The rebirth and evolution of Bezymianny volcano, Kamchatka after the 1956 sector collapse

Volcanic sector collapses are major instability events affecting active volcanoes, ranging from slow to catastrophic failures and mobilizing volumes up to 10^11 m³. Such collapses generate devastating hazards (directed blasts, pyroclastic flows, lahars, tsunamis) and modify subsurface conditions by unloading, thereby changing magma reservoir pressure, geometry, and the pathways of magma ascent. Following collapses, many volcanoes regrow cones and domes, but continuous documentation of the post-collapse regrowth, vent migration, and subsequent vent focusing has been scarce. Bezymianny volcano (Kamchatka), an active andesitic system that suffered a major sector collapse in 1956 (~0.7 km³ removed), offers a unique opportunity to chronicle edifice rebirth. The study aims to reconstruct the seven-decade morphological evolution of Bezymianny’s edifice using high-resolution photogrammetry and to test, via numerical models, how stress changes due to collapse and subsequent regrowth can explain observed shifts and centralization of eruptive vents. Understanding these processes informs broader questions of volcano evolution, hazard recurrence, and the mechanics of magma ascent in changing topography.

Previous work documents sector collapses and post-collapse regrowth at volcanoes such as Mount St. Helens, Soufrière Hills, Santa María/Santiaguito, Ritter Island, Shiveluch, and Bezymianny, as well as prehistoric cases (Socompa, Parinacota, Vesuvius, Tungurahua). Modeling and observations suggest that collapse-induced unloading can deflect and bifurcate magma pathways, shifting eruption sites into collapse embayments. However, despite geological reconstructions and intermittent monitoring elsewhere, detailed decadal observations that capture shifting and progressive focusing of vents during regrowth have been lacking. Prior studies at Bezymianny show changes in eruptive style and magma composition (progressively more mafic since 1956), and at analogous domes (e.g., Mount St. Helens, Shiveluch, Santiaguito) document high initial extrusion rates that later decline. This study builds on these insights by providing a continuous, high-resolution morphological record and linking it to mechanical modeling of magma pathways under changing loads.

