Some STREVA researchers keep an online blog. 

Our 'Opinion' page reproduces posts that are related to STREVA. 

Understanding Volcanic Risk: how well are we doing?

This blog is a reflection of discussion around a 2 day workshop held at the University of East Anglia 7th-8th January 2015.


    (Image by Banksy)

Embracing Risk?

Rationale: For the past 2+ years the STREVA project has been engaged in research focussed on understanding dynamic volcanic risk. Some of this work involves new research into individual problems that contribute to risk (for example improved monitoring and forecasting of volcanic hazards). A core team have also been engaged in the interdisciplinary analysis of past or ongoing volcanic eruptions to understand what drives and changes volcanic risk. This work is very challenging: it requires the combination of knowledge from a wide variety of disciplines; careful strategies for collaborations with partners and work with ‘at risk’ populations.

We are starting to question whether some of the most important (and dynamic) questions of risk can be meaningfully partitioned into ‘hazard’ ‘vulnerability’ or ‘exposure’ or whether new ways of thinking about risk might help us target new research more effectively.

The purpose of this meeting was to report some of these findings but, more importantly, reflect on and share the challenges of working at the ‘risk interface’. We wanted to share experience and consider where common themes are developing with other experts in the field.


The first afternoon was spent in ‘charrettes’ (revolving mini discussion groups). Each group was focussed around a theme that has emerged across the settings where we have been analysing past volcanic risk. These themes has been broadly identified as something that could drive risk over the course of a crisis. We have become particularly interested in the intersection (similarities and differences) between risk in the immediate crisis phase (risk to life) and the longer term consequences of risk (risk to livelihoods) and how each of these contributes to population resilience:

 The four themes discussed were

  1. The role of science innovation: its practical application in a crisis and improving forecasting in the longer term.
  2. The impact of coping with long time-scales and multiple hazards        
  3.  The impact of warnings and communication processes
  4. Management, ownership and transfer of risk: who protects life and livelihood? 


There was a strong acknowledgement that each situation could be highly context dependent. It was clear that the application of scientific innovation depended on its appropriateness to that specific time and place and as a consequence of long-term and trusted collaborations.

Similarly, risk messages around volcanic activity could equally only be relevant when placed in the context of daily risk ‘portfolios’ and in many places mediated by previous local experience. Important messages about risk were often thus dependent on trusted sources, perceived to understand what they were asking of the population in terms of action and recognising other priorities.

It was acknowledged that the most important risk drivers (and indeed ownership of risk) might change over time – this was a theme that recurred over both days.

Volcanoes are sometimes capable of generating activity so intense that populations need to move and resettle (temporarily or permanently). The attachment to place (economic and emotional) is hard to break. In this regard volcanoes are somewhat unique in that they intensify the emotional attachment to place, are often a fertile source of agricultural or tourism-focussed livelihoods and provide a very identifiable hazard source. Processes that improve the pull of new locations (and acknowledge this upheaval is difficult) through participation are helpful. Consequently ‘political’ processes have a strong bearing on decisions to act to reduce risk by permanent or temporary movement.

Although there are some  issues around re-settlement that are unique to volcanic activity, there are strong parallels with other risks e.g.  drought, landslides and some conflict settings.

What messages to learn for research focus?

Clearly little advance in ‘risk reduction’ will be made if we maintain our mono-hazard focus. How do we improve the multi-hazard approach ?

How do we innovate in forecasting and monitoring, while maintaining credibility and continuity of information?

How can we place hazard information in the context of everyday risks (e.g. acknowledging risk tolerability thresholds) while maintaining scientific independence?

How do we understand the vulnerabilities (physical and social) that come with attachment to place?

Do we actually have any tangible evidence for the power of participation in addressing some of these issues?


The second day focussed on current progress in mapping hazards and risk and challenges in monitoring, forecasting and warning of activity.


Moving the State of the Art in Understanding Risk


Mapping Risk

The first part of the morning focussed on vulnerability. In transforming hazard maps from a description of likely hazard footprint to risk there is a need to tackle vulnerability. Volcanic hazards are highly spatially dependent and maps are a natural way to encapsulate that information.

Vulnerability expressed as ‘exposure’ (population numbers, physical vulnerability and demographics) is comparatively ‘easy’ to include. The opportunity for innovation comes around social vulnerability. ‘vulnerability indicators’ can accommodate the practice of mapping, however these are necessarily aggregates of several processes or criteria that have lead to these attributes. With social vulnerability the most important distinctions arise from differentiation for individuals across these indicators or broad criteria. Do mapping approaches and indicators then encourage us to leave important groupings or dynamic processes out - would we be in danger of not recognising the most vulnerable members of a population as an eruption evolves?

