Intensifying human-driven heatwaves characteristics and heat related mortality over Africa

Heatwaves in Africa are expected to increase in frequency, number, magnitude, and duration. This is significant because the health burden is only expected to worsen as heatwaves intensify. Inadequate knowledge of the climate’s impact on health in developing nations such as Africa makes safeguarding the health of vulnerable groups at risk challenging. In this study, we quantify possible roles of human activity in heatwave intensification during the historical period, and project the future risk of heat-related mortality in Africa under two Representative Concentration Pathways (RCP26) and (RCP60). Heatwaves are measured using the Excess Heat Factor (EHF); the daily minimum (Tn ) and maximum (Tx ) are used to compute the EHF index; by averaging Tx and Tn . Two heat factors, significance and acclimatization are combined in the EHF to quantify the total excess heat. Our results confirm the intensification of heatwaves across Africa in recent years is due anthropogenic activity (increase in greenhouse gas concentration and changes in land use). The Return event highlights the potential future escalation of heatwave conditions brought on by climate change and socioeconomic variables. RCP26 projects a substantial rise in heat-related mortality, with an increase from about 9000 mortality per year in the historical period to approximately 23 000 mortality per year at the end of the 21st century. Similarly, RCP60 showed an even more significant increase, with heat-related mortality increasing to about 43 000 annually. This study highlights the potentially growing risk of intensifying heatwaves in Africa under different emission scenarios. It projects a significant increase in heatwave magnitude, number, duration, frequency, and heat-related mortality. Africa’s low adaptive capacity will amplify the impact, emphasizing the need for emissions reduction and effective adaptation measures.


Introductions
As a result of global warming, heat waves are becoming more frequent and intense, which is particularly concerning for Africa (Russo et al 2016, Déqué et al 2017, weber et al 2018), with devasting consequences for human health (Gasparrini et al 2017).Around 29 936 896 deaths have been attributed to heat-related causes globally from 1991-2016(Vicedo-Cabrera et al 2021).A recent study by the World Health Organization (WHO) shows that heatwaves are responsible for an average of 12 000 deaths per year in Africa, and this number is expected to increase as the average surface temperature increases.This is a significant health burden that is only expected to worsen as temperatures continue to rise.Climate change is a basic component that presents broad and persistent threats to public health, such as the rising death toll due to high temperatures (Li et al 2015, Weinberger et al 2017).It is projected that Africa will see higher-than-average temperature rises by the end of the century (Engelbrecht et al 2015, Baker and Anttila-Hughes 2020), considering the sizable sociodemographic and rising greenhouse gases (GHG) emissions from human activities.In such an event, the African continent's future population is in jeopardy.
Human activities are largely responsible for the present increase in global average temperature (Gillett et al 2008), which is reflected in higher atmospheric quantities of greenhouse gases such as carbon dioxide and methane.Heat extremes have grown more common and severe, primarily due to climate change attributed to human activities and local variables like urbanization (Perkins and Alexander 2013, Adigun and Koji 2022, Seun et al 2022).The continent is also undergoing rapid demographic and economic shifts, posing severe health challenges.There has been a 50% rise in heatwaves in Africa since the 1960s (WMO).The intensification of heatwaves in Africa has been attributed to the global average temperature, which has risen by about 1 degree Celsius since the pre-industrial period, and this warming is projected to continue, contributing to the worsening of heatwaves in Africa (Lennard et al 2018, Masson-Delmotte 2018).Lost of vegetation cover in the last several decades, contribute to the worsening heat waves (Jaafar et al 2020, Palafox-Juárez et al 2021, Adigun and Koji 2022).Since urban regions often have more heat-emitting concrete and asphalt, they tend to be warmer than their rural counterparts.(Seun et al 2022).
Previous studies (Liu et al 2023, Vicedo-Cabrera et al 2021, Zhao et al 2021) have highlighted the significant impact of rising temperatures on mortality.The connection between temperature and mortality can vary significantly depending on the region.Various factors, such as urbanization, climate change-related effects, population density, geographic factors, social structures, the built environment, and the ability of communities to adapt, are likely to play a role in influencing mortality rates (Zeng et al 2022).Inadequate knowledge of the climate's impact on health in developing nations such as Africa makes safeguarding the health of the most at-risk people challenging.Cities in Africa are especially at risk of experiencing greatly increased exposure to dangerous heat in the coming decades (Rohat et al 2019).This study aims to examine the synergistic contributions of anthropogenic activities and Shared Socioeconomic Pathways scenarios to the intensification of heatwave characteristics throughout the African continent.This entails an examination of the impact of anthropogenic forcing in heightening the prevalence and intensity of heat waves throughout the region.Also, to examine the projected changes in heatwave patterns under various socioeconomic development scenarios, our study thoroughly explores the intricate interplay between anthropogenic factors and socioeconomic pathways in shaping heatwave dynamics.Our Secondary focus is quantifying future risks associated with heat-related mortality in Africa.Leveraging the ISMIP2b experiment, we employ a forward-looking approach to assess the evolving landscape of heat-related mortality.This dual-pronged investigation advances our understanding of the complex nexus between anthropogenic influences, socioeconomic trajectories, and heatwave characteristics.Providing valuable insights into the potential health ramifications of such climatic shifts.

