4.6 Theme 5: Possibilities for future VAL systems
4.6.1 A shifting of foundations
New Zealand’s VAL system is predominantly based on currently occurring phenomena, such as ‘apparent seismic, geodetic, thermal or other unrest indicators’ and ‘minor eruptions
commenced’. It also acknowledges hazards and risk, such as ‘hazardous local eruption in progress’ and ‘significant risk over wider areas’ (although risk in this context was undefined). This section discusses five potential foundations underpinning future VAL systems. Some were suggested directly by participants (and have been member-checked), while others are the result of my interpretation of interview and observation data. The strengths and weaknesses of each system are described, along with an example of a system based on each of the foundations. The ‘foundation’ of the VAL system is essentially the theme used to divide the levels.
4.6.1.1 Phenomena-based
The current VAL system’s foundation is based on volcanic phenomena largely due to the era in which the system was formed, according to one of the participants involved in its creation. The focus on hazards, risks, and social impacts by volcanologists that can be observed currently was not emphasised in the early to mid-1990’s, and volcano-specific models were few and far
Chapter 4 An exploratory investigation of New Zealand’s VAL system
152
between, at least in New Zealand. Volcano scientists at various branches of DSIR had only recently been combined into one organisation (now known as GNS Science), with a less cohesive nature of working as a team than is apparent now. All-in-all, basing the VAL system on currently observable phenomena was the most “comfortable” foundation for a scientific group not yet confident in eruption forecasting. Aspects of volcanology and scientific knowledge have developed substantially in the 20 years since the VAL system’s formation, along with a paradigm shift of acknowledging societal needs in the communication of scientific information. These developments prompt the need to carefully consider whether the
foundation of current phenomena, containing little to no eruption forecasting or hazard information, is still adequate – some participants think it is not (although it may still be the most “comfortable” foundation).
The benefits of retaining the phenomena basis to the system were identified as including a lower level of uncertainty in communicating physical monitoring data than in communicating either hazard or risk information that is associated with societal systems, or underlying
magmatic models. The phenomena-based system is thought to be "the system that is truest to the science and conveys what the volcanoes are doing without added layers of interpretation" (Sc23).
Another benefit is that descriptions of the physical phenomena can be seen as the first step in the communication process, prior to their interpretation and relationship to hazards and possible future activity. By communicating this first step, the subsequent interpretation, hazards, and future activity information can be tailored to suit the wide range of end-users, volcanic environments, and situations. Additionally, by basing the VAL descriptions directly on the observable phenomena, the opportunity for subjectivity to influence the VAL decision is minimised, and the time it takes to determine the VAL may be shorter.
An example of a phenomena-based VAL system is presented in Table 4.4. An additional column relating to (simplified) indicative phenomena may be added, and there are six levels included in this example in accordance with the end-users’ wishes.
Table 4.4. Hypothetical VAL system for New Zealand based on a foundation of currently occurring volcanic phenomena. This example demonstrates the use of a numerical system, from low (at the top) to high (at the bottom).
Hypothetical Phenomena-Based Volcanic Alert Level System
Volcanic
Alert Level
Description of volcanic activity
0
No volcanic unrest1
Minor volcanic unrest2
Moderate to heightened level ofvolcanic unrest
3
Minor volcanic eruptionhas recently occurred or is in progress
4
Moderate volcanic eruptionhas recently occurred or is in progress
5
Large volcanic eruptionhas recently occurred or is in progress
Potential issues with retaining the phenomena foundation as identified by interview participants are:
1) some end-users find it difficult to interpret the current phenomena into meaningful information for hazard planning and decision-making
2) the reliance on summarising observable phenomena at various levels of activity, when a wide range of activity is possible (including for the volcanoes that have not had witnessed eruptions, and for de-escalation)
3) the grouping of volcanoes with a range of hazard environments and possible behaviours into one system based on currently observable activity
4) it is very difficult, if not impossible, to accurately set the VAL during a short-lived eruption (where observable phenomena are rapidly changing) using the current phenomena-based system (discussed further in section 4.5.4). This is the reason the example VAL system in Table 4.4 contains the words ‘has recently occurred or is in progress’.
