Abstract The ability to rapidly disseminate scientific knowledge across the world has never been greater. This is paralleled by the ability of rumors to spread equally far and fast. Popular misconceptions can take root and, without correction, take on the appearance of facts. Commonly misunderstood topics include those with attention-grabbing phrasing (e.g. “Pacific Ring of Fire”, “supervolcanoes”, or “mega-tsunamis”), the intersection of volcanic research with high-profile science topics (climate or human health), aspirational science goals (eruption forecasts), and basic terminology (volcanic ash versus smoke, or magma “chambers”). In this paper, we describe eighteen misconceptions grouped in six themes (seismology, volcanoes and climate, systems and structures, volcanic hazards, timescales, photographs and footage) commonly encountered by volcanologists who are actively engaging with the public and media both during and between eruptions. Each misconception description is accompanied by takeaway points and summaries of recent research to assist the reader with the information necessary to describe the issues. Finally, we present seven case studies where incorrect information was widely shared and counteracted at well-known volcanoes and recent eruptions at Mount Agung (Bali, Indonesia), Mount Etna (Sicily, Italy), Kīlauea (Hawaii, U.S.A.), Tajogaite (La Palma, Spain), La Soufrière (St. Vincent), Taal Volcano (Philippines), and Yellowstone caldera (U.S.A.). This work is intended to help both volcanologists and non-specialists get the story straight, and to effectively communicate processes and scales of volcanism and volcanic hazards, impacts, and risks.
‘Mountain Aglow’ is a co-produced disaster risk reduction initiative which incorporates the lived experience of Soufrière Hills Volcano into official channels of volcanic risk communication in Montserrat, Eastern Caribbean. By using a multi-media exhibit format, this project collates and celebrates the wealth of eruption-related experiential knowledge and artistic products – e.g. song, poetry, and stories – created by Montserratians as they lived through and coped with the upheaval of volcanic disaster. Important scientific knowledge pertaining to the volcano is intertwined with this experiential story of the eruption. This article first outlines the rationale and structure of the initiative, highlighting our methods of practicing co-production with local people and institutions. It then presents an empirical mixed methods evaluation of the project to examine key outcomes and the potential longer-term impacts of this process on disaster risk reduction on the island and reflect on the practical benefits and challenges of co-production. We show that our approach of blending experiential and scientific knowledge has had many positive outcomes and has served to: 1) improve the repertoire and reach of disaster risk education activities in Montserrat; 2) enhance engagement with young people in particular, and in turn improved their awareness and responsive capacity; and 3) enhance community awareness and capacity at large by encouraging cross-demographic/generational sharing of disaster experience. We also demonstrate that while the practical application of co-production comes with significant challenges, the benefits it brings can outweigh these difficulties.
Abstract The volcanic islands of the Aeolian chain (Italy) in southern Tyrrhenian Sea have been sources and receptors of local tsunamis triggered by landslides from the same islands several times during the last century. This study focuses on two of the Aeolian Islands, Vulcano and Lipari, to develop a first approximation evacuation model for cases where there is very little information on potential inundation zones, flooding depths and wave arrival times. To set up zones in need of evacuation, buildings and road characteristics were collected in the field as well as data for the spatial and temporal distribution of population and boats. Tsunami impact zones were determined from historical events which showed likely wave run-ups of one to ten meters and the source of the landslide triggering tsunami was set up on the flank of La Fossa volcano. In the absence of detailed run-up mapping or modelling, contour lines extracted from DEM were used for a first approximation of likely inundation zones. In the case of Vulcano, inundation zones of up to 270,000 m2 can be expected for even a small 2 m run-up scenario. In the tourist season, 2500 people can be present in the potential flooded area, especially on the beaches. Due to the proximity of these beaches to the north-eastern flank of La Fossa volcano landslide source (≤500 m), threshold times are 1–3 minutes. This means that for horizontal evacuation only 13% and 55% of the exposed area can be evacuated in 1 and 3 minutes respectively. However, if we consider using vertical evacuation involving roof access we increase to 19% and 71%. At Lipari, we have a case of the industrial port, so evacuation has to consider the heavy boat traffic which includes passenger boats, car ferries, tourist boats and cargo ships. Given typical boat speed and threshold time, even the slowest boats can reach a distance of 3–4 km offshore before the tsunami arrives, if reaction time is immediate. Due to the presence of hazardous materials at sea level we need also to consider industrial accident in our plan which can add 200–400 m to the impact area. While we provide a first approximation model for evacuation of flat areas and ports, this analysis highlights some issues related to harbours and inhabited areas located at sea level.