- Data sets and acquisition: Pre-collapse topography was derived from 1949 aerial photography. Post-collapse surfaces were reconstructed using near-nadir aerial photogrammetry from campaigns in 1967, 1968, 1976, 1977, 1982, 1994, 2006, and 2013. Additional tri-stereo Pléiades imagery (1 m) acquired 09.09.2017 and UAV (DJI Mavic Pro) images (July 2017) complemented the series. 2002, 2005, 2009, and 2010 overflights supported visual interpretation of lava flows. - Aerial imaging and scanning: Analog cameras included AFA 41-10 (1967, 1968, 1976, 1977; focal length 99.086 mm), TAFA 10 (1982, 1994; 99.120 mm), and AFA TE-140 (2006, 2013; 139.536 mm), each with 18×18 cm frames. Flight altitude was 1500–2500 m. Film negatives were scanned at 2400 dpi (~0.01 mm/px), yielding ground resolutions ~10–20 cm/px depending on scale. - Photogrammetric processing: Interior/relative orientations were performed in Erdas Imagine 2015 v15 and Photomod 5 with ~25 tie points (RMSE ~0.1 px). Twelve GCPs (theodolite-derived, 1977 fieldwork) anchored the 1977 stereo model to the USSR State Geodetic Network (GCP RMSE ≤0.06 m). Because some benchmarks were buried, six stable topographic features were used as GCPs to georeference other years to the 1977 model (triangulation RMSE 0.4–1.9 m). Terrain extraction used Erdas eATE (normalized cross-correlation, 11×11 px window; correlation range 0.2–0.7). Point-cloud noise filtering used CloudCompare v2.9.1 (spherical radius = 0.75 neighbors), yielding ~300,000 points per ~5 km² (~0.06 pts/m²). Manual stereo compilation in Photomod DTM filled gaps caused by fumaroles/shadows; final point-cloud resolutions ranged from ~2 to 30 m. - Satellite and UAV processing: Pléiades tri-stereo images were processed in Erdas Imagine with 45 tie points (relative orientation RMSE = 0.1 px) and RPC-based exterior orientation; DEM extracted via eATE and filtered, producing ~10 million points over ~60 km² (~0.16 pts/m²). UAV images were processed in Agisoft Metashape v1.6.4 (high-quality alignment; dense cloud generation; DEM building, WGS84). The Pléiades and UAV point clouds were merged in CloudCompare (alignment RMSE ~0.8 m) and subsampled to 2 m. - Alignment and referencing: All point clouds were aligned in WGS84/UTM 57N using common points on the 1956 amphitheater rim. The 2013 cloud was aligned to 2017 (RMSE 1.3 m), and earlier clouds were sequentially aligned to the next younger (RMSE 1.3–1.7 m). - Volumes and rates: DEM differencing (CloudCompare 2.5D volume tool) provided net added/removed volumes within mapped dome/flow extents. Growth rates were computed as positive volume change divided by inter-survey time. The 1956–1967 dome volume used prior published value; 1967–2017 volumes were computed within eruption-affected areas. For 2017, the base surface was the 2013 DEM aligned to 2017. - Acquisition date estimation (1967, 1968): Sun angles derived from shadow measurements and NREL’s Solar Position Algorithm constrained dates to 24 Oct 1967 and 9 Sept 1968 (±1 day), consistent with partial snow cover. - Error estimation: Volume uncertainty sources included triangulation RMSE (Z-component dominant), point-cloud resolution, and alignment RMSE. TRMSE-derived volume uncertainties ranged from ~0.5×10^6 m³ (2017) to ~3.2×10^6 m³ (1967), i.e., ~0.1–1.4% of dome volume. Resolution-induced uncertainties were 3,900–7,900 m³ (>0.001% of dome volume). 2017 DEM merger alignment contributed ~344,000 m³ (0.06%); 2013–2017 alignment contributed ~2.9×10^6 m³ (0.5%); total 2017 volume error ~0.7%. Rate errors were computed via root-sum-squared adjacent volume errors divided by time; relative errors ranged 2.1–31.2% (largest for 1967–1968 due to short interval/small increment). - Morphological mapping: High-resolution stereoscopic interpretation (Erdas Imagine Stereo Analyst), DEM hillshades, and GIS (Surfer 10, QGIS 3.2.3, Inkscape) were used to map endogenous domes, exogenous shear lobes, lava flows, lava plugs, craters, talus/pyroclastics, and collapse scars. Vents were identified per period-specific criteria (endogenous dome centers, thickest shear-lobe parts, lava vent areas/plugs, summit crater centers). Vent migration was analyzed along NE–SW and NW–SE profiles to compute maximum inter-vent distances (Dmax). - Magma pathway simulations: 2D boundary-element mixed-mode crack models in plane strain simulated magma pathways under three stress scenarios: pre-collapse loading (triangular force, EVL 0.5/0.6/0.7), post-collapse unloading (triangular force), and regrowth reloading (trapezoidal dome). Topographic forces applied on a reference horizontal surface at ~1900 m asl. Nineteen trajectories started 2 km beneath the reference surface across a 3-km-wide window centered at x=0. Magma/rock parameters: magma density 2400 kg/m³; compressibility 10 GPa; intrusion cross-sectional area 4×10^-4–9×10^-4 km²; host-rock density 2500 kg/m³; rigidity 20 GPa; Poisson’s ratio 0.25. The algorithm advanced cracks along directions maximizing elastic plus gravitational energy release. Viscosity and heterogeneity effects were neglected.

