Changes in the Seasonal Cycle of Heatwaves, Dry and Wet Spells over West Africa Using CORDEX Simulations

: This study analyzes the potential response of the seasonal cycle of heatwaves (HWDI) and dry (CDD) and wet (CWD) spell indices over West Africa for the near-(2031–2060) and the far-future periods (2071–2100) under RCP4.5 and RCP8.5 scenarios using Coordinated Regional Climate Downscaling Experiment (CORDEX) simulations. Despite the fact that some relative biases (an underestimation of 30% for CDD, an overestimation of about 60% for CWD, and an overestimation of about 50% for HWDI) exist, during the historical period (1976–2005) in general, the CORDEX simulations and their ensemble mean outperform the seasonal variability in the above-mentioned indices over three deﬁned subregions of West Africa (i.e., the Gulf of Guinea and Western and Eastern Sahel). They show high correlation coefﬁcients (0.9 in general) and less RMSE. They project an increase (about 10 and 20 days) in heatwave days for both the near-and far-future periods over the whole West African region under both RCP scenarios. In addition, projections indicate that the Sahel regions will experience a decrease (about 5 days) in wet spell days from March to November, while in the Gulf of Guinea, a decrease (about 3 days) is projected throughout the year, except in the CCCLM simulation, which indicates an increase (about 5 days) during the retreat phase of the monsoon (October to December). Our results also highlight an increase (about 80%) in dry spells over the Sahel regions that are more pronounced during the March–November period, while over the Gulf of Guinea, an increase (about 40%) is projected over the entire year. Moreover, the months of increasing dry spells and decreasing wet spells coincide, suggesting that countries in these regions could be simultaneously exposed to dry seasons associated with a high risk of drought and heatwaves under future climate conditions.


