Introduction
Located at the western end of the South Shetland Islands, Deception Island is bordered to the south by the Bransfield Strait, which separates the island chain from the Antarctic Peninsula, and to the north by Livingston Island (Fig. 1). Although there are other volcanic islands and submarine volcanoes along the Bransfield Strait, Deception Island is considered the most active volcano within the South Shetland Islands. With numerous historical eruptions, the most recent in 1970, evidence of volcanic activity is frequently observed in the form of volcanically induced seismicity, episodes of crustal deformation and fumarolic and hydrothermal activity.
Throughout its history, Deception Island has hosted several research bases, of which only the Spanish Antarctic Gabriel de Castilla Base (BGdC) and the Argentine Antarctic Base Deception remain active. Currently, both bases are operational only during the austral summer.
Figure 1. Regional location map of Deception Island.
Unlike most nearby islands, Deception Island is an active volcano. Formed by successive submarine eruptions, the island spans ~30 km in diameter from its seafloor base, with only the caldera of the stratovolcano emerging 540 m above sea level, reaching a total elevation of 1500 m from the ocean floor (Geyer et al. Reference Geyer, Álvarez-Valero, Gisbert, Aulinas, Hernández-Barreña, Lobo and Martí2019). The exposed section - a horseshoe-shaped caldera ~15 km in diameter - is open to the sea on the island’s south-eastern side, forming a shallow inland bay, Port Foster, with a depth of 180 m and a diameter of 6–10 km (Baker et al. Reference Baker1969, Martí et al. Reference Martí, Geyer and Aguirre-Diaz2013).
Dating back to the Quaternary period, Deception Island originated in a complex tectonic setting. Located south-west of the South Shetland Islands, it lies within the remnant subduction zone between the Phoenix Plate and the Antarctic Plate. This subduction, active during the Upper Mesozoic and Cenozoic eras, gave rise to the South Shetland Islands as a series of island arcs aligned parallel to the convergent plate margin, along with the back-arc basin south-east of the islands (Martí et al. Reference Martí, Geyer and Aguirre-Diaz2013). As subduction nearly ceased ca. 4 million years ago (Barker et al. Reference Barker, Dalziel and Storey1991, Lopes et al. Reference Lopes, Caselli, Machado and Barata2014), a north-west to south-east extensional process perpendicular to the subduction zone began, forming what is now the Bransfield Strait and causing crustal thinning and rift initiation. These extensional forces are thought to have resulted either from rollback of the Phoenix Plate (Maldonado et al. Reference Maldonado, Larter and Aldaya1994) or from the left-lateral motion of the Antarctic and Scotia plates and their interaction with fracture zones and plate boundaries (Gràcia et al. Reference Gràcia, Canals, Lí-Farràn, Prieto, Sorribas and Team1996). Deception Island, along with Penguin and Bridgeman islands and several submarine volcanoes, aligns along this rift in a north-east to south-west direction, although there is no indication of oceanic crust formation (Christeson et al. Reference Christeson, Barker, Austin and Dalziel2003).
Geochemical analysis of subaerial volcanic deposits on Deception Island shows basaltic to rhyolitic magmas. This composition, differing from that of typical island arc and subduction zone magmas, resembles that of oceanic rift magmas, indicating a mantle origin with partial melting influence from the subduction zone (Geyer et al. Reference Geyer, Álvarez-Valero, Gisbert, Aulinas, Hernández-Barreña, Lobo and Martí2019). These compositional findings correlate with the tectonic context to which the island belongs.
The geological evolution of Deception Island has been shaped by the volcanic edifice’s collapse and the formation of its caldera, marking a distinction in its stratigraphy between pre- and post-caldera units. This event, estimated to have occurred ca. 3980 bce (Antoniades et al. Reference Antoniades, Giralt, Geyer, Álvarez-Valero, Pla-Rabes and Granados2018), separates the two main stratigraphic groups on the island:
-
1) Pre-caldera deposits: Primarily basaltic-andesitic to basaltic-trachyandesitic pillow lavas (Geyer et al. Reference Geyer, Álvarez-Valero, Gisbert, Aulinas, Hernández-Barreña, Lobo and Martí2019), these form the shield structure base below sea level (Martí et al. Reference Martí, Geyer and Aguirre-Diaz2013, Smellie Reference Smellie2001). These magmas are the least evolved on the island, originating directly from the mantle (Geyer et al. Reference Geyer, Álvarez-Valero, Gisbert, Aulinas, Hernández-Barreña, Lobo and Martí2019). Hydrovolcanic eruption deposits and pyroclastic materials are also included (Lopes et al. Reference Lopes, Caselli, Machado and Barata2014).
-
2) Post-caldera deposits: Subaerial deposits consisting of lava flows, scoria cones and several monogenetic craters with strombolian and phreatomagmatic deposits, ranging from basaltic to andesitic compositions (Lopes et al. Reference Lopes, Caselli, Machado and Barata2014, Jiménez-Morales Reference Jiménez-Morales2022).
Both groups exhibit normal magnetic polarity, restricting the age of the rocks to a maximum of 0.75 million years (Baraldo et al. Reference Baraldo, Augusto, Rapalini, Böhnel and Mena2003).
Deception Island remains volcanically active, with historical eruptions recorded since the mid-nineteenth century (Grad et al. Reference Grad, Guterch and Sroda1992). The most recent eruptions occurred in 1967, 1969 and 1970, consisting of short-duration strombolian and phreatomagmatic eruptions that primarily affected the island’s eastern side. Seismic events preceding each eruption by weeks acted as precursors.