- Bezymianny’s post-1956 regrowth followed four stages: (I) 1956–1967 endogenous growth of two lava domes from vents ~400 m apart; (II) 1967–1976 exogenous extrusive growth via multiple viscous shear lobes from migrating vents; (III) 1977–2006 exogenous extrusive–effusive activity with lava flows, shear lobe remnants, and lava plugs from multiple craters; (IV) 2006–present stratocone formation with a centralized summit crater and interbedded lava and pyroclastic deposits. - Vent migration and focusing: Early activity involved distributed vents within the collapse amphitheater (vents migrating within ~200 m by two decades post-collapse), progressively focusing toward a single, stable summit-crater vent position established by 2006 and maintained through 2017. - Morphological evolution: The amphitheater was progressively infilled; by 2013 the edifice had become a symmetric stratocone. The cone height reached ~3020 m in 2017 (~820 m relative height), ~90 m below the 1949 pre-collapse summit. Pyroclastic deposits thickened by up to ~140 m (western amphitheater) by 2006 and increased by an additional ~56 m after 2013; lava flows from 2009–2012 covered S–SE slopes; 2016–2017 flows filled the summit crater and descended W–SW. - Volumes and rates: Total rebuilt central-cone volume (1956–2017) is ~0.591 km³. Average long-term growth rate is ~26,400 m³/day. Period-specific averages: 1956–1967 ~56,600 m³/day; 1967–1968 ~43,600 m³/day; 1977 onward ~15,500–17,000 m³/day, with relatively constant rates during lava-flow dominated phases. - Numerical modeling: Collapse-induced unloading deflects magma pathways into the collapse embayment, shifting post-collapse vents several hundred meters from the pre-collapse conduit. As the new dome loads the edifice, magma pathways increasingly converge toward the center of the new topography, providing a mechanical explanation for observed vent focusing. A corridor of relatively lower compressive stress beneath the growing cone facilitates ascent and centralization. - Magma composition and style: Whole-rock silica content decreased smoothly from ~60.4% (1956) to ~56.8% (2012), consistent with a transition toward lower-viscosity lavas and the onset of lava flows in 1977. The trend may reflect increasing dominance of a deeper (~18 km) reservoir over time, compared to an initial shallow (~7 km) source. - Comparative context: Bezymianny’s long-term rate (~26,400 m³/day) is comparable to Mount St. Helens 1980–1986 (~30,000 m³/day) and Santa María/Santiaguito 1922–2000 (~38,000 m³/day), but lower than Mount St. Helens 2004–2008 and Shiveluch 2001–2012. Unlike those, Bezymianny shows clear stratocone formation linked to persistent effusion and edifice stabilization by lava flows. - Forecast: At the measured average growth rate, the edifice was projected to regain its pre-collapse size within approximately 15 years.

The study addresses how edifice-scale mass redistribution after a catastrophic sector collapse governs subsequent vent locations and the morphological trajectory from a dome to a stratocone. Decadal photogrammetry demonstrates that Bezymianny’s post-collapse activity initiated with distributed, endogenous domes in the collapse amphitheater, transitioned to extrusive shear-lobe growth, then to mixed extrusive–effusive activity, and finally to sustained stratocone building with a centralized vent. Numerical models attribute the initial vent shift and subsequent focusing to stress changes caused by unloading (collapse) and reloading (regrowth): unloading reduces horizontal compression within the collapse embayment, favoring magma ascent there; progressive reloading by the growing cone generates a stress field that converges magma pathways toward the edifice center, consistent with the observed stabilization of the summit crater since 2006. The mechanical picture aligns with petrological trends indicating decreasing silica content and inferred deeper magma contributions, which plausibly facilitated lava-flow emplacement. The persistence of lava flows appears to armor and stabilize slopes, mitigating partial collapses and enabling centralized vent establishment, in contrast to systems dominated by thick, unstable lobes and frequent collapses. These findings generalize to other regrowing andesitic systems, explaining observed post-collapse vent shifts and, where effusion is sustained, the potential for relatively rapid stratocone re-formation.

This work provides the first high-resolution, seven-decade chronicle of edifice rebirth after a major sector collapse, documenting a dome-to-cone transition and progressive vent centralization at Bezymianny. By integrating detailed photogrammetry with mechanical modeling, it demonstrates that collapse-induced unloading shifts vent locations into the amphitheater and that subsequent reloading by regrowth focuses magma pathways toward a centralized summit crater. The edifice has nearly regained its pre-collapse height and volume at a long-term rate of ~26,400 m³/day, implying a relatively rapid recovery of cone morphology compared to many analogs. The results have direct implications for hazard assessment at Bezymianny and analogous volcanoes, highlighting phases of edifice stabilization by lava flows and the potential for renewed instability as the cone steepens. Future work should couple mechanical models with rheology (viscosity, degassing), reservoir dynamics, and 3D heterogeneity, and extend multi-sensor monitoring to better resolve transient changes in extrusion rates, vent migration, and slope stability.

- Numerical models neglect magma viscosity, volatile exsolution, and rock heterogeneities, and apply 2D plane-strain approximations, which may simplify true 3D magma pathway behavior. - Mechanical coupling with reservoirs is not explicitly modeled; stress effects may vary with reservoir geometry and depth. - Photogrammetric uncertainties arise from triangulation errors, film distortion, scanner artifacts, fumarolic obscuration (requiring manual compilation), and inter-epoch alignment; although quantified, they contribute up to ~1.4% of volume estimates and up to ~31% relative error in short-interval growth rates. - Exact dates for some aerial surveys (1967, 1968) were inferred from solar geometry, introducing small temporal uncertainty (±1 day). - DEM resolution varies across epochs (2–30 m), potentially limiting detection of fine-scale features and small-volume changes.