Introduction
West Africa is one of the regions mostly affected by the effects of climate change due to its vulnerability, poor economy, and low capacity for adaptation [1][2][3].Consequently, climate change has some societal impacts in many areas including agriculture, transportation, infrastructure, tourism, and economy [4][5][6][7].The global temperature increase as a result of anthropogenic climate change is one of the main drivers of the weather, thus impacting the gross domestic product (GDP).The authors of [4] showed that the least developed and developing countries are the most affected by high temperatures compared to developed countries.In addition, they also revealed that weather impacts per-capita GDP growth through all its factors of production, with the largest impacts being on total factor productivity.Beyond the connection between weather and the above sectors, human civilization evolution is also impacted by short-term adverse weather conditions [5,8].The Intergovernmental Panel on Climate Change (IPCC)'s framework recommendations highlight that governments, nongovernmental organizations (NGOs), and stakeholders must take appropriate actions to reduce greenhouse gas emissions in order to minimize their effects in response to global warming.Moreover, climate change has enhanced the occurrence of climate extremes, including heatwaves and dry and wet spells, considered among the most dangerous hazards as they have drastic impacts on human and natural ecosystems as well as on anthropogenic activities, like infrastructures and the economy [4].For instance, exceptional or abnormal temperature values recorded over a region or area during at least three consecutive days can cause exposure to a long dry spell, which can cause a risk of drought.However, the same region can experience wet spells, which lead to flooding.Therefore, sectors including human health, agriculture, livestock, and the environment are the most vulnerable to such risks.For example, in the tourism sector, there is an alteration of human behavior and desires due to atmospheric thermal conditions [6].
Several studies have revealed the drawback of heatwaves worldwide, particularly in Asia [9], in the Middle East [10], in America [11,12], in Europe [13][14][15], in France [16,17], in Russia [13,18], in Africa [19,20], and in West Africa [21][22][23].These investigations have shed light on important damage, losses of life, and downturns in economies, as highlighted by [24].Heatwaves are also exceptional events associated with a mean temperature increase and its variability [25] as well as the influence of large-scale atmospheric circulation.Unfortunately, heatwaves, dry and wet spells, and climate extremes (i.e., flooding, warm spell days, very warm day frequency, and warm night frequency) are projected to increase in intensity, duration, and frequency worldwide [26,27] and, to some extent, with different magnitudes or intensity from one area to another [28] in the context of global warming.
Heatwaves are generally lethal phenomena, but when combined with the effects of humidity, they become a serious threat [29][30][31].Heatwaves can also be modulated by sea surface temperature (SST) anomalies, which can remotely influence summer heatwave variabilities over land by generating severe convection and triggering atmospheric teleconnections [32].In the same vein, [30] underlined that hot and humid conditions prevailing in coastal regions can be more dangerous than equivalently hot but dry conditions [33].Consequently, heatwaves can be classified into two categories: wet heatwaves, which are the most dangerous for human health, and humid heatwaves [29,30,33].Furthermore, [34] also defined heatwaves based on the 95th percentile of the local temperature probability density function and additional criteria, including spatial and temporal extensions in Europe.They concluded that heatwaves can be classified into six classes (Russian, West European, East European, Iberian, Scandinavian, and North Sea) based on cluster analysis and several classes related to different physical mechanisms over Europe.
Thus, heatwaves are generally defined as consecutive days of extremely hot temperatures that exceed temperature thresholds [9,31].However, this definition is not universal because a heatwave is defined based on several factors including the research application or activity sectors (health, infrastructure, and agriculture), the geographic and climatic conditions, and the thresholds [24,34].
Hence, heatwaves do not have a unique definition and occur during a certain period of at least three or six consecutive days; no study on them is documented over the West African region.Moreover, studies related to heatwaves have not analyzed the composite effects between heatwaves and other climate extremes.This study aims to provide a comprehensive analysis of heatwaves and wet and dry spells over West Africa.To this end, this work aims to evaluate the strength of the CORDEX simulations and their ensemble mean in capturing the seasonal cycle of heatwaves and wet and dry spells under the present climate  conditions, with the observations from the Climate Prediction Center (CPC) as a reference, and then analyzes the projected changes in their seasonal cycles over two future periods (2031-2060 and 2071-2100) with respect to the present climate, using CORDEX simulations and their ensemble mean.In fact, CORDEX simulations were mainly used in many studies conducted over West Africa for a validation of the mean climate state, its variability, and extreme events [35][36][37][38][39][40], as well as to better understand relevant regional/local climate phenomena and their variability and changes through downscaling [41].This study aims to provide information about the future climate changes under different Representative Concentration Pathways (RCPs) and global warming levels [23,[42][43][44][45][46][47].The RCPs are a set of scenarios containing emission, concentration, and land use trajectories based on radiative forcing in w/m 2 [48,49].RCP4.5 and RCP8.5, referring to a radiative forcing of 4.5 w/m 2 (stabilization scenario) and a radiative forcing of 8.5 w/m 2 (high-emission scenario), respectively, are used in this study.The paper is organized as follows.First, we describe the materials and methods in Section 2. This part includes detailed information about the study domain, data, and methods, with a brief description of the defined Expert Team on Climate Change Detection and Indices (ETCCDI) indices related to heatwaves and wet and dry spells.Section 3 is the results and discussion, which is followed by the summary and conclusions in Section 4.

Study Domain
This study was conducted over West Africa, covering the domain (5 • -15 • N, 15 • W-15 • E) and spanning the Atlantic coast to Chad and the Gulf of Guinea to the southern fringes of the Sahara [33].This domain is subdivided into three subregions, coherent with [35] the Western Sahel and the Eastern Sahel, located in the Sahel area and the Gulf of Guinea on the border of the Atlantic Ocean (refer to Figure 1).The identification of these subregions is based on their location, the climate variability, and other physical features including the ecosystem, the elevation, and the soil occupation.The West African region exhibits high climate variability at regional and local scales [50].This climate is mostly influenced by the West African monsoon flux, which governs the rainy season and thus the rain-fed agriculture [50].This region has a semi-arid and hot climate with a dry season, corresponding to an alternation between a short wet season and a very long dry season [33].It is important to indicate that the Sahelian subregions (e.g., the Western and Eastern Sahel) have a unimodal rainfall regime, while the Gulf of Guinea region is characterized by a bimodal regime [51].The unimodal rainfall regime of the Sahelian zone is centered around the month of August, with an annual rainfall amount between 400 and 600 mm.Also, the bimodal rainfall regime over the Gulf of Guinea region occurs during the months from June to October, which represent the peaks of the long and short rainy seasons, respectively.In general.the total annual rainfall amounts over the Gulf of Guinea region range between 1500 and 2000 mm.Overall, these rainfall regimes follow the seasonal migration of the Intertropical Convergence Zone (ITCZ), which itself is linked to fluctuations in the trade wind system and the oceanic variables over the tropical Atlantic [52,53] and also associated with the northward penetration of the monsoon flux into the continent [53].The influence of these mechanisms on the climatic zones depends largely on large-scale atmospheric circulations as well as continental and meteorological conditions [54][55][56].

