construct a timeline of The Hunga Tonga – Hunga Ha’apai eruption on 15 January 2022 through analyses of seismic, barometric, infrasonic, lightning, and satellite data. Satellite imagery at 04:00 UTC showed no ash in the air, but by 04:10 UTC, a plume had risen to 18 km. Over the next 20 minutes, the plume rose to 58 km. USGS determined that Mw5.8 volcanic earthquake of unknown mechanism had occurred at 04:14:45. Gravity waves were observed in satellite imagery, and barometric and infrasound stations around the world recorded ultra-low frequency pressure variations of more than 100 Pa, inducing ground-coupled airwaves around the globe, and meteo-tsunamis in the Caribbean Sea and Mediterranean Sea. Tsunami waves were recorded in coastal areas around the Pacific Ocean. From record sections, we determined speeds of 3.9 km/s and 299 m/s for the initial seismic and infrasound signals respectively, converging to an eruption onset time of ~0402 UTC ± 1 minute. The global pressure pulse has a speed of ~314 ± 3 m/s, consistent with theoretical models for Lamb waves (Bretherton, 1969), suggesting an origin time of ~0415 ± 2 minutes (consistent with the Mw5.8 volcanic earthquake, and sharp increases in lightning flash rates), and peaking around ~0429 ± 2 minutes. We suggest that Surtseyan volcanic activity commenced at ~04:02, building to a sub-Plinian eruption ~7 minutes later, before a phreato-Plinian eruption commenced at ~04:14. The peak Lamb wave amplitude at the closest station (757 km from HTHH) was 780 Pa. Assuming geometrical spreading like 1/√r (where r is the source-receiver distance), we estimate a lower bound of ~23 kPa for reduced pressure by extrapolation back to 1 km. Adding a near field term that decays like 1/r, we estimate an upper bound of 170 kPa for reduced pressure. Comparison of these values with those from other eruptions (McNutt et al. in this session) suggests the 15 January HTHH eruption was in the VEI 5-6 range.

Seismic activity during the eruption of Soufriere Hills volcano comprised various transient signals, which were classified visually by the Montserrat Volcano Observatory (MVO), considering waveforms recorded at several stations. For 217,290 transients detected on the MVO digital seismic network between 1996/10/21 and 2008/10/16, five main classes have been identified: rockfall (ROC: 58%), hybrid (HYB: 19%), long-period (LPE: 11%), lp-rockfall (LP-ROC: 5.8%), and volcano-tectonic (VT: 3.1%). Temporal trends in the rate and energy release of these different transients (in addition to swarms and tremor) were key to short-term forecasting of eruptive activity. However, visual classification is highly subjective and non-repeatable, and the inconsistency of the catalog is a barrier to research. In a pilot study, we automatically removed waveforms with dropouts, and manually verified transient classifications until we had approximately 100 transients of each class (total 522). We found ~21% of these transients were incorrectly classified at MVO. Our re-labelled dataset was then used as a starting point for supervised learning, using code from http://github.com/malfante/AAA. This code was used by Malfante et al. (2008) to classify 109,609 transients at Ubinas volcano with a 93.5% accuracy. They transformed each waveform into a set of 102 features: 34 features for each of three domains (time, spectral, cepstral). We added 6 frequency features of our own, including band ratios, peak frequency, median frequency, bandwidth, and frequency change. The resulting 108-point vectors of features were then used for modeling. The dataset is randomly divided 50 times into training and testing datasets, to produce a robust model. One model is produced per channel. We use the Random Forest Classifier algorithm from the scikit-learn library. For each waveform, a probability is computed for each class. Initial results are promising. Separate models for 3 channels yield accuracies of 76-80%. If the LP-ROC class is omitted (following Langer et al, 2006), accuracy rises to 82-85%. If only VT and LP classes are considered, accuracy is 96-99%. We intend to expand our labelled dataset to 1000 events, add new features, build models for each channel, and reclassify the catalog of 217,290 transients by a weighted average of probabilities.