Heat waves are considered among the deadliest extreme weather events worldwide (e.g. [
1]). A significant number of deadly heat waves has been observed over the last three decades. The ones of Chicago and Pakistan in July 1995 generated a mortality toll estimated respectively at 670 and 523 deaths [
2,
3]. One of the most famed heat waves was observed in several European countries in August 2003, causing an excess estimated at 45,000 deaths in 12 countries [
4]. In July 2010 in Russia, the heat waves increased the number of death by 11,000 more than the previous year [
5,
6]. In Quebec, during the five-day heat wave of July 2010, the excess daily mortality reached around 33% in the Greater Montreal area and four other public health regions [
7]. In early July 2018, a six-day heat wave caused 30% excess mortality in the same geographical region and 23% excess ambulance transportation [
8].
The increase in the number and severity of heat wave events led several countries to establish heat-health watch and warning systems (HHWWS) or early warning systems [
9]. These systems are usually based on meteorological indicators (generally maximum, minimum, or mean temperatures, and in some cases the humidity level) or on air masses (in case of the synoptic systems [
10]), and a threshold above which a significant increase in mortality is expected [
2,
11‐
14]. As in the case of the definition of heat waves, there is no universal threshold for warning systems. This is due to the fact that they reflect local weather/climate conditions and specificities of the local population [
2,
15‐
19]. Moreover, many of these thresholds are still not evidence-based on human heat-related health mortality or morbidity data [
2]. In addition, almost all the existing HHWWSs are established with a single constant threshold for the whole summer season, usually the four or five hottest months [
9,
12,
20‐
24]. The system in Spain is an exception with thresholds that vary in time throughout the year [
9]. On the other hand, according to climate projections and due to climate change, the probability of heat waves occurring early or late in the season should increase [
25‐
31]. Ouarda and Charron [
32] studied over 50 years of heat waves in six stations across the Province of Quebec. They found a non-negligible increasing trend of the intensity, magnitude, and duration of these events. Another study reported that the number of heat-wave days could increase by up to 13 days in the period 2021 to 2050 and even by up to 40 days in the period 2071 to 2100 in the Iberian Peninsula and the Mediterranean region [
33]. Acclimatization is an essential element of the human adaptation mechanism to variations in environmental heat exposure. Several studies have shown that the level of human heat acclimatization varies throughout the season, explaining why deadlier heat waves are often detected in June or July [
34‐
39]. For instance, Lee et al. [
35] have demonstrated that, over 148 cities in the U.S., heat effects of increased temperatures were larger in the spring and early summer. It is thus of public health importance to take into account human acclimatization through seasons and develop an early warning system where health-based thresholds could evolve over time, with a monthly resolution, for instance.
In Quebec, The HHWWS proposed by Chebana et al. [
12] is already implemented and integrated into public health practice in the province of Quebec. Indeed, the results of this HHWWS constitutes the basis of the automated System for Surveillance and Prevention of the Health Impacts of Extreme Weather Events (Système de surveillance et de prévention des impacts sanitaires des événements météorologiques extremes, SUPREME) [
40]. The latter is a source of information allowing regional and departmental stakeholders in the public health network to have access to health and meteorological information relating to the health impacts of extreme weather events.
The objective of the present study is to establish an extended data-driven HHWWS that evolve over the season, based on each month’s meteorological and health data (April to October in the studied case). To this end, In the available systems, including but not limited to Chebana et al. [
12], the thresholds of the climate variables are constant (the same value) over the whole season. In the present paper, the thresholds are considered not constant but evolving from month to month (each month has its own threshold) within the summer season. In addition, the proposed system is established over a time period beyond the usual hottest summer months in an extended season. Therefore, the proposed system is more realistic since it reflects the climate variability over the season, accounts for early and late heat waves as well as the population adaptation throughout the season. The purpose is to anticipate earlier, longer and hotter summers in the coming decades for the northern countries such as Canada [
41].