Volcanoes

The Rabaul caldera is an active Pleistocene to recent volcanic complex on the north-eastern point of New Britain Island, Papua New Guinea (PNG). It contains several small Holocene eruptive centres within the footprint of a large nested caldera structure. In the last ~200 years these small volcanoes have erupted on average every 20-60 years. Historically the most active are Tavurvur and Vulcan, located within the Rabaul harbour. The Rabaul caldera represents the highest risk of all PNG volcanoes since the town of Rabaul is located within the caldera structure. A significant twin eruption of the Tavurvur and Vulcan vents occurred in 1994 that destroyed large parts of the town and resulted in five casualties. Since 1994 Vulcan has remained dormant, but a major eruption occurred at Tavuvur on 6 October 2006. A simultaneous drop in surface height of nearly 30 cm was observed in GPS data collected at Matupit Island, the closest land mass to the centre of the main sub-caldera magma chamber. Following this eruption, a period of almost continuous minor eruptions of Tavuvur ensued, accompanied by a non-linear subsidence signal. These eruptions ceased at the beginning of 2010, at which time subsidence changed to uplift. Uplift prevailed until another significant eruption of Tavuvur on 29 August 2014, when an instantaneous subsidence of ~7 cm was observed. 

We have processed 21 ALOS PALSAR fine-beam SAR images acquired between 27 February 2007 and 10 March 2011 using the GAMMA software. L-band data is required in the highly vegetated environment of PNG to overcome temporal decorrelation that affects shorter wavelength data such as Sentinel-1. Analysis of continuous GPS data from four stations within the caldera was undertaken using the scientific Bernese software V5.2 with solutions tied to the ITRF2008 reference frame. We perform a joint inversion of a connected network of 20 interferograms and the continuous GPS observations to determine the temporal variation in volume change for the best-fitting Mogi point-source model of the deformation field. The weighted least squares inversion is solved using Singular Value Decomposition where the weights are derived from noise within each interferogram in the far field of the caldera. The spatially dense InSAR observations are subsampled in order to enable an efficient computation on desktop machines. The procedure involves first conducting coarse and fine grid searches over 3-dimensional space to find the best-fitting source location. Once the minimum has been located the temporal variation in volume change of the source is estimated.

The best fitting Mogi point source location is situated south of Matupit Island and north-west of Vulcan at 4.26°S, 152.18°E at a depth of 4.5 km, which is well within the geophysically-imaged sub-caldera magma chamber. We find that the inferred volume change for this source amounts to a deflation rate of ~10x10⁶ m³ yr^-1 during 2008 and 2009 and an inflation rate (uplift) of ~7x10⁶ m³ yr^-1 during 2010 and early 2011. Transient deformation signals in the interferograms localised around the Tavuvur vent may be useful in helping to derive a better understanding of the plumbing system connecting the magma chamber with the eruptive centres. This in turn may enhance short term eruption prediction. The Rabaul Volcano Observatory (RVO) is responsible for issuing eruption alerts for all PNG volcanoes. Future work to apply InSAR analysis on a national scale will assist the RVO to monitor the eruptive state of other active volcanoes that otherwise remain un-monitored.

Mapping eruptive deposits is essential for long term volcanic hazard assessment because it helps quantifying the magma eruptive rate and to calibrate numerical models predicting lava flows or pyroclastic density currents paths. Besides when unconsolidated deposits are emplaced on steep slopes in tropical areas, there is a crucial need to rapidly map them in order to mitigate the risk induced by their remobilization leading to destructive lahar formation. Here, we show that SAR data, providing an information even in cloudy conditions, are of significant interest to rapidly map eruptive deposits (both effusive and explosive) based on results obtained at two active andesitic stratovolcanoes: Merapi in Indonesia and Colima volcano in Mexico. During the period studied (2010-2016), large pyroclastic density currents deposits were emplaced respectively in November 2010 at Merapi (VEI 4, runout distance: 16.5 km) and in July 2015 at Colima (VEI 2, runout distance: 10 km). In addition, several lava flows were emplaced near the summit of Colima volcano from November 2014 to March 2015. The results obtained by SAR data have been compared with field data as well as optical images. Whereas the coherence evolution gives the best results to map effusive lava flows, amplitude changes detection appears to be the most efficient way to identify explosive deposits. We also show that the amplitude changes related to explosive deposits emplacement for co-polarized data follow the same trend for X-band (TerraSAR-X), C-band (Sentinel-1) and L-band data (ALOS). Finally, our study clearly shows the ability of radar data to identify, map and classify the pyroclastic deposits together with the additional value of considering dual-polarization and multitemporal images to combine the amplitude time series information with temporal decorrelation. 

Askja is an active volcano in the Northern Volcanic Zone of Iceland, lying within a spreading segment of the mid-Atlantic ridge. Its eruptions can be very powerful such as the 1875 VEI-5 caldera-forming Plinian event, however the current state of the complex magmatic system and the probability of an eruption in the near future are poorly understood.

Steadily decaying subsidence centred on the main caldera has been recorded using different geodetic measurements since at least 1983. It has been postulated that rifting extension and shallow magmatic processes, e.g. outflow and/or crystallisation, could be responsible for this subsidence. All models using surface deformation data agree that there is at least one shallow magmatic source at 3-3.5 km depth, undergoing volumetric changes at a rate of approximately -0.0014 to -0.0021 km³ yr^-1. However, recent results from seismic tomography revealed the presence of two melt storage regions at about 6 and 10 km depth.

