Communicating hazard and risk information products to meet user needs

This week I’m excited to be in Queenstown, New Zealand, for the international Landslide Risk and Geo-Education conference, or ‘LaRGE’.  The conference theme brings together the topics of landslide risk assessment, communication, and geo-education, with a focus on helping information support decisions and reduce impacts.

I’m contributing to a communication panel and presenting evidence-based tips for communicating landslide information with users. This is a topic I care a lot about because hazard and risk information is only useful if it helps someone reduce their risk. For this talk, I’m drawing on guidance I recently created alongside Danielle Charlton (Earth Sciences New Zealand) and Emma Hudson-Doyle (Massey University), available for free from here.  This post summarises both the guidance and the talk for anyone wanting the 7-minute summary.

If you are creating data, products, tools, models, or information for other scientists, decision-makers, and/or communities, regardless of the hazard – this one is for you.

1  Start with the user, not the product

Scientists often begin with the data, model, map, forecast, dashboard, or tool. They ask: What can we produce? What is scientifically interesting? What is technically possible? What does the model show? How should people use it?

A better starting point is: Who are the key potential users of this information, what decision(s) are they making, what do they need, and when do they need it?

An effective product starts with understanding the audience and their needs. The audience might be emergency managers, roading agencies, council planners, lifeline utilities, iwi and hapū, local communities, researchers, consultants, or insurers.

These groups often need different information, at different times.

A technical user might need GIS layers, metadata, model assumptions, and uncertainty ranges. A non-technical decision-maker might need a short summary of what has changed, where the greatest concern is, what the information means, and what action is recommended. A community member might need clear local place names, practical guidance, and reassurance that the information is coming from a trusted source.

Schematic indicating formats of information to consider, depending on how technical the users are. From Potter, Charlton & Doyle (2026), Fig. 2.

Audience needs should shape the data sources, models, level of detail, format, time window, update frequency, delivery channel, and supporting information for the product. For example, a rainfall-induced landslide forecast for emergency managers will need a different time window and update frequency than long-term landslide susceptibility information used for planning.

One product may not be enough. A dataset might need a dashboard, or a link to an explainer website. A technical map might need to be accompanied by a plain language summary. This point repeatedly came up in our Hōretireti Whenua Sliding Lands Endeavour Research Programme interviews with key users of landslide forecasts, and has led to changes in the way Earth Sciences New Zealand now communicates these trial forecasts to Emergency Managers.  

2   Make the information useful, specific, clear, and trustworthy

In the landslide communication guidance, we describe effective landslide risk information as being useful, specific, clear and accurate, effective, consistent, timely, and trustworthy.

For landslide information, this means including the practical details people need, such as the hazard, location, time of issue, forecast period, possible impacts, a link to more information, and whether it is an update. It also means being specific enough that people can tell whether the information applies to them.

A map labelled with meaningful town names, roads, and regions, and accompanied by a clear statement or key points, is often more useful than one that assumes everyone can interpret technical spatial boundaries with tiny cell resolutions.

Clear language matters. Avoid acronyms and technical jargon unless the audience has asked for them. Accuracy matters too, including spelling, grammar, place names, and consistency across products (e.g. ensuring legend colours schemes and ranges are the same across multiple products).

Trust depends on transparency. People do not necessarily need every detail of a model, but they do need enough information to understand what the product is based on, what it can and cannot tell them, and where the main uncertainties are.

3     Communicating probability clearly

I fully support communicating probabilities, because many people find them useful. However, they can also be confusing to many people, so it does need to be done carefully and clearly. Here are some tips:

• Consider the audience of the product, and the purpose. Include both the verbal term (e.g., ‘very unlikely’) and the number (e.g. 10%), unless your audience is very technical. However, verbal terms should not be used on their own because people interpret them differently.  There are probability translation tables available to help you choose the most appropriate verbal term for your number (see example below).

  
  

• Numerical probabilities can be given as percentages (e.g., 10%) and/or frequences (e.g., 3 in 10). Keep the denominator consistent when multiple frequences are in the same product (i.e. don’t use 1 in 10,000 and 1 in 100,000 in the same message). You can also use visual ways of showing probabilities, such as with error bars on graphs, or gradational boundaries on maps.

  
  

• Clearly describe the reference class, which is what the probability refers to, such as the type of events, the spatial area, or the time frame. If a landslide forecast says there is a 70% chance of landslides, what does that mean? Does it mean landslides will occur across 70% of the mapped area? That the conditions occur 70% of the time? That 70 out of 100 similar situations would result in landslides? Explain it carefully in the accompanying text statement.

