Consistency is frequently identified as a key benefit of standardised hazard and risk communications. A narrow focus on consistency, however, fails to account for transactional and other factors contributing to the effective use of VALS in practice. VALS works well in operation, with all observatories using it to relay the status of volcanic activity. These practices rely heavily, however, on prior experience to familiarise all involved with the meaning of different VALS alert levels. Similarly, the limitations on the capacity of VALS to provide useful and usable information are mitigated by the largely multi-valent communication techniques and cross-boundary networks that have developed between scientists and VALS users. To appreciate and understand the workings of the standardised VALS model, in other words, requires that it be recognised as context dependent, relying heavily on everyday practice. That is, the new standardised VALS ‘works’ precisely because it continues to afford scientists the flexibility to conduct their usual communication practices in order to translate scientific information into a form that is accessible to bureaucratic rationalities.
Scientists and users commonly regard VALS as a catalyst that initiates the communication (both uni- and multi-valent) required to facilitate required discussions and operational decisions. As one scientist commented:
[An] alert level system is a shorthand, is the vehicle, it is the excuse to get into communications and dialogue, that gives you a justification and purpose […] that provides you the entry into having a discussion with very busy people who are otherwise occupied with other duties they have (VHP manager 4).
In practice, a VALS is a communication initiation tool, an instrument to develop coordination plans and to provide general awareness about the state of the volcano, rather than about a specific hazard. If this communication occurs regularly, then it may actually be surplus to requirements. That is, VALS can appear overly complicated given that the concept is simply to gain attention to an event and its anticipated impacts, and valuable time can be spent on deciding alert levels that might better be used to initiate the necessary communication to provide scientific information. It is through multi-valent communication outside of the VALS that producers and consumers can establish meaningful interpretations of warnings, even if they are based in different contexts.
Using Cash et al.’s (2003) model, we break down the various processes and issues that make the VALS and other communication tools work in practice by classifying them under (i) the need to ensure they are scientifically robust, driven by the scientific credibility requirement; (ii) the need to generate salient knowledge relevant to the needs of decision makers; and (iii) to need to ensure that (i) and (ii) are balanced, in order for both processes and information to be perceived as legitimate. This classification is summarised in Fig. 5.
To be successful, a VALS needs to meet scientific credibility criteria. However, there are a number of challenges facing scientists responsible for ensuring the accuracy of the warnings they provide, many associated with negotiating the uncertainty of the phenomena and of the monitoring data. In order to provide timely volcanic hazard warnings to communicate to the users, it is important for the scientists monitoring the volcano to accurately interpret scientific data, provide the best information about current activity and if required, generate reasonable forecasts for potential hazards; this is addressed fully in Fearnley (2013). Before scientists discuss what alert level volcanic activity should be assigned, there is a rigorous process of establishing exactly what is going on at the volcano. This process is often dependent upon the monitoring capabilities of each observatory to provide scientific data, the technology and staff to interpret this data and the need to form a consensus about what the volcano is doing. Whilst global databases such as WOVOdat and GVM bring together global expertise and comparative examples, each volcano has its own ‘personality’ (CVO senior scientist 7). Therefore, understanding a volcano is ‘part science, partly an art’ (CVO senior scientist 7), since volcanoes can behave in unexpected ways, and recognising patterns of behaviour for a particular volcano is critical to understand what that volcano is doing and to generate accurate forecasts. The difficulties of establishing a consensus concerning volcanic activity are further exacerbated by the often-short timeframes in which scientific evaluations need to be conducted during a crisis and by the fact that these evaluations often have a shared, rather than individual value. Although procedural protocols can contribute to the credibility of scientific evaluations, it is also important to recognise the subjective components in these evaluations.
Further complications arise when the scientists review possible forecasts. Forecasts are highly relevant to users. Forecasting volcanic behaviour involves much greater uncertainty than simply determining volcanic activity, since given volcanoes can sustain unrest or eruptions for long durations. These uncertainties make assigning a volcano alert level and deciding when to ramp-up or reduce the levels a difficult, highly complex and concerning process, with real-world consequences. Essentially, this ‘forces the scientist to think about the alert levels rather than the science’ (HVO senior scientist 5), driving a process in which scientists rank the possibilities of what is likely to happen and then release this information, rather than conducting in-depth scientific discussions. We concur with Papale’s (2017) concerns about the position this puts scientists in. We differ from him, however, in that we understand this position to be created by the use of a linear process to try to manage a highly complex situation, and decision-making processes.
