Shallow permeability structure and gas flow in hydrothermally altered soils at the Rotokawa Geothermal Field, New...

Hydrothermal alteration exerts strong control on shallow permeability and degassing dynamics in geothermal systems. Here, we investigate how soil alteration influences near-surface gas flow by combining in situ petrophysical measurements with horizontal and vertical subsurface gas-concentration profiles at the Rotokawa geothermal field, New Zealand. These data are compared against unoccupied aerial system (UAS) thermal surveys of collapse structures across the field. Soil permeability at Rotokawa ranges from 8.7 × 10^-14 to > 6.5 × 10^-13 m², highlighting strong meter- to decimeter-scale heterogeneity in shallow soil properties. Pumice-rich horizons are the main conduits for CO₂ and CH₄ flow, whereas clay-rich horizons locally act as semiconfining layers that promote CO₂ accumulation and lateral diversion (leading to concentrations of > 4 × 10⁴ ppm). Since CO₂ does not condense under near-surface conditions, the clay layers promote horizontal flow along permeable beds until gas encounters a high-permeability escape route or collapse-related discontinuity. Collapse structures locally disrupt and reorganize flow by acting as conduits or sinks that focus, capture, or redistribute gases near their margins. The gas profiles reveal patterns consistent with a shallow gas-steam decoupling zone in which steam condensation may contribute to sealing processes, as previously observed in steam-heated geothermal systems worldwide. These coupled effects of soil type and structural and alteration controls explain the spatial heterogeneity of surficial degassing at Rotokawa and provide a framework for interpreting evolving degassing patterns in similar steam-heated geothermal systems. In the context of the reported decrease in diffuse CO₂ emissions at Rotokawa, progressive shallow sealing and gas refocusing may have contributed to apparent changes in emission patterns, alongside reservoir processes, recharge variability, environmental effects, and methodological uncertainty.

The online version contains supplementary material available at 10.1007/s00445-026-02002-7.