Categorization and multi-criteria analysis of heat wave vulnerability in Senegal

Recent variations in global climate have been accompanied by a recurrence of thermal extremes (Seneviratne et al. 2012). On a global scale, IPCC forecasts indicate that heat waves are expected to become more severe and recurrent and will go hand in hand with an increase in hot days and nights (Stocker et al. 2013). Thus, there appears to be an increase of between 1 and 6 °C by 2100, with the persistence of extremes, particularly in the Sahel (Kovats & Kristie 2006), which will have a considerable impact on the Sahelian zone (Smith et al. 2014; WMO 2015). Fontaine et al. (2013) indicate a 1–3 °C temperature rise in the sub-Saharan continent over the time sequence 1979–2011.

This dynamic evolution of the world's climate explains the scientific community's growing interest in studying and understanding meteorological extremes. Despite this scientific trend, heat waves are little known in Africa and the Sahel. This scarcity of studies is partly explained by the availability and quality of climate data, which reduces the confidence level of studies between 60 and 90% (Diouf 2018).

Despite these constraints, work has been done to study recent climate changes on a global scale. For example, the IPCC has identified an estimated global warming of 1.5 °C (Rome et al. 2019). This same trend has been observed by several authors (Fontaine et al. 2013; Moron et al. 2016; Oueslati et al. 2017). It is in this sense that Ringard et al. (2016) established a correlation between these climate changes and the severity of heat waves in the Gulf of Guinea and the central part of the Sahel.

Extreme events often referred to as mega-heat waves have been observed in other parts of the world, such as in Chicago (USA) in 1995, causing about 365 deaths (Naumova et al. 2007; Sy et al. 2022). In France, the heat waves of 1983 and 2003 caused more than 300 deaths (Cadot & Spira 2006; Fouillet et al. 2006) and 15,000 deaths, respectively (Besancenot 2005; Sahabi-Abed & Kerrouche 2017). In the Sahel region, Ringard et al. (2014) reported extreme temperatures in April 2010 that reached 47 °C in the shade, particularly in Niger. This event caused considerable damage to populations unprepared for this type of hazard. This heat wave resulted in around 40 deaths per day, generally among the elderly and children (Barbier 2017). In Senegal, in 2013, hot extremes were noted, with temperatures exceeding 45 °C, leading to the deaths of around 27 people (Sy et al. 2022). As a result, it appears that heat waves seriously affect the elderly, the chronically ill people, children, and infants (Sahabi-Abed & Kerrouche 2017). In addition, these extremes significantly affect economies (Schaller et al. 2018; Li 2020; Yao et al. 2020).

Given the above, it is undeniable that heat waves are omnipresent in different parts of the world, especially in Africa, where forecasts are confirmed daily. The heat waves detected can lead to varying impacts in different areas depending on the thresholds. Therefore, it is not advisable to analyze the severity of these phenomena in terms of absolute temperatures but rather in terms of differences from threshold values, often called percentiles (Besancenot 2002). This explains the need to analyze the characteristics of heat waves and break them down into individual categories. This work thus provides a scientific basis for a better understanding of these climatic phenomena, with a view to implementing comprehensive, endogenous, appropriate, and sustainable adaptation strategies.

In this context, the present study seeks to examine the categories and hierarchy of extremely hot events in Senegal. The aim is to categorize and analyze the vulnerability of the Senegal region to heat waves using a multi-criteria GIS approach.

The methodology of this work revolves around two main axes: the categorization of heat waves recorded over the 1984–2020 period and an assessment of community vulnerability to heat waves based on a multi-criteria analysis.

This study uses NASA climate reanalyses as a data source (https://power.larc.nasa.gov) produced by combinations of ground observation data and climate model outputs (Ndiaye 2021). This method makes it possible to obtain the various climate variables (Poccard-Leclercq 2000; Barbier 2017) with quality data and a better global spatial distribution (Ruan et al. 2015; Martins et al. 2016). Thus, from 1981 to the present day, several climatic parameters (minimum, maximum and mean temperatures, precipitation, relative humidity, wind speeds and directions, and insolation) are available at daily, monthly, and annual time intervals. In addition, these data are at a spatial resolution of 0.5° × 0.625°, thus ensuring better land coverage. This is why, in this work, reanalyses were chosen to circumvent the difficulties of accessing observed data, which can often present several problems. The coordinates of 40 stations across Senegal were used to extract average daily temperatures over the period 1984–2020 (Table 1).

