An introduction to impact-based warnings for natural hazards
- Sally Potter
- Dec 3, 2024
- 11 min read
In this article, I introduce impact-based warnings for natural hazards and give some of my own thoughts on them. I describe:
I can’t fit everything into this post (it would quickly turn into a thesis!), so I will write a short series on impact-based warnings. In future posts, I’ll talk more about how to develop such a system, and what some of the challenges are that people around the world are facing when doing so – and potential solutions, based on our recent research through the World Meteorological Organization. If you have any questions or suggestions about what to write about on impact-based warnings, please leave a comment below.
Short on time? For a quick summary of impact-based warnings, check out this video:
What are impact-based warnings?
An impact-based warning for natural hazards is a type of warning that goes beyond being based on the intensity of a hazard (like wind speed). Instead, it focuses on the potential impacts of that hazard on people, property, livelihoods, infrastructure, and so on. This approach combines data on the hazard(s) with information about the exposure and vulnerability of people and assets in the affected area to provide a more comprehensive understanding of the risks involved.
For example, instead of just forecasting ‘heavy rainfall’, an impact-based warning would be triggered when that heavy rainfall could lead to flooding in specific areas, causing road closures, property damage, and/or risks to personal safety. Impacts of rainfall-induced landslides could also be included in the assessment, to take a multi-hazard approach. The resulting information helps communities and decision-makers take more effective actions to mitigate the various impacts, such as evacuations or preparing for interruptions to infrastructure services. These are sometimes called 'anticipatory actions', and the key thing is they take place before the event occurs, using the forecast and warning.
Many people think that including impact information in a warning is the key requirement of an impact-based warning. In my view, it’s not. A warning can be triggered entirely on the intensity of the phenomena (e.g. rainfall) or resulting hazard (e.g. flooding) – and therefore is not an ‘impact-based warning’ – and still include useful impact and guidance information. By definition, an impact-based warning needs to actually be triggered by the expected impacts.
To some degree, this is semantics - regardless of how it is triggered, an effective warning should include information on the source, hazard, impacts, guidance information, location, time, and a link to further information - see this post for more details.
Can you give me some examples of impact-based warnings?
While impact-based warnings have largely stemmed from the meteorological sector, they are just as applicable across other types of hazards, including volcanic eruptions, earthquake shaking, tsunami, and landslides.
Taking it to the extreme to help understand about impact-based warnings, let’s take a severe storm that is forecast to move across a city, such as Auckland (New Zealand), and then over the surrounding rural area with fewer people and assets. Assume the storm’s intensity, spatial extent (at any one time), and duration all stay the same throughout its journey. Using a more traditional hazard-based warning, both the urban and rural areas would receive the highest level of warning, as all areas are expecting strong wind and heavy rain. Flood warnings could be issued for all waterways expecting to flood.
If an impact-based warning system was being used, the urban area of Auckland would receive the highest level of warning, as there are a lot of people and assets exposed to the hazard (wind, rain, flooding), and many of those people and assets are vulnerable to it. Flood warnings would only be issued for areas where there are people and assets exposed to the forecasted extent of the floodwater.
However, the rural area has less people and assets exposed, and let’s assume the vulnerability is about the same as the urban area. This means the rural area has less overall risk and therefore would receive a lower level of warning. Areas receiving flooding in these areas might not receive any warnings, due to the low exposure.
Do you think this is a problem? This 'urban-rural' bias is one of the key challenges in impact-based warnings (Potter et al., 2021). New Zealand agencies have told us that everyone should receive a warning, if they are potentially in danger. Solving this problem essentially means taking out the element of 'exposure' from impact-based warnings and focusing more on vulnerability. In a joint research project involving Canary Innovation, GNS Science, NIWA, and the UK Met Office, we are currently exploring the role of vulnerability and exposure information on impact-based warnings.
Here are a few more examples of impact-based warnings for other natural hazards:
Heat warnings
Impact-based heat warnings could highlight the expected health impacts, such as increased risk of heatstroke, that are targeted at vulnerable populations like the elderly and young children. The rest of the population might just receive an ‘advisory’ or similar lower level of warning. This differs to the other examples where there is no spatial ring fence around the area at risk – it depends on the demographics and vulnerability of the people (and infrastructure). Another example is heat wave impact-based warnings having dynamic thresholds based on when in the season they occur (e.g. towards the start, middle, or end) and where they occur around the country, as the impacts are likely to differ (Pascal et al., 2006).
