USING GPS SIGNAL-TO-NOISE RATIO (SNR) OBSERVATIONS TO DETECT AND CHARACTERIZE THE VOLCANIC PLUME ASSOCIATED WITH THE 2003 SOUFRIÈRE HILLS VOLCANO DOME COLLAPSE

GPS signals have been shown to attenuate when passing through volcanic plumes. This attenuation can be detected through the analysis of signal-to-noise ratio (SNR) observations. The capabilities of GPS SNR observations for constraining plume geometry, however, have not been fully explored. In this study, we investigate the temporal and spatial aspects of the volcanic plume resulting from the July 12-13, 2003 Soufrière Hills Volcano (SHV) dome collapse by analyzing GPS SNR residuals from three continuous receivers that logged data at 30 s.

Nominal performance is determined for each satellite-receiver pair by averaging over a series of days when no plume is present and then calculating a standard deviation about the mean SNR observations. SNR residuals are then computed by subtracting the mean SNR (plume absent) from the observed data during the dome collapse, using a threshold for plume detection of four standard deviations below the mean. Attenuation beyond the threshold is observed by all three GPS receivers during the peak eruption on 13 July 2003 at ~03:35 UTC. Attenuation is also observed during the day before the main collapse supporting independent observations that pyroclastic flows cascading down the Tar River Valley were occurring before the phase of dome collapse. Furthermore, we observe azimuthal variations in SNR residuals during and after the dome collapse event, which imply the plume traveled north-northeast, consistent with available GOES satellite images. Analysis of plume signal for satellite tracks crossing above the SHV yields an estimated plume height of ~3.3km directly above the vent. Geometry of the operational cGPS network was poor during the collapse and explosion sequence, therefore additional satellite tracks would be needed to determine the maximum plume elevation. Finally, timing of the plume signal on 13 July corresponds to an increased zenith Wet Tropospheric Delay, estimated independently through GIPSY-OASISII kinematic processing of the code and phase observables, suggesting that a plume signal may be included within the tropospheric delay data. Our results demonstrate the ability to detect volcanic plumes using GPS SNR for the massive dome collapse and explosion event at SHV in July 2003 and further provide a preliminary basis for constraining plume geometry with this method.