Abstract Volcanic ash is a significant hazard to aviation due to its potential to cause physical impacts and disrupt aviation networks through airspace and airport closures. Until now, ashfall impacts on airport operations have received much less attention than airborne ash impacts on aircraft and aviation. The time and resources taken to remove ashfall from runways and airport infrastructure can cause extended disruption to the aviation network beyond the presence of ash in the atmosphere. Using publicly available information, we have compiled a global dataset of 334 volcano-related airport impact events between 1944 and 2024. The dataset includes events at 141 unique airports in 44 different countries and territories, caused by eruptions from 73 different volcanoes. The countries with the most events are Indonesia (52), Mexico (27), and New Zealand (27), and the most impacted airports are Catania-Fontanarossa International Airport (20), Puebla International Airport (15), and La Nubia Airport (12). The volcanoes responsible for the most airport impact events (besides Eyjafjallajökull) are Etna (24), Soufriere Hills (22), and Popocatépetl (22). We used 74 data-rich, ashfall-induced events from this dataset to categorise ashfall impacts into five states of closure: <1 day, 1–2 days, > 2–7 days, > 1 week (finite), and permanent, and present fragility curves for each state. Data points are concentrated between 0.1- and 100-mm thickness such that fragility curve uncertainties are greatest at the smaller (< 0.1 mm) and larger (> 100 mm) thicknesses. The curves show that there is an > 80% probability of airport closure of any duration even for trace (defined here as 0.1 mm) amounts of ashfall. Closures of more than one day under trace ashfall are much more probable than closures lasting more than two days (30% vs. 6%, respectively), and once ashfall of 10 mm thickness is reached closures of more than two days are highly probable (> 85%). The development of continuous fragility curves from empirical airport closure data represents an advance that will prove useful for hazard management and long-term forecasting closure at airports. To illustrate the potential impacts of varying closure durations on flights and passengers, we provide a case study of the simulated closure of Ninoy Aquino International Airport in the Philippines. In addition to ashfall thickness, closure duration is affected by airport or national emergency management policies, resources available (personnel and equipment), and environmental conditions. This dataset and the derived curves provide a starting point and global evidence base for better understanding the impact of volcanic ashfall on airport operations.
Abstract Around 5,000 people were successfully evacuated by fishing fleet from the island of Heimaey to the Icelandic mainland within the first 15 h of the Eldfell eruption in 1973. Pets, livestock, fishing industry supplies, community services and personal belongings were left on Heimaey however, with many believing the evacuation only a precautionary exercise. As the severity of the eruption became clear, thoughts turned to the evacuation of the non-human, or more-than-human, but was stalled by the island’s location and poor planning by authorities. This paper uses interviews and archived historical documents to investigate the non-human evacuation from Heimaey and suggests combining More-Than-Human and Plural and Relational Values approaches to disaster management planning, to consider at-risk communities through a more holistic lens.
Abstract Volcanic gas emissions and fluxes are crucial inputs to hazard assessments and global volatile budgets. However, for many volcanic areas (such as the Auckland Volcanic Field [AVF], a distributed basaltic system that underlies New Zealand’s largest population centre) neither directly measured nor analogue gas emissions data are readily available. In lieu of measured gas emissions data, we apply the petrologic method, using data from crystal-hosted melt inclusions and volcanic glass to calculate volatile emissions for a compositionally representative set of five AVF eruptions. Our results indicate that emissions from small-volume eruptions can rival those from polygenetic eruptions. Extrapolating this data for all 53 eruptive centres suggests that the AVF has emitted ~26,000 kt CO2, ~9,000 kt ~SO2, ~470 kt HCl, and ~2220 kt HF over its ~200 ky eruptive history. We extend our analysis to develop an estimate of volatile emissions during open- and closed-system degassing and model daily fluxes for eight previously developed AVF eruption scenarios. Scenario unrest timelines were coupled with a thermodynamic magma degassing model, EVo, to estimate open-system fluxes. For closed-system fluxes, we distributed scenario emissions over a simple log-normal magma discharge curve defined by the scenario eruption durations and data from the literature. We find that maximum, time-averaged daily SO2 fluxes from distributed eruptions can be as high as those that have posed significant volcanic gas hazards to nearby populations, such as at La Palma in 2021. However, this is heavily dependent on scenario model inputs. Our results and the methodology may be used to estimate volcanic gas emissions and fluxes in other long dormant or unmonitored volcanic areas.
