Potential changes in temperature extreme events under global warming at 1.5 °C and 2 °C over Côte d'Ivoire^*

Côte d'Ivoire is a country in West Africa located in the Gulf of Guinea (figure 1). Its economy is based largely on rain-fed agriculture and rainfall is driven by the WAM. The regional climate is characterised by the north–south shift of the inter-tropical convergence zone, which is warm and humid, ranging from equatorial in the southern coasts, to tropical in the central regions and semiarid in the far north. In the southern coastal regions, precipitation patterns are correlated with air–sea interactions in the equatorial Atlantic known as the Atlantic Niño index, rather than the WAM index (Nnamchi and Li 2011). The dry and hot season is from November to May, while the wet and hot season is from June to October. The country is the world's top exporter of Cocoa with almost 70% of the population involved in agriculture. With a total population of 22 million in 2015, the country's population rate is increasing and is expected to reach 48 million inhabitants in 2050.

Zoom In Zoom Out Reset image size

In this study we consider three climatic regions in the country, the southern equatorial region from the coast 4° N to 6° N (R1), the central tropical region from 6° N to 8° N (R2) and the northern semi-arid climatic region from 8° N to 10.5° N (R3) (figure 1).

2.2.1. CORDEX-Africa data

2.2.2. Meteorological observation data

This present work focuses on the changes in temperature based on the GWLs (1.5 °C and 2 °C). The term GWLs is defined as the average GWL above some baseline period (of e.g. 1.5°, 2°, 3°, 4°). Various definitions, depending on a pre-industrial baseline with a mean period of 15, 20 or 30 years, have been used to describe the GWLs. In this work, the methods used to determine the GWLs (1.5 °C and 2 °C) and the robustness of a climate change signal follow Nikulin et al (2018), who report on the timing (i.e. years) of when these levels are crossed, and an assessment of the robustness of the changes are detailed in Nikulin et al (2018).

The timing of each warming level crossing is computed for each downscaled GCM and defined as the centre year of a long