Watch Forcing Data ERA5 (WFDE5) Dataset
The minimum and maximum temperature datasets utilized are retrieved from the WFDE5 dataset (Cucchi et al 2020).The WFDE5 dataset provided bias-corrected minimum and maximum daily temperature data from the ERA5 reanalysis over 1980-2014.The WFDE5 is a bias-corrected version of the ERA5 reanalysis dataset.The European Centre for Medium-Range Weather Forecasts (ECMWF) developed the ERA5 dataset, a high-resolution global atmospheric reanalysis dataset (Hersbach et al 2020).To improve the reliability of the ERA5 dataset, especially in the representation of extreme events like heatwaves, WFD methodology (WATCH Forcing Data) was applied to the ERA5 dataset.The WFDE5 dataset has a spatial resolution of 0.5 degrees because it was generated by aggregating the higher-resolution ERA5 dataset rather than interpolating the lower-resolution WFDE4 data.  1.The Large ensembles enabled robust quantification of human contribution to observed heatwave intensification and projections of future changes.To quantify human contribution to heatwave intensification, sensitivity experiments isolating the effects of individual external forcings (i.e.GHG, anthropogenic aerosols, natural forcing, land use change) were analyzed and compared to the historical ALL forcing runs.20 ensemble member of transient simulations driven by historical ALL forcing (hist-ALL), which is forced with (well-mixed greenhouse gas(WMGHG), black carbon (BC), organic carbon (O.C.), Sulfur dioxide (SO2), sulphate (SO4) Nitrogen oxides (Nox), ammonia (NH3), carbon oxide (CO), Non-Methane volatile organic compounds (NMVOC), solar irradiance, nitrogen deposition, stratospheric aerosols, land use, and Ozone).10 ensemble members for each of the external forcing scenarios, which included historical aerosol (hist-AER), which is forced with (SO2, BC, NOx, O.C., SO4, NH3, CO, and NMVOC).Historical natural (hist-NAT) forcing simulation, which is forced with (solar irradiance and stratospheric aerosols).Historical greenhouse gasses (hist-GHG) which are forced with well-mixed GHG (Gillett et al 2016).The historical Land-land only (hist-Land) depicts how land management, land use, and land cover changed over time.Relevant land models such as nitrogen deposition, carbon dioxide (CO2) concentration, aerosol deposition, and population density are used to force the hist land changes (Lawrence et al 2016).Two Shared Socioeconomic Pathways are employed, SSP245 to represent a moderate mitigation scenario and SSP585 to represent a high emission scenario.

ISMIP2b
Daily minimum and maximum temperature projections from four models, including GFDL-ESM2M, HadGEM2-ES, IPSL-CM5A-LR, and MIROC5 (refer to Table 1 for full description of the models) are retrieved from the Inter-Sectoral Impact Model Intercomparison Project (ISIMIP) Phase 2b, which are used to calculates heatwave characteristics under Representative Concentration Pathways (RCP2.6)and (RCP6.0)(Warszawski et al 2014).The outputs generated by the ISIMIP2b models, are used for estimating total annual heat-related mortality under two Representative Concentration Pathway (RCP26) and (RCP60).These models provide climate-related data, which, in turn, informs the health impact models.By integrating this climate data with health impact modeling, we can derive accurate projections of heat-related mortality.In essence, the ISIMIP2b model outputs serve as a crucial input for our analysis, allowing us to assess the potential impacts of climate change on heat-related mortality.The ISIMIP2b combines climate and health impact models to estimate historical and projected heat-related deaths which combines temperature information with mortality, which is freely available at www.isimip.org/protocol/3/.The experiment annually examines climate change's effects on heat-related mortality and then estimates the number of fatalities that will occur due to heat stress using a health impact model (Honda et al 2013).The mortality data in the ISMIP2b dataset are records or statistics relating to human fatalities (Honda et al 2013).

Heatwave metrics
Heat waves are measured and recorded using the Excess Heat Factor (EHF), EHF is based on the concepts of overheating and thermal stress (Varghese et al 2019).The daily mean temperature is used to compute the EHF index; by averaging the daily maximum (T x ) and minimum (T n ) temperatures.The EHF formulation has been enhanced by including daily minimum temperatures, which implicitly correct for changes in humidity (Perkins and Alexander 2013), allowing for more chances to study the effects of heat on human health.An increase in average temperature over the recent historical norm might cause heat stress, even if this temperature increase is temporary.Two heat factors, significance (EHIsig) and acclimatization (EHIaccl), are combined in the EHF to quantify the total excess heat over three consecutive days (Perkins and Alexander 2013).Which is written as where the daily maximum temperature is denoted as T x , the daily minimum temperature (T n ), the daily mean temperature is denoted as (T), the temperature on the ith day is denoted (T i ), the climatological temperature at the 90th percentile is denoted as (T90).The EHIsig indicates the current temperature deviation over the last three days compared to the comparable climatological threshold.The EHIacc indicates the departure in temperature from the preceding month's average for the current three-day period.
In this study we focus primary on four heat wave metrics which are heatwave magnitude (HWM), heatwave frequency (HWF) heatwave number (HWN) and heatwave duration (HWD).In the estimation of the Return event, the generalized extreme value (GEV) distribution is a limit distribution that results from the maximum of a series of independent and identically distributed random variables after they have been properly normalized such that the cumulative density function of the distribution is written as: where µ, σ, ξ and are the distribution's location, scale, and shape parameters.If ξ =0, the distribution becomes Gumbel since it dictates the tail behavior; positive denotes heavy tail, while negative denotes light tail (Wang et al 2017).Each grid box's extreme is determined based on the GEV distribution.Extreme value analysis on HWM, HWD, HWN, and HWF was performed by involving the application of a GEV distribution model at every grid point, as outlined by Coles (2001).The return period is defined as the value that exceeds the annual extreme at least once yearly and was obtained by inverting the fitted GEV distribution.We then estimate the 2-year and 20-year return periods under the historical period  and SSP245 (2050-2100).Estimation of heat-realted mortality was leverage on historical heat-related mortality dataset from the ISMIP2b experiment spanning the period from (1980 to 2005), serves as a baseline for our analysis.To calculate future mortality, we considered projections from 2006 to 2099.Our approach involved assessing the change in heat-related mortality relative to the historical period, effectively quantifying the difference between past and future mortality rates.To achieve this, we aggregate the outputs of four climate models (GFDL-ESM2M, HadGEM2-ES, IPSL-CM5A-LR, and MIROC5) under RCP2.6 and RCP6.0 scenarios.The projected changes in annual heat-related mortality relative to the historical period  are then calculated; this allows us to quantify the projected changes in mortality attributed to future heat stress increases under both scenarios.This comparison allowed us to gauge the anticipated increase or decrease in mortality due to heat stress as projected for the 21st century.We further developed an ensemble machine learning model to predict heat-related mortality as a function of heatwave characteristics (duration, number, frequency, and magnitude) and population size.We utilized the historical dataset   and RCP26  .Scenario data of mortality, population, and heatwave characteristics is provided by the ISMIP dataset.Our projection analysis was restricted to RCP26 scenerio, given that it was the only scenario with available population projections in the ISMIP dataset.
Random forest model was trained on the historical and RCP26 scenarios to capture nonlinear relationships between the target mortality as a feature of heatwave characteristics and population.These predictors which are (heatwave characteristics and population) were chosen based on prior domain knowledge indicating their influence on heat-related mortality risk (Rohat et al 2019, Weber et al 2020).The model took the mathematical form, written as Y = f (X1, X2) + ε.
Y denotes annual heat-related mortality.f () denotes the Random forest modeling function, X1 denotes Annual heatwave characteristics, X2 denotes annual population.ε denoted the error term.Specifically, the random forest modeling function can be represented as f (X1, X2) = ∑ k = 1 to K tk(X1, X2)/K.Where: tk is the prediction from the kth decision tree, and K is the total number of trees (K = 100 trees).Each constituent decision tree made a separate mortality prediction based on splits of the predictor data: tk(X1, X2) = Predict mortality by recursively splitting X1 and X2 to minimize prediction error.The individual tree predictions were averaged to obtain the overall random forest prediction.The random forest ensemble combined 100 decision trees trained on bootstrapped samples.Each tree splits the samples recursively based on the predictor variables to minimize prediction error.This multivariate regression modeled complex variable interactions and provided robust mortality projections compared to individual decision trees.Predictive performance was evaluated using a temporally stratified training-test split; 20% of data for each was held out as test sets.The model was trained on the remaining 80% and used to predict mortality for the test sets.This enabled the assessment of out-of-sample predictive accuracy over time.Additional cross-validation was undertaken to tune model hyperparameters and minimize overfitting.
For future projections under the RCP26 scenario, the trained Random Forest model for historical was applied to estimate annual heat-related mortality.The same model structure was employed.The decision trees predicted mortality by recursively splitting the projected heatwave characteristics and population.The ensemble average of these individual tree predictions provided the overall projection for future heat-related mortality.By combining a robust ensemble modeling technique, we estimate the integrated model to predict heatwave mortality to the end of the 21st century as a function of changes in heatwave duration and population size under climate change.The model provides data-driven mortality projections to support risk assessment and planning.