Data
In this study, we used daily precipitation and maximum temperature (see Table 1) data from an ensemble of four (04) Coordinated Regional Climate Downscaling Experiment (CORDEX) RCMs [41,57,58] available on the Earth System Grid Federation (ESGF) website (https://esg-dn1.nsc.liu.se/projects/esgf-liu/(accessed on 31 March 2018)) at a resolution of 0.44° (~50 km) (see Table 2).CORDEX is a coordinated regional climate downscaling experiment program elaborated by an ensemble of research centers.It provides reliable and highly accurate climate change scenarios, enabling climate change impact studies at regional and local scales.It enables researchers to understand and analyze uncertainties in climate projections using regional climate models [41,[57][58][59].The data span the historical period of 1950-2005 and the future period of 2006-2100 under the scenarios of RCP4.5 and RCP8.5 [48,49,60].These simulations are performed over a domain covering the whole of Africa.However, our domain of interest in this study was West Africa (see Figure 1).CORDEX simulations are useful in providing climate information using several meteorological parameters (i.e., temperature, rainfall, wind speed, etc.) and some extreme climate events (i.e., heatwaves, dry spells, etc.) over a specific region (i.e., Africa) at many time scales [37,[60][61][62].We also used the gridded gauge-based global daily precipitation and maximum temperature from Climate Prediction Center (CPC) observation data at a 0.5° (~50 km) resolution from 1979 to the present [63,64].This dataset enabled us to evaluate the performance of the CORDEX simulations (Table 1).

Data
In this study, we used daily precipitation and maximum temperature (see Table 1) data from an ensemble of four (04) Coordinated Regional Climate Downscaling Experiment (CORDEX) RCMs [41,57,58] available on the Earth System Grid Federation (ESGF) website (https://esg-dn1.nsc.liu.se/projects/esgf-liu/(accessed on 31 March 2018)) at a resolution of 0.44 • (~50 km) (see Table 2).CORDEX is a coordinated regional climate downscaling experiment program elaborated by an ensemble of research centers.It provides reliable and highly accurate climate change scenarios, enabling climate change impact studies at regional and local scales.It enables researchers to understand and analyze uncertainties in climate projections using regional climate models [41,[57][58][59].The data span the historical period of 1950-2005 and the future period of 2006-2100 under the scenarios of RCP4.5 and RCP8.5 [48,49,60].These simulations are performed over a domain covering the whole of Africa.However, our domain of interest in this study was West Africa (see Figure 1).CORDEX simulations are useful in providing climate information using several meteorological parameters (i.e., temperature, rainfall, wind speed, etc.) and some extreme climate events (i.e., heatwaves, dry spells, etc.) over a specific region (i.e., Africa) at many time scales [37,[60][61][62].We also used the gridded gauge-based global daily precipitation and maximum temperature from Climate Prediction Center (CPC) observation data at a 0.5 • (~50 km) resolution from 1979 to the present [63,64].This dataset enabled us to evaluate the performance of the CORDEX simulations (Table 1).A detailed list of regional and global climate models used from CORDEX simulations are presented in Table 2.In this paper, we used a special denotation of the CORDEX simulations composed of RCMs and the GCMs in column 3 of Table 2.

