*3.2. Syn-Explosive SO2 Emissions during the Lava Fountaining Event*

The 18 May lava fountaining event, entirely captured by the SO2 camera (Figure 9), allowed us to explore to what extent SO2 cameras could resolve the degassing dynamics associated with a lava fountaining event. Lava fountains are of special concern at Etna since the volcanic ash they inject into the atmosphere is a potential threat to aviation and population living in the surroundings [73]. These events, while very well monitored and understood [34,36–42,50,74–77], are poorly characterized in terms of their associated gas emission rates and volumes. Our 18 May results (Figure 9), therefore, represent one of the first syn-explosive gas records on the volcano [28,30].

**Figure 9.** SO2 camera record during the 18 May lava fountain compared with the thermal signal recorded by the co-located INGV-OE TIR camera. The upper panels show thermal (**a**), absorbance (**b**), and visible (**c**) images of the summit area in specific time intervals. SO2 fluxes are shown in (**d**) as raw measurements (gray line) with the associated 10 second-window moving average (red line), while the thermal signal is shown in (**e**).

Onset of the lava fountain at 10:57 UTC, as constrained by co-acquired images of the INGV-OE thermal and visual cameras (Figure 9), was clearly detected as a visible SO2 flux increase up to ~260 kg/s (22,000 t/d), relative to a pre-eruptive level of ~12 kg/s (~1000 t/d). In the following hour, while activity escalated to peak at ~ 11.30–12.00 (see thermal records), a fluctuating and irregular SO2 flux trend is registered (Figure 9). Co-acquired thermal and visual images (see panels in Figure 9), clearly indicated that negative peaks in the SO2 flux time-series were systematically associated with the presence of ash. The latter severely impacts SO2 detection via UV cameras [21], particularly in near-vent measurements where plumes can be very ash-rich and, thus, optically opaque. Thermal and visual observations showed that ash caused high-frequency fluctuations (short-lived negative peaks) in the SO2 flux record particularly during the paroxysm climax (~11.30–12.00), but also prior to the lava fountain onset, e.g., after 09.00 when the visual camera captured the first ash emission with no thermal anomaly yet detected (Figure 9).

#### **4. Discussion**

Our results demonstrated the ability of the novel automatic processing routine to capture fluctuations in the SO2 regime, which responded to changes in Etna's activity style (Figure 7). The automatically processed SO2 data were consistent with those manually obtained (Figure 6). However, requiring no operator time and being obtained/delivered in nearly real-time, they represented a clear advantage for monitoring purposes.

Our 2016 UV camera-based dataset also provided novel insights into the relationship between rates/modes of SO2 release and eruptive/degassing styles. These latter were quite diverse in 2016 since they ranged from non-eruptive quiescent degassing to intense paroxysmal activity in May. Nicely, our automatically processed SO2 fluxes peaked during heightened activity (Figures 7 and 8) during the May 2016 eruptive sequence, and during the July-August degassing unrest that led to opening of the VOR incandescent vent on 7 August.

One important observation is the mild but significant SO2 flux increase in April 2016, as demonstrated by a clear change in the slope of the cumulative SO2 mass (see orange-coloured area in Figure 7a). The SO2 flux increase started more than one month before onset of the May eruptive sequence, which is a period when the seismic tremor was at average levels and no significant thermal anomaly was detected (Figure 7). We interpreted these escalating SO2 fluxes as reflecting the slow but systematic increase in the magma supply rate to the shallow (<3 km [78]) Etna's magma feeding system that triggered the May 2016 paroxysmal sequence [68]. However, while the relatively subtle changes in SO2 passive degassing in April 2016 may have represented a precursory sign for the imminent (May 2016) eruption, we still noticed that a similar SO2 increase was observed from July to September 2016 (orange-coloured sub-interval in Figure 7a). This did not culminate into an eruption yet (if not for the opening of the 7 August summit vent). Thus, the volcano's feedback to increased shallow magma emplacement (as indicated by increasing SO2 fluxes) may be different from time to time, perhaps in response to distinct stress regimes prevailing on the upper part of the edifice, and/or temporally varying feedbacks between magma ascent and rates of gravitational spreading of the mobile eastern flank [79,80].

We also characterized SO2 emissions associated with a lava fountaining event at a high spatial and temporal resolution [28,30]. Even though ash is a serious issue for ground-based SO2 remote sensing during explosive eruptions, we still noted that the entirety of 18 May eruptive episode was well marked by elevated SO2 fluxes well above background emissions in the pre-event and post-event phases. In fact, SO2 emissions manifestly dropped down at only 14:00 UTC, when declining thermal emissions consistently marked lava fountaining termination. From such, we found it useful to tentatively estimate the cumulative SO2 mass released by the 18 May lava-fountain, by integrating the signal over the eruption duration. We obtained a total SO2 mass released in the event of ~1700 tons, and an average flux of 158 kg/s (13,600 t/d) for a total duration of ~3 h. We caution this inferred mass corresponds to a lower range estimate, due to the presence of volcanic ash that severely depressed the measured SO2 signal during the eruption climax.