Microgravity data have been acquired at Askja since 1988. A residual gravity decrease (mass loss) was observed during 1988-2003 and a residual gravity increase (mass gain) from 2007 to 2009. These gravity signals, which were both accompanied by ongoing steady subsidence, were interpreted as being due to magma drainage and magma intrusion, respectively.

Here, we present new models constraining (1) the location, depth and volume change of the shallow deformation source beneath Askja from 2002-2016, using InSAR timeseries analysis of radar data from ERS-2, ENVISAT, COSMO-SkyMed and Sentinel-1, and (2) the location, depth and magnitude of the mass changes over the same period of time, using microgravity data, which we extended to 2016.

In comparing our geodetic and microgravity models, we investigate the connection between mass and volume changes to constrain the potential physical processes that produced these signals.

Volcanic eruptions are the ultimate surface expressions of magma movements at depth. Analyzing ground deformations associated with volcanic eruptions contributes to understand the mechanisms of magma emplacement and characterize the magma plumbing systems. 

InSAR is a particularly well-suited tool to measure ground displacement in areas that are difficult to access because of geographical, economical or political reasons, and/or where little or no ground-based monitoring systems are available. This is the case of Nyamulagira, an active shield volcano with a central caldera, located in the eastern part of  the Democratic Republic of Congo along the  western branch of  the East African Rift System.

Nyamulagira shows a particular eruptive cycle characterized by short-lived flank eruptions (sometimes accompanied  with  intracrateral  activity) every 1-4 years, and less frequent long-lived eruptions usually emitting larger volumes of lava from eruptive vents located >8 km from the central caldera. The 2011-2012 Nyamulagira eruption is one of that last  type. This eruption began on November 6 2011 and ended in late April 2012.

In  the  present  study  we  use  InSAR data  from  various satellite (Envisat, COSMO SkyMed, Terrasar-X and Radarsat) to  measure pre-, co and post-eruptive ground displacement associated with the Nyamulagira 2011-2012  eruption. In  particular ground deformation time series obtained with the short revisiting time COSMO SkyMed satellites allowed us to detect a very fast (one day) magmatic intrusion below the Eastern flank of the caldera two days prior to the eruption. It also allowed the detection of the subsequent intrusion that brought the magma up to the two eruptive vents located 11 km ENE from the caldera.

To evaluate the source parameters and the mechanisms of magma emplacement we used analytical models  jointly  inverting  two  interferferograms (COSMO SkyMed in descending orbit and Envisat in ascending orbit) that cover the intrusive period. We tested different type of sources to find the most suitable   for this eruption given the observed deformation and the volcano-tectonic context. Considering also the few geophysical (seismic  and  GPS) data available for this area during the eruptive period, our analysis suggest that the eruption is a complex sequence of a deflation of a shallow magma chamber (~3km below the caldera) that fed a sill intrusion toward the ENE direction that twisted into a dyke and brought the magma up to the surface. 

Furthermore, GPS, InSAR and seismic datasets suggest the presence of a deep magmatic source that possibly fed the shallower magmatic system. This mechanism, involving a deep source for this large eruption, contrasts with the usual shallow plumbing system identified during the classical flank eruptions.

Galápagos volcanoes are among the most active volcanoes in the world and offer a unique natural laboratory to study volcanic processes. Many volcanic eruptions have occurred there during the past two decades with InSAR observations providing key information about the associated dike intrusions and underlying magma reservoirs. Several recent studies focusing on co-eruptive deformation in the Galápagos have reported co-eruption deflation consisting of focused intra-caldera subsidence superimposed on wider and weaker deflation signals. These coupled subsidence patterns have been modeled using pressure decrease in two magma reservoirs located at different depths. It is not clear, however, why magma would accumulate in reservoirs at different crustal depths.

We processed Sentinel-1 and ALOS-2 data to study the co-eruptive deformation of the most recent volcanic eruption in the Galápagos, the 2015 Wolf eruption. The resulting deformation patterns are complex, indicating multiple deformation sources during two main eruptive phases. The eruption started energetically on 25 May 2015 from a circumferential fissure located near the caldera rim and further activity followed in mid-June within the caldera. Localized deformation near the circumferential and intra-caldera eruptive fissures provides information about the dikes that fed these two phases of the activity. In addition, strong subsidence signal is observed within the Wolf caldera that is superimposed on a larger-scale subsidence signal that is weaker in magnitude and larger in spatial extent, extending well down the flanks of the volcano. We can model these two subsidence patterns as being due to pressure decrease in two shallow magma reservoirs at ~1 km and ~5 km depths below sea level. The concurrence of the subsidence patterns suggests that the magma reservoirs are hydraulically connected.

As it is not clear why magma would accumulate in reservoirs at different depths, we explore whether the two observed deformation patterns could be explained with a single reservoir and ring fault activity. The long-term uplift since the last 1982 Wolf eruption likely caused tension across the caldera rim, making conditions for a circumferential dike favorable. This first phase of the eruption caused depressurization of the reservoir and subsidence. This might have activated an outward-dipping reverse fault, with fault slip initiating above the depressurizing reservoir and not reaching the surface. In this case, the depressurization of a single magma reservoir combined with reverse, buried ring faulting will predict localized subsidence embedded within a broader deflation signal. This might therefore explain the observed co-eruption deformation at Wolf and at other Galápagos volcanoes.