Calculation of HWDI, CWD, and CDD Indices
In this study, we calculated three climate indicators defined by the Expert Team on Climate Change Detection and Indices (ETCCDI): the consecutive dry days (CDD), the consecutive wet days (CWD), and the heatwave duration index (HWDI).These indices describe the duration and number of dry and wet spells and heatwaves [65,66] (Table 3).
Table 3. ETCCDI indices used in this study [66,67].• CDD (consecutive dry days) is the greatest length of consecutive days when the precipitation amount is less than 1 mm/day.This is an indicator of dry spells (day) and drought.A further variable is the number of consecutive dry day periods of more than 5 days in a given time period.

•
CWD (consecutive wet days) is the greatest length of consecutive days when the precipitation amount is greater than 1 mm/day.A further variable is the number of consecutive wet periods of more than 5 days in a given time period.• HWDI (heatwave duration index) is the maximum period of at least three (or six) consecutive days where the daily maximum temperature exceeds the daily mean maximum temperature during the period of 1976-2005 + 5 CDD, CWD, and HWDI indices are calculated on a grid for each simulation, then interpolated onto a common regular grid (0.5 × 0.5) degree of resolution using bilinear interpolation to enable the use of the multi-model ensemble (MME) approach.This approach allows the calculation of the mean of an ensemble of simulations to reduce existing uncertainties between them [39,42,69] using the following Equation (1): where Simulation i represents each simulation and N represents the number of simulations.

An Evaluation of the Simulations' Ability to Represent the Indices
The ability of simulations to capture the variability in the annual cycle of CDD, CWD, and HWDI is assessed to better identify their respective strengths and weaknesses [70].This technique is performed using the Taylor diagram [71,72] indicating three statistical quantities: the correlation coefficient (CC), the centered root-mean-square error (RMSE), and the amplitude of the variations represented by their standard deviation (SD).The aim is to identify the strengths and weakness of the different simulations through an easy and straightforward visual comparison [71,72].In this study, the CORDEX simulations were compared to the unified gauge precipitation (CPC) in order to demonstrate their ability to reproduce the indices of CDD, CWD, and HWDI over the three subregions of West Africa for the historical period of 1976-2005.

Climate Change Signal
Climate change signal was evaluated using historical  and projected data under the RCP4.5 and RCP8.5 scenarios to investigate changes in the future.The changes were estimated as a percentage (%), with the historical mean period of 1976-2005 used as a reference.To this end, two future periods of thirty (30) years were considered, namely, the near-future period of 2031-2060 and the far-future period of 2071-2100, to obtain a more detailed description of the future climate, taking into account its evolution [73][74][75][76].The climate change signal was evaluated using Equation (2) or Equation (3): Climate change signal = projection − re f erence where projection refers to the mean value of the variable (index) over a future period (2031-2060 or 2071-2100) and re f erence refers to the mean value of the variable (index) over the historical period (1976-2005).

Historical Seasonal Cycle of CDD, CWD, and HWDI Indices over the Different Subregions of West Africa
In this paragraph, the annual cycles of the different indices (CDD, CWD, and HWDI) over the three subregions (the Western Sahel, the Eastern Sahel, and the Gulf of Guinea) of the West African domain are intercompared using the observation data (CPC) and the four CORDEX simulations.This enables us to highlight the ability of the CORDEX simulations to outperform the annual cycle.

CDD Index and Its Number
Figure 2 presents the annual cycle of CDD and CDD number over the three subregions of West Africa.CDD and its number present the same variability during a year over the different subregions, as shown by CPC as well as the CORDEX simulations.The period from November to March is characterized by longer dry spells (about 30 days) and the occurrence of the maximum number of dry spells (about 2 days), while during the period of May-September, a shorter dry spell (about 10 days) is depicted, with a relatively smaller number of dry spells occurring over the whole West African region.During the period of May-September, considered the monsoon season in the Sahel regions (western and eastern), a lower occurrence of dry spells is indicated, while during the period of April-October, known as the monsoon season in the Gulf of Guinea region, there is a higher occurrence, as revealed by previous studies [52,53,59].Overall, the annual cycle of CDD was well captured by the CORDEX simulations over all the three West African subregions, consistent with results from other studies over West Africa [36,77] and over Central Africa [78].The latter pointed out the accuracy of CORDEX simulations in capturing the annual cycle of CDD.
Atmosphere 2023, 14, x FOR PEER REVIEW 7 of 22 Central Africa [78].The latter pointed out the accuracy of CORDEX simulations in capturing the annual cycle of CDD.