Among these historical eruptions, the 1969 event produced pyroclastic emissions and lahars, resulting in the destruction of the nearby Chilean Antarctic base and severe damage to the British Antarctic base at Whalers Bay (Baker et al. Reference Baker1969).
To ensure the safety of the active Antarctic bases on Deception Island, as well as of vessels and personnel, a continuous monitoring network is essential for tracking volcanic activity and providing volcanic alerts.
Spain has conducted seismic monitoring on Deception Island since the establishment of the BGdC. This has primarily taken place during the austral summers, employing both short-period and broadband seismic stations and using seismic arrays (Carmona et al. Reference Carmona, Almendros, Serrano, Stich and Ibáñez2012, Reference Carmona, Almendros, Martín, Cortés, Alguacil and Moreno2014, Jiménez-Morales et al. Reference Jiménez-Morales, Almendros and Carmona2022).
In addition to seismic monitoring, geodetic techniques have been employed, primarily using Global Navigation Satellite System (GNSS) space geodesy, although classic geodetic techniques such as levelling and the use of total stations (i.e. a surveying instrument that combines an electronic theodolite and a distance meter) have also been applied (Rosado et al. Reference Rosado, Fernández-Ros, Berrocoso, Prates, Gárate, de Gil and Geyer2019). The Volcanic Surveillance Network on Deception Island, maintained by Spain’s National Geographic Institute (IGN), remains among the few distributed volcanic monitoring networks across Antarctica, including others such as the networks on Mount Erebus on Ross Island and Mount Melbourne, close to Italy’s Mario Zucchelli Station, both being active volcanoes (Smellie et al. Reference Smellie, Panter and Geyer2021, Hakim et al. Reference Hakim, Sakina, Fadhillah, Lee, Park, Kim and Lee2024).
Field equipment infrastructures
Station enclosure
To withstand the extreme climatic conditions of Deception Island, the infrastructure of scientific stations relies on prefabricated shelters made of marine-grade stainless steel, with 2 mm-thick walls supported by a 30 × 30 × 2 tubular frame that adds rigidity. These shelters are designed to maximize durability and energy efficiency. Integrated into the exterior are solar panels for power generation, while the interior houses batteries, an electrical control panel and dataloggers for recording measurement equipment data.
Two models of shelters have been implemented. The initial model was designed to support a 360 W solar panel powered by two 135 amp-hour (Ah) batteries (lead-acid AGM (absorbent glass mat) batteries). However, as this configuration proved insufficient, a second, slightly larger model was developed to accommodate a 545 W solar panel along with a series of four 250 Ah batteries, totalling 1000 Ah, providing enhanced energy autonomy (Fig. 2).
Figure 2. Diagram of the infrastructure built to house the volcanic monitoring network stations. It consists of a steel hut with a solar panel. Inside the hut are housed the electronics of the seismic station, the Global Navigation Satellite System (GNSS) station, the batteries, the regulator and the Wi-Fi communications antenna.
To maximize solar energy capture, the panels must be installed at an angle corresponding to the site’s latitude, with variations of ± 15° to optimize performance across seasons. Given that energy shortages on Deception Island are most severe during winter, the optimal tilt for the solar panels is 63° ± 15° (Duffie et al. Reference Duffie, Beckman and Blair2020).
Due to the limited sunlight during the Antarctic winter, with daily sun exposure often below 5 h, solar energy alone cannot sustain the stations. To address this deficit, a methanol fuel cell has been integrated as a supplementary energy source. This system provides continuous and reliable power in tandem with the solar panels, ensuring uninterrupted operation year-round, even in adverse conditions.
The combination of photovoltaic systems and methanol fuel cells (Fig. 3) represents an efficient and sustainable energy solution ideal for remote scientific installations. While solar panels harness available energy during daylight hours, the methanol fuel cells complement this by generating continuous electricity, regardless of solar radiation or weather conditions. This setup is particularly advantageous during winter months when daylight hours are significantly reduced.
Figure 3. a. Photograph of the methanol fuel cell station house. Attached to the hut is the box with the electronics associated with the methanol stack along with the methanol cylinders. b. Interior of the box containing the methanol stack.
The chosen sites for the installation of the scientific stations are historically significant locations previously used by the University of Granada and the University of Cádiz. Installation and maintenance were made possible with logistical support from Spanish Army personnel stationed at the BGdC (Fig. 4).
Figure 4. Detail of the logistics used for the installation of the booths. The booths were transported with the help of inflatable boats and with the support of Spanish Army personnel from the Gabriel de Castilla Base. Photographs: Marcos Rozalen (Spanish Army).
Due to the extreme wind gusts in the region, the seismological stations are affected by increased seismic noise. To mitigate this phenomenon and ensure a stable measurement environment, boreholes of up to 2 m deep were drilled, allowing for sensor installation at depths shielded from ambient noise.
Inside the shelters, in addition to the power system, a charge controller is programmed to cut off power to both the station and the transmission when battery voltage drops below a predefined threshold. The system reactivates once it surpasses a preset charge threshold, creating a type of hysteresis cycle that enhances the station’s robustness. The initial shelters did not transmit photovoltaic system data to the cloud, but the two most recent shelters now include an energy consumption monitoring system that aids in equipment programming. Data are transmitted using a 2.4 GHz Wi-Fi antenna to the base, located up to 6.5 km away.