An analysis of heat waves reveals a wide range of definitions whose effectiveness varies geographically. Thus, it is necessary to categorize heat waves to see which categories are most common in the study area. In the present study, the heat wave naming method of Hobday et al. (2018) was used. Once heat waves have been detected, we need to check whether they meet specific criteria (intensity, duration, and proportion). Hobday et al. (2018) have established four categories to classify heat waves.

Thus, in category I (moderate), the heat waves detected must have an intensity that does not exceed more than twice the amplitude between the average climatology and the threshold value, which corresponds to the 95th percentile in this study. For category II (strong), the maximum intensities must be at least twice the difference between the threshold and the climatology for the period. Category III (severe) lists heat waves whose intensity is at least three times greater than the amplitude between the percentile and the average climatology for the period. Finally, category IV (extreme) lists extreme events whose intensity exceeds four or more times the amplitude between the threshold and the average for the period.

The heat waves that are detected are then subjected to the heat wave naming method of Hobday et al. (2018) to highlight their category as well as their seasonality. This work was carried out using the heat wave R package of the R software: https://www.rdocumentation.org/packages/heatwaveR/versions/0.4.5. This method has been used by Hobday et al. (2016); Hobday et al. (2018); Schlegel & Smit (2018); Asuquo & Oghenechovwen 2019; Tassone et al. (2023).

The methodology applied to multi-criteria analysis is based on the identification of relevant criteria that form the basis of this assessment. In this work, a set of eight criteria was chosen: the number of extreme events detected per site, the number of heat wave days recorded over the period, the average duration of an event, the maximum duration, the maximum intensity, the average intensity, the proximity to the ocean in terms of distance, and the 95th percentile for the period.

The 95th percentile is one of the criteria used to identify heat waves. It is thus a threshold value above which a climatic event is considered extreme. The number of extreme events corresponds to the total number of heat waves recorded per station, i.e., the number of times the daily temperature exceeded the 95th percentile over a minimum time sequence of 3 days over the period 1984–2020.

In addition, heat waves are defined by their intensity or severity and their duration. The latter is divided into two criteria: average duration and maximum duration. The former identifies the average time a heat wave lasts in the study area, while the latter reflects the longest period an extreme event has lasted over the 1984–2020 period. Intensity, on the other hand, refers to the extent to which the identified threshold value is exceeded. The maximum intensity criterion lists the highest temperatures, while the average intensity represents the mean temperature recorded during the heat wave.

In keeping with this logic, in order to integrate the effects of local and geographical factors in the characterization of heat waves, proximity to the ocean was chosen as a criterion. The study area has a very marked ocean/continent climate duality. Therefore, the distance of the various stations from the ocean must be estimated.

These different criteria are weighted, normalized, and aggregated using the analytic hierarchy process (AHP). The AHP is an analytical method using several criteria to facilitate decision-making. It is a method that allows the criteria to be compared in pairs and weights to be assigned to each (Table 2).

Table 2

Appraisal value

View Large

The AHP analysis is based on the adjustment of the evaluation criteria, which, based on the peer comparison (Table 3), makes it possible to assign a weight to each by summing the columns of each row. The weighting of the various criteria is obtained by the linear combination method, based on pairwise comparison according to Saaty's (1977) AHP. This method was subsequently used by Youan Ta et al. (2011) and Doumouya et al. (2012). Table 3, for example, provides the weights of the selected criteria. This allows us to rank the criteria according to their contribution to the final index.

Table 3

Weighting of criteria

View Large

This section looks at assessing the consistency of the matrix before using the weights. The idea is to assess the coherence ratio (CR) in order to see whether the judgement or the matrix is coherent. This assessment requires the determination of various indices and ratios: the random index, the consistency index, and the consistency ratio. Compliance with this procedure guarantees the reliability of the decision matrix, which reinforces the authors' judgements and choices in relation to the evaluation criteria chosen. This evaluation involves satisfying the following stages:

• - Determining random coherence

Saaty establishes a maximum acceptable error value of 10%. Beyond this value, the work will have to be repeated. Thus, if CR ≤ 0.1 or CR ≤ 10%, the matrix is considered sufficiently coherent; if this value exceeds 10%, the assessments require revisions. Therefore, the closer the consistency ratio is to 0, the more consistent the assessment. In this study, the consistency ratio obtained is 0.059 (5.9%) which indicates the comparison matrix is chosen is consistent to continue with the aggregation of criteria with the corresponding weights.

• - Criteria aggregation

Over the period 1984–2020, between 44 and 74 heat waves were recorded per station (Figure 2). Specifically, minimum temperatures were recorded in the west (Dakar and Rufisque), south (Ziguinchor, Sédhiou, and Kolda), east (Tambacounda and Goudiry), and center, corresponding to the Koungheul and Koumpentoum stations. The highest temperatures were recorded in the country's central, northern, and eastern parts, at stations in Podor, Ranerou, Kanel, and elsewhere.

