Supervised deep learning models have become a popular choice for seismic phase arrival detection. However, they don’t always perform well on out-of-distribution data and require large training sets to aid generalization and prevent overfitting. This can present issues when using these models in new monitoring settings. In this work, we develop a deep learning model for automating phase arrival detection at Nabro volcano using a limited amount of training data (2498 event waveforms recorded over 35 days) through a process known as transfer learning. We use the feature extraction layers of an existing, extensively-trained seismic phase picking model to form the base of a new all-convolutional model, which we call U-GPD. We demonstrate that transfer learning reduces overfitting and model error relative to training the same model from scratch, particularly for small training sets (e.g., 500 waveforms). The new U-GPD model achieves greater classification accuracy and smaller arrival time residuals than off-the-shelf applications of two existing, extensively-trained baseline models for a test set of 800 event and noise waveforms from Nabro volcano. When applied to 14 months of continuous Nabro data, the new U-GPD model detects 31,387 events with at least four P-wave arrivals and one S-wave arrival, which is more than the original base model (26,808 events) and our existing manual catalogue (2,926 events), with smaller location errors. The new model is also more efficient when applied as a sliding window, processing 14 months of data from 7 stations in less than 4 hours on a single GPU.

Lava­­-water interactions (LWIs) are rarely considered in lava flow hazard assessments or emergency planning scenarios, though they can generate a range of secondary hazards, including tephra blasts, rootless eruptions, disruption to water supplies, and flooding. These hazards may endanger life, damage property, and hinder evacuation or rescue efforts, so identifying the signs of LWI in the products of past eruptions may help emergency planners identify potential hazards for future eruptions. The physical products of LWI, such as abundant hyaloclastite, high proportions of fine ash, lava pillows, and irregular columnar jointing, have long been recognized in the field. However, remote sensing offers the opportunity to assess whole lava fields relatively quickly and cheaply, and allows investigation of inaccessible lava fields and planetary volcanism. In addition, the large-scale view can reveal features that are not immediately visible in the field, and tools like LiDAR can be used to strip away vegetation to show hidden morphology and structure. We present features indicative of LWI that can be identified by remote sensing techniques and discuss what they can and can’t tell us about LWI in past eruptions. We illustrate these with data from the well-documented 1783-84 Laki fissure eruption, supplemented with other case studies from Iceland, Hawai’i and the Pacific NW. In particular, the size, type and spacing of rootless cones can tell us about the availability of water and intensity of rootless eruptions. When examined in conjunction with lava flow morphology and local topography, we can learn about the local lava flux and the likely water sources and pre-eruptive landscape. In the absence of rootless cones, dendritic textures on the lava flow surface may indicate passive LWI. These textures are found across the Laki lava field, commonly in areas where the lava encountered rivers or floods, and match those at other lava basaltic flows known to have interacted with water, including the 2018 eruption of Kilauea. Together, these features are useful indicators for identifying and interpreting past LWIs, both as a complement to field observations and when field studies are not feasible.