The two primary challenges in remotely monitoring volcanic activity at the BGdC have historically been energy and communications. For years, the infrastructure could not support real-time data transmission from all seismic stations. Beginning in 2007, a system was designed to send daily information on seismic activity through an image generated from waveform data. Although this information did not allow for in-depth signal analysis or earthquake localization, it did provide insights into activity near to the base (Carmona et al. Reference Carmona, Almendros, Martín, Cortés, Alguacil and Moreno2014).
Energy
The installation of a new photovoltaic station during the 2022–2023 campaign significantly increased the base’s energy capacity, enabling the deployment of more powerful communication equipment with higher energy demands. The system consists of 16 monocrystalline solar panels, each generating 385 W, with two additional wind turbines of 3 and 5 kWp, respectively, added in the 2023–2024 campaign. It includes a battery system with four units, each containing six 2 V cells, providing a total storage capacity of 2600 Ah. The average consumption is 270 W, allowing for ~10 days of autonomous operation without recharging (Fig. 5).
Figure 5. Details of the photovoltaic installation.
This new energy system is also designed for remote control, with integrated technologies that store data in the cloud, enabling real-time management and monitoring.
Communication
The communications infrastructure between the BGdC and the central headquarters of the IGN in Madrid consists of two satellite antennas, each with complementary functions to ensure system stability and redundancy.
The first antenna employs Hughes technology (www.hughes.com), providing connectivity through a satellite solution optimized for remote environments. This system includes a router that manages the network and balances traffic via the Wide Area Network (WAN) port to the Hughes satellite connection and through Wi-Fi to the Starlink antenna (www.starlink.com). Technically, it operates in the Ku band, offering up to 25 Mbps download and 3 Mbps upload speeds, although these values may vary depending on atmospheric conditions. The typical latency for this geostationary solution is ~600 ms. The router connected to this antenna enables interconnection with other devices through the Local Area Network (LAN) connection.
The second antenna uses Starlink Business, offering greater transmission capacities and lower latencies due to its low-Earth orbit satellite constellation. This antenna is also equipped with a router that manages internet access through Starlink while serving as a Wi-Fi link to the Hughes router to balance both connections. Its technical specifications include a Ka band operating frequency, with up to 200 Mbps download and 20 Mbps upload speeds. The Starlink low-Earth orbit provides a latency range of 20–40 ms.
The installed routers play an essential role in network management and load balancing between the Hughes and Starlink satellite connections. The Hughes router connects to its network via the WAN port, while its Wi-Fi interface links to the Starlink router. This configuration allows traffic to be distributed efficiently between the two connections, maximizing the use of available resources. Both routers are configured with OpenVPN (https://openvpn.net), establishing encrypted tunnels that ensure secure data transmission and enable port forwarding for remote access to scientific services on the island (Fig. 6).
Figure 6. Operational diagram of the communications system, outlining the data flow from the field stations, through the Gabriel de Castilla Base (BGdC) and onwards to their final reception in Spain. IGN = Spanish National Geographic Institute.
The system prioritizes the use of Starlink, due to its lower latency and higher capacity, for critical applications such as real-time scientific data transmission and Voice over Internet Protocol (VoIP) communications. Hughes acts as a backup, maintaining connectivity in the event of Starlink instability. Dynamic load balancing allows switching between the two connections based on availability or demand.
This architecture has significantly improved the stability, availability and security of communications. The implementation of OpenVPN protects data from interception, and port forwarding facilitates the remote management of scientific stations such as seismographs or meteorological sensors. Overall, the system ensures continuous and secure connectivity, even in adverse weather conditions or during maintenance work.
The data received at the central headquarters of the IGN show how energy and communication challenges have impacted the monitoring stations in various ways. The MECO station, located at the main base, has maintained a steady data flow due to the solar panel bank installed in the workshop module. In contrast, the BASE station, equipped with a solar panel and a methanol fuel system, has operated stably for the most part, although data transmission stopped in August 2024, an issue to be investigated in the next field campaign. Other stations have shown expected performance, with interruptions during the austral winter due to reduced sunlight in their enclosures. The FUM and CHI stations, installed during the latest campaign with larger enclosures, managed to operate longer due to their higher-capacity solar panels and greater energy storage capabilities. This variety in station performance highlights the effectiveness of recent upgrades while also identifying areas where future enhancements may be necessary to ensure consistent data flow throughout the year (Fig. 7).
Figure 7. Schematic diagram of the operation of the surveillance network stations. Comparison is made with the hours of insolation of the solar panels.
Seismic monitoring
The seismic network installed by the IGN on Deception Island is based on the extensive experience that the IGN has in volcanic monitoring in its national territory, especially in the Canary Islands, where it has managed the last two eruptions that occurred at El Hierro in 2011 and La Palma in 2021 (López et al. Reference López, Blanco, Abella, Brenes, Cabrera-Rodríguez and Casas2012, Del Fresno et al. Reference Del Fresno, Cesca, Klügel, Domínguez-Cerdeña, Díaz-Suárez and Dahm2023). It has also relied on the accumulated experience of the University of Granada in its seismic monitoring of the region around the island (Carmona et al. Reference Carmona, Almendros, Serrano, Stich and Ibáñez2012, Almendros et al. Reference Almendros, Carmona, Jiménez, Díaz-Moreno and Lorenzo2018, Jiménez-Morales et al. Reference Jiménez-Morales, Almendros and Carmona2022). Historically, the monitoring network on Deception Island was primarily a temporary network, deployed annually at the start of each Antarctic campaign around December and withdrawn at the end of March. However, with the recent assignment of responsibilities to the IGN by the Spanish Polar Committee, the monitoring has evolved into a permanent network. The new stations, mostly located in the same sites as before, are now designed to operate continuously throughout the year, allowing real-time monitoring from the IGN headquarters in Madrid and significantly improving the response capacity and volcanic analysis regarding Deception Island (Fig. 8).