Present day heatwave characteristics
Four heatwaves' metrics are used to describe heatwave characteristics over Africa, which include: heatwave magnitude (HWM), heatwave number (HWN), heatwave duration (HWD), and heatwave frequency (HWF), which are shown in figures (1-4).The key heatwave metrics used to characterize heatwaves are defined respectively as the mean excess heat, annual count of events, duration of heatwave in days, and frequency of heatwave days.A categorical scale of non-extreme, moderate extreme, and super extreme heatwaves is used based on threshold values of the EHF, an index quantifying excess heat relative to climatological baselines.Figure 1 shows the spatial distribution of heatwave magnitude (HWM), which is defined as the average excess heat in • C during events, for the historical period from 1980 to 2014 (Figures 1(a) and (b)).Figure 1(c) show HWM projections for the period (2050-2100) under the Shared Socioeconomic Pathway 2-4.5 (SSP2-4.5)scenario, while Figure 1(d) illustrates HWM projections for (2050-2100) under the SSP5-8.5 scenario.Both figures depict the second half of the 21st century but under different future pathways.WFDEA5 and CMIP6 historical ALL forcing simulations show that the African continent has been subjected to severe heatwaves magnitude over the historical period .These were particularly prevalent in the northern, sub-saharan and southern African region.In contrast, moderate magnitude was seen to impact West and Central Africa.Under SSP245 and SSP585 emission scenarios, this trend is projected to persist into the future, as seen in figures 1(c) and (d).Super extreme occurrences are expected to become increasingly common in Africa.By the end of the century, it is predicted that over 70% of the area will be affected yearly by high-impact, super-extreme heat waves.Similar to the historical period, SSP245 and SSP585 shows heatwaves are projected to be more severe in the northern and southern, and sub-saharan region as compared to central and western regions, according to both emission scenarios.HWM is expected to rise significantly for both emission scenarios by the end of the 21st century.This implies that the region is presently encountering notable magnitudes of heatwaves, which is anticipated to persist into the future.The mean spatial distribution of HWN is shown in (Figure 2), HWN is the total count of events per year exceeding the 90th percentile temperature threshold for 3+ days.WFDEA5 (a) and CMIP6 historical ALL forcing simulation (b), which reveals that the HWN across Africa during the historical period varies from 1 to 5 HWN events per year.The SSP245 (c) and the SSP585 (d) scenarios suggest that the HWN would increase more than double between 2050-2100.The SSP245 scenarios show that the number of HWN events would have increased by 2−7 HWN events per year.While SSP585 project that the number of HWN incidents would have grown by 4-14 HWN events per year.The most significant escalation is expected to occur prominently in West and Central Africa.This implies an anticipated increase in HWN acrossAfrica, with a higher frequency expected continent-wide as shown by SSP245, and SSP58 (Figures 2(c) and (d)).Heatwaves duration (HWD) is defined as the number of heatwave days per year.HWD has varied significantly from region to region during the historical period.The continent has had an HWD of about one week averaged from .Figures 3(a) and (b) show the spatial distribution of mean HWD, for the WFDEA and CMIP6 historical ALL forcing simulation respectively.HWD is shown to occur during the historical period, on average, of 1-6 days.Both the SSP245 and SSP585 emission scenarios project significant increases in HWD.HWD is expected to rise most dramatically in northern and some parts of Eastern Africa.The SSP245 and SSP585 scenarios projected an increase ranging between 11days to 50 days, with prolonged heat of more than a month (Figures 2(c) and (d)).Since the tropics have less temperature change, HWD will rise dramatically in the saharan, and northern region durinng 2050 and 2100.End-of-the-century heatwave estimates indicate extended episodes lasting several weeks (Figures 2(c) and (d)), with the greatest duration events lasting for more than a month in certain regions of North and East Africa.The typical length of a heatwave in West, central, and South Africa is predicted to be between 11 and 21 days (three weeks).Extremely hot years are predicted by SSP585, with significantly longer heatwaves.
Figure 4 shows the temporal evolution of the frequency of heat wave days (HWF) over the period  for the WFDEA5 dataset (black line), historical ALL forcing (purple line), as well as the future period from (2015 to 2100) for the SSP245 (blue line) and SSP585 (red line).During the historical period, according to WFDEA5, the mean HWF over Africa was about 20 heatwave days per year, while CMIP6 historical ALL forcing simulation shows average HWF of 23 heatwave days per year from .SSP245 and SSP585 demonstrate that the HWF will have increased to around 51 heatwave days per year on average in 2050.This rise will take place in the near future in 2050, according to SSP245, HWF is projected to increase to around 55 heatwave days per year on average by the end of the 21st century.However, according to SSP585, the HWF is projected to grow to approximately 80 heatwave days per year.Under the high-emission scenario, the HWF from the years 2050-2100 for the SSP585 scenario suggests that the frequency of heatwaves would dramatically increase.A substantial increase in HWF over Africa under high emissions scenarios is noticed.This dramatic increase in projected heatwave frequency is evidenced by the continuously rising HWF trends in Figure (4).This underscores the significant risks African nations could face from more frequent and intense heatwaves if high GHG concentrations continue to materialize.Urgent global efforts to curb GHG emissions and limit global warming below 2 • C preferably (1.5 • C) are thus critical to containing future heatwave escalation.The CMIP6 projections provide a sobering glimpse of potential future heatwave regimes and emphasize the importance of adapted planning to enhance resilience.
Figure 5(a) presents the projected heatwave magnitude over the 21st century under RCP2.6 and RCP6.0 scenarios., HWM is projected to increase from around 37 • C to 39 • C by 2099 under both RCPs.However, the rate of increase is greater under RCP6.0,reflecting more intense heatwaves expected from higher emissions-nonetheless, even the low emissions RCP2.6 projects substantial intensification of heatwave severity by the late century.The projected annual HWN is shown in figure 5(b) for RCP2.6 and RCP6.0, under RCP2.6,HWN gradually increases from around 3 event per year to 5 event per year in 2099.For RCP6.0, heatwave counts will rise sharply from 3 event per year to 9 event per year by 2099, illustrating the dependence on emissions and radiative forcing.Still, even the mitigation scenario projects a near doubling of annual heatwave events by 2100. Figure 5(c) presents projections of annual HWD.Under RCP2.6, HWD rises from around 7 days currently to 12 days by 2099.For RCP6.0, heatwaves substantially lengthened from 13 days to 36 days in 2099.HWF projections are shown in figure 5(d) under RCP2.6 and RCP6.0,Frequency increases more rapidly under RCP6.0,reaching 63 heatwave days per year in 2099 versus 32 heatwave days per year for RCP2.6, compared to ∼20 heatwave days per year observed during the historical period.Key heatwave characteristics will intensify through the 21st century, even under low emissions pathways (RCP26).Mitigation can substantially reduce amplification, but adaptation strategies should still account for potential growth in heatwave number, magnitude, duration, and frequency.By the late century, under high emissions scenarios like RCP6.0, extreme heatwave events may last over a month with magnitudes exceeding 40 • C. Our ensemble projection indicates that even under the moderate emission scenarios, heatwave frequency could double and duration increase relative to the present day.