Tsunami warnings
Impact-based warnings for tsunami would be triggered for areas where there are people and assets, and not for unpopulated areas. In my view, this is downright dangerous as just one person in those unpopulated areas would be more at risk because they weren’t warned, than if a traditional hazard-based warning had been issued.
Another take on this could be that it is assumed that there is always one person on every beach, so every area exposed to the hazard gets a tsunami warning. A higher level of warning could be reserved when a certain number of people are at risk. This highlights the difference between when the audience of the warning is the response sector (using societal risk) or the public (when individual risk would be more suitable).
Wildfire warnings
When a wildfire starts, impact-based warnings would be issued only when the fire is forecast to affect people and assets. If it is in an unpopulated area, no (public-facing) warning might be issued.
Earthquake shaking warnings
As earthquake early warning systems are developed, the algorithms can start to calculate whether the intensity of the shaking is likely to produce impacts in a certain location, and alert just those people. This would use the shaking intensity rather than the magnitude of the earthquake and distance to source, and perhaps in future use high resolution location data in conjunction with building information to determine the appropriate tailored advice for actions.
Landslide warnings
Impact forecasts for landslides triggered by both earthquake shaking and rainfall are currently being trialed by GNS Science. These are spatial maps showing areas that are more susceptible to having landslides occur (the hazard forecast), in addition to the number of landslides per kilometer of road, and the number of properties expecting to have landslides occur on them. This is a big step forward towards impact-based warnings for geohazards and informs responding agencies about where to focus their mitigation and response efforts. We are currently evaluating these products through the Hōretireti Whenua Sliding Lands Endeavour research programme.
Volcanic eruptions
Impact-based warnings for eruptions could be triggered when infrastructure outages are expected as a result of ashfall, for targeted populations at risk of health impacts from ashfall, for flight disruptions due to certain concentrations of tephra in the air, or when lahar (volcanic mud flows) are expected to cause impacts to roads or houses.
For near-vent hazards that are more lethal, impact-based warnings would be issued for any eruptions in or near urban or populated areas, whereas remote islands with no people on them might not have any (ground-level) warning issued because no one is at risk. We did consider this as an option when reviewing New Zealand’s Volcanic Alert Level system as part of my PhD in 2014. However, we decided that no matter where the volcano is, we needed to be able to give it the highest level of alert so people know that it is dangerous if they are planning to go near it. Also taking into account that this alert level system does not incorporate forecasting – it indicates the current status of the volcano - we went with a phenomena/hazard-based approach at the time.
What are the essential elements of an impact-based warning system?
Firstly, warnings are only one part of the risk management system – the last resort in reducing impacts before an event occurs. Land-use planning and restricting development on at-risk areas, having building standards that improve resilience to hazards, plus a raft of other risk mitigation measures, go a long way to save lives. Narrowing in on warnings, any successful warning system needs to consider the whole warning value chain, as pictured below.
The people-centred warning value cycle. Created by Sally Potter (Canary Innovation) based on WMO (2018), Golding (2022), and the Early Warnings for All Executive Action Plan.
Setting up an impact-based warning system requires an understanding of current institutional arrangements, roles and responsibilities, and the wider (legal, political, cultural, social) context. The people and assets at risk of each hazard need to be assessed, and those communities and sectors should be part of and central to the warning system development.
The data required to produce impact-based warnings includes the vulnerability and capacities of people, the built environment, infrastructure, economic sectors (etc.); exposure of these elements to the hazard(s); and the hazard forecast, including the severity/intensity, spatial extent, duration/frequency/timing, and likelihood of occurrence (e.g. Harrison et al., 2021, 2022). The hazard forecast needs underpinning monitoring and modelling data about the phenomena (e.g., weather, volcanic unrest), emphasising the importance of ongoing funding for these activities. Observations can be enhanced through crowdsourcing, where people report what they are seeing, hearing, feeling (and smelling!), which is especially useful where monitoring equipment is sparse. Understanding vulnerabilities to the hazard involves the collection and management of impact data and working closely with communities.