Figure 8. Location map of the new Spanish National Geographic Institute (IGN) seismic stations.
For this purpose, in addition to stable shelters for housing the instrumentation, the seismic recording processes have been improved. The seismic signal is highly influenced by the strong winds that the island endures throughout the year, which can reach speeds of 160 km/h (https://antartida.aemet.es/informes/2023). To minimize these atmospheric effects, a series of boreholes have been made to house the sensor. The equipment used to excavate these boreholes consisted of medium-sized drilling equipment (Fig. 9). The installation of seismic sensors using boreholes provides several essential advantages to improve the quality of recording in environments with high levels of ambient noise (Trnkoczy et al. Reference Trnkoczy, Bormann, Hanka, Holcomb, Nigbor and Bormann2012), as is the case regarding the Deception Island equipment. Burying the sensors at depth provides significant isolation from external vibrations, such as wind or human activity. This improves data accuracy compared to sensors installed at the surface level (Aderhold et al. Reference Aderhold, Anderson, Reusch, Pfeifer, Aster and Parker2015).
Figure 9. Details of the soundings carried out at the stations.
Analysis of the average amplitude of the ground velocity recorded by the seismometers (real-time seismic amplitude measurement, or RSAM; Endo & Murray Reference Endo and Murray1991) indicates significant improvement in the signal recorded when the seismic sensor is installed inside a borehole. Figure 10 shows the RSAM record of MECO station, where we can observe some peaks recorded due to the anthropogenic activity of the base reducing substantially from 5 February onwards. In addition, the analysis of the seismic noise in the vertical component of the MECO station data for a period before (1–20 January 2023) and after (1–20 March 2023) the installation of the sensor inside the borehole (Fig. 11) also indicates significant improvement, with decreases in the power spectral density at a maximum of 10 dB for frequencies above 0.3 Hz.
Figure 10. Real-time seismic amplitude measurement (RSAM) of the MECO station. It can be observed how the peaks disappear from 5 February onwards.
Figure 11. Analysis of the seismic noise in the vertical component of the MECO station for a period before (1–20 January 2023) and after (1–20 March 2023) the installation of the sensor inside the borehole. Peterson (Reference Peterson1993) maximum and minimum background noise curves are included.
Furthermore, buried sensors enjoy greater thermal stability, minimizing the impacts of temperature fluctuations at the surface on these measurements. As we go deeper into the ground, the temperature becomes more constant, preventing thermal deformations that could introduce additional noise into the recorded data (Quigley & West Reference Quigley and West2023).
The drilling conducted on Deception Island has been conditioned by strict environmental regulations in Antarctica and by the particularities of the terrain at each measurement station. Depending on the characteristics of the soil at each location, different methodologies were employed to carry out the drilling. In areas where permafrost (frozen soil) was found just a few centimetres below the surface, a first layer of pyroclastic material was removed manually, and a portable drilling machine equipped with a 20 cm-diameter drill bit was placed on top of the permafrost. After completing this drilling, a 16 cm polyvinyl chloride (PVC) pipe with a concrete base was installed, inside which the seismic sensor was placed, improving its stability and protection against environmental conditions. In locations where permafrost was not detected, the drilling was conducted manually. The depth achieved in both cases was ~180 cm.
The sensors used in the various installations are all broadband sensors with sampling frequencies of 100 samples per second. Products from different manufacturers have been used, including three Quanterra Q8 - MBB2 devices from Kinemetrics, two Minimus devices from Güralp (one with a Radiant sensor and another with a Certis sensor), a Centaur station from Nanometric with a Trillium Compact Post Hole sensor and a Cube station with a Trillium Compact PH sensor. All equipment transmits data in mseed format via Wi-Fi to the BGdC. There, the data are received on an industrial mini-PC running Linux, integrated into SeisComp, and sent via the internet to the IGN headquarters in Madrid (Table I).
Table I. Chronology of the installation of the monitoring stations (seismic, Global Navigation Satellite System (GNSS), webcam and thermal) carried out during the Antarctic campaigns from 2020–2021 until 2023–2024. The table indicates the type of station, location, incorporated technologies and improvements implemented during each campaign. The table details the name, the type of station with the commercial brand, the digitizer model and the sensor used.
aSame place and name as the 2021–2022 campaign, but the instrumentation differed.
Permanent GNSS station network
A permanent GNSS station network began to be installed in the 2022–2023 campaign and is currently made up of six stations (Fig. 12). In the first campaign (2022–2023), four stations were installed (BASE, OBSG, CR70 and RONG), and in the second (2023–2024), another two were installed (FUMG and CHIG; Table I). The receiver installed was a Leica GR50 device, except at station CR70, where a Leica GR10 device was used. As for the antennas, a Leica AR20 device with a dome has been used at all of these stations.
Regarding the stations’ distribution, they have all been installed in the vicinity of the seismic stations in order to share in the power and communications infrastructure. In locations where there was already infrastructure that had been used previously by other research groups, such as that of the University of Cadiz (Rosado et al. Reference Rosado, Fernández-Ros, Berrocoso, Prates, Gárate, de Gil and Geyer2019), this has been used during the installation process (e.g. at OBSG and CR70). For the rest of the stations, the antennas have been installed on steel pillars between 0.8 and 1.2 m high. Where possible (e.g. RONG), the pillar has been anchored to the bedrock (Fig. 13). For the remainder of the cases (BASE, FUMG and CHIG), concrete cubes weighing between 200 and 250 kg have been built and buried in a hole made in the pyroclast.