CWD Index and CWD Number
The annual cycle variability in CWD and number of CWD over the three subregions of West Africa are presented in Figure 3. CWD and its number present bimodal variability during the year over the three subregions of West Africa, with an overestimation by the CORDEX simulations during the period from March to November.CCCMA overestimates CWD compared to other simulations, while HIRHAM simulates a similar pattern of CWD to other models compared to the CPC observation.

CWD Index and CWD Number
The annual cycle variability in CWD and number of CWD over the three subregions of West Africa are presented in Figure 3. CWD and its number present bimodal variability during the year over the three subregions of West Africa, with an overestimation by the CORDEX simulations during the period from March to November.CCCMA overestimates CWD compared to other simulations, while HIRHAM simulates a similar pattern of CWD to other models compared to the CPC observation.

HWDI and Number of HWDI
Figure 4 shows the variability in the annual cycle of HWDI and HWDI number for each West African subregion.HWDI and its number show the same seasonal cycle over each subregion.Over the Eastern Sahel, the seasonal variability in HWDI and its number exhibits three peaks in February, June and November.The Western Sahel presents two peaks in the months of March and May.In contrast, the observed seasonal cycle of HWDI and its number in the Gulf of Guinea indicates a marked peak in April.In summary, the seasonal variability in HWDI is well captured by the CORDEX simulations and their ensemble mean over the three subregions of West Africa, although some biases exist (Figure 4).
region.Over the Eastern Sahel, the seasonal variability in HWDI and its number exhibits three peaks in February, June and November.The Western Sahel presents two peaks in the months of March and May.In contrast, the observed seasonal cycle of HWDI and its number in the Gulf of Guinea indicates a marked peak in April.In summary, the seasonal variability in HWDI is well captured by the CORDEX simulations and their ensemble mean over the three subregions of West Africa, although some biases exist (Figure 4).

Validation of the CORDEX Simulations of CDD, CWD, and HWDI over West African Subregions
Figures 5-7 show the performance of the CORDEX simulations in simulating the seasonal cycle of CDD, CWD, and HWDI over the three subregions of West Africa using the Taylor diagram during the period of 1979-2005 with respect to the CPC observation data used as a reference.The different CORDEX simulations are characterized by differentcolored dots, as indicated in the legend (Figures 5-7).The green circles centered at the reference point represent the loci of constant RMS distance and the circles centered at the origin represent the loci of constant standard deviation.The correlation is represented as the cosine of the angle from the X-axis.The models with as much variance as the observation data, the largest correlation, and the smallest RMS error are considered the best performers in the Taylor diagram [79].

CWD
The Taylor diagram of CWD and CWD number is presented in Figure 6.For each West African subregion, the CORDEX simulations almost present the same standard deviation (SD) and root mean square difference (RMS) with a correlation coefficient value greater than or equal to 0.6 for both CWD and its number, except for the HIRHAM model over the Gulf of Guinea, which indicates a negative correlation coefficient (Figure 6).Furthermore, the HIRHAM simulation outperforms the others for CWD and its number over the Eastern Sahel, while CCLM, MME, and RCA outperform the others for CWD and its number over the Gulf of Guinea and Western Sahel.Overall, the CORDEX simulations overestimate the CWD and its number over the whole West African region.These results are in agreement with study [59], which underlined that most of the CORDEX simulations overestimate precipitation over Africa, although some biases exist based on the region, the season, and the evaluation metric.

HWDI
As with the CDD and CWD indices, the CORDEX simulations reproduce the HWDI and its number with the same standard deviation over West Africa.However, the amplitude of variation is well captured by the CORDEX simulations over the Eastern and Western Sahel, whereas an overestimation of the HWDI and its number is observed over the Gulf of Guinea subregion for the 1976-2005 period.In addition, MME agrees reasonably well with the CPC observation data in reproducing the HWDI over the Eastern Sahel and the HWDI number over the Gulf of Guinea, whereas the HIRHAM simulation outperforms the others in both the HWDI and its number over the Western Sahel, the HWDI  Overall, the correlation coefficient values between the CPC observation data and the different CORDEX simulations are summarized in Table 4.