Spatially, moderate heat waves are highly localized in the Saint-Louis, Matam, Louga, and Diourbel regions. These regions experienced between 56 and 72 heat waves over the study period. The southern part of the country experienced between 33 and 49 moderate heat waves (Figure 3(a)). Severe heat waves were more widespread in the southeastern part of the country, in the regions of Tambacounda, Kolda, and Kédougou, where between 9 and 15 extreme heat events were recorded. The rest of the country experienced fewer than six such heat waves on average (Figure 3(b)). Furthermore, severe heat waves were mainly encountered in the south of the country, in the regions of Kolda, Sédhiou, and parts of Kaffrine, Kaolack, and Ziguinchor.

One can deduce that moderate heat waves are widespread in the region, strong ones are more frequent in the southeast, and severe ones are localized in the central south.

In spring, the country's north is subject to heat waves, with 17–24 extremes recorded per station (Figure 5(a)). This situation extends to Diourbel, Fatick, and part of Tambacounda. The rest of the country is affected between 8 and 14 extreme heat waves, with minima recorded at stations in the Kolda region.

During winter, heat extremes are rare and more concentrated in the north, covering parts of the Saint-Louis, Matam, and Louga regions (Figure 5(b)). The number of heat waves detected ranged from 12 to 17 extremes. The center, parts of the north, east, and south recorded between 9 and 12 heat waves per station from 1981 to 2020. The south, around Ziguinchor and Sédhiou, is not much affected by heat waves during this period. The occurrences of heat waves at these southern stations are less than eight heat waves per station.

In autumn, between 16 and 22 heat waves are recorded in parts of the north (Saint-Louis, Matam, and Louga), south-west (Ziguinchor and Sédhiou), and southeast (Kédougou and part of Tambacounda) (Figure 5(c)). The west, center, and east are less affected by heat waves, having experienced fewer than 16 hot extremes.

Finally, during the summer, the north and part of the Kédougou region are more prone to heat waves. Between 17 and 23 extremes are recorded during this season. The west, mid-west, and much of the south are less affected during this period.

Overall, in spring, between 25 and 35% of heat waves are recorded in the northern and central parts of the country, compared with less than 25% in the southeast (Figure 6(a)). In winter, between 16 and 25% of extremes are recorded over most of the country, compared with less than 15% in the west (Figure 6(b)). In autumn, between 30 and 38% of heat waves occur in the southern part of the country at the Ziguinchor, Sédhiou, and Kolda stations (Figure 6(c)). In summer, between 28 and 38% of heat waves are detected in the southeast and north around the city of Saint-Louis (Figure 6(d)). In the rest of the country, only less than 28% of extremes are recorded during this season. To sum up, heat waves in Senegal are somewhat irregular from one season to the next. In spring, they are more frequent in the north. In autumn, they are more recurrent in the east; during the summer, they are highly localized in the south. In summer, hot extremes are more common in the southeast.

Analysis of Figure 7 reveals an east–west gradient in the spatial evolution of vulnerability. There is very little or no vulnerability in the west, which corresponds to the Dakar region. This is due to the rarity of hot extremes in this region and its characteristics, which are less severe than in other parts of the country. After Dakar, the coastline is characterized by low vulnerability to hot extremes. This part covers the coastal regions, even reaching parts of Louga, Diourbel, Fatick, and Sédhiou, which are marked by fewer heat waves and often have more moderate characteristics. This situation in the western regions and parts of the center can be explained by the presence of a Sahelian coastal climate marked by incursions of oceanic air masses (the maritime trade winds). The country's center is subject to average vulnerability to extremes, with occurrences and intensity averaging between 35 and 37 °C. The east of the country is marked by high vulnerability to heat waves, whose frequency, intensity, and duration characteristics are harsher than in western regions. Extremes are more recurrent (an average of between 333 and 412 heat waves over the 1984–2020 period), which are longer, lasting up to 20 days, and are more severe, exceeding the 95th percentile by an average of 4 °C. Finally, a high vulnerability to hot extremes is noted in this part of the east, corresponding to part of the Tambacounda region at the Gamon station. The statistical properties of heat waves there are the harshest compared with other parts of the east. Overall, the east is highly vulnerable, which is justified in view of the climatic and geographical factors of the environment. This region is subject to hot and dry winds from the Saharan desert. Vulnerability is high due to the continental Sahelian climate that prevails and the plain-type relief that generates greater solar absorption.

