New paper published on tropical cyclones and warning systems: the extraordinary among the commonplace

Many of you may know well what it means to live through recurring hazards, such as annual seasons of tropical cyclones. Some of you will know how to protect yourselves and your families against the frightening but smaller storms. Some will know the catastrophic danger and absolute fear created by the larger ones (all in relative terms of course). Some will know what it means to live in evacuation centres and to be displaced in emergency shelters for weeks or even months at a time.

Whatever your experiences, imagine for a moment that you’ve never experienced a Category 5 hurricane before, unaware of what it could do to your family, friends and home; a person living in a wooden home on stilts over the ocean and unsure of what “storm surge” means; a farmer whose life depends on the pigs he keeps on his land around his home; an elderly woman having experienced a deadly disaster years ago but who is now completely dependent on her family to ensure her safety. Does experience or naivety help you make safer decisions? What happens if you want to leave for a shelter but the rest of your family doesn’t? What do you do if you keep animals on your homeland and can’t leave them behind? Or what if you hear a message on the radio that conflicts with advice you hear on the TV?

In our latest paper published in Disasters journal, Joanna Faure Walker and I have drawn on our fieldwork studies in the Philippines and Dominica to investigate what warnings people heard, when and where from in relation to how they then reacted before major tropical cyclones. In the Philippines, we took Super Typhoon Yolanda in 2013 (internationally known as Haiyan) as a case study, and in Dominica, we studied Major Hurricane Maria in 2017. The Philippines and the Caribbean experience annual tropical cyclone seasons, and so are accustomed to events that usually range between tropical depressions and Category 1-2 storms. However, we are particularly interested in examining what happens on rarer occasions, when these locations experience large Category 4 and giant Category 5 storms.

We found that among the people we surveyed in the two locations, a warning that both Yolanda and Maria were approaching was heard by all but one person before both storms arrived. These were often received with more than a day’s notice, however, over three quarters of our populations chose to either remain at home throughout the storms, leave for shelter during them, or leave for shelter once they had passed, not complying with direct instructions from the authorities to evacuate. Not the intention of those issuing the warnings, and not the safety seeking behaviours we would associate with a successful warning system.

Single sources of warning, such as a message through a radio only, failed to reach everyone in both locations, and so warnings issued across several media platforms were often the best way of ensuring as far as possible that the most people received a warning advisory. This is intuitively sensible, especially as some may fail at critical stages. However, in the Philippines, this had practical implications. Even though only around half of respondents heard a warning from two or more sources, slightly more people evacuated before Yolanda arrived when they heard two sources, rather than only one.

So if the warning system technology works, why did the desired human response not follow? We know from other studies that evacuation is tricky because of the complexities of people’s lives, and that people stay at home to protect their possessions, their livestock, to adhere to social pressures etc. But revealed in our surveys were a number of key elements that also deprived our respondents of a full appreciation of the heightened danger in these two cases. These tropical cyclones were more deadly than the average storm, but not realising the implications of “storm surge” because the term was widely unknown among respondents in the Philippines, signalled a failure in the messaging that almost certainly resulted in a higher death toll. Similarly, radio network breakdown during Maria’s very late and rapid intensification near Dominica meant that warning messages were confusing and Category 5 impacts were not expected. In such situations, people defaulted to their usual behaviour: stay at home and ride it out, it’s what normally works. And because both information pictures were incomplete, people were caught unaware.

In both locations, messages were reported to have been inconsistent and unclear, for example to evacuate if you live close to the water or in “vulnerable housing” (what does this even mean?) in the Philippines. Often these required people to exercise considerable levels of subjective judgement over several risk profiles, most notably their own and that of their locale. This necessitates, at the very least, a full hazard information picture. Additionally, evacuation and shelter infrastructure that should support warning messages and promote safety seeking behaviour was often so substandard that it was a deterrent. The inadequacy of many emergency shelters discouraged people from their use, being overcrowded, lacking in resources, offering little personal safety, and incurred physical damage themselves by the storms.

Our paper demonstrates that within the social processes of warning mechanisms, a failure at any stage can render them decidedly less effective in saving lives. It shows that warning systems require the support of accurate forecasting and message dissemination technology (improved hazard modelling, the acknowledgement of scientific risk uncertainty, robust and consistent communications networks, and context appropriate language), solid infrastructure (e.g. fit-for-purpose evacuation shelters) and an inherent consideration for the idiosyncrasies of populations at risk, taking into account “foreground” and “background” constraints and assumptions (these are explained in the paper, so go read it). It also suggests that experiencing more regular, lower intensity tropical cyclones may in itself not help reduce vulnerability to the more deadly effects of rare, higher-intensity storms.

Our full study and findings in more detail can be found here:

Yore, R., Walker, J.F. (2020). Early Warning Systems and Evacuation: Rare and Extreme vs Frequent and Small-Scale Tropical Cyclones in the Philippines and Dominica. Disasters, doi:10.1111/disa.12434