For VALS to be effective, assessments conducted by scientists must be relevant to the needs of the key decision makers. The relevance requirement has been found to drive associated demands for timeliness and for simple accessible alert information (Sarkki et al. 2013; Parker and Crona 2012). With reference to VALS, this includes demand for timely simple and accessible alert information, that is usable subject to a range of contingent factors.
When providing warning information, the key issue for end-users is to provide timely information. During times of non-crisis, those involved can spend time deliberating plans and protocols. During a crisis, information is required quickly, regardless of scientific uncertainties, and with guidance on what the information means and how to act upon it. Scientists can be reluctant to disseminate guidance due to scientific uncertainty, since it is often impossible to accurately predict what a volcano is going to do until days/hours/minutes before it happens. In addition to concerns about the credibility of the information, scientists are also concerned about the legal context of such warnings. Numerous users reported in interviews that volcano observatories are discouraged from issuing alerts until there is greater certainty, because a VALS is a legal ‘formal warning’ under the USGS mandate provided by the USA Stafford Act. This places pressure on the scientists to get the decision ‘right’ before issuing alerts (AVO scientist 3).
Users however are much more concerned with the relevance of information. It was common for users to state that it is better to communicate what is known to the users irrespective of how certain or clear scientists were about the situation. For users, it is much better to say something albeit uncertain rather than ‘nothing at all’ (LVO scientist 2). A user in the Cascades, for example, expressed frustration with the time needed to debate and initiate a VALS and accompanying information statement,
Basically, after the action in 2004, I said I thought that it was dangerous actually; that they got state emergency managers and people like ‘x’ just sitting around deciding what the words [for the information statement] are going to say, and I said ‘you know we need to call ‘y’, and let them know. I don’t care what it [information statement] says we just need to know that something has changed’. To be sitting worrying about what the three sentences are is silly. […] There is this tension between wanting to have everything be just right and needing to get the word out (CVO user—USFS).
Another lesson here is that informal methods of communication such as telephone calls were a valuable means of facilitating timely interaction and the translation of scientific uncertainty one on one to end-users, since they did not technically involve issuing official warning information.
Establishing meaning in alert levels
The explicit aim of introducing standardised VALS, as noted above, has been not only to ‘fix’ the meaning of the information they convey but also to fix the meaning of a VALS itself. Such efforts are complicated by the fact that each VALS is understood in relation to prior learning experiences, that ‘school’ users to respond in particular embodied ways to the medium provided. One such example, recalled at both AVO and at HVO as illustrative of the role of prior experience, concerned a commercial Alaskan pilot flying from Alaska to Hawaii. The pilot, used to flying in Alaska and dealing with the aviation colour code frequently in place there, was concerned that the Kilauea volcano on the island of Hawaii was assigned an orange alert level. Based on his experience of warnings issued in Alaska, he anticipated that the volcano would be exhibiting unrest with increased potential for eruption with ash. When the pilot arrived in Hawaiian airspace, he expected some form of diversion or information (such as a Volcanic Ash Advisory) regarding Kilauea, but received nothing and landed with no problems. He later discovered that Kilauea was erupting, but emitting such a small ash plume that low-level flying was only prohibited within close proximity of the volcano.
Generating a meaningful warning or ensuring that information that is used as intended is challenging for a number of reasons. The ambiguity of alert levels is acknowledged. In particular, the orange/watch alert level presents a problem, because it involves two descriptions for the VALS, with one predicting imminent eruption and the other describing an already occurring but non-threatening eruption. This is a specific issue demonstrating the importance of VALS design and criteria. There is general concern that an alert level can generate complacency. If at a single status other than green/normal for too long, the alert level loses meaning and impact, as seen with the Homeland Security Terror Alert System which has been historically disregarded due to perceived lack of efficacy, often staying at one alert level for years rendering it meaningless (CVO user—emergency manager 2). For places like Hawaii, constant eruption means alert levels remain at orange/watch. For users, this can become the ‘status quo’, prompting them to think ‘I do not need to be worried about it’. In the case of pilots, this kind of thinking can put the lives of passengers in danger if the volcano suddenly erupts, as per the orange/watch alert level (AVO user—FAA).