Heatwave return event
Figure 6 shows both the maximum and minimum return periods (2 years and 20 years) of the four heatwave characteristics examined in this study during the historical period  and under the SSP2-4.5 scenario (2050-2100).Here we evaluate variations in and their potential return event to understand the necessity of adaptive measures in response to changing climate conditions.The 2 year return period for HWM indicates minimal variations between historical and projected scenarios.The best estimate of maximum HWM for historical period stands at (40.42) while the best estimate for SSP2-4.5 stand at (40.95).Despite the marginal changes in magnitude, the consistently high values of HWM during the historical period are similar to the projected period, indicating a significant risk to various sectors.The 20 year return period for maximum HWM also shows minimal change between the historical and SSP2-4.5 scenarios.The best estimate for the historical period is around (41.82), while SSP2-4.5 is approximately (49.92),which suggests a potential intensification of HWM in the future.This highlights the importance of continuous monitoring and the implementation of adaptive measures to mitigate the impacts of heatwaves on Africa's ecosystems and human populations.The best estimate return event of the 2 year and 20 years return period for maximum HWD shows a robust increase in likelihood under the SSP245 scenario compared to the historical period.The 2 year return period best estimate of maximum HWD during the historical period is (5.04) days, while the SSP2-4.5 scenario projected to rise to (31.7), reflecting the potential increase in heatwave occurrences and their duration due to the projected socioeconomic and climatic conditions associated with SSP2-4.5.In comparison the minimum HWD has a best estimate of (5.05) days during the historical period, while the best estimate of SSP2-4.5 is (47.1)considerable increase in HWD is noticed by comparing return events under historical and SSP245 scenarios.These differences signify the possible effects of climate change and socioeconomic variables on the HWD.Under the SSP2-4.5 scenario, best estimate return event for maximum and minimum HWD in Africa are much higher than the historical return event.
The 20 year return event of maximum HWD during the historical period has a best estimate of (6.8), the best estimate of (38.9) is seen under under the SSP2-4.5 scenario.Minimum HWD follows a similar trend, with the historical return event indicating best estimate of (5.7), while SSP2-4.5 scenario indicate best estimate of (61.5), indicating a significant rise.This large increase highlights the probable future escalation of heatwave conditions brought on by climate change and socioeconomic variables.The possible amplification of heatwave occurrences throughout the selected period range highlights the urgent need for adaptation and mitigation methods to meet the escalating hazards of heatwaves and preserve human health and well-being across Africa.Best estimate of maximum HWF during the 20 year return period is (13.5) while the SSP2-4.5 HWF is estimated to be (90.7)as shown in figure 6.The increasing frequency and severity of heat waves in Africa are having a catastrophic effect on the continent's population and environment.Increase in GHG emission and changes land-use are some of the major drivers contributing to the increasing heat wave frequency (Figure 7).Rising temperatures due to climate change increase the likelihood of heat waves.Since vegetation is important in local climate regulation, its loss adds to heat waves.The best estimate of the 20 year return period HWF is a grim reminder of the threat that heatwaves pose to the continent.The likelihood of future heat waves may be reduced if we take measures to reduce greenhouse gas emissions and protect our vegetation.Intensification of HWM, HWD, HWN and HWF is extremely dangerous to human health for several reasons, including the potential for heat-related diseases, worsening preexisting ailments (Cheng et al 2019, Liu et al 2022), and even mortality.Agriculture (sun et al 2019), water supplies, and ecosystem (Aminzadeh et al 2021) health across africa Continent are all threatened by the projected intensification.These results emphasize the need to lessen the effects of more frequent and severe heat waves.Heatwave early warning systems, heat health action plans, urban design to reduce the urban heat island effect, and sustainable practices to lower greenhouse gas emissions are all possibilities.