Abstract The city of Goma in the Democratic Republic of Congo faces significant volcanic hazards from Nyiragongo volcano, yet these dangers are largely overlooked in urban planning and land use strategies. Effective evacuation plans for potential volcanic disasters depend on a thorough understanding of the geospatial characteristics of the evacuation environment, which can change over time. This study assesses risks associated with volcanic hazards in Goma, focusing on exposure to lava flows and the dangers of CO2-rich gas emission areas known locally as Mazuku. Using QGIS and Q-LavHA tools, mapping and simulations were conducted based on historical eruption data from 1977, 2002, and 2021 to model lava flow paths towards Goma from eight eruption vents. The findings indicate that over half of Goma is at high risk from volcanic hazards, particularly in areas where the majority of the population and critical infrastructure are located. To enhance community resilience in line with the eleventh Sustainable Development Goal (SDG 11), this paper proposes evacuation scenarios and mitigation strategies based on a detailed hazard and exposure assessment. The result of this study shows that eleven neighborhoods in the east of Goma city are located in the potential lava flow corridor and that roads, lake ports and part of the airport are potentially exposed to volcanic hazards. This study contributes to global methodologies for assessing volcanic hazard exposure in data poor regions. In addition, the proposed geospatial environment model is crucial for developing agent-based simulation studies to improve evacuation effectiveness during volcanic events.
Abstract Indonesia, home to some of the world’s most active volcanoes, faces a challenge in balancing tourism growth with disaster risk mitigation. This study explores the critical role of communication in reducing disaster risks in volcanic tourist destinations, focusing on Mount Merapi in Yogyakarta and Mount Agung in Bali. A qualitative research approach with a comparative case study design was employed. Data were collected through in-depth interviews with disaster management authorities, tourism agencies, community leaders, tourism village managers, and tourism operators in both study locations. Additional data were obtained from policy documents, disaster communication guidelines, and official reports related to volcanic risk management and tourism. Thematic analysis was applied to identify patterns of communication practices, stakeholder relationships, and coordination dynamics. Data triangulation across sources and document analysis was conducted to enhance the validity and reliability of the findings. The findings reveal that disaster risk communication in volcanic tourism destinations remains fragmented, particularly between disaster management institutions and tourism actors. While culturally embedded communication practices and community-based mechanisms effectively enhance preparedness among local residents, they are not systematically translated into communication formats accessible to tourists. Consequently, tourism actors often assume informal intermediary roles in conveying risk information to visitors without sufficient institutional support. This structural gap generates uncertainty during volcanic crises and poses challenges for tourist safety and destination trust. This study recommends the integration of tourism stakeholders and culturally grounded communication practices into formal disaster risk communication frameworks. By positioning disaster risk communication at the intersection of disaster governance, tourism governance, and local cultural systems, the study offers an original empirical and conceptual contribution to the literature on disaster risk reduction and sustainable tourism. The findings provide evidence-based insights to support the development of more inclusive, context-sensitive, and sustainable risk communication strategies for community-based tourism destinations in volcanic regions.
The Chiles-Cerro Negro volcanic complex (CCNVC), located on the Ecuador-Colombia border, has exhibited persistent seismic unrest since 2014. This activity has significantly impacted the nearby community of Tufiño, increasing both concern and interest in volcanic hazards. In response, the Instituto Geofísico de la Escuela Politécnica Nacional (IG-EPN) developed a participatory monitoring and risk communication strategy inspired by the successful Vigías (community volcano observers, or “watchmen”) program implemented during Tungurahua’s prolonged eruption (1999 – 2018). Motivated by the period of increased seismicity at Chiles–Cerro Negro, this study documents the decade-long (2014–2024) co-creation of a Vigía network in Tufiño, examining how collaborative engagement between scientists, local authorities, and community members enhances volcanic risk preparedness. Through iterative workshops, interviews, and surveys, the program trained residents to systematically record and report volcanic phenomena while integrating local knowledge with technical monitoring. The Vigías became vital actors in early warnings, fostering trust and bidirectional communication between stakeholders. To understand the long-term impact of the Vigias network, we evaluated with specific interventions in 2015 and 2024. Key findings demonstrate that this participatory approach strengthened community resilience by: (1) improving hazard awareness and response capacity, (2) establishing reliable communication channels between scientists and at-risk populations, and (3) creating a sustainable framework for citizen science in volcanic monitoring. This article not only recounts the activities undertaken but also serves as an evaluation of the entire process. Despite its successes, the initiative faces challenges, notably maintaining volunteer engagement in the absence of an eruption. Nevertheless, the Vigías network remains an active component of the CCNVC communication strategy, underscoring the importance of sustained cooperation between scientific institutions, local authorities, and communities. The CCNVC case study provides a transferable model for volcanic risk reduction, emphasizing that community-scientist partnerships are not merely supplementary but foundational to effective disaster risk management. This partnership approach offers a replicable model for fostering resilience and preparedness in regions exposed to volcanic hazards. Lessons from this initiative highlight the critical role of local knowledge, the importance of institutional commitment, and the value of pre-crisis relationship-building in volcanic regions worldwide.