Near each of the permanent stations, two survey nails have been placed. In this way, the differences in level between each of the nails and the structure in which the GNSS antennas are anchored are calculated during each campaign, with the purpose of detecting any local deformation that could affect the calculated position in our stations.
In addition to the network of permanent stations, there is another network of points distributed throughout the interior of Puerto Foster Bay, consisting of a piece of threaded rod embedded in rock or concrete cubes. Some of these points have been established for the observation of this network, and others were already observed previously by other research groups and have been reused. This network began to be observed in the 2023–2024 campaign following the real-time kinematic (RTK) methodology using two Topcon Hiper HR GNSS receivers. For this purpose, two of the network’s permanent stations (BASE and RONG) have been equipped with radio antennas that emit differential corrections in RTCM 3.2 format. This design allows all points in the network to have direct line of sight with at least one of the two stations that emit the corrections. Observation (60 s at each point) is carried out following an itinerary between all of the points, with the journeys being made using a zodiac-type boat, which allows the itinerary to be completed in ~6 h. In the 2023–2024 campaign, observations were made on three occasions (Fig. 12), with the intention of establishing a references level, and three to four observations per campaign are planned for subsequent campaigns (Table I).
Thermometry measurements and surveillance cameras
Thermometry and surveillance cameras play a crucial role in volcanic monitoring, as they allow for the detection of early signs of eruptive activity or changes in hydrothermal systems. Thermometry is used to measure and monitor temperature variations in critical areas, such as fumaroles, fractures or thermal anomaly zones. These changes in temperature can be indicators of magmatic movements or the circulation of hot fluids, which can warn of possible imminent eruptions or of increases in volcanic activity.
The surveillance cameras, both visible and infrared, allow continuous observation of the volcano surface, recording events such as explosions, gas emissions, lava flows or changes in the morphology of the crater. These cameras provide real-time information, facilitating a rapid response to any significant change in the volcano’s activity. Together, the combination of thermometry and visual surveillance provides a comprehensive monitoring tool, helping us to better understand the volcano’s behaviour and reduce associated risks.
For all of these reasons, during the last campaign (2023–2024), a thermometry monitoring system and a visible-spectrum camera were installed within the volcanic surveillance network of the island (Fig. 14).
Figure 12. Location map of the new Spanish National Geographic Institute (IGN) permanent Global Navigation Satellite System (GNSS) stations and the points for the real-time kinematic (RTK) itinerary.
Figure 13. RONG permanent Global Navigation Satellite System (GNSS) station.
Thermometry
The existence of thermal anomalies on Deception Island since the beginning of human presence on the island has always been a point of interest from both a scientific and a cultural point of view. One of the common practices of Antarctic tourism has been to enter Deception Bay to bathe in the hot springs. Many of the thermal anomaly points are located in the intertidal zone. Apart from these points that are difficult to monitor continuously, there are other such points in the interior of the island in fracture zones such as Cerro Caliente or Monte Pound. Some of these points have been monitored for years (Berrocoso et al. Reference Berrocoso, Prates, Fernández-Ros, Peci, de Gil and Rosado2018). Since the 2023–2024 campaign, the Cerro Caliente point has been monitored, with data being sent in real time to the IGN headquarters in Madrid. This had never been done before. During the same campaign, a team from the University of Navarra also implemented a monitoring system that transmits data in real time via satellite to a local server in Spain (Astrain et al. Reference Astrain, Pascual, Catalán, Araiz, Alegría, Rosado and Berrocoso2025).
For the monitoring of thermal activity, equipment developed by the IGN team has been installed, consisting of four PT100 temperature probes installed in a vertical profile with a separation between sensors of 10 cm in order to study the temperature gradient (Fig. 15). The design consists of very low power consumption equipment that has been successfully used in other volcanoes, such as Mount Teide in the Canary Islands (Awadallah et al. Reference Awadallah, Moure and Torres-González2019). To achieve minimum data consumption, the equipment is designed not to store data and can be powered for 1 year with a battery. The data are transmitted using LoRa (Long-Range) technology (Agustín et al. Reference Agustín, Yi, Clausen and Townley2016) of very low power consumption. For discrete data transmission with a low sampling rate, the system acquires data every 5 min. The data are received from the base and stored within a Raspberry Pi-based system (www.raspberrypi.org) that manages and forwards the data to Spain using real-time Starlink-based telemetry (Fig. 16).
Surveillance camera
The use of webcams for volcanic surveillance has become widespread at numerous volcanoes around the world, providing real-time visual monitoring that complements geophysical and geochemical data. They make it possible to observe changes in surface activity, such as eruptions and gas emissions, which is crucial for the early detection of eruptive events and emergency management. They also help to identify deformations in the volcano and other morphological changes that could indicate imminent activity. The images obtained can be used to create time-lapse videos, facilitating the study of volcanic evolution.
Figure 14. Location map of the new surveillance camera and thermometry monitoring system.
Figure 15. Installation of a vertical temperature profile at four depths.
Figure 16. Recording of the four temperature sensors installed at Cerro Caliente. The gaps are due to communication failures.
The surveillance camera installed at the BGdC is a visible spectrum-type camera with pan, tilt and zoom capabilities. This camera has the ability to pan both vertically and horizontally and has an optical zoom of up to 32×, allowing detailed tracking of points of interest. It operates continuously, capturing high-resolution images at 1920 × 1080 pixels every minute. The images are transmitted in real time to the headquarters of the IGN in Madrid, facilitating constant monitoring of activity on the island. In addition, these captures are used to generate an hourly time-lapse covering the last 24 h, providing a dynamic visualization of this activity. The camera can also be remotely manipulated and reorientated, allowing flexible focusing on different areas of interest (www.ign.es/web/ign/portal/camara-web-antartida; Fig. 17).
