Meteorological and Biometeorological Evaluation of a Population-Aware Heat Warning Framework for Mauritius

Heat is not a single thing. To a person standing in the sun, it is a mix of the air temperature, the humidity that keeps sweat from evaporating, the wind that might cool the skin, and the direct radiation from the sun beating down. For scientists who study how weather affects human health, this means there is no single number that can tell the whole story of a dangerous heat day. A heatwave that lasts for three days feels different from a single scorching afternoon, and a hot night that offers no relief is a different kind of stress than a hot day with a breeze. In places like Mauritius, a humid island nation in the Indian Ocean, the challenge is to build a warning system that understands these differences. The goal is to know when the heat is truly dangerous for the people living there, not just when the thermometer hits a high number.

Yashvir Ramlukon, a researcher with the Mauritius Meteorological Services, set out to build and test a new way to measure this heat. He created a framework that looks at the island through six different zones, from the coastal lowlands to the central highlands, and asks a simple but difficult question: how many people are actually experiencing dangerous heat, and what kind of heat is it? To answer this, he did not rely on just one type of data. He combined direct measurements from weather stations on the ground with a sophisticated computer model of the atmosphere called ERA5-Land. This model fills in the gaps between stations, creating a complete picture of the weather across the whole island for every hour of the day. However, the researcher knew that computer models are not perfect. They can smooth out the sharpest peaks of heat, so he treated the real station measurements as the most important evidence for what actually happened on the ground.

The study covered a long period, from 1991 to 2025, creating a massive archive of daily heat records. The researcher tested several different ways to define a heat warning. One method looked at how hot the days were compared to what is normal for that specific time of year. Another measured how hot the nights were, because a lack of cooling at night is a major health risk. A third method, called the Excess Heat Factor, checked if the heat was a sudden shock to the body or a persistent event that the population had already started to get used to. Finally, he used a complex index that simulates how the human body reacts to the combination of temperature, humidity, wind, and sun. This index, known as the Universal Thermal Climate Index, tries to calculate the actual thermal stress a person feels, rather than just the air temperature.

When the researcher compared the computer model to the real weather station data, he found that the model was good at showing the general pattern of temperature changes but often underestimated the hottest days. At some lowland stations, the model was cooler than reality by nearly two degrees Celsius. This meant that if the warning system relied only on the computer model, it might miss the most dangerous heat events. The model also struggled to catch the very hottest days that happen only rarely. Because of this, the framework was designed to trust the real station data when checking for historical extremes, using the computer model only to fill in the gaps where no station existed.

The study also looked at how many people were affected by these heat events. Instead of averaging the temperature across the whole island, which would hide the fact that some areas were much hotter than others, the researcher calculated the fraction of the population living in zones that exceeded the danger thresholds. He used the official population counts from the 2022 census to weigh these zones. This approach revealed that different heat events affect people in very different ways. For example, a heat event in December 2013 was severe because it combined hot days, warm nights, and a high level of thermal stress that lasted for several days. In contrast, a heat event in February 2025 was intense enough that the thermal stress index reached dangerous levels across the entire island, yet it did not trigger the "persistence" warning because it was not part of a long, multi-day buildup.

One of the most important findings was that no single warning rule could catch every dangerous situation. The researcher tested different combinations of rules to see which ones matched the real heat events best. He found that a system that only looked at how unusual the temperature was compared to the past would miss events that were physically very stressful but not statistically rare. Conversely, a system that only looked at the thermal stress index might miss events that were dangerous because of their duration. The best approach was to keep these different measures separate and look at them together. This way, a warning could be issued if a heatwave was persistent, or if the thermal stress was high, or if the nights were unusually warm, without forcing all these different types of danger into a single score.

The research also highlighted a specific event in March 2019 that served as a crucial test. The computer model suggested the heat was not severe, but the real weather stations showed a strong, multi-zone heat event. This discrepancy proved that relying solely on the computer model could lead to a failure to warn people. The framework successfully identified this event as a case where the model had failed, reinforcing the need to keep real observations at the center of the warning system.

Ultimately, this work does not claim to have solved the problem of heat-related illness. The researcher explicitly states that the warning levels developed here are based on meteorological and physical evidence, not on health data. There is no record of how many people got sick or died during these heat events in the data used for this study. Therefore, the system cannot yet say with certainty which specific temperature or stress level causes a health crisis. Instead, the study provides a solid, reproducible foundation. It offers a clear, detailed record of how heat has affected the population of Mauritius over the last thirty-five years, separating the different types of heat stress and showing exactly how many people were exposed to each. This archive is now ready for health experts to use. They can take this detailed map of heat exposure and match it with future health records to find the exact point where the weather becomes a medical emergency. Until that health calibration is done, the system stands as a scientifically rigorous way to measure the heat, ensuring that when a warning is issued, it is based on a complete and honest picture of the weather, not just a single number.