Chapter 4 An exploratory investigation of New Zealand’s VAL system
154
As a result of these identified issues, participants considered shifting the foundation of the system to other options, and including response advice and eruption forecasting.
4.6.1.2 Hazard-based
Some parts of the current VAL system are based on hazards ‘in progress’ (e.g., 'hazardous local eruption in progress', the original meaning of which was based on a subjective level of hazard in a spatial area (Table 4.3). It was suggested by interview participants that the foundation of the future VAL system could be based entirely on the level of hazard, and referred to as a ‘Hazard Level’ instead of a VAL system. Hazard assessments are based on information of past activity (from the geological and historical records), and the understanding of underlying processes and models. The method used to ascertain the level of short-term hazard may include the interpretation of monitoring data, and its application to conceptual models. This in turn would suggest styles of potential future eruption activity with associated hazards. The level of hazard can then be based subjectively on this understanding.
Having an indication of the hazards associated with various levels of volcanic activity is an important influence on the planning and response decisions made by end-users in various roles and risk environments. By creating a VAL system with a foundation in hazards, one system could be applied to all of New Zealand’s volcanoes, as a hazard system is thought to make all volcanoes “directly comparable” (Sc4). Using this foundation, response advice and systems incorporating spatial zonation of hazards could be associated with each level, depending on the volcano and situation. An example of a hazard-based system is provided in Table 4.5, and includes a description for each Hazard Level, largely based on the spatial extent of the volcanic hazards. A simpler Hazard Level could be used consisting of solely the left hand column (extremely high to very low), to remove the influence of the spatial extent.
Potential issues with a hazard-based system, as identified by interview participants, include:
1) Identifying appropriate terminology to enable applicability to the wide range of New Zealand’s volcano types and spatial extents may be a challenge. For example, if a general hazard description of ‘hazardous on the volcano’ is used, it may be difficult to define the perimeter of ‘the volcano’ at a caldera or in a volcanic field where the location of the new volcano is highly uncertain. It is a similar situation for the terms ‘crater’ and ‘vent’. Predefining these terms and associated spatial extents for each volcano will be beneficial.
Table 4.5. Hypothetical VAL system for New Zealand’s volcanoes based on a foundation of hazards. This example demonstrates the use of ordinal words instead of a numeric system, with a high to low order of activity from top to bottom.
Hypothetical Hazard-Based Volcanic Alert Level (or ‘Hazard Level’) System
Hazard Level
Extremely
high
Very hazardous on and near volcano
(hazards depend on eruption style)
e.g., widespread ash, lava flows or domes, pyroclastic flows, lahars, flying rocks
High
Hazardous on volcano
(hazards depend on eruption style) e.g., ash, lava flows or domes, lahars, flying rocks
Moderate
e.g., unpredictable small eruptions, poisonous gas, flying rocks, hot Hazardous at areas near crater geysersLow
Low level of hazard, associated with volcanic unrest
e.g., unpredictable small steam eruptions, gas emissions, earthquakes
Very low
No volcanic hazards2) Having a Hazard Level in addition to maps with associated spatial hazards may be confusing, particularly as they are both dynamic systems which change over short periods of time.
3) A wide range of volcanic activity would be included in each level due to the emphasis on hazard rather than magnitude of activity. For example, an ‘extreme’ level of hazard is likely to include an eruption at Raoul Island, a pyroclastic flow on Ngauruhoe, or an eruption at AVF. The level of risk (described further in section 4.4.1.2), on the other hand, would be comparatively low at Raoul Island and Ngauruhoe compared to AVF due to the relatively low level of exposure of human life and property to the volcanic hazards.