Research into Eldfell 1973 typically favour scientific studies, with human aspects of the eruption overlooked. The first night, and the subsequent evacuation to the mainland are particularly neglected, their details often incorporated into the larger overall story of the eruption. This research uses a “Small Stories” narrative approach to reconstruct the events of the first night of the Eldfell eruption. Archived documents, written personal accounts and interviews highlight the communication and planning involved in the evacuation, the lived experiences of individuals during the event. Understanding the community response to the eruption supports present and future generations of islanders, scientists, and emergency responders to understand and prepare the community for future eruptions on the island and further afield. We answer the call to better utilise historical documents in volcanological studies but also pose our own call to action that hazard scientists establish and maintain long-term meaningful relationships with at risk communities, even during times of quiescence.
Abstract Pyroclastic density currents represent one of the deadliest hazards posed by active volcanoes. Analysis of their deposits provides valuable insights into their internal dynamics and informs numerical simulations of pyroclastic density currents which underpin many volcanic hazard assessments. We present PDCD-DAT, a global database of pyroclastic density current deposit characteristics compiled from peer-reviewed literature. The database includes both quantitative datasets (e.g., grain size, density, bedform dimensions, thickness) and qualitative descriptors (e.g., sedimentary structures, lithofacies). PDCD-DAT includes data from 85 source publications, covering 97 eruptions or eruptive phases, and 214 individual depositional units from 55 globally distributed volcanoes. Eruptions recorded in the database range from VEI 1–8. We highlight examples of potential applications of the database, which include (i) comparison of single deposit case studies to global datasets, (ii) informing input parameters and conditions for numerical and analogue models of pyroclastic density currents, (iii) validation of numerical and analogue models against a wide variety of natural case study deposit architectures, and (iv) estimating hazard impact metrics of pyroclastic density currents from past eruptions. We show that the database represents a useful tool for improving our ability to model pyroclastic density currents, predict their associated hazards, and understand the relationships between the internal dynamics of pyroclastic density currents and the properties of their deposits. PDCD-DAT is integrated with the FlowDat Mass Flow Database, which provides a sustainable platform for the database. We aim for PDCD-DAT to be expanded in the future through addition of pyroclastic density current deposit datasets from new field studies conducted by the volcanology research community.
Modern operational eruption forecasting methods rely heavily on human judgment in the face of uncertainty and are thus susceptible to myriad cognitive biases and errors by the scientist-forecasters. Recent developments in the behavioral sciences have elucidated cognitive biases across a wide spectrum of human behaviors and found ways to mitigate them. These insights have led to significant gains in human forecast accuracy across a range of disciplines. However, such gains have yet to widely penetrate volcanic eruption forecasting efforts. In this study, we review recent progress in these fields as relevant to improving current eruption forecasting methods as practiced by volcano observatories worldwide. We group cognitive biases into 1) information and selection biases, 2) group effects and social-emotional biases, and 3) framing and decision biases and highlight the numerous places in which these biases may permeate eruption forecasts. We present a framework for improving group discussion and forecasting processes at volcano observatories, considering recent gains from behavioral sciences along with a mitigation checklist to effectively reduce bias in operational eruption forecasts. Finally, we present an updated forecasting methodology for use in our own group, the US Geological Survey (USGS) Volcano Disaster Assistance Program, based on previous “multiple datasets” methods, that includes clear and deliberate efforts to minimize cognitive biases and thus improve eruption forecasting accuracy.