Legitimacy relies on the perception that the processes through which information has been generated and disseminated have included and balanced the interests and knowledge of all involved (Cash et al. 2003; Sarkki et al. 2013). This can be difficult to do using VALS, since they are not a static tool but rather change in purpose as the volcano ramps up and down from an eruption, and during various phases of eruptive activity. VALS have in effect been used to encompass at least two roles, driven by demand from users, consisting of a forecasting tool with warnings and a reportage tool describing what was happening at the volcano, such as an eruption, also seen in Quito, Ecuador, in 1998 (Metzger et al. 1999). This combination of roles can help to make a range of users become more expert, increasing understanding of the history of the relevant volcano and awareness of what to watch out for, so that once an alert level is issued they know how to respond (AVO user—NWS 2).
The overriding consensus among all interviewees as to what VALS are intended to accomplish was that VALS were an awareness-raising medium, or ‘flag’, providing a “heads up, pay attention, something has changed, you need to look at reports, updates for information statements, or listen to advice from federal agencies” as one interviewee put it (CVO Senior Scientist 2). This applied across the interface between communities of experts and of end-users. The VALS helped to “ramp up situational awareness” among users in that it “dictates and drives our situation awareness and staffing” (AVO user—NWS 3), and provided an equivalent “flag for users for how often the users should be looking for information” (CVO scientist 12). Interviews suggested that scientists similarly responded to VALS as an awareness raising mechanism, since the rate at which they “process and disseminate information is dependent on the alert level” (CVO scientist 12), implying that at red alert, information statements would be issued faster than at yellow alert.
Expectations on the role of the VALS
Defining a VALS is critical to establish whether or not it serves its purpose and to determine whether stakeholders share a common understanding of what the alert levels are. The overriding consensus among the expert community was that VALS were used to alert users, communities, and individuals to the state of the volcano in a simple and concise message that allowed them to gear an appropriate response (AVO senior scientist 1/HVO scientist 1). Scientists regarded VALS as a tool to translate scientific assessment information about the nature of volcanic unrest and possible hazards quickly to non-scientists, excluding technical details, so that the use of one word, such as ‘Red’, would allow a large and diverse range of people to know what the conditions are (HVO scientist 2). At the same time, it was also widely accepted that an alert level alone cannot provide all the information required for users to make decisions:
[The] complicated reduction of all of these factors (risk, hazard, activity) and boiling that down to a simple number [means] inevitably if you do that, something is going to be lost. You can’t just project a ten-dimensional problem down to one dimension and expect it to retain all its complexity (AVO scientist 4).
How an alert level is defined and what it means depends on a balance between the meaning conveyed by the scientists and that understood by the users, which relies on their institutional requirements for the alert. Interviews with users suggested that they regarded the VALS as a scale to determine the importance and relevance of the information being distributed. This is at odds with the understanding of those issuing the VALS, who regarded it as a concise indication of the science—which is to say the eruptive activity of the volcano. Users tended to work in busy government agencies that were often already overwhelmed with other duties. This made establishing the urgency of warning information a key priority, since they needed to determine “where we are in terms of imminent danger” (LVO user—Mammoth Lakes town 2). The VALS helped to “ramp up situational awareness” and “dictates and drives our situation awareness and staffing” (AVO user—NWS 3). Consequently, although VALS provided information about the physical hazard, they were primarily useful and used for planning purposes. Some senior scientists felt that VALS were in effect a distraction from the need for more communication between scientists and non-scientists:
I think the whole alert level thing is […] an attempt to better communicate with the public, media [and] help scientists convey the message. Most people put too much emphasis on that and not enough with the basic problem, which is communication between scientists and non-scientists (HVO senior scientists 4).
User groups tended to agree with this assessment, but from the other direction. Those in charge of making decisions about people’s safety had difficult problems to deal with like “should I evacuate or not?” (LVO user—emergency manager 1), “where are people going to go/live?” and “will people actually pay attention or ignore the warnings?” (HVO user—emergency manager). Using the VALS alone cannot facilitate discussion of these real-world issues. Where scientists focused on the need to translate scientific information to users, emergency managers expressed the need for scientists to develop a better understanding of the problems they face in order to help with their decision-making processes, and limitations in knowledge (Fearnley 2013).