Impact forecasts can integrate this data either through sophisticated quantitative risk modelling software (such as New Zealand’s RiskScape), or by using a more qualitative approach to estimate the likely severity of impacts. The impacts of a hazard (or multiple cascading or compounding hazards) can be on a whole raft of different elements, including but not limited to fatalities and injuries, building damage, impacts on culturally significant sites, infrastructure outages and travel delays, economic damage, and environmental damage. These all need to be considered when issuing an impact-based warning, although many agencies at this stage focus on saving lives. Impact tables can help to compare the level of expected impacts across these different elements.
Once the overall level of impacts is ascertained, a risk matrix is one way to combine the likelihood of the hazard occurring in a certain location with the severity of the expected impacts in that location, as shown in the image below.
Example risk matrix, with the tick showing a moderate likelihood of a high impact event occurring in a certain location, resulting in an ‘orange’ warning being issued.
The resulting forecasts and warnings need to be effectively communicated through multiple channels. Importantly, they need to have appropriate guidance information within them that relate to the level and type of impacts. These should be supported by standard operating procedures, education and training initiatives, and response plans. The forecasts and warnings should be evaluated (a whole topic in itself, and a challenge for impact-based warnings) and improvements made to the system.
What is the future of impact-based warnings?
One of the key challenges with impact-based warnings is providing meaningful impact information. Warnings triggered by (or containing information about) ‘road closures’ may not be useful for someone in a retirement home who isn’t planning on going anywhere. ‘School closures’ are not so useful unless you have children at school or if you an employer facing staffing issues. Advice about wind on ‘high-side vehicles’ are not relevant for people driving sedans or who walk to work. Therefore, to make these warnings more meaningful, I believe the future of impact-based warnings will be tailoring impact information and guidance advice right down to the individual level.
How to do this? Technology advancements are making this possible through using our digital footprint data, or we could be taking a user-led preference approach, where individuals set their warning thresholds. You can check out my post for more on these ideas, or my TEDx talk (below) on the topic.
This leads to the question of who issues these personalised impact-based warnings – should this be the role of science agencies, such as meteorological services? Or emergency managers/local government? Will the tech giants, who hold all this personal data, issue the warnings instead?
One option is science agencies stick with issuing hazard-based information in future, so that other agencies who hold the relevant data issue the warnings using their own thresholds. For example, the meteorological service issues a spatially-varying forecast for severe wind, emergency management use their own impact-based thresholds (ideally co-created with at-risk communities) to issue relevant warnings for public safety, electricity companies use their thresholds to warn people who will be affected by power outages, and transport agencies use their thresholds to give warnings for certain stretches of road or ferry service cancellation. This is instead of the meteorological service issuing impact-based warnings themselves, and needing to know all those different thresholds and the underpinning data. This approach demonstrates why strong partnerships between agencies is necessary, so that warnings and guidance messaging are given in a coordinated way.
Time will tell how the movement towards impact-based warnings pans out, and how legal roles and responsibilities of forecast and warning agencies influence the outcome.
What do all these terms mean?
Just so we’re all speaking the same language, this is the way I’m thinking about the risk-related terms in this article. These definitions are from the United Nations Office for Disaster Risk Reduction Glossary.
[Disaster] risk
The potential loss of life, injury, or destroyed or damaged assets which could occur to a system, society or a community in a specific period of time, determined probabilistically as a function of hazard, exposure, vulnerability and capacity.
Hazard
A process, phenomenon or human activity that may cause loss of life, injury or other health impacts, property damage, social and economic disruption or environmental degradation.
Exposure
The situation of people, infrastructure, housing, production capacities and other tangible human assets located in hazard-prone areas.
Vulnerability
The conditions determined by physical, social, economic and environmental factors or processes which increase the susceptibility of an individual, a community, assets or systems to the impacts of hazards.
Capacity
The combination of all the strengths, attributes and resources available within an organization, community or society to manage and reduce disaster risks and strengthen resilience.
Where to find out more
In addition to the references linked to in this article, you can check out some more papers and guidance documents here:
Sara Harrison's raft of papers on impact-based warnings data and partnerships (proud PhD supervisor alert!):
Finally, this Resilience to Nature's Challenges webinar from 2022 on impact-based warnings may also be useful. You can enjoy the whole thing or skip to 19:13 where I give an overview of how effective these types of warnings are, and describe some of the benefits and challenges. Then Sara Harrison gives a description of her PhD research on the underpinning data needed for impact-based warnings. Finally, we describe the potential for impact-based warnings for geohazards.
Please do let me know in the comments below if there are any topics relating to impact-based warnings that you are particularly interested in.
Commentaires