Figure 17. Photographs taken at the same hour of the day on different dates: a. 19 March 2024, b. 21 March 2024, c. 9 April 2024, d. 27 July 2024, e. 11 August 2024 and f. 29 August 2024.
First results
Seismic data analysis
For the analysis of seismic events in the vicinity of Deception Island, SeisComp version 3 software has been used (www.seiscomp.de/doc/base/citation.html). SeisComp is a seismic software widely used in many seismological observatories around the world for acquisition, processing, distribution and interactive analysis. It has been developed by the GEOFON Program at Helmholtz Centre Potsdam, GFZ German Research Centre for Geosciences and gempa GmbH. It is a free software distributed for non-commercial use.
Through SeisComp’s seedlink module, the signals from the Deception Island seismic equipment reach the IGN’s headquarters in Madrid via a Transmission Control Protocol/Internet Protocol (TCP/IP).
Through the implementation of this calculation system, all local and regional seismic events detected by our stations are subjected to real-time analysis. This capability is enabled by a dedicated monitoring room that operates continuously, staffed by qualified personnel who perform this task 24 h a day, 365 days a year, at the headquarters of the IGN. Within this framework, we categorize and identify distinct types of seismic signals, including volcano-tectonic (VT) earthquakes, hybrid earthquakes, and long-period (LP) earthquakes.
This calculation system analyses in real time all local and regional events detected by our stations, as well as possible signals of volcanic origin. The IGN has a 24/7 shift operating every day of the year at its headquarters in Madrid, and the staff are in charge of routinely analysing both the signals at the national level and the data collected on Deception Island. During recent campaigns, different types of signals have been identified: VT earthquakes, hybrid earthquakes, LP earthquakes and whale song (Fig. 18a–d). In order to characterize the volcanic signals, it is very important to observe not only the waveform but also its spectrogram to be able to identify the different signals. For this purpose we use Swarm software to visualize the signals. Swarm is a Java application designed to display and analyse seismic waveforms in real-time, developed by the United States Geological Survey (USGS; https://volcanoes.usgs.gov/software/swarm/index.shtml).
Figure 18. Examples of waveforms of events recorded in the vertical component at BASE and OBS stations and their spectrograms: a. low-frequency event, b. hybrid event, c. tremor event and d. volcano-tectonic earthquake with whale song.
Low-frequency or LP earthquakes have their origins in the resonance of water, magma or gas in crustal fractures (Chouet Reference Chouet2003). They generate earthquakes without clear phases, have a very emergent onset (making it difficult to define both primary (P) and secondary (S) waves) and have a characteristic shape. Their spectral content is limited to low frequencies, generally between 0.5 and 5.0 Hz. They can occur in isolation (Fig. 18a), but they most commonly occur in the form of seismic swarms.
Hybrid earthquakes are events whose origin is associated with the rupture of new fractures in the Earth’s crust that are filled with volcanic fluids. The main difference between hybrid and LP earthquakes is that the fluid resonance is triggered by the rupture process of the faults involved in hybrid earthquakes. They are characterized by initial high-frequency arrivals in which it is possible to identify P and S phases followed by low frequencies characteristic of LP earthquakes (Fig. 18b).
VT earthquakes are events originating in fractures in the Earth’s crust in volcanic environments. They are tectonic earthquakes related to the dynamic processes of the volcano. They are characterized by generally impulsive P phases and identifiable S phases. Their spectral content is wide, reaching frequencies of 20–30 Hz (Fig. 18d).
Whale songs are occasionally recorded when individual whales enter the inner bay of the island, exhibiting a characteristic acoustic pattern with dominant frequencies at ~16 Hz (Fig. 18d; Vergara-Gonzalez et al. Reference Vergara-Gonzalez, Almendros and Teixido2025).
In the context of the volcanic monitoring of Deception Island, the implementation of a real-time remote analysis system, operated from Madrid with 24 h shift personnel, represents an important advance compared to previous methods. Previously, seismicity analysis was performed exclusively on the island, in parallel with the installation and maintenance work of the seismic stations. This new approach allows for greater agility and efficiency in responses to crisis episodes, as seismic-volcanic activity can be evaluated continuously and accurately, which strengthens the capacity to respond to possible increases in activity. In recent years, near 800 seismic events have been recorded, the number of which correlates with the number of active stations transmitting data to the IGN in Madrid. However, during the austral winter there is a significant decrease in the recording of earthquakes due to interruptions in data transmission caused by extreme local conditions that affect the ability of the stations in the monitoring network to transmit data. An example of the event locations observed during the 2022–2023 surveys is presented in Fig. 19.
Figure 19. Regional seismicity map in the vicinity of Deception Island in the period 2022–2024.
Discussion of deformation data
Currently, the stations record data with periods of 1 and 30 s, and the data are stored in hourly and daily files, respectively. Observations from four constellations are used: Global Positioning System (GPS), Glonass, Galileo and Beidou. These data are sent via Wi-Fi from the stations to the BGdC, from where they are sent to Spain as described earlier. The receivers are equipped with a 32 Gb SD card, which allows data storage during the austral winter and data recovery during the following campaign in case data transmission fails.