Human contribution to historical heatwave intensification
This section examines human contribution to the historical intensification of heatwave waves.The investigation focuses on evaluating linear trends associated with various components, namely ALL (all forcings), GHG, aerosols (AER), and land use and land cover changes (LAND).By scrutinizing these distinct heatwave metrics, our objective is to elucidate the nuanced roles of anthropogenic and natural factors in shaping observed changes in heatwave characteristics throughout the historical period spanning from 1950 to 2014.Examining these linear trends provides valuable insights into the complex interplay between human activities and natural influences in driving changes in heatwave patterns.We examine contribution of various external forcing components which allows us to understand their impacts in driving the intensifying heatwave characteristics.The increasing impact of heatwaves is proportional to the adjustments made by different forcings.A clear upward trend in the model simulated response in historical ALL forcing is consistent for HWM, HWN, HWD and HWF (Figure 7).For HWM, the linear trend for hist-ALL is (3.01), indicates a positive trend, suggesting that combined forcing, including anthropogenic and natural factors, has increased heatwave intensity.The (hist-GHG) forcing shows a higher linear trend of (3.17), slightly exceeding the (hist-ALL) forcing run trend.The Positive trend in HWM seen in the hist-GHG only simulation provides clear evidence of the significant role the increase in greenhouse gas in the atmosphere has played in escalating HWM over recent decades.On the other hand, the historical aerosol (histAER) forcing exhibits a negative linear trend of (−1.50).This suggests that aerosols(SO2, BC, NOx, O.C., SO4, NH3, CO, and NMVOC), such as pollutants and particles suspended in the atmosphere, have had a mitigating effect on heatwave magnitudes, likely due to their cooling properties (Myhre et al 2013).The historical natural (hist-NAT) forcing, representing natural factors, also has a negative linear trend (−0.23), indicating a minor dampening effect on heatwave magnitudes by (stratospheric aerosols and solar irradiance).The histLAND forcing indicate a relatively high positive linear trend of (4.54), indicating changes in land use have pronounced impacted increasing heatwave magnitudes across the continent.This suggests that alterations in land use and land cover, such as increased urbanization (Adigun et al 2022) or deforestation (Wolff et al 2021), have contributed significantly to the intensification of HWM.In terms of the minimum HWM, the trends follow a similar pattern.The hist-ALL, hist-GHG, and hist-LAND forcings exhibit positive linear trends, while hist-AER and hist-NAT show negative trends.These trends suggest that human activities, particularly greenhouse gas emissions and land use changes, have amplified the minimum HWM, while aerosols and natural factors have had a mitigating effect.The linear trend for the maximum HWF under all forcings (hist-ALL) show a significant increasing trend of (7.25), the positive trend indicate , anthropogenic and natural forcing combine have contributed to increase in the frequency of heatwave events during the historical period.HWF under greenhouse gas (hist-GHG) indicate a linear trend of (8.53), figure 7. The higher positive trend compared to hist-ALL, suggest increase GHG concentration in the atmosphere have played a significant role in driving the increasing frequency of heatwave events.The linear trend for the maximum HWF associated with AER forcing shows a decreasing trend of (−5.48).This negative trend suggests that aerosol emissions and their associated cooling effects may have partially offset the overall positive trend in HWF driven by other factors.However, it is important to note that other factors, such as regional variations in aerosol emissions, could also influence the negative trend.The linear trend for the maximum HWF attributed to hist-NAT is −1.31.This negative trend suggests that natural variability alone has slight damping effect on the increase frequency of heatwave events during the historical period.However, the negative trend is relatively weaker compared to the positive trends associated with human-induced forcings (i.e hist-GHG and hist-LAND).The linear trend for the maximum HWF related to land use changes (hist-LAND) is 6.02.This positive trend suggests that land-use modifications, such as deforestation or urbanization, have contributed to increased HWF.Land use changes can affect local climate patterns and exacerbate heatwave conditions.considering minimum HWF, similar patterns can be observed.hist-ALL 6.36 and hist-GHG 8.32 indicate positive linear trends, suggesting an increase in the minimum frequency of heat waves.This further supports the notion that greenhouse gas emissions play a significant role in driving the observed changes in heatwave occurrences.Interestingly, hist-AER (−5.32) and hist-NAT (−1.08) still exhibit negative trends for the minimum heat wave frequency, indicating a decrease in minimum heatwave occurrences associated with aerosol and natural forcings.These findings suggest that factors such as aerosol emissions and natural variability have slightly mitigated the minimum frequency of heat waves.Maximum HWD during the historical period indicates positive linear trends for hist-ALL (2.89), hist-GHG (6.10), and hist-LAND (3.94).The ALL forcing runs show an increasing trend across Africa for HWM, HWN, HWD, and HWF.The increasing pattern is primarily due to anthropogenic forcing since the ALL forcing is trend is higher than NAT forcing.The increasing trend is controlled by the overall increasing in GHG forcing during the historical period (Figure 7).Hist-LAND experiment shows land use over Africa has also contributed significantly to the intensifying heatwave characteristics.These comparisons highlight the complex interplay between different forcings and their contributions to heatwave magnitudes.The positive trends observed in hist-ALL, hist-GHG, and hist-LAND underscore the significant role of human activities, especially greenhouse gas emissions and land use changes, in driving the intensification of heat waves.These positive trends suggest that antropogenic forcing have increased the maximum and minimum frequency of heatwave events over the historical period.This is consistent with the understanding that anthropogenic activities, such as burning fossil fuels, (Menon 2004) have led to the accumulation of greenhouse gases in the atmosphere, resulting in a warmer climate and more frequent heat waves.