The role of contingency
Every volcano has a diverse range of hazards in different spatial and temporal combinations, making the individual behaviour of each unique. This can make understanding the activity and issuing a warning for a volcano alert a highly complex and context-specific process. Many hazards can occur within close proximity of a volcano, whether it is active or not, in different locations (geographically), and at different times. Most are excluded from the VALS, which relates only to the occurrence of volcanic/eruptive unrest/activity, and must apply to every volcano. Many scientists stated that VALS should convey information about all volcanic hazards, whether they proximal to the volcano, i.e. volcano-centric, or distal. Some expressed the view that a warning can only be truly issued after the event has begun (CVO collaborator 2), which means that the only way to measure if a lahar has developed, or where an ash cloud is moving, is to monitor them individually. A number of the observatories have developed independent alert level systems tailored to the nature of a range of these hazards, including volcanic gases (in particular seen at HVO), lahars (CVO), volcanic ash clouds, volcanic ashfall (AVO) and hydrothermal activity (YVO). The unique individual behaviour of a volcano, each with differing hazards in differing spatial and temporal relations makes monitoring, understanding the activity and issuing a warning for a volcano alert highly complex processes.
In addition, contingent institutional factors need consideration as they are shaped by local cultural, political and judicial systems. The dynamics of USGS policy, governance and operations had a profound effect on the resources required to provide an effective VALS, such as funding for monitoring capabilities, staff resources and protocols for issuing warnings. Education and outreach were essential activities to ensure that stakeholders are aware of VALS and how they work, but these required significant staff time and resources. In addition, each user group will also have their specific institutional and legal remits, factors and limitations to consider. Inevitably these factors, along with securitisation, influence how communities will respond to a warning.
The diverse range of contingencies arising out of the institutional and geophysical context associated with any given volcano further underlines the importance of including VALS as one among a suite of boundary objects used to facilitate the co-production of knowledge and decision-making about volcanic risk. To make sure that the decoding of a warning was accurate and meaningful, all the actors involved worked very hard external to the VALS. This was done using other boundary objects, including multi-valent communication products (via protocols) and collaboratively developed coordination plans and meetings that gave user groups opportunities to discuss what the hazards and potential risks were, and to prepare by co-producing response plans. This open, multi-valent communication allowed mutual understandings of events to emerge, helping to bring together knowledge and expertise from both sides of the science and decision-maker interface. It was this process that over time helped all those involved (including the observatory scientists) to recognise and understand each other’s needs and concerns, thereby incorporating local context into the dialogue process. Such a dialogue depended, however, on preparatory work in order to make the communication network ‘work’.
These communication techniques function differently from VALS. They fostered a sense of trust based on dialogue, for example, rather than implying the top-down authority created by the uni-valency of VALS. A number of users expressed the view that trying to get “facts out of scientists” was difficult, but by “building trust ahead of time”, it was possible to trust each other and understand each other’s limitations, despite institutionally “different cultures” (CVO scientist 5). This was also the conclusion of Peterson (1988, p. 1467) who pointed out after interviewing journalists and scientists following the 1980 Mt. St. Helen eruptions that journalists found that “scientists are too long-winded; they talk all around the subject and never get to the point; they do not understand that we need to use straightforward, simple statements; we have to convert the complicated discourses to words that people can read” (after Peterson 1988, p. 4167).
The development of coordination plans—as implemented in Alaska, the Cascades, Mammoth Lakes and in Hawaii (Hill et al. 2002, 2017; Madden et al. 2008)—was key to this process. The plans were drawn up to provide background information about the volcano, its history and potential hazards, the different land owners, stakeholders and federal or state agencies involved with the land, and the plan for a crisis. It is through meetings, developing preparedness plans and reconvening frequently that the actors get to know one another and create the potential to rapidly generate “situational awareness” between scientists and users (CVO scientist 5). This enabled scientists to communicate issues around uncertainty, for example, so that “they [users] have some sense of our level of anxiety on escalating unrest on a volcano so they can understand the alert levels better” (LVO senior scientist 1). Furthermore, coordination meetings provided the opportunity to clarify the nuances of the terms as understood by the users and scientists, so each group was confident about what each alert level is likely to mean to each user for that specific volcano.