The aim of this work was to categorize the characteristics of heat waves in Senegal and analyze the geographical vulnerability to this extreme heat. Combining these two objectives has enabled us to better understand the spatial component of heat waves and their properties.

The results show that Senegal was affected by between 44 and 72 heat waves per station over the period 1982–2020. Significant disparities remain between east and west. Indeed, the West African coast is relatively rarely subjected to heat waves (Rome et al. 2019). As a result, hot extremes are more frequent in the eastern part of the country than in the west. This situation is corroborated by the work of Sy et al. (2022), for whom the north–east is subject to recurrent heat waves with significant variations of −2 to 3 °C between 1971 and 2000. This situation is quite similar in the Sahel, where heat waves are more frequent (Ringard 2014). The presence of hot extremes in the Sahel and Sahara is explained by the opposition of wind anomalies to the mean circulation of the upper and middle tropospheres (Fontaine et al. 2013).

Analysis of the categories shows that heat waves are present in all parts of the country but to varying extents. Severe heat waves are very rare (202 events between 1981 and 2020) and occur mainly in the southeast of the country. Those in the severe category are spread across the south. It should be noted that severe or even extreme events are fairly rare throughout Senegal. These severe extremes are localized over short periods (Rome et al. 2019).

Regarding seasonality, it is clear that heat waves are counted year-round, without distinction of season. However, peaks are noted during summer and spring, with 28.34 and 27.58% of heat waves, respectively. Rome et al. (2019) have shown that heat waves are persistent in the Sahel in spring, covering between 20 and 30% of days in this season. Moreover, this recurrence is perceived as a public health problem (Sy et al. 2022). Diouf (2018) corroborates this finding by showing an increase in the discomfort index during spring and summer.

The results of the heat wave vulnerability analysis reveal significant spatial variability. Unlike the western regions, the eastern areas are highly vulnerable to hot extremes and their characteristics. This result aligns with the work of Ndiaye (2021), who found relatively low temperatures along the coast. Concerning hot extremes, Sy et al. (2022) showed that there was an increase in consultations and excess mortality of +25 people on hot days in the departments of Matam, Bakel, Dagana, and Louga. The populations of the northeast are thus highly exposed to heat waves and the associated health risks (Sy et al. 2022).

In addition, these thermal extremes are known to affect various socio-economic domains. In the field of fisheries and agriculture, heat waves have a negative impact on biodiversity and marine ecosystems through the loss of production and inhabitants through the development of diseases, parasites, and harmful algae (FAO 2023). In agriculture, prolonged heat waves disrupt photosynthetic activity through prolonged water stress, resulting in reduced crop yields (Lobell et al. 2008). Chakraborty et al. (2000) show that these extreme events greatly influence the frequency and distribution of pests and pests, thus increasing the risk of agricultural losses. With regard to animal husbandry, Hahn & Mader (1997) show that heat waves cause heat stress in animals, causing problems with growth, milk productivity, and susceptibility to certain diseases.

This work aimed to categorize heat waves detected in Senegal from 1981 to 2020 utilizing the methodology of Hobday et al. (2018) and analyze the vulnerability of territories to hot extremes. In addition, multi-criteria analysis was used to understand the regions more vulnerable to heat waves in Senegal.

The results show that heat waves are almost present throughout Senegal, with 2,378 events detected over the period 1984–2020. These are divided into 2,166 moderate heat waves, 202 strong, and 10 severe heat waves. In addition, there is considerable seasonal variability in heat extremes. This seasonal frequency is around 28.34% of heat waves in summer, compared with 27.58% in spring, 27.10% in autumn, and 16.82% in winter. The peaks are observed in spring (April–May–June) and summer (July–August–September). In addition, heat wave areas have a high degree of variability, with inland regions, especially in the east, being more severely affected. The vulnerability analysis confirms this spatial trend where the gradient of vulnerability varies from the lowest in the coastal regions and some central parts to the strongest in the interior regions, especially in the east. This work of categorization and vulnerability analysis meets two objectives. On the one hand, it lays the foundations and scientific arguments necessary for the knowledge of climate extremes. It is thus a task that makes it possible to verify in a local framework the similarities and differences in the trends identified at the global level or in other regions. On the other hand, the study area is characterized by a scarcity or even an absence of studies on extreme climatic phenomena related to temperature due to the problem of observed climatic data. Therefore, the study aspires to be an essential scientific reference for the analysis of the characteristics of heat waves in West Africa and Senegal in particular.

Based on the results of this study, it would be interesting to analyze future trends in heat waves in Senegal, as well as their socio-economic and health impacts.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.