Heat related mortality over Africa
Boxplot whiskey of total annual heat-related mortality is shown in figure 8(a) historical ) (b) far future (2050-2099) rcp26 and (c) far future (2050-2099) RCP60.Annual heat-related mortality for the historical period is, based on four climate models, gfdl-esm, hadgem2, ipsl-cm5a, and miroc5, along with the multi-model ensemble (MME).The median mortality value across all four models indicates annual mortality of approximately 9412 mortality per year during the historical period.A considerable variation in mortality estimate estimations is seen across different models.gfdl-esm shows the highest median mortality estimate of approximately 11,248 mortality per year.Indicating a higher number of heat-related mortality in Africa compared to the other models during the base period.Under RCP26, (figure 8(b)), the number of mortality attributed to heat exposure each year in Africa is projected to rise dramatically.The MME estimates an annual mortality rate of 23060 mortality per year, miroc5 project the highest annual mortality rate of 26159 mortality per year, while the gfdl-esm projects the lowest, at 18 924.However, the projected mortality varies somewhat between models.Although the general trend is apparent, heat-related mortality are expected to increase dramatically in Africa under RCP26.Under RCP60 median mortality for MME is around 42000 mortality per year, (figure 8(c)) ranging from 36,000 to 54,000.Hadgem2 projects a mean annual heat-related mortality of 57,259 mortality per year; the lowest value is from the gfdl-esm model, which stands at 34,209 mortality per year.
To identify the regions of Africa most at risk of heat related mortality, we divide Africa into four sub regions (table 3) which are East Africa (EAF), Southern Africa (SAF), Saharan Africa (SAH), and West Africa (WAF).The historical period shows the median annual heat-related mortality is highest in EAF (13994) and lowest in SAH (8648) (figure 8(d)).SAF and WAF show approximately the same median annual heat-related mortality rate of 12,405.Multi-model ensemble projections under RCP2.6 and RCP6.0 scenarios suggest greater future heat-related-mortality risk across all regions but highest in both EAF and WAF compared to the historical baseline.This is evidenced by increases in the estimate of annual heat-attributable mortality under low (RCP2.6)and high (RCP6.0)emissions trajectories for these regions, as shown in (figure 8(e), (f)).
The magnitude of the increase varies by model, but all models show an increase in mortality.The increase is likely due to rising temperatures, socio-economic development, and a growing African population (Rohat et al 2019).It is important to note that these are just the results of four climate models.Other models may predict different mortality levels.However, mortality in Africa is a serious threat.Under RCP60, annual heat-related mortality is projected to increase significantly compared to the RCP26.gfdl-esm projects a pronounced increase in annual heat-related mortality of over 80% between RCP26 and RCP60 (table 2).Since RCP60 assumes that emissions will continue to rise rapidly, leading to significant warming and more intense heatwaves by 2100, which will directly increase the risk of heat stress mortality.The rise in annual heat-related mortality under RCP60 relative to lower scenarios like RCP26 highlights the significant implications for human health if mitigation action is not taken.These results are consistent with the findings of other studies, which have shown that climate change is increasing the risk of heat-related mortality in Africa (Scovronick et al 2018).The substantially amplified heat-attributable mortality rates projected under greenhouse gas-intensive scenarios, including RCP6.0, underscore potentially dire public health consequences if climate change remains unchecked.Recent studies indicated heat-related premature fatalities could rise two to three-fold by mid-century globally under strongly mitigated trajectories up to six-fold under high emission baselines due to enhancements of extreme heat event magnitude, duration, and frequency (Coffel et al 2017).Regionally, Sub-Saharan Africa could witness ten times more annual heat fatalities under 4 • C warming scenarios than 1.5 • C pathways (Dosio. 2017).These projections remain highly sensitive to socioeconomic differences in adaptive capacity.However, the non-linear amplification of risks posed at higher increments of global mean temperature rise highlights the urgency of transitioning societal infrastructure, practices, and energy systems towards carbon neutrality by 2050 in alignment with +1.5 • C trajectories.Without substantial mitigation, continuing high emissions put us on track for deadly heatwaves that exceed human physiology's ability to cope.Such projected impacts suggest that our current approaches to prevention are woefully insufficient to avoid catastrophic consequences.The continent is particularly vulnerable to heat stress due to its high population density, low adaptive capacity, and limited access to healthcare.It is estimated that under RCP60, heat-related mortality could increase up to 43,632 mortality per year by the end of the century.The increase in heat-related mortality will depend on several factors, including the rate of emissions reductions and the effectiveness of adaptation measures.Time evolution of projected change in annual heat-related mortality relative to 1980-2005 for RCP26 (pink line) and RCP60 (red line) over Africa and its subregion is shown in figure 9.Over Africa, the projected increase in heat-related mortality in Africa is significant, and the increase is even greater under the RCP6.0 scenario, which has a projection of 47,369 at the end of the 21st century figure 9(a).This is because RCP6.0 represents a more severe pathway of climate change, with higher greenhouse gas emissions resulting in more warming.In the EAF region, The projected change in annual heat-related mortality for the RCP2.6 scenario at the end of the 21st century is projected to be 14,509 while RCP6.0 scenario, indicate a projected increase of about 88,015 figure 9(b).In this case, the result indicates a significantly higher increase in heat-related mortality under the RCP6.0 scenario compared to RCP2.6.In the WAF region, RCP2.6, scenario project annual heat-related mortality of 18,645 figure 9(e) indicating a greater vulnerability to heat-related conditions in this region.RCP6.0 scenario projects an even higher increase in annual heat-related mortality 68,487 compared to RCP2.6.The lowest annual mortality is projected across the SAF and SAH region for both emission scenarios examine figure 9(c), (d).Under the lower-emission mitigation scenario represented by RCP2.6, heat-related mortality estimates show smaller increases across African regions than high-emission scenarios like RCP6.0.Adhering to the lower RCP2.6 emissions pathway and restricting global warming below 2 • C, the East Africa region is projected to avoid up to approximately 73,506 heat-related mortality at the end of 21st century.Similarly, avoided mortality under RCP2.6 is estimated to be about 49,842 in West Africa(figure 9) Therefore, current projections indicate substantially higher heat mortality burdens across major African regions under higher-emission scenarios, while a low-emissions mitigation pathway could prevent hundreds of thousands of heat-related deaths at the end of the 21st century.RCP60 predicts increased greenhouse gas emissions, which raises average global temperatures and increases the likelihood of severe heat events.The pace of greenhouse gas emissions, the success of adaptation strategies, and the sensitivity of the population are all variables that will influence heat-related mortality.Action is needed to decrease and adapt to the consequences of climate change, which poses a serious threat to human health across Africa and the subregion under study.
Figure 10 shows the mean projected change in annual heat-related mortality in some selected African countries for RCP6.0 during 2080-2099.Most African countries are already experiencing high temperatures and high heat-related mortality rates (Manyuchi et al 2022).At the end of the 21st century, some countries are projected to experience a significant increase in heat-related mortality.For example, Nigeria, the most populous country in Africa, is projected to have a staggering 472 340 annual heat-related mortality by 2080-2099 under the RCP6.0 scenario (figure 10).This highlights the potential impact of climate change on densely populated regions-considerable variation in projected mortality across different countries.For instance, Malawi and South Africa are projected to have high annual heat-related mortality of (275 340 and 292 270, respectively), indicating a higher vulnerability to heat-related mortality in those countries.On the other hand, some countries, such as Senegal and Djibouti, show a relatively lower (15 830 and 13 324, respectively) projected change in heat-related mortality.These variations may be influenced by geographical location and existing infrastructure.It is essential to acknowledge that climate model projections involve uncertainties.The results are based on complex models that consider various factors, and the actual outcomes may differ due to technological advancements, policy changes, and socioeconomic developments.
A random forest model containing 100 decision trees was developed to predict annual heatwave-related mortality based on population and several heatwave characteristics, including duration, number, frequency, and magnitude.The ensemble predictions from the random forest model closely track the observed mortality values (figure 11).This demonstrates the model's ability to effectively capture the relationship between mortality, population, and heatwave metrics.While all heatwave predictors provide reasonable mortality estimates, some differences in their predictive capability are apparent.1998, for instance, the model  using HWN as the predictor (hist_HWN-mortality) most closely matches observed mortality with only a 1.7% underestimate.On the other hand, the prediction based on HWF (hist_HWF-mortality) underestimates 1998 mortality by 3.1%.This indicates that frequency provided less predictive power for that year's mortality.In 2005, however, models using HWD and magnitude overestimated observed mortality by 11% and 7%, respectively, while number and frequency remained within 1%-2%.This highlights how different heatwave properties may have variable relationships with mortality in a given year.The nonlinear random forest approach can capture these complex predictor interactions.Overall, the close fit between observed and predicted mortality across model iterations demonstrates that heatwave duration, number, frequency, and magnitude can serve as informative predictors.By integrating multiple decision trees built on random samples, the model accounts for nonlinearity and avoids overfitting.Incorporating changes in population demographics and heatwave characteristics into ensemble models can help inform predictions of heatwave-related mortality.Figure 11 validates the capability of the random forest model for heat related mortality estimate using these predictors.
Model projections under the RCP2.6 lower-emission mitigation scenario still reveal worrying trends of steadily rising heat-related mortality over time, underscoring concerns that even mitigation efforts will be insufficient to reverse increasing extreme heat-related burdens in African nations, with total mortality (rcp26_Mortality) nearly doubling from around (15 000) in 2020 to ( 29  lengthen, people endure prolonged exposure to dangerously hot conditions.In contrast, the projected mortality associated with HWN and population (rcp26_HWN_Mortality) fluctuates between (16 000-24 000) over the years.Projected mortality linked to HWF and population (rcp26_HWM_ Mortality) aligns closely with the overall heat mortality, reaching nearly (23 000) by 2098.While projected mortality linked to heatwave magnitude (rcp26_HWM_Mortality) aligns closely with the overall heat mortality, reaching nearly (22 000) mortality per year by 2099.The variable relationship between frequency and deaths may reflect mortality also depending significantly on heatwave magnitude and individual health factors.As heatwaves become more intense under climate change, high temperatures pose greater health risks, particularly among vulnerable groups.Our result reveal escalating heatwave-associated mortality driven by heatwave characteristics and growing populations.A multipronged adaptation approach can help curtail loss of life.Efforts should focus on extending protections during lengthy heatwaves, building general resilience across groups, and providing emergency relief during extremely hot spells to create a more heat-adapted populace (Harlan et al 2014, Jänicke et al 2018).Our projections underscore the urgency of implementing policies and interventions, before climate and population trends result in unchecked increases in mortality.Unlike the steady rise in overall mortality, this variable relationship indicates factors beyond just frequency to determine health impacts.The inconsistent increases in projected mortality may reflect yearly variations in heatwave intensity duration, number as well as population vulnerability.