4) Defining the terminology used will be important to reduce ambiguity and subjectivity in the decision-making process. For example, defining at what point a situation is deemed ‘hazardous’ will influence the outcome. The ‘moderate’ Hazard Level in Table 4.5 may include heightened unrest (in progress), or rapid fluctuations between geyser
Chapter 4 An exploratory investigation of New Zealand’s VAL system
156
activity in a crater lake and heightened unrest, where no further warning of a sudden eruption can be given. Due to the subjective nature of the hazard-based system, intentions for every level (potentially for every volcano) should be defined to retain consistency in the scientists’ decisions over time.
5) To retain simplicity, it will not be possible to include all types of hazards in the VAL system, particularly with levels defined by the spatial extent of hazards. As identified by one science participant,
“[in] the case of Ruapehu you can have a minor eruption … down one particular valley [where] the impacts could go a hundred kilometres, but … I think if you try and allow for every possible permutation and combination you’ll get your hands tied up in knots” (Sc9).
In the future, the situation might arise where a new or additional system is required to deal with a continuously high level of volcanic eruption activity from one volcano. In this situation, specific hazards over time that may occur in various locations become a pressing issue, rather than a reliance on VALs stating the current level of activity. This occurred at Montserrat during the 1990’s and 2000’s, according to interview participants involved in the revisions of the VAL system there. While eruptive activity continued at a fairly steady state, the areas impacted changed requiring the combination of a colour-coded VAL system with a hazard map associated with evacuations. This situation may occur at a New Zealand volcano, and may cause complications due to the existence of other volcanoes also in the nearby area, which would presumably use the original, more generic VAL system. Whereas Montserrat
communities just had one volcano and VAL system to understand. In this case it is
hypothesised that the overall VAL system could remain the same, but a volcano-specific hazard map be utilised. Consideration should also be made for the development of a more detailed localised warning system (which may cause difficulties if used in conjunction with a hazard- based VAL system) in this situation to suit the needs of the local end-users.
4.6.1.3 Process-based
As discussed in section 4.5.1, once volcanic monitoring data have been collected at a volcano, they are interpreted to understand the underlying processes and used to develop a
theoretical, conceptual model of the volcanic system. As further data are collected, the model is tested and refined until it becomes as accurate as possible. Accurate models, along with an understanding of volcanic processes, contribute greatly towards eruption forecasts and hazard predictions. According to one interview participant, this is similar to the diagnosis of a sick
patient by doctors to ascertain the outcome of a disease. An initial diagnosis of a disease is based on the patients' symptoms, and over time, the diagnosis is tested as new symptoms occur. The diagnosis may be used to predict what the outcome is for the patient, and the symptoms which may be likely along the way. Likewise, a volcano model is based on the observable phenomena, and tested over time. Forecasts of future activity can then be made with increasing accuracy. By basing the future VAL system on underlying volcanic processes, it is thought by participants that determining resulting phenomena and associated hazards may be possible, and can therefore be included in the system.
An example of a potential process-based VAL system based on one participant’s thoughts is presented in Table 4.6. An additional level could be added representing ‘moderate extrusion of magma’ to bring the number of levels into alignment with the majority of participant’s wishes. During the feedback process, a number of participants suggested changes to the wording of these levels – if it is to be used in the future, further investigation of the terminology is needed, with consultation of the system’s users.
Other scientist participants identified a number of issues relating to this potential VAL foundation. These consist of:
1) A reliance on having accurate models for all volcanoes in New Zealand. Most volcanoes do not have a model at all, and those that do are highly uncertain. It is thought by a number of GNS Science participants that “we haven’t got enough science and understanding of the volcanoes to create those models" (Sc3). It was also identified that using this system “would imply that at any time we know where the magma is”, whereas it is thought GNS Science has not “ever been confident that magma is in a particular place underground until it is virtually at the surface” (Sc8). However, regardless of the existence of models, the ability to transfer the understanding of volcanic processes may permit the use of this foundation.