  
  

• Consciously choose the directionality of your message, as this influences the receiver’s perceptions and actions. For example, ‘there is a 1 of 20 chance of x occurring’ is a positively-framed message, and ‘there is a 19 in 20 chance of x not occurring’ is a negatively-framed message.  

  
  

• Choose and communicate the time window for probabilities. Ideally match the time window to the user’s decision needs. For example, if a roading agency needs 12 hours to make decisions, then a forecast should ideally provide at least that amount of lead time.

  
  

• Uncertainties can be communicated by using ranges (e.g. 5 to 15% chance), and it’s recommended that you are transparent about where the uncertainty has come from, as well as how and when it will be reduced.

  
  

• Risk comparisons can be used carefully. But be aware that people have different tolerances for risks. If you use them, ensure both risks being compared are either voluntary (e.g., between skiing or driving), or involuntary (e.g., death from natural hazards), and that the contexts are the same (e.g., hiking in place A vs place B, but not hiking in place A vs driving).

This all connects strongly with my previous posts on communicating warnings, impact-based warnings, and geoethics in natural hazard communication.

GeoNet's probability translation table, from Doyle & Potter (2016).

4     Tips for maps

Landslide information is often spatial, so maps are a natural format. Good maps can reveal patterns and support situational awareness far more quickly than a table or paragraph. But maps can also mislead if they are too complex, poorly labelled, inaccessible, or not matched to the decision being made.

Start with a clear purpose and user. Is the map for rapid situational awareness, detailed technical analysis, public communication, or planning? Each purpose requires different choices. Simple maps may only get a few seconds of attention, while complex data visualisations may require minutes of careful interpretation.

Maps should include essential elements such as a clear title, scale, legend, date, source, and appropriate spatial reference information. They also need accessible colours and symbols, suitable base maps, and a clear visual hierarchy so the most important information stands out. Uncertainty can be shown directly on the map, in an adjacent map, through symbolisation, or in supporting text.

Interactive maps (such as using GIS) can be very useful when they are feasible and when users have the time, technology, and confidence to use them. Static maps, summaries, and printable resources may be more accessible for some audiences, and can be easier for users to share with colleagues. The fanciest dashboard in the world is not useful if the person who needs it is on a slow rural connection, in a noisy operations room, or trying to make a decision between meetings.

5       Accessibility is not an optional add-on

Hazard and risk information needs to be accessible to as many people as possible (see Lawson et al., 2025, for more details). About 17% of New Zealand’s population have a disability, including many disabilities that are not immediately visible. Barriers can include language, low literacy, colour vision deficiency, hearing impairments, dyslexia, ADHD, memory impairments, limited internet access, and low trust.

Practical steps include using plain language, providing information in multiple formats, adding alternative text to visuals, using captions and transcripts for videos, choosing accessible colours, avoiding rainbow colour scales, using readable fonts, and considering multiple languages, including te reo Māori.

Accessibility is not just about compliance - it is about whether the information can actually be found, understood, trusted, and used.

Checklist

Before releasing a risk or hazard information product, it is worth asking:

• Is this information that the intended user needs?

• Is it useful, specific, clear and accurate, effective, consistent, timely, and trustworthy?

• Is the title clear and understandable?

• Have you considered the accessibility guidelines?

• Would the user know how uncertain it is, and what parameters it uses?

• Is there a link for further information, or contact details for them to follow up through?

Communication is not the decorative ribbon tied around the science at the end. It is part of the science-to-action pathway. If we want models, maps, forecasts, and tools to support safer decisions, we need to design them with users from the start.

That is the real landslide communication challenge: not just moving information downhill, but making sure it lands where people can use it.

Don't forget to check out the full Guidance document below for more tips, including in relation to ethics, and using scenarios, risk matrices, graphs, charts, and photos.

References

Doyle, E. E. H., Potter, S. H. (2016). Methodology for the development of a probability translation table for GeoNet (GNS Science report 2015/67). GNS Science report 2015/67. http://shop.gns.cri.nz/methodology-for-the-development-of-a-probability-translation-table-for-geonet/

Lawson, R.V., Charlton, D.H., Harrison, S.E., Kaiser, L.H. (2025). Enhancing accessibility in science communication: best practices and recommendations for geohazard advice. Lower Hutt (NZ): Earth Sciences New Zealand. 59 p. GNS Science miscellaneous series; 151. https://doi.org/10.21420/ZZDW-HX90.

Potter, S.H.; Charlton, D.H.; Doyle, E.E.H. (2026) Guidance for communicating landslide information. Horetireti Whenua Sliding Lands Endeavour Programme resource. Lower Hutt, NZ.: Earth Sciences New Zealand. Hazard and Risk Data Guidance 2026/01. 13 p.; doi: 10.21420/SW07-KM24