Volcanic eruption of La Soufrière volcano :assessing volcanic hazards in real-time
The Soufrière de Saint Vincent volcano entered a major explosive eruptive phase on April 9, 2021, covering the entire island of Saint Vincent with a layer of ash several centimeters thick. Fortunately, the precursory signs of the eruption were correctly interpreted by the volcanologists in charge of monitoring, which enabled them to convince the authorities to trigger an order of preventive evacuation for nearly 20,000 people living in the “red zone”, the area under highest risk. To date, despite the power of the eruption, associated with devastating pyroclastic flows and mud flows, no casualties are to be deplored.
As part of crisis management and scientific response to the eruption, in collaboration with teams from the Seismological Research Center (SRC), the University of the West Indies (UWI) and the Montserrat Volcano Observatory (MVO), researchers from the ForM@Ter community (IPGP, BRGM, Univ. Strasbourg) were able to provide assistance to the teams on site by providing real-time analyzes of satellite observations of the eruption.
Ahead of this major crisis, a pre-existing cooperation between the IPGP and the UWI-SRC had already made it possible to follow the growth of a new lava dome at the top of the volcano, which started in December 2020. This dome of lava flowed peacefully, without causing explosions, for almost 3 months. A digital terrain model of the volcano, with a resolution of 2 meters, was thus calculated from Pleiades images acquired in 2014 using the MicMac software suite [Rupnik et al., 2016], with the aim of providing a very precise 3D image of the topography preceding the placement of the new lava flow [Grandin and Delorme, 2021, https://zenodo.org/record/4668734]. Regular drone flights, operated by the SRC/UWI/MVO, make it possible to update the topography, and to quantify the volumes of lava emitted over time. Meanwhile, the COPERNICUS Sentinel-1 and Sentinel-2 satellites provided regular monitoring of the thermal “hotspot” associated with the eruptive vent, as well as ground deformations. Finally, the installation of a webcam at the top made it possible to follow qualitatively the evolution of the growth of the dome, until the destruction of the station in April.
At the beginning of April 2021, the situation begins to deteriorate. Accelerated dome growth, increased outgassing, changes in gas composition, and the detection of more and more intense earthquakes, provide growing indications suggesting that a major eruption is in preparation.
At the request of IPGP scientists, alerted by UWI-SRC of these precursory signs, the “CIEST2” rapid response mechanism was triggered on Friday, April 9 at 8:00 a.m. UTC. A first acquisition by the Pleiades constellation (Airbus ADS / CNES) took place the same day, at 2:40 p.m. UTC, just two hours after the volcano’s first explosion, illustrating the reactivity of the device.
The acquisition strategy chosen consisted in pointing the Pléiades-1A and -1B satellites systematically at each pass over the area. For 10 consecutive days, daily acquisitions in “stereo” mode took place, exploiting the agility of the satellites, which are capable of pointing their optical system towards any target located in their field of view. Each acquisition consists of a pair of two images, taken with different viewing angles, less than a minute apart during the same orbital overflight, in order to increase the chances of obtaining a visual observation, and, if applicable, to be able to calculate the topography of the volcano by stereophotogrammetry. This strategy is implemented without taking into consideration the cloud cover, deliberately deviating from the nominal specifications of the Pleiades system, which in normal times make it possible to predict in advance the successes of exploitable acquisitions depending on weather conditions. The goal is clear: to try everything to provide observations at very high resolution, in order to to shed some light on the situation on the ground, and help local scientists as quickly as possible.
The first attempts to observe the ground failed, due to the thick cloud cover, a usual situation in tropical island conditions, worsened by the presence of an opaque eruptive plume, loaded with ash, aerosols, gas and ice crystals. Under such adverse conditions, it is impossible to see the volcanic edifice! Nevertheless, it is still possible to exploit the image of the volcanic cloud provide by Pleiades to extract key parameters of the eruption. The use of these images made it possible to determine the altitudes of the plume, as well as the speeds of movement of cloud parcels thanks to a method developed at BRGM (by Michele et al., 2016), subsequently adapted to Pléiades (Figure 1).
During a temporary lull in explosive activity, a Pléiades acquisition, carried out on April 15, 2021, around 2:40 p.m. UT (10:40 a.m. local time), finally made it possible to obtain detailed observations of the situation at the top of the volcano. These images reveal the formation of an explosive crater nearly 600 meters in diameter, probably formed during the paroxysmal phase of April 9–10. A plume of gas and ash continues to rise from the bottom of the crater, partially masking the surroundings which can be seen covered with a layer of uniformly gray ash: this weak contrast makes image interpretation difficult. Nevertheless, the high spatial resolution of Pleiades, associated with a high radiometric sensitivity, makes it possible to discern the existence of a thick pile of ash deposited around the explosive crater. The filling of the valleys incising the summit of the volcano, the result of pyroclastic flows, is also clearly visible.
Using the DSM-OPT on-demand service [D. Michéa and J.-P. Malet / EOST; E. Pointal, IPGP], which allows the automatic creation of digital terrain models with the MicMac software suite and ad-hoc downstream developments, a very high resolution topographic model is quickly calculated, thus making it possible to make a difference with the digital elevation model calculated from images acquired before the eruption (Figure 2, bottom right). This difference reveals the colossal thickness of ash accumulated near the crater: more than 100 meters! The crater itself, the bottom of which cannot be seen because of the plume, is at least 100 meters deep. The volumes of material displaced are in the order of several million cubic meters, indicating an eruption magnitude of 4 on the explosive index scale, comparable to the 1902 eruption of Mount Pelee (Martinique).
Mapping the thickness and distribution of ash deposits is essential to be able to anticipate the risks faced by the local population, as the arrival of spring rains will inevitably trigger destructive mudflows into the valleys that drain the volcano (“lahars”). Observing the consequences of the eruption makes it possible to better predict the preemptive measures to be deployed in the field.
Reactivity in times of volcanic crisis involves actors at several levels, from volcanologists in charge of observatories to space agencies operating satellites in orbit. The information chain involves computer scientists ensuring the rapid transmission of data, engineers developing image processing methods, under the leadership of scientists who are trying to better understand the eruptive dynamics, and better inform decision-makers, in the aim to better protect populations.
Scientific contact
Raphaël Grandin (grandin@ipgp.fr) – Institut de Physique du Globe de Paris – University of Paris
Participants
- Arthur Delorme (IPGP)
- Marcello de Michele (BRGM)
- Raphaël Grandin (IPGP)
- Jean-Philippe Malet (EOST)
- David Michéa (EOST)
- Elisabeth Pointal (IPGP)
References
Grandin Raphael, & Delorme Arthur (2021). La Soufrière volcano (Saint Vincent) Fusion of Pleiades (2014, 2 m) and Copernicus (2018, 30 m) digital elevation models (Version v1.1) [Data set]. Zenodo. http://doi.org/10.5281/zenodo.4668734
Ewelina Rupnik, Mehdi Daakir, and Marc Pierrot Deseilligny (2017). Micmac-a free, open-source solution for pho- togrammetry. Open Geospatial Data, Software and Standards, 2(1):1-9. https://opengeospatialdata.springeropen.com/articles/10.1186/s40965-017-0027-2
de Michele, M.; Raucoules, D.; Arason, Þ (2016). Volcanic plume elevation model and its velocity derived from Landsat 8. Remote Sens. Environ. 176, 219-224.@