Monitoring changes of geochemical parameters in the subsurface around volcanoes is crucial for hazard assessment and early warning. This study establishes a baseline of soil CO _2 emission from its flanks of the Mount Scenery stratovolcano, on Saba in the Caribbean Netherlands, during its period of quiescence. The soil CO _2 flux and carbon isotope variation were mapped across the island to find patterns and locations of higher flux. Additionally, soil gas was analyzed for helium isotopes at ‘Green Gut’, a location which has an anomalous soil temperature. Results reveal spatial variations in temperature and soil CO _2 fluxes across the island, with higher temperature and fluxes observed along a transect from the southwest to the island’s north side. The anomalous sites identified from soil CO _2 flux and carbon isotopes measurements align with known alteration and temperature anomalies, including an abandoned mine in The Bottom and cracks along the road in ‘Green Gut’. The identified transect partially overlaps with a previously proposed ancient sector collapse, but our results suggest an extension northward. Integrated soil CO _2 flux and carbon isotope data across the island indicate that the emitted gas is a mixture of biogenic, atmospheric, and hydrothermal-magmatic sources.
Scoria cones are the most common volcanic landform on Earth. While eruptive styles range from effusive to explosive, the low-energy spectrum of activity is best documented. Hazards from these styles of eruption, including effusive and Strombolian styles as well as fountaining, are typically restricted to a relatively small area about the cone. However, it is the more explosive, violent Strombolian to sub-Plinian styles of eruption that result in sustained ash columns capable of disrupting aviation and distributing ash on the ground over larger regions. To increase our understanding of these higher-explosivity events, we must come to a better understanding of how commonly they occur, both temporally and geographically. Given that the scoria cone edifice itself is the longest-lasting remnant of any scoria cone eruption, it is important to use existing technologies to examine cone formation processes for evidence of sustained ash columns (indicative of regional hazards) within the geologic record. To this end, ground penetrating radar (GPR) techniques were employed at Crater Flat volcanic field, NV. While radar has previously been shown to be a valuable tool on fresh ( < 150 year old) scoria deposits, this work uses the technique on older ( > 100 ka) cones. Results indicate complex eruptive and erosive histories, and clearly indicate the utility of the GPR method in elucidating the stories of older cones and contributing to the growing body of knowledge concerned with high energy scoria cone volcanism.
Volcanic eruptions produce plumes of ash, gas and aerosols that present a risk to aviation at all standard flight levels. Here, we investigate atmospheric dispersal of volcanic emissions, whether and how they infiltrate aircraft, and whether ground-level public health exposure thresholds can be related to the pressurised cabin environment. We then review the limited evidence for physical and mental health, and behavioural impacts, resulting from volcanic emissions entering aircraft. Serious health risks are considered low for healthy individuals, but respiratory irritation is likely for a high exposure scenario to sulfur dioxide (SO2). Asthmatics are particularly sensitive to SO2, with even relatively low, short exposures, potentially resulting in severe respiratory impacts. Negative group behaviours are not expected but individual distress is possible. Communicating this evidence to the aviation industry may result in more informed decision-making on flightpath alterations and triggering of emergency protocols, both before and during volcanic emission encounters.
The U.S. Geological Survey’s Hawaiian Volcano Observatory (HVO) has developed a new method to continuously monitor lava lake elevations. Since 2018, HVO has stationed a laser rangefinder on Kīlauea’s caldera rim. The instrument automatically measures lava lake elevation each second, with centimeter accuracy. A stream of elevation data flows to HVO’s database and public website, contributing a valuable channel to HVO’s volcano monitoring network. The data display is intuitive for users, providing essential information with a new level of clarity. HVO has used this method to track Kīlauea’s changing lava lake elevations over a series of eruptions, and the time series data show several volcanic processes: crater refilling, gas pistoning, lava lake surface behavior, and endogenous crater floor uplift. This technique is versatile, nimble, and easy to use. Continuous laser rangefinders may also prove useful for tracking lava lakes elsewhere, and for monitoring other hazards such as growing lava domes and debris flows.