CDD
Figure 5 shows the Taylor diagram of CDD and CDD number over the three subregions of West Africa.For the period of 1976-2005, all the CORDEX simulations overestimate the annual cycle of CDD and its number over West Africa with almost the same standard deviation but different root mean square differences (RMS).The CORDEX simulations have the correct standard deviation in capturing the CDD number over the Western Sahel and the Gulf of Guinea (Figure 5).In fact, the MME captures the amplitude of the CDD number over the Western Sahel and the Gulf of Guinea, with the RCA model being the best performer for CDD number over these subregions.These simulations indicate correlation coefficients of about 0.97 and 0.99, respectively; the CCLM simulation is the best performer over the Eastern Sahel with a 0.9 correlation coefficient.The Canadian simulation CCCMA shows the worst performance, with correlation coefficients of 0.9, 0.5, and 0.7 over the Gulf of Guinea, Eastern Sahel, and Western Sahel, respectively.Nevertheless, MME outperforms the others for CDD over the Gulf of Guinea subregion, with a correlation coefficient of more than 0.99, while RCA reproduces the CDD over the Eastern and Western Sahel reasonably well, with a correlation coefficient of more than 0.99.These results are in line with study [37].

CWD
The Taylor diagram of CWD and CWD number is presented in Figure 6.For each West African subregion, the CORDEX simulations almost present the same standard deviation (SD) and root mean square difference (RMS) with a correlation coefficient value greater than or equal to 0.6 for both CWD and its number, except for the HIRHAM model over the Gulf of Guinea, which indicates a negative correlation coefficient (Figure 6).Furthermore, the HIRHAM simulation outperforms the others for CWD and its number over the Eastern Sahel, while CCLM, MME, and RCA outperform the others for CWD and its number over the Gulf of Guinea and Western Sahel.Overall, the CORDEX simulations overestimate the CWD and its number over the whole West African region.These results are in agreement with study [59], which underlined that most of the CORDEX simulations overestimate precipitation over Africa, although some biases exist based on the region, the season, and the evaluation metric.

HWDI
As with the CDD and CWD indices, the CORDEX simulations reproduce the HWDI and its number with the same standard deviation over West Africa.However, the amplitude of variation is well captured by the CORDEX simulations over the Eastern and Western Sahel, whereas an overestimation of the HWDI and its number is observed over the Gulf of Guinea subregion for the 1976-2005 period.In addition, MME agrees reasonably well with the CPC observation data in reproducing the HWDI over the Eastern Sahel and the HWDI number over the Gulf of Guinea, whereas the HIRHAM simulation outperforms the others in both the HWDI and its number over the Western Sahel, the HWDI alone over the Gulf of Guinea, and the HWDI number alone over the Eastern Sahel (Figure 7).
Overall, the correlation coefficient values between the CPC observation data and the different CORDEX simulations are summarized in Table 4. Future changes were evaluated with respect to the historical mean period (1979-2005) for each index over West Africa in order to determine the future evolution of these indices' characteristics from the present.The changes were evaluated over the near-future period (2031-2060) and far-future period (2071-2100) under the RCP4.5 and RCP8.5 scenarios.

Seasonal Changes in the HWDI
Figure 8 displays changes in the HWDI derived from the CORDEX simulations and their multi-model ensemble mean (red) during near-(2031-2060) and far-(2071-2100) future periods over West African subregions.A substantial increase in the HWDI is observed over the three subregions of West Africa during both periods.Under RCP8.5, the changes are more pronounced during the far future compared to the near future and under the RCP4.5 scenario.The projected changes in heatwaves may be related to possible changes in the dynamical features of the West African monsoon and the teleconnection with El Niño/ENSO in the context of global warming [80,81].Especially in the coastal regions of West Africa, the moisture flux from the ocean associated with evapotranspiration over land increases the water vapor content entering the atmosphere, which could play a major role in heatwaves through an enhanced greenhouse effect [80,82].However, the mechanism of heatwaves in Europe is different.In Europe, heatwaves are often associated with anticyclonic conditions, stronger sensible heat fluxes, and deeper boundary layers [80].The fact that all CORDEX simulations indicate an increase in the HWDI suggests that the changes are robust [83,84].These results suggest that heatwaves will be amplified in West Africa in future periods.