These data are processed using different methodologies in order to allow the monitoring of the island’s volcanic activity, both in quasi-real time and over the long term. To this end, in addition to the data from the IGN network stations on Deception Island, data are downloaded from the International GNSS Service (IGS) network stations in the area (PALM and IOHI2/OHI3) and from the IGN station (BJCI) near the Juan Carlos I Antarctic station on the neighbouring Livingston Island, which operate as reference stations outside Deception Island.
The first process we perform is a baseline calculation between the various stations in the network using the open-source program package RTKLIB (www.rtklib.com; Fig. 20). To do this, we use the daily data at 30 s intervals and obtain a daily solution that gives us the variation in the three components of the movement (East, North and Up) between each of the stations in the network (Fig. 19). This processing is carried out in situ by the group deployed to Deception Island during each campaign, and it has been designed to be able to work with rapid, final or predicted ephemerides, which allow the calculation to be maintained for a period of several days even if communication with the BGdC is lost.
Figure 20. RONG-OBSG baseline series for the 2023–2024 campaign. A 3 day moving average has been applied to eliminate gross errors.
The second process allows the calculation of station coordinates for each observation epoch. It utilizes reference stations outside Deception Island, and the variation in coordinates with respect to the reference stations is obtained by performing a network adjustment (Lamolda Reference Lamolda2017). It uses the most precise ephemerides available at the time. A Kalman filter is applied to the time series (Fig. 21). Similarly to the previous process, this process uses the RTKLIB software, and, in this case, it is carried out at the IGN calculation centres in Madrid.
Figure 21. Three month sub-daily series (East, North and Up components) for the OBSG permanent station. The grey line represents the raw solution for each observation epoch (30 s), whereas the green, red and blue lines represent the series after applying the Kalman filter.
Other processing strategies using double differences (Dach et al. Reference Dach2015, Del Fresno et al. Reference Del Fresno, Cesca, Klügel, Domínguez-Cerdeña, Díaz-Suárez and Dahm2023) and precise point positioning (PPP) methodologies (Bertiger et al. Reference Bertiger, Bar-Sever, Dorsey, Haines, Harvey and Hemberger2020) are being developed, although they are not yet operational.
In addition to observation using GNSS techniques, surface deformation analysis of the entire island is carried out using radar interferometry (interferometric synthetic aperture radar, or InSAR; Hanssen Reference Hanssen2001). Currently, radar images from the Sentinel-1 and PAZ satellites are used for this purpose. The Sentinel-1 satellite operates in the C band (wavelength of 5.56 cm) and acquires images in ascending orbit every 12 days for this area, whereas the PAZ satellite operates in the X band (wavelength of 3.10 cm) and acquires images in both ascending and descending orbits every 11 days. The periods between acquisitions are 11 and 12 days, respectively. With each new satellite acquisition, three products are generated using SNAP software (http://step.esa.int): an interferometric coherence map, a phase interferogram and a displacement map. In the case of the Sentinel-1 satellite, products are automatically generated for the periods of 12, 36 and 96 days prior to the date of the last acquisition throughout the year (Fig. 20). In addition, products are also generated using PAZ images every 11 days for 5–6 months per year, coinciding with the opening of the BGdC.
Advanced InSAR (A-InSAR) workflows are also currently being developed to obtain time series. Different methodologies, such as Persistent Scatterer and Small Baseline (Hooper Reference Hooper2008), are particularly interesting for monitoring long-term slow deformations, but they are challenging to implement on Deception Island due to the long periods of time during which the ground is covered by snow or ice (Fig. 22).
Figure 22. Interferometric synthetic aperture radar (InSAR) displacement, phase and coherence maps generated using SNAP for the 20240216_20240228 Sentinel-1 pair of images. The images contain modified Copernicus Sentinal-1 data from 2024. GNSS = Global Navigation Satellite System.
Collaborations
Deception Island is also home to the Argentine Antarctic Base Deception, which has been operational since 1947 and has a long history of volcanological studies. During the 2022–2023 campaign, the Argentine Geological Mining Service (SEGEMAR) installed three seismic stations and two GNSS stations to monitor tectonic and volcanic activity in real time. Through a collaboration agreement signed by the two institutions, the IGN collaborates with the volcanic monitoring group of SEGEMAR, which is the institution that manages the Argentine Antarctic Base Deception. SEGEMAR has volcanic surveillance instrumentation deployed at Deception Island, whose data are transmitted via satellite in real time to its facilities in Buenos Aires. As a result of the collaboration established, the data acquisition and processing systems of IGN and SEGEMAR have been connected, so that all of the data from the monitoring stations of both institutions are being shared in real time, enabling the achievement of a high level of cooperation (Fig. 23).
Figure 23. Map showing the locations of stations deployed by the Argentine Geological Mining Service (SEGEMAR). The green dots correspond to Spanish National Geographic Institute (IGN) stations.
Conclusions
Volcanic monitoring on Deception Island is crucial for ensuring the safe and continuous development of the numerous and diverse scientific activities conducted in this region of Antarctica. As one of the most active volcanic zones on the continent, Deception Island represents not only an exceptional site of scientific interest but also a constant high-risk environment that demands continuous surveillance. The implementation of a monitoring network on the island - one of the first on the continent - constitutes a significant advancement in Antarctic safety. This system enables real-time tracking of volcanic phenomena that may affect both scientific bases and local ecosystems, becoming an essential tool for the international scientific community and other stakeholders (such as those involved in tourism) operating in these extreme conditions.
