Bubbles in Basalts: Measuring and Modelling Basaltic Degassing

Basaltic degassing is driven by the release of CO2, H2O, and SO2. Hitherto, the measurement of SO2 has been commonplace due to the lack of significant ambient atmospheric content. UV camera technology is currently among the best of techniques to measure this SO2 release from volcanoes given its high spatial and temporal resolutions. Given that an elevated CO2 flux can be an indication of magma movement at depth, a reliable method of measuring this species at similarly high temporal resolutions would be valuable. A technique making this possible is described here. This technique combines measurements of SO2 flux at Mt. Etna, using a UV camera, with CO2/SO2 gas ratios, which when multiplied together allow the creation of a contemporaneous CO2 flux datasets at a time resolution of ≈ 1 Hz. This also allowed the comparison of degassing with infrasonic and seismic datasets. This comparison was facilitated by the development of a new analysis technique to investigate correlative trends between noisy environmental datasets. The technique works by combining the continuous wavelet transform of two separate signals, with correlation of their respective coefficients at matching timescales using Spearman’s rank to produce a visually intuitive graphical plot. This revealed intriguing links between CO2 degassing and seismicity. Stromboli is renowned for its regular explosive activity. Through a permanent network of UV cameras at the summit area, a large number of explosive (120) and puffing events (80) were characterised in terms of their explosive and coda masses, termed the total strombolian event mass. Through this analysis, it was discovered that a large proportion of gas for each strombolian event is contained within the coda, ≈ 53 to 75% and for hornito events ≈ 70 to 84 %. The events were also characterised into six separate groups according to gas release pattern following the main eruptive burst. Through computational fluid dynamical simulations, for a range of appropriate strombolian eruption gas masses, the results demonstrated that there is potential for the release of daughter bubbles from the base of rising slugs. These daughter bubbles act to reduce the mass of slugs and can make slug flow unsustainable. Models were initiated over a suitable range of event masses, which demonstrated that ≈ 43 to 69% of the initial slug masses was released into the daughter bubble train. By applying the average mass loss rate, of ≈ 13.2 kg s-1, with total event masses, slugs are unlikely to be self-sustainable below depths of ≈ 740 m. A non-linear relationship between the dimensionless inverse viscosity term, N_f, and mass loss rate was also discovered. Also noted for its explosive activity is Mt. Etna. This activity includes hard to measure strombolian activity. During a rare period of activity at the Bocca Nuova summit crater ≈ 27 minutes of frequent but mild strombolian behaviour was captured using a UV camera. Given the unorthodox use of a rock background for the reflectance of light, calibration was tested and performed successfully on a basaltic background at the summit. Results show an SO2 mass range of ≈ 0.1 – 14 kg and a total gas mass range, on combination with measured Multi-GAS ratios, of ≈ 0.2 – 74 kg. Compared to events at Stromboli the activity was more frequent with an ≈ 4 s modal repose and with much lower overall masses. On investigating temporal trends between events it was observed that the largest mass events were followed by longer repose periods before another event occurred, smaller events occurring more frequently, a feature which is termed repose gap behaviour. Given the rapidity and mass of events it is reasonable that this activity was driven by gas slugs and that they were travelling in close proximity to each other. Using existing fluid dynamical models for the wake interaction length, an area behind a slug where a trailing slug can begin to interact with a leading one, it is possible that slugs are close enough to interact and coalesce. Indeed, this would provide a plausible mechanism for the repose gap. Building on the observations in the field at Mt. Etna a series of analogue laboratory experiments and computational fluid dynamics models were devised to investigate rapid strombolian activity, that driven by slugs. Behaviour of slugs acting independently of one another in a single-slug volcanic regime have been investigated thoroughly, however, the behaviour of slugs in a multi-slug volcanic regime have been neglected, largely a result of its comparative complexity. Laboratory experiments allowed the investigation of a series of average gas flow rates and hence slug lengths (i.e. overall gas volume fractions). The rates of expansion were also varied to simulate slug flow at depth and nearer to the magma surface. In particular, the process of coalescence was investigated. By comparing slug length at burst with repose time the repose gap feature was also identified. Given that values for rise speed, liquid, and conduit dimensions are known, this enabled the definition of the minimum period of repose as the wake length plus the length of the slug all divided by the rise speed of the base of the slug. This relation is validated successfully on the laboratory data and also on the collected Etna data. Additionally the laboratory analysis identified a previously unidentified feature whereby coalescence can occur between rising slugs, even when the trailing slug base is rising at a slower speed than the leading. This is likely related to the expansions of gas slugs. Computational fluid dynamics identified similar processes whereby the gap between identically massed slugs was maintained by slug expansion which acted to increase the speed of slugs above them. It is only when slugs are initiated within the wake length that coalescence occurs. Further relationships were discovered between slug rise speed and gas volume fraction, whereby the average rise speed of a slug increases with regime volume fraction, and burst slug length and volume fraction. Finally, building on the repose gap observations and developed relation, the observed relationships between slug length and gas rise speed with gas volume fraction are used to develop two separate models categorising the styles of volcanic activity which will be prevalent. The first, slug length model, is based upon repose time and slug lengths, with the second based upon overall volume fraction and repose time. The slug length model splits activity into: passive, puffing, strombolian explosive and strombolian rapid. This model performs well when applied to strombolian events, successfully differentiating between explosive and passive events. The volume fraction model applies fluid dynamical relationships for transitions to churn and annular flow, in addition to the already defined strombolian relationships, assumed here to play some part in defining the transition to hawaiian lava fountaining activity. This allows the definition of critical volume fractions above which large gas slugs or pockets can burst with increasing frequency until full lava fountaining behaviour is realised. Both models allow eruption parameters to be estimated via the delay time between events or vice versa. On comparison of known correlative relationships between gas emissions and seismicity a log relationship is discovered when all events are normalised to comparable parameters, suggesting that seismicity could also be incorporated into such a model in the future. In particular, the latter volume fraction model is the first step in developing a unifying theory of basaltic degassing based on a varying delay between events and could be particularly useful when used in tandem with real time gas emission data for eruption forecasting and understanding the fluid dynamical flow processes occurring in the sub-surface.