Discussion
This study provides crucial insights into the intensification of heatwaves across Africa and the escalating threats to human health.Our multi-faceted analysis aligns with evidence that extreme heat events have become more frequent and intense in recent decades (Wehner et al 2018).We advance understanding by quantifying trends across diverse heatwave properties, including magnitude, number, duration, frequency, and return periods.This reveals nuanced changes in heatwave characteristics beyond temperature thresholds alone (Li et al 2018).Our projections indicate continued escalation across heatwave metrics under moderate and high emissions scenarios, with over 70% of Africa potentially experiencing frequent, extreme heatwaves by 2100.A unique aspect of our modeling is examining heatwave duration, which exhibits a sharp increase in length, potentially lasting over a month in North and East Africa by 2100.Prolonged heatwaves allow little respite and are especially detrimental to health (Gasparrini et al 2015).Our mortality model identifies duration as the most robust predictor of heat-related deaths.This highlights the need for relief interventions during extended hot spells, such as cooling centers (Bai et al 2018).We advance heat-mortality modeling by harnessing nonlinear machine-learning approaches.Our integrated random forest model incorporates diverse heatwave metrics and demographic trends, capturing complex interactions influencing mortality (Chen et al 2021).This ensemble approach addresses the limitations of conventional linear models (Liu et al 2021).We provide new quantified estimates of potential heat-related deaths under varying emissions pathways.Regional analyses have revealed varying heatwave and mortality patterns across Africa historically.East African countries like Ethiopia, Kenya, Somalia, and Uganda have endured the most significant heatwave mortality burdens.While this region has faced the highest heat mortality in the past, which is projected to continue, we see Nigeria, a country in West Africa, will experience the most severe future health burden from heatwaves under climate change.Estimates suggest over 400,000 heat-related mortality in Nigeria by the end of the 21st century.This elucidates regional nuances and contexts within broader trends (Dosio et al 2019).Multi-timescale assessment, from historical to end-of-century projections, facilitates targeted adaptation planning aligned with location-specific risk trajectories.Our analysis reveals robust evidence that human-induced climate change and land use changes are key drivers of the observed African heatwave increases (Ciavarella et al 2018), while natural forcings have effects are minimal.Our integrated approach clearify the interacting effects of escalating heatwaves and demographic changes on mortality (Jones et al 2018).The findings underscore the critical need for preventative policies, climate-smart urban planning, early warning systems, and other protective measures to build heat resilience (Otto et al 2018).Our projections also highlight the enduring importance of mitigating climate change by reducing greenhouse gas emissions.Here, we provide new insights into the threats of intensifying heat extremes across this highly vulnerable region.The results make an urgent call for evidence-based heat adaptation and mitigation strategies to prevent the loss of human lives from escalating heatwaves in Africa.

