Warning fatigue - Why human-centred warning systems are a game changer

In floodplain management, flood risk studies are an important part of understanding the risk profile for a community and how we can use that information to develop a plan and prepare ourselves for what may come next. Flood studies are also instrumental for the purposes of infrastructure design, development control, and to assist in strategic land-use planning and controls. The challenge with these static studies is that they:

• are limited to a finite number of modelled event scenarios (for example the 1% Annual Exceedance Probability). As flood event occur, they can sit between modelled scenarios, be larger than a modelled scenario, be a coincident event with different magnitudes or be a combination of all these scenarios. Rarely does the weather play by the rules and it can be difficult to determine the area of impact as an event is unfolding due to the numerous variables at play;

• may rely on the use of techniques, such gauge rating curves and flood frequency analysis, to derive probability statistics for event based modelled scenarios. The limited available historical record of rainfall and river data (usually a maximum of 100 years) may not be sufficient to extrapolate an accurate probability of what might happen in the future; and

• may be limited in geographical reach with not all areas of interest within the modelled area or may not be reflective of significant changes in topography due to development or natural processes.

Industry accepted hydrologic and hydraulic modelling techniques for event-based scenarios are valid and a crucial element in developing flood risk management plans. However, for emergency managers relying on event-based modelling, it can be difficult to understand the complexities of the unfolding event and select the most appropriate model scenario to determine the likely impact. Without having situational awareness in real time, the outcome is usually lack of warning (due to lack of confidence in what modelled scenario to apply) or to broadcast warnings that cover too broad of an area, are unspecific, and in hindsight seem either overkill or inadequate.

For anyone who lives in an area affected by frequent disasters, there can be the danger of falling into a cycle of warning fatigue.

The cycle of warning fatigue might look something like this:

• a disaster event starts to unfold. People are alerted to the disaster, but the warning is inadequate for a range of reasons: the warning does not cover the area affected; the warning underplays the severity of the event, or the warning comes too late to allow action;

• a major catastrophic disaster seems to come from nowhere, taking many by surprise and causing widespread devastation;

• as the emergency subsides, recriminations ensue, and systems are put in place to create better warnings next time;

• another disaster is forecast;

• this time, warnings are over the top. They are well ahead of time (too far ahead); they cover the area affected but in an abundance of caution cover a vast area; the severity is overestimated. General feeling that the warnings caused too much angst since no widespread impact was observed;

• another disaster is forecast;

• this time, while warnings are still forthcoming, the people in affected areas have lower trust in those providing the warnings. Most people in the warning area assume they won’t be impacted, as was observed in the last event;

• this disaster is a major catastrophe with many people underprepared, despite the warnings;

• the cycle then repeats.

It can be an uphill battle to get the balance right. The unpredictability of disasters means that warnings need to be given close enough to the event to be relevant, but there also needs to be time for people in affected areas to take action and be able to carry out their emergency action plan.

An area for hope may be in turning towards a more impact-based warning model, whereby action-orientated warnings become hyper-localised and impact based. Instead of just one blanket warning for a large area, imagine if flood warnings could be provided street by street.

While general warnings need to cover a broad area, having an idea of specific areas that will be more affected means emergency managers can direct resources where they are most needed, evacuate people most likely to be affected ahead of time, and switch off essential infrastructure to prevent damage and maintain public safety. Localised real-time alerts can

• empower communities to help each other, taking personal action to prepare;

• be flexible and updated as the latest situational awareness comes to hand; and

• expand and contract impact areas to provide the most effective allocation of response resources.

Most importantly, accurate localised alerts can help create more trust for warning systems in general. People may be more likely to act in future because they can understand the impact associated with the warning. It also means people new to an area do not have to rely on historical and local knowledge to understand if they are at risk. Finally, for major infrastructure and essential services, scalable, local, real-time impact-based warnings provide the chance to evacuate, shut down and/or divert resources to ensure public safety, staff safety and continuity of service. This helps not only during an emergency but in the immediate aftermath of recovery as well.

There are many factors why we are seeing more “unprecedented” events in the world. By taking the focus from a broad one-size-fits-most approach to creating hyper-local real time information enabled through technological innovation, we can help save time, money, and lives.

FloodMapp is working with Humanitech on a pilot project that enables emergency managers to access scalable and real-time information to prepare and plan for emergency response activities to flooding.