2) Due to the uncertainty associated with processes and models, it is thought that there would be significant delays for the scientists to decide on the most appropriate model, and therefore on the VAL. “This is too dependent of knowledge of process. As we saw at Te Maari it might take months to get a handle on that. Adequate knowledge may come well after the time an alert system is most needed” (EU12). Monitoring sample results (e.g., of magma inclusion in eruption deposits or crater lake chemistry changes to determine the eruption classification as phreatic (VAL 2 in Table 4.6) or
Chapter 4 An exploratory investigation of New Zealand’s VAL system
158
phreatomagmatic (VAL 3)) can take days to weeks to be analysed, contributing to delays in determining the model and thus the VAL.
Table 4.6. A hypothetical VAL system with a foundation of volcanic processes, designed with input by an interview participant. The levels of ‘hazard’ and ‘activity’ are based on the ‘underlying process’ column.
3) Models can be difficult to understand – even senior scientists can have difficulties comprehending the discussions around specific phenomena outside of their specialities, and the impact they have on the model. All staff involved in the voting
Hypothetical Process-based Volcanic Alert Level System
VAL
Hazard Activity Underlyingprocess Model
0
No volcanic hazard No unrest oreruptions No magma
1
Low level of hazard, associated with volcanic
unrest
e.g., steam eruptions, gas emissions, earthquakes Minor volcanic unrest with no eruptions Shallow, stable magma in rock beneath volcano
2
Hazardous at areas near vent
e.g., small eruptions, poisonous gas, flying rocks, hot geysers
Heightened unrest with possibility of minor eruptions Intrusion of fresh magma into rock beneath volcano
3
Eruption hazards on volcano and downwind(hazards depend on eruption style)
e.g., ash, lava flows, lava domes, pyroclastic flows, lahars, flying
rocks Minor to moderate volcanic eruption Extrusion of magma (explosive or effusive)
4
Very hazardous near volcano
(hazards depend on eruption style)
e.g., widespread ash, large lava flows, unstable lava domes, pyroclastic flows, lahars, flying
rocks Large volcanic eruption Large extrusion of magma
(which includes technicians) would need to be able to comprehend the models; the "system needs to be based around something they can understand" (Sc9).
Additionally, the inclusion of underlying volcanic processes and models are likely to be incomprehensible to the vast majority of end-users, leading to questioning the
purpose of their inclusion in the system. However, the ‘underlying process’ and ‘model’ columns could be decoupled from the system, removing it from the public arena, whilst being used by the scientists to determine the level. Voters who may not understand the models or who have an empirical focus could use the ‘activity’ column to determine the VAL, which should be predominantly cohesive with the ‘underlying process’ column.
4) Systems based on interpretations have much more room for error than those based on the observable data, and can be proven retrospectively to be incorrect. It is thought, however, that scientific decisions are mainly based on the currently available
information (and therefore defensible), as would be the case for this VAL foundation. It is unlikely that scientists would be comfortable with this increased likelihood of being proven ‘wrong’ retrospectively.
5) There is a concern that a process-based system will be very difficult to use during de- escalation, largely due to lengthy magma residence times. Adopting a process-based system would therefore require a change in the concept of how a VAL system is used. The lower levels of the process-based system would be fairly static, more of a label for each volcano than a system reflecting short-term changes, as the underlying system would take in the order of years to de-escalate after an eruption using this system. Allocating a lower level after an eruption using this system would be problematic due to the use of categories such as ‘stable magma’. Levels may also be missed during escalation (e.g., going straight from VAL 0 to 2), which may go against expectations of end-users with response planning repercussions.
6) A range of hazards will be apparent for each level, but particularly levels 1 and 2 in Table 4.6. Level 1 (shallow, stable magma) may involve a range from no hazard at all to hydrothermal eruptions and poisonous gas emissions. Level 2 could include periods with no signs of unrest or associated hazards, through to unpredictable phreatic eruptions (or even magmatic eruptions until the magma inclusion results are obtained, by which stage the eruption may be over). This wide range of hazards within each level may not be very useful for end-users, although it is not too dissimilar to the range of hazards in the current frequently active volcanoes system for VALs 1 and 2. During the feedback process, one end-user participant described it as “the scientists’ system, and