Seasonal Changes in CWD
Seasonal changes in CWD exhibit a general decrease from March to November over the Eastern and Western Sahel regions, with the CCCLM simulation showing the maximum decrease.In the Sahelian regions, almost no change is observed during the dry season ranging from December to February.However, the models project that the Gulf of Guinea region will experience a substantial decrease in CWD throughout the year, except for CCCLM, which projects an increase during September-December corresponding to the retreat of the monsoon in the region (Figure 9).The discrepancy shown by the CCCLM simulation during the September-December season over the Gulf of Guinea and the projected large changes in CWD over the Sahelian regions (eastern and western) are due to the convective scheme used by this model.Indeed, CCCLM employs the Tiedtke convective scheme [85], causing an overestimation of spells over the Sahel and flat terrains of the Guinea region [61].

Seasonal Changes in CDD
Figure 10 shows the projected changes in the annual cycle of CDD over the three subregions of West Africa during near-and far-future periods under the RCP4.5 and RCP8.5 scenarios.The simulations project a general increase in CDD during March-November over the Sahel, except RCA, which projects a decrease.In addition, CCCMA indicates a decrease in September during the near-and far-future periods under both RCP scenarios.Over the Gulf of Guinea region, discrepancies exist between the CORDEX simulations regarding the projected changes in CDD.Indeed, during the year, RCA and CCCLM show both a decrease and an increase in CDD, while other simulations clearly indicate only an increase, revealing some uncertainties in projected CDD.Overall, the multi-model ensemble (MME) mean shows an increase in CDD over West Africa but a slight decrease over the Gulf of Guinea in November-December during the late century under the RCP8.5 scenario.It is worth noting that the periods of CWD changes coincide with the periods of changes in CDD over the Sahelian regions (eastern and western).Our results are in line with the results of [78], which suggested that the increases in dry spells coupled with the decrease in wet spells and wet-day frequency could have strong consequences for seasonal rainfall onset, along with the total yearly rainfall amount over Central Africa.This implies that the increase in dry spells coupled with the decrease in wet spells could have negative effects on the environment and society.For instance, this situation could induce strong and severe drought and human discomfort in the future.

Summary and Conclusions
In this study, the projected changes in the seasonal cycles of heatwave and wet and dry spell indices were investigated over West Africa under RCP4.5 and RCP8.5's forcing scenarios.To this end, an ensemble of four CORDEX simulations and their ensemble mean with observation data (CPC) were used.Changes were calculated with respect to a reference period of 1976-2005 for near-(2031-2060) and far-(2071-2100) future periods.Firstly, the performance of the CORDEX simulations and their ensemble mean was evaluated in comparison to the CPC observation data in terms of the seasonal cycle of indices including heatwaves and wet and dry spells.The results of our analysis indicate that the CORDEX simulations realistically reproduce the seasonal cycle of heatwaves and wet and dry spells over the three defined subregions of West Africa (the Eastern Sahel, the Western Sahel, and the Gulf of Guinea) in the present climate, although some biases exist.In general, the latter are characterized by an underestimation of 30% for CDD and an overestimation of about 60% and 50% for CWD and HWDI, respectively.The different biases observed in individual CORDEX simulations are due to the use of different convective schemes for the parametrization of the simulations.In addition, CORDEX simulations and their ensemble means reproduce the different indices with the pattern variations of a similar order of magnitude reasonably well.On the one hand, individual simulations outperform their ensemble mean in terms of the variability in the indices.On the other hand, the projections indicate an increase of about 10 to 20 days in heatwave duration and number over the whole West African region under both forcing scenarios and in future periods.There is an increase (of about 80%) in dry spells, whereas there is a decrease (of about 40%) in wet spells observed over the West African region.However, the changes in dry and wet spells span the period from March to November over the Sahelian regions (the Eastern Sahel and Western Sahel), while over the Gulf of Guinea, the changes span the whole year, except a slight decrease in dry spells during the retreat phase of the monsoon (November-December) in the far future under RCP8.5.Almost all the CORDEX simulations agree on the projected signal of the seasonal cycles of the different indices, except CCCLM, which shows a discrepancy in the projected signal of CWD during the September-December season over the Gulf of Guinea.This suggests that the projected changes are robust.Finally, the period of increasing dry spells coincides with that of the decreasing wet spells over West Africa, indicating that West African countries will be at risk of severe drought, flooding, and heatwaves.Thus, governments and stakeholders must take action to suitably reduce greenhouse gas emissions in West Africa.

