José Coelho / Lusa
Days, sometimes weeks, in advance, scientists can warn of an increased risk of storms, floods, droughts or other dangers. However, the cycle of destruction invariably repeats itself — because systems are built to “wait for certainty”.
In Portugal, as in other countries, after major natural disasters, public debate often tends to treat them as unexpected or unprecedented.
This reaction does not necessarily result from the absence of warnings. Rather, it reflects the way in which societies process the shock – and how authorities often explain disruptions as inevitableinstead of presenting them as having been the result of previous choices.
Extreme weather events are rarely unpredictable. With days, sometimes weeks in advancescientists are able to warn of an increased risk of storms, floods, droughts or other dangers.
However, the cycle repeats itself.
To understand why, Jeff Da Costahydrometeorology researcher at the University of Reading, and his colleagues, reconstructed the scientific warnings and official responses to those recorded in July 2021 – the most destructive catastrophe ever recorded in your home country.
These floods caused much higher damage to those who would have provoked if they had been taken early measuresbut Luxembourg is not an isolated case: many other countries suffer from the same problems that the study authors identified.
It is worth noting that the problem is not the absence of warningsnotes the Luxembourg researcher in an article in .
The problem is that alert systems are often designed to act based on certainty rather than probability – and that’s not how predictions work. By the time warnings become visible to the public, it is often too late.
Weather forecasts may seem definitive on your phone, but they are probabilistic in nature. They are created by running a series of computer simulations of future time.
O degree of agreement between the results of different simulations provides the probability of dangerous conditions, not guaranteed results. This allows meteorologists to identify high risks long before impacts occur, even if the precise location of an event and its size remain uncertain.
Crucially, uncertainty is generally greater in the longer termwhen preventive action would be more effective. Acting early therefore almost always means acting without certainty. This is not a weakness of sciencebut an inherent characteristic of anticipating complex systems under changing conditions.
O real challenge lies in the way institutions are organized to interpret, trust and act based on these probabilities.
Act based on certainty
Most alert systems are based on predefined procedural thresholds: alert levels, activation protocols and emergency plans that come into action when specific criteria are met.
Forecasts may indicate that floods are increasingly likelyfor example, but measures such as evacuations or road closures can only be triggered after formal thresholds have been exceeded.
Before that moment, risk information passes through many layers of interpretation and evaluationwhere early signs are often registered but do not generate action.
Thresholds serve important purposes. They help coordinate the response, clarify chains of command and reduce unnecessary disruptions.
But they also incorporate a structural preference for certainty. The action is only authorized when the risk is considered imminenteven when credible evidence already points to a growing danger.
This attitude was evident in the days leading up to the July 2021 floods.
Da Costa’s, published on January 20th in European Geosciences Uniondemonstrates that multiple forecasts at European level and in Luxembourg indicated a high probability of extreme precipitation and floods, in some cases a week in advance.
This information was available in different parts of the alert system. At that time, there was naturally uncertainty about the precise impacts. But what really mattered was how the warning system was designed to deal with this uncertainty.
Too early to warn
As Luxembourg’s response measures were linked to procedural thresholds, the Early signs could not translate into action anticipatory.
The country’s water administration and national meteorological service had access to relevant information, but operated within a structure that did not allow for a collective interpretation of what was happening. nor encouraged action before thresholds are exceeded.
This was not a scientific miscalculationnor was it necessarily an operational failure of individual bodies. The meteorological and hydrological services probably did as much as their mandates allowed.
A decision to wait for formal triggers it was human and institutional, not technical, reflecting a system designed to privilege procedural certainty over sensible decision-making.
When the action was authorized, For many people it was already too late. Evacuations or installing floodgates have become much more difficult, particularly for communities with limited experience of such severe flooding.
From the perspective of those affected, the warnings seemed late or they simply didn’t arrive – even if the risks had been identified earlieror throughout the system.
Luxembourg is a particularly instructive illustration of what could go wrong, because It is a small, rich and well-connected country. The problem was not necessarily the lack of resources or scientific capacity, but of institutional design and social preparation to act in the face of risk.
Learning and resilience
The effectiveness of early warning systems over time depends on their ability to learn from extreme events. This requires an open and independent analysis of what worked, what didn’t and why.
In several neighboring countries affected in 2021, such as Germany and Belgium, formal inquiries and external reviews were carried out. In Luxembourg, no.
When a expert criticism is discouraged or avoidedlearning slows down. Questions about system performance remain unresolved and the same structural vulnerabilities tend to persist.
This creates a systemic risk alone: Societies become less able to adapt warning systems, interpret uncertainty and act sooner in the face of emerging threats.
As someone who has worked within these systems and continues to research disaster risk governance, Jeff Da Costa has seen how to ask difficult questions can be treated as destabilizing rather than constructive. Resilience depends on facing uncomfortable truths, not avoiding them.
O risk of extreme weather events is set to increase across Europe and beyond. Early warning systems are rightly central to reducing the risk of disasters. But its effectiveness depends on the way societies authorize action in the face of uncertainty. This is a choice, not an inevitability.
Uncertainty cannot be eliminated. The challenge is deciding how much uncertainty is acceptable when lives and livelihoods are at stake.
Os systems designed to wait for certaintyfor procedural, organizational, financial or reputational reasons, are more likely to issue warnings that arrive too late to be felt as warnings.
If resilience to future climate risks is to be sustainable, warning systems must be designed to learn, adapt and act sooner when faced with credible risks.
As lessons learned from the study by the Luxembourg hydrometeorologist, published just 3 weeks ago, arrive too late to prevent the effects of the natural disasters that have hit Portugal in recent days. Maybe they can arrive in time to stop them in the future — so that the cycle does not repeat itself.
