The majority of natural hazards that trigger disasters are foreseen, which helps to avoid catastrophic effects. This blogpost, however, explores efforts to go further, looking at the challenges associated with forecasting not only the hazards themselves, but also their impacts, to inform anticipatory action that protects many more people from harm.
Imagine you are planning to go on a hike. You might want to look at your favourite weather app to see what the temperature variations will be and whether it will rain. Based on the forecast, you can decide whether to bring an extra layer or even cancel your trip. This could “impact” your decision.
We can predict the weather in the next 24 hours with increasing confidence these days, but such forecasts alone do not allow the disaster risk and humanitarian communities to prepare properly before a disaster strikes. How can those working on the ground act more efficiently with trusted information? How can they get useful information ahead of a shock and mitigate its effects? Impact forecasts move on a step from conventional weather forecasts to give information on how the weather will affect us.
Anticipatory action is a smart way to respond to potential crises when they can be forecast1. It includes early warning systems designed to protect families and their assets and it supports decision makers and affected communities in making informed choices before a hazard strikes. How many shelters need to be opened? How much food assistance will be needed?
These are some of the questions that IDMC, in partnership with the Weather and Climate Risks group at the Swiss Federal Institute of Technology in Zurich (ETHZ), have tried to answer since late 2020 through the lens of displacement under our Pacific Response to Disaster Displacement project.
Effective early warning systems are vital in reducing the number of people exposed to dangerous and often life-threatening hazards. This reminds us that displacement is not always a negative outcome. Pre-emptive evacuations, or “short-term” displacements, save lives and are an effective resilience measure. They should not, however, detract from the fact that people whose homes are severely damaged or destroyed are still likely to be forced into medium to long-term displacement.
The risk of an impending weather event displacing people for weeks or months is predicted by combining information on its likelihood, its potential severity, the number of people or dwellings in exposed areas and their vulnerability. In other words, risk is determined by far more than just the predicted severity of an event. It is also driven by settlement and construction types, their resilience or otherwise to the hazard in question, and the economic and other resources of people in potential danger.
Like any other type of prediction, displacement risk determined from an impact forecast system also comes with a range of uncertainties. One contributing factor is the nature of weather forecasts. Meteorological services often give a range of scenarios for an impending event instead of making a single deterministic prediction.
Figure 1 shows the example of the European Centre for Medium-Range Weather Forecast (ECMWF)’s predictions for tropical cyclone (TC) Yasa two days before it made landfall in Fiji in December 2020. It includes 51 possible scenarios. Each could involve completely different impacts that the decision-making process for anticipatory action needs to try to take into account.
At this stage, however, the impact forecasting modelled by CLIMADA and calibrated using displacement vulnerability from our empirical data suggested that between 3,000 and 400,000 people were at risk of being displaced by Yasa. This range is clearly far too large for planning purposes, and current research aims to address this by reducing the uncertainties.
Imagine your hike was planned on the Mount Korobaba trail in the outskirt of Suva on Viti Levu, Fiji’s main island. You might have to cancel it. Now imagine you work for the National Disaster Management Office in close collaboration with the Fiji Meteorological Office and you have to decide how to protect your population. Based on the forecast, who will you decide to evacuate, how, and where will these people go?
Uncertainties in exposure and vulnerability information and the accuracy of associated data inputs into an impact forecast could also influence its overall performance. Research efforts are being made to include all sources of uncertainty in our impact forecasting tool to enable better-informed decisions.
TC Yasa made landfall on Fiji's main islands of Viti Levu and Vanua Levu as a category 5 cyclone on 17 December 2020. It caused flooding and damage to buildings and crops on Vanua Levu, where schools and homes were destroyed and residents took shelter in public facilities. The Fijian government reported that 23,413 people were sheltering in 456 evacuation centres. Many other people were hosted by relatives, but their number is unknown so we do not have a full picture of the displacement TC Yasa triggered. Some 2,141 homes were destroyed across the country’s four divisions. 2
The implementation of impact forecast tools involves a visual presentation of information that includes warning maps, user-specific graphics and symbols to accompany and support decisions that authorities may have to take to reduce the number of people exposed to dangerous and often life-threatening hazards. Collaboration between modellers and stakeholders is important in co-designing such tools to ensure their use for the planning of anticipatory action.
IDMC will keep working closely with partners to obtain improved data on risk exposure and rethink how to assess vulnerability in the displacement risk equation. Given that “riskscapes” evolve constantly, we need to understand population and socioeconomic patterns, and fluctuations in the frequency and intensity of hazards linked to climate change.
Furthermore, the Weather and Climate Risks group will continue to pilot impact forecasting for displacement risk related to tropical cyclones, and verify the forecast predictions and accuracies with past events. A scientific paper that details the methodology will be published in 2023. Stay tuned!
Guest author, Pui Man (Mannie) Kam is Doctoral Student at the Weather and Climate Risks Institute for Environmental Decisions - ETH Zurich