Figure 1 .
Figure 1.West African domain elevation in (m) including its three subregions (Western Sahel, Eastern Sahel, and Gulf of Guinea).

Figure 1 .
Figure 1.West African domain elevation in (m) including its three subregions (Western Sahel, Eastern Sahel, and Gulf of Guinea).

Figure 2 .
Figure 2. Seasonal cycle of CDD (a-c) and CDD number (d-f) over the different subregions for the observation data (CPC) and the CORDEX simulations.

Figure 2 .
Figure 2. Seasonal cycle of CDD (a-c) and CDD number (d-f) over the different subregions for the observation data (CPC) and the CORDEX simulations.

Figure 3 .
Figure 3. Same as Figure 2, but for CWD (a-c) and num

Figure 3 .
Figure 3. Same as Figure 2, but for CWD (a-c) and number of CWD (d-f).

Figure 4 .
Figure 4. Same as Figures 2 and 3, but for HWDI (a-c) and number of HWDI (d-f).

Figure 4 .
Figure 4. Same as Figures 2 and 3, but for HWDI (a-c) and number of HWDI (d-f).

Figure 5 .
Figure 5.Taylor diagrams displaying a statistical comparison of CORDEX simulations and the CPC observation data for CDD and CDD number during the historical period (1976-2005) over Gulf of Guinea (GG; a,d), Eastern Sahel (ES; b,e), and Western Sahel (WS; c,f).

Figure 5 . 22 Figure 6 .
Figure 5.Taylor diagrams displaying a statistical comparison of CORDEX simulations and the CPC observation data for CDD and CDD number during the historical period (1976-2005) over Gulf of Guinea (GG; a,d), Eastern Sahel (ES; b,e), and Western Sahel (WS; c,f).Atmosphere 2023, 14, x FOR PEER REVIEW 11 of 22

Figure 6 .
Figure 6.Same as Figure 5, but for CWD and CWD number.

Figure 7 .
Figure 7. Same as Figures 5 and 6, but for HWDI and HWDI number.

Figure 7 .
Figure 7. Same as Figures 5 and 6, but for HWDI and HWDI number.

Figure 9 .
Figure 9. Same as Figure 8, but for CWD.Figure 9. Same as Figure 8, but for CWD.

Figure 9 .
Figure 9. Same as Figure 8, but for CWD.Figure 9. Same as Figure 8, but for CWD.

Figure 10 .
Figure 10.Same as Figure 8 and Figure 9, but for CDD.Figure 10.Same as Figures 8 and 9, but for CDD.

Figure 10 .
Figure 10.Same as Figure 8 and Figure 9, but for CDD.Figure 10.Same as Figures 8 and 9, but for CDD.

Table 1 .
List of the different data (CORDEX simulations and CPC observation) with their resolutions and periods.

Table 1 .
List of the different data (CORDEX simulations and CPC observation) with their resolutions and periods.

Table 2 .
The details of the CORDEX simulations composed of the RCMs and the GCMs used as boundary data.

Table 4 .
Correlation coefficients of CDD, CWD, and HWDI between the CPC observation data and the different CORDEX simulations over West African subregions.

Table 4 .
Correlation coefficients of CDD, CWD, and HWDI between the CPC observation data and the different CORDEX simulations over West African subregions.