The dynamics of Earth’s largest, most voluminous eruptions (≥ 100 km3 ejecta; VEI 7) are poorly understood. Here, we explore the question of whether these very large eruptions can be treated as scaled-up versions of moderate volume historical eruptions (0.1–10 km3 ejecta; VEI 4–6), or whether they should be treated as fundamentally different Earth system phenomena. To examine this, we compile fall deposit and ignimbrite volume data for 74 explosive eruptions worldwide that are magnitude 4 or greater, and use this dataset to assess how material is partitioned into buoyant plumes versus pyroclastic density currents as a function of eruption magnitude. Importantly, we filter our results by overall distance from seas/oceans, so that we can focus in on the eruptions for which preservation of the deposits is as reliable as possible. After filtering, we find that the largest eruptions are dominated by ignimbrites and not fall deposits, implying that, co-ignimbrite plumes notwithstanding, there may be little or no buoyant plume component to eruptions of the highest known magnitudes. This result is consistent with model simulations showing that the pyroclastic materials produced during larger events can be emplaced in density currents alone, and highlights important considerations for contemporary eruption simulations, the fate of volcanic gases relative to solid mass, and subsequent appraisals of the climatological and environmental impacts of explosive volcanism on Earth.
Volcanic eruptions can inject ash into the atmosphere, which is then advected by meteorological winds, potentially affecting large volumes of airspace. Volcanic Ash Advisory Centres (VAACs) issue volcanic ash advisories (VAAs) when airspace is likely to contain ash above a concentration threshold. Much research has been done to improve operational ash forecasts of volcanic ash location in the atmosphere, but until now the paths of aircraft around erupting volcanoes and when VAAs have been issued, and the impact these routes have on flight schedules and diverted aircraft’s fuel consumption, have not been closely examined. Here, we investigate the behaviour of commercial aircraft during times of volcanic ash emissions as reported in VAAs. We use publicly available flight trajectory data during several ash-rich eruptions at Etna, Sakurajima, Marapi, Sheveluch, Klyuchevskoy and Ubinas volcanoes in 2022 and 2023. We examine a range of geographic locations and eruption sizes. Flight trajectories during periods when VAAs were issued are compared with flight trajectories during periods when no VAAs were issued. We find that the aircraft largely avoided the air space shown to be affected by ash by VAAs, indicating that they adopt a range of strategies to avoid ash. We also find that, in general, by avoiding ash aircraft also avoided volcanic SO2 plumes. Our results confirm that the greater the volume of airspace affected by volcanic ash the greater the deviation of aircraft from their usual flight paths. Rerouted aircraft may travel significantly further distances to avoid ash, which results in longer air travel time and delays, suggesting greater fuel consumption and carbon emissions. Further long-term systematic studies of the impact of volcanic eruptions on flight routes and timing would help to characterise ash-related aircraft disruption over time. Air traffic is likely to grow in the coming years and VAAC advisory strategies will also evolve so understanding how such changes affect disruption trends may be useful.
Although disasters disproportionately affect vulnerable people, their experiences often go untold. Disaster researchers increasingly use participatory research methods to recognize and amplify these missing voices. Arts-based participatory methods promise both empowerment for local people and opportunity for researchers to work creatively and reflexively. They also present challenges of participation and representation. This paper describes the process of co-producing a “zine” describing local people’s experiences of eruptions of Fuego volcano (Guatemala) and the impacts of these eruptions that caused disaster. The zine is narrated by local people and is illustrated by the researcher (me) who conducted semi-structured interviews of residents to better understand their experiences. The motivations for this project are explored together with lessons learned in turning participatory research design into practice. Taking a constructively critical perspective on my process, I document challenges I encountered in community participation and representation and discuss how they may be addressed in project design and execution. I also describe the tensions and value in occupying both researcher and artist roles. This paper describes an open method of research exploration where the opportunities and limitations of visual representation to contribute to shared and widening understandings of volcanic disaster experience is documented and reflected upon.
All historical eruptions at Ruapehu have occurred from its Crater Lake, Te Wai ā-moe. This study aims to better understand Crater Lake dynamics by using visible light and long wavelength infrared images of the lake. Over 10,000 images from 1902 – 2021 were analysed to produce a time-series of lake observations. Our results show that visible light observations reveal colour changes on the entire Crater Lake surface from blue to grey, and localised grey, yellow, and black discolourations. Grey discolourations are interpreted as localised upwellings of lake-floor sediment, and yellow and black material to comprise vent-hosted sulphur/sulphides, both transported by volcanic fluids from subaqueous vents to the surface. The locations of upwellings were used to identify five vent locations beneath Crater Lake, three more vents than were previously recognised. Upwellings appeared and disappeared in 10 min. Steam above the lake surface was controlled by both lake temperature and cloud conditions. Blue lakes were most common in summer and autumn, implying a relationship with ice or snow melt entering the lake. Grey lakes were observed in the month before 97
