Brenna A. Halverson | Geology Reports

USGS2026

Lava flow emplacement dynamics during the Mauna Loa 2022 eruption

On November 27, 2022, Mauna Loa erupted for the first time in 38 years, threatening a major highway on the Island of Hawaiʻi. The opening phases of the eruption were focused at the volcano’s summit, before magma propagated into the upper Northeast Rift Zone to erupt at four fissure segments that localized at fissure 3 by December 2. This fissure produced a braided ʻaʻā flow that traversed the steep north flank at >1 km/hr, before slowing as slopes shallowed. Effusion rates peaked in the early hours of the eruption and then steadied at around 100 m3/s at fissure 3. Lava effusion became variable and dropped abruptly on December 7. The eruption ended on December 10 with a final flow length of 18 km. We tracked lava effusion rates, flow advance, channel dynamics, and 3-D flow field development with field and remote sensing observations throughout this eruption to inform lava flow forecasts and investigate ʻaʻā flow dynamics. Flows were mapped using optical and infrared imagery, uncrewed aircraft systems, and airborne single-pass InSAR. Video- and time-lapse cameras were deployed to record lava fountaining and channelized flow at frequent intervals, supplemented by field measurements. Analyses of these data built a detailed record of effusion rates and associated flow dynamics. We combine these high frequency observations of the proximal flow field with broader thermal and 3-D mapping of the evolution in the medial and distal flow field to examine ʻaʻā flow dynamics from source to toe. Within the steep, medial section of the flow field, channel blockages caused back-ups and overflows as the eruption progressed. Disrupted lava supply to the distal flow field and low slopes resulted in stalling, widening, and thickening at the toe. Overall, 142 million m3 of lava covered 36 km2 over 13 days. Our results inform the controls on development and evolution of lava flow fields, with implications for forecasting primary flow advance and secondary flow field growth.

Hawaii↗

USGS2025

Case study of deep learning image segmentation for the purposes of rapid 2D petrographic analysis in volcanic rocks

Automation using deep learning methods is a useful alternative to manual methods of petrographic segmentation, but often requires user familiarity with coding and/or algorithms. We examine the Dragonfly TM program's deep learning tools for application by users with a variety of skill levels as a method for petrographic image segmentation. An image processing methodology, bimodal image stacking, was created for low-input-data, high-efficacy training of models which can then be applied to varied samples. Using backscatter electron images we show that the resulting model segmentations agree with manual segmentation total and modal crystallinity values within 5%, and calculated plagioclase crystal size distribution (CSD) values within 2σ, despite limitations in discriminating mafic phases. Model creation and training takes <24 hours, 1–3 hours of which are supervised, and the resultant model can then be applied to new uncharacterized samples in <15 minutes per image. This allows for non-experts to create and utilize deep learning models to segment images of variable brightness and texture, at low user-time cost and resulting in size and shape data which are within uncertainty of manual segmentation. While some limitations are noted (for example, sieve-textured phases may need manual correction, and different minerals with similar BSE intensity may not be resolved as separate phases), this methodology can be utilized for general application of models to wide ranges of volcanic crystalline and bubble textures, and to create a library of models for rapid petrological analysis during volcanic eruptions.

Alaska↗