Conclusion
In this study, our result reveals a concerning trend of increasing magnitude, frequency, duration, and number of heatwaves in Africa.The historical period  has already witnessed severe heatwaves, predominantly affecting the continent's northern, sub-Saharan, and southern regions.Looking ahead, under the SSP245 and SSP585 emission scenarios, it is projected that the severity of heatwave magnitude will continue to escalate towards the end of the century.Super extreme heatwaves are expected to become more common, affecting over 70% of the African continent annually.Heatwave duration varied across regions during the historical period, averaging around one week for the entire continent.However, the future projections indicate a substantial increase in HWD, particularly in northern and eastern Africa.HWD is expected to rise between 11days to over a month, with certain regions experiencing heatwave episodes lasting several weeks.This extended duration of heatwaves poses significant challenges to various sectors and emphasizes the urgency for adaptation measures.The frequency of heatwave days (HWF) is also predicted to rise significantly.The historical period saw an average of 20-23 heatwave days per year, but by 2050, this is expected to increase to around 51 heatwave days per year under the SSP245 scenario.By the end of the 21st century, the HWF could reach approximately 55 heatwave days per year under SSP245 and 80 heatwave days per year under SSP585.These findings highlight the alarming increase in the frequency and duration of heat waves, especially under high-emission scenarios.Return periods further confirm the intensification of heatwave characteristics in the future climate.Although the changes in heatwave magnitude (HWM) between historical and projected periods are relatively minimal, they still pose significant risks to human health, agriculture, and ecosystems.Adaptation measures and sustainable practices are crucial to address Africa's challenges of extreme heat.The projected changes under different emission scenarios emphasize the urgent need for proactive measures to mitigate the impacts of heatwaves on vulnerable populations, ecosystems, and key sectors.Understanding the characteristics of heatwaves is crucial for informed decision-making, policy formulation, and planning for climate change adaptation in Africa.The study's results highlight the increasing threat of heat-related mortality in Africa due to climate change which is consistent with previous study (Scovronick et al 2018, Vicedo-Cabrera et al 2021).The analysis of heat-related mortality data from different climate models reveals significant variations in annual mortality estimates, with the highest numbers observed in EAF and WAF.However, all regions of Africa are vulnerable to heat-related mortality.Under the RCP26 the projected increase in annual heat-related mortality is substantial, with the MME estimating a rate of 23 000 mortality per year.These findings underscore the urgent need for adaptation and mitigation strategies to address Africa's escalating risks of heat-related mortality.The projected increase in heat related mortality in africa is primarily due to rising temperatures and the growing population in the region (Ebi et al 2021, Rohat et al 2019).Vulnerability to heat stress is compounded by factors such as high population density, limited healthcare access, and low adaptive capacity (Mushore et al 2018, Voelkel et al 2018).The study's results align with previous research, confirming that climate change poses a significant threat to African public health (Sarfaty et al 2014, Opoku et al 2021, Sy et al 2022).The projected rise in heat-related mortality emphasizes the importance of effective emissions reductions and adaptation measures to mitigate the impacts.Without timely action, it is estimated that heat-related mortality in Africa could reach 43 000 mortality per year by the end of the century under the RCP60 scenario.Policymakers, healthcare providers, and communities must prioritize measures that address heat-related health risks, including heatwave early warning systems, improved access to healthcare, and heat mitigation strategies in urban areas.By implementing proactive measures, Africa can work towards reducing the devastating impacts of heatwaves on human lives and maintaining the well-being of its populations in a changing climate.A limitation of the current modeling approach is the assumption of constant relative risk over time without accounting for potential acclimatization to heat.Physiological adaptations and societal changes could modify vulnerability over time as populations are increasingly exposed to higher temperatures (Rupert Stuart-Smith et al 2023).For instance, increased access to air conditioning and heat warning systems may reduce risks, causing mortality projection curves to flatten compared to a fixed temperature-mortality relationship.Capturing these acclimatization dynamics could improve projections and avoid overestimating future impacts.Assessing changes in vulnerability and developing models that allow the temperature-mortality curve to evolve based on adaptation efforts would be valuable areas for advancement.Studies incorporating time-varying exposure-response relationships are needed to provide more realistic heat-mortality projections in an adapting society.
We employed the Canadian Earth System Model version 5 (CanESM5; Swart et al 2019) and Model for Interdisciplinary Research on Climate version 6 (MIROC6; Tatebe et al 2019) large ensemble, from the Coupled model intercomparison project phase 6 (CMIP6) model as shown in table

Figure 6 .
Figure 6.Best estimate return period of 2 and 20 years return event for heatwave magnitude in ( • C), heatwave number (annual event), heatwave Duration (days), heatwave frequency (event) over Africa for historical period (black bar) and SSP245 scenario (red bar).

Figure 7 .
Figure 7. Modelled trend of heatwave characteristics heatwave magnitude in ( • C), heatwave number (annual event), heatwave Duration (days), heatwave frequency (event) over Africa in response to response to natural and human forcing during 1950-2014.

Figure 9 .
Figure 9.Time evolution of Projected change in annual heat related mortality relative to 1980-2005 over Africa and its subregion for rcp26 and rcp60 from 2020-2100.

Figure 10 .
Figure 10.Mean Annual annual heat related mortality for 2080-2099 relative to 1980-2005 over some selected countries in Africa under rcp60.

Figure 11 .
Figure 11.Predicted Mortality as a function of population and heatwave characteristics under historical period for the years 1980-2005 000) in 2076 and approximately (21 871) in 2099 figure (12).When examining the contributions of different heatwave characteristics, HWD and population (rcp26_HWD_Mortality) accounts for up to (27 000) mortality by 2076.As heatwaves

Figure 12 .
Figure 12.Predicted Mortality as a function of population and heatwave characteristics under RCP26 for the years 2020-2099.

Table 2 .
Summary of the median estimates of mortality for different climate models and scenarios over Africa.

Table 3 .
Median mortality estimates in various African regions for multimodel ensemble fro historical, RCP26 and RCP60 scenario.