During the past four campaigns, significant progress has been made towards establishing a robust and efficient volcanic monitoring system on Deception Island. This network’s infrastructure has been enhanced through the installation of more stable and resilient sites, which has strengthened data transmission both during and outside Antarctic campaigns. Furthermore, the energy capacity at the BGdC has been considerably increased through integrating a high-capacity photovoltaic station supported by wind generation systems.
Communication capabilities have also been substantially expanded, providing significantly greater bandwidth that enhances real-time data transmission. From a monitoring perspective, a thermometry system has been installed to detect possible anomalous temperature variations in real time. Additionally, a visual monitoring system featuring a high-precision robotic camera has been established, allowing for the creation of time-lapse videos for rapid visualization of volcanic activity on the island.
Moreover, the personnel dedicated to this monitoring have been bolstered by the support of the 24/7 monitoring shift from the IGN, enabling continuous and real-time tracking of volcanic activity from the IGN headquarters in Madrid. These combined efforts have optimized the quality and capacity of volcanic monitoring in this remote and challenging environment.
As a result of this endeavour, an open-access website has been created for both the scientific community and the general public (www.ign.es/web/ign/portal/vigilancia-volcanica). This platform provides real-time access to the volcanic activity on the island, offering various monitoring perspectives. Through this website, users can observe the waveforms and spectrograms of the seismic signals from one of the stations on the island, facilitating real-time tracking of seismic activity. Additionally, an interactive viewer allows for the visualization of the locations of regional earthquakes, providing a geospatial representation of seismic events.
Regarding potential ground deformation, analyses based on GNSS data and the InSAR technique are included. Daily and sub-daily GNSS solutions and interferograms every 11–12 days allow possible displacements to be detected. The site also integrates visual monitoring of the island through a surveillance camera, enabling the observation of real-time conditions and the detection of any unusual volcanic phenomena. Moreover, the system generates time-lapse videos that depict the meteorological state of the island, thus providing a comprehensive view of environmental and volcanic conditions.
All generated data, in addition to being stored at the IGN headquarters, are periodically uploaded, to be made available to researchers who request access through a platform created by the Marine Technical Unit (UTM) of the Spanish National Research Council (CSIC; http://cndp.utm.csic.es/portal).
The fact that the IGN has assumed responsibility for volcanic monitoring on Deception Island marks a milestone in the management of safety in this Antarctic territory. The establishment of this volcanic monitoring network on Deception Island represents one of the most comprehensive and sustained efforts on the Antarctic continent, marking a fundamental advancement for polar research and operational safety in the region. The creation of this infrastructure sets an important precedent that could serve as a reference for the implementation of similar systems in other volcanic areas of Antarctica, promoting a greater understanding of volcanic hazards and improved preparedness for potential volcanic events in extreme conditions. Furthermore, the success of this network could encourage future international collaborations for the development of monitoring networks in other remote areas of the continent, thereby contributing to the strengthening of safety and scientific knowledge in an environment of global relevance.
Dedication
In memory of Andrés Barbosa and Mane Catalan, two great losses to the entire Antarctic family.
Author contributions
R. Abella participated in the conception and design of the project, in the deployment and configuration of field equipment, in the interpretation and analysis of results and in the writing and review of the article, after completing four campaigns. A. Fernández-García contributed to the conception and design of the project, the deployment and configuration of field equipment, the interpretation and analysis of results and the writing and review of the article, also after completing four campaigns. S. Blanca collaborated in the deployment and configuration of field equipment, completing two campaigns. E. Carmona participated in the deployment and configuration of field equipment, as well as in the writing and review of the article, and completed two campaigns. R. Martín was in charge of the deployment and configuration of field equipment, provided informatics support for the project and reviewed the article, having completed two campaigns. G. Sosa collaborated in the deployment and configuration of field equipment, participating in one campaign. G. Contreras completed one campaign and contributed to the deployment and configuration of field equipment, the interpretation and analysis of results and the writing and review of the article. V. Martín participated in one campaign, working on the deployment and configuration of field equipment. M. Abella, with one campaign, participated in the deployment and configuration of field equipment, in addition to collaborating on the writing and review of the article. R. Antón, without having participated in campaigns, contributed to the writing and review of the article, as well as to the interpretation and analysis of results. J. Barco provided informatics support for the project and was responsible for the configuration of communications. D. Minguez collaborated on the informatics support for the project, the writing and review of the article and the configuration of communications. M.V. Manzanedo participated in the interpretation and analysis of results, as well as in the writing and review of the article. D. Moure contributed to prototype development for the project, as well as participating in the writing and review of the article. M.F. de Villalta was responsible for the cartographic work of the project. H. Lamolda participated in the interpretation and analysis of results and in the review of the article. Finally, C. López contributed to the conception and design of the project.
Acknowledgements
The authors thank the Spanish Polar Committee for the trust placed in IGN, allowing us to carry out this work over the past 4 years. The authors are also grateful for the logistical and personnel support provided by the Spanish Army during the 2020–2021, 2021–2022, 2022–2023 and 2023–2024 campaigns at the BGdC, and for the technical and logistical support from UTM staff and the crews of the BIO Hespérides and Sarmiento de Gamboa research vessels. The authors also thank the University of Granada and the University of Cádiz for their foundational contributions and all colleagues at the BGdC. The authors thank Lupo for their commitment to our project and the SEGEMAR team for initiating a collaboration that will strengthen both groups. The authors acknowledge the use of ChatGPT by OpenAI (2024) to help with editing parts of this manuscript for language refinement. Lastly, the authors thank to Juan Rueda, whose efforts have been invaluable; without his work, this project would have been much more complicated.
Financial support
None.
Competing interests
The authors declare none.
Open access
