Supplementary text
Several external factors need to be considered for the stringent evaluation of semi-continuous
scan DOAS measurements. The intensity of scattered UV light significantly changes throughout
the day, and affect the signal-to-noise ratio of the recorded spectra. Low light intensity causes
low signal-to-noise ratios and consequently increases the spectroscopical error of the
measurements. Thus we set the limits of the daily data acquisition interval to 1 hour after sunrise
and 1 hour before sunset, in order to avoid low light conditions and to assure a more constant
data quality (Fig.S1). Further external parameters that need to be considered are those used to
calculate the fluxes such as wind speed and wind direction at plume height. Various statistical
tests were conducted on wind speed and wind direction time series obtained from the GDAS1
model (Fig.S2). Systematic variations similar to those found in SO2-flux time series are also
observed in wind speed and wind direction time series. Especially the intra-day wind speed
variations coincide with changes in strength of the gas emissions. However, even if we remove
the wind speed from the calculations and use a constant wind speed of 1 meter per second to
calculate fluxes, the systematic degassing variations still persist (Fig.S3). Thus the daily wind
speed variations cannot alone account for the observed variations of the sulfur dioxide emissions.
Of course we cannot generally rule out the influence of the wind speed on volcanic activity, as
strong winds generate windward surface loads, and may trigger rockfalls and landslides. And
indeed comparatively high SO2-fluxes were measured during stormy conditions. The influence of
systematically changing wind directions on the flux measurements mainly depends on the spatial
coverage of the respective scan DOAS array, because the NOVAC-software (Galle et al. 2010)
automatically estimates the plume coverage of each scan. This estimate together with the wind
direction from the GDAS1 model was used to distinguish between 1) scans which missed the
plume, i.e. where the plume was not within the range of the instrument and 2) scans recorded
during very low degassing activity. The DOAS stations at both Villarrica and Llaima are located
east of the volcanoes, because westerly winds prevail in large parts of Southern Chile, due to the
passage of frontal systems which approach from the Pacific and move eastward along the midlatitude stormtrack (Garreaud 2009).
Garreaud, R.D. (2009). The Andes climate and weather. Adv. Geosci., 22, 3–11
Fig. S1 Comparison between sunrise and sunset times in Temuco and the timing of Mini-DOAS
scans at Llaima volcano used for the present paper.
Fig. S2 Sequence of vertical atmospheric profiles (GDAS1 soundings) showing the variation of
wind speed (top) and wind direction (bottom) obtained for location 38.69 S, 71.73 W (Llaima
volcano). The profiles comprise 19 pressure surfaces in the range from 900 to 20 hPa, which are
spaced at intervals of 50 hPa. The GDAS1 model has a temporal resolution of 1 sounding every
3 hours and a latitude-longitude grid spacing of 1 degree, i.e. it does not account for volcanoscale perturbations of the local wind field. Note that systematic daily variations of wind speed
and direction appear as vertical “ripples”. Wind speeds at summit height (at a pressure level of
about 700 hPa) ranged from 0.3 to 37.9 m sec−1 during the depicted period, and were at average
12.7 m sec−1 . We note that wind directions are largely uniform throughout the whole
atmospheric column. Strong westerly winds (orange colors) prevail particularly in the upperlevels of the atmosphere and especially during austral winter. Short phases of sustained easterly
winds (light blue and cyan colors) caused some of the gaps in our SO2-flux time series, since
westward drifting plumes are out of range for the scan DOAS stations.
Fig. S3 Chronologic compilation of calculated SO2 fluxes divided by wind speed data used for
the flux calculations of Llaima volcano (top) and enlargement of the time series shown for the
months June and July 2010 (bottom). The systematic daily variations and the scaling demonstrate
that the observed patterns of variations are not controlled by the wind speeds used.
Fig. S4 Time series of the first derivatives of the solid Earth tidal movements and the degassing
rates, respectively, for Llaima volcano. a) The complete time series signals and their patterns. b)
Sliding window cross-correlation of linearly interpolated and detrended SO2 –flux and solid
Earth tidal signal. The time series reveal a shift of roughly an hour per day between the signals.
c) Enlargement of the time series shown in a) for the month of May 2010.