With mapping there is an inherent tension between maps for emergency planning purposes (determining mass temporary evacuations) versus longer term planning (re-settlement, land-use development). Here this intersects with issues around attachment to place  (Day 1) above. We also need to acknowledge that any mapping which results in ‘lines on the ground’ can be interpreted as a political act. A key question we did not answer was whether there are useful components or purposes where vulnerability can be expressed spatially via indicators.

 More Questions: Are there many examples of completely successfully re-settlement in the face of physical hazards. What contributed to the success?

Are there examples of times when an incomplete understanding of vulnerability led to increased risk/loss of life or income? What was incomplete about it? What mapping scales and level of data completeness worked?

If we can’t describe reliable indicators for vulnerability then when and how can we define risk spatially?


Time-series analysis of risk

A novel approach to analysing risk using  time-series was described where risk response was defined via evacuated area;  hazard impact via the total cumulative runout of pyroclastic flows; and hazard ‘potential’ via seismic energy associated with sub-surface magmatic activity. This has been done quantitatively for Montserrat and was in the process of being trialled with less data around Tungurahua (Ecuador).

In discussion the argument that evacuation decisions are often politically motivated returned and it is important not to assume causality from correlation. However this provided a new way to analyse data without only relying on recollection from differing view points and so was worth pursuing.


Questions: How can we reliably integrate quantitative (measured outcomes from hazardous activities) with qualitative (reported decision-chains, rationales) data to identify effective targets for  improvements in reporting and forecasting risk?


Monitoring Data and Forecasting Activity

The key to this still lies with the gathering and interpretation of geophysical data. Gathering and analysing complex seismological data is computationally expensive and methods that develop data mining techniques to improve and automate the detection of patterns would improve how this information is used. New spectral analysis techniques are being trialled in STREVA locations.

Fundamentally the ‘interpretative component’ remains challenging – deciding what these time series data mean in terms of deterministic models for the Earth’s subsurface is extremely challenging. This became a discussion  point in recent simulations (VUELCO) as in real crisis situations.

Proper cooperation and collaboration are the key to coupling developments in understanding to developments in forecasting.

A significant but overlooked aspect is the resilience of observing networks and the observers themselves to the demands of volcanic activity. This has also been a recurring theme during our workshops.

New technology (e.g. USGS SPIDERS) can bring redundancy to the observing network but attention to that problem in network design can and should help.

The USGS places a strong emphasis on developing and maintaining partnerships both during quiescence and activity using Volcano Working Groups. During a crisis all the different research groups will be brought together – the gathering of a consensus around physical discussion, often using tools such as Event Trees to focus that discussion. It is important that communication networks provide consistent messages.

In New Zealand risk is managed via the following process pathway: Risk identification – risk analysis – risk evaluation  (including establishing tolerable risk) – risk treatment (transfer, avoid, mitigate). A recent revision of the warning system focuses largely and deliberately on the ‘ongoing’ situation (no forecasting element). New methods are being developed to include formalised probabilities and elicitation but as yet this has been focussed on traceability of decisions and the safety of field teams.

In STREVA we are conducting research on ways in which more collaborative partnerships are developed between populations at risk and those responsible for monitoring the hazard (largely volcano observatories). Here, again,  clear understanding of context and the building of trust have a strong role to play in the success of these initiatives.

New developments in probabilistic approaches to forecasting activity were presented, including a Bayesian approach. Both of these in the STREVA Project are currently focussed on forecasting lahar risk – in Montserrat and at Tungurahua.


Challenges in working practices: interdisciplinarity and risk

Several key challenges were identified:

  1. Big (discovery) science versus practical science. In our field (volcanology) this is easy to counter because the two are intrinsically linked.
  2. Partnering with volcano observatories in understanding risk. Our research partners are also a stakeholder and we need to acknowledge the difficulties this brings from the beginning, for everyone! The motivation for the study needs to be clear (practical or theoretical development); the role that everyone has outlined at the start and the expectations for outcomes clearly outlined. This is very hard with ‘discovery’ science where research may move in unanticipated directions.
  3. Interdisciplinarity: not everyone needs to be an interdisciplinary scientist in the field of risk but respect for other fields and those who are key for successful exchanges. Continued communication and clear shared goals and language are very important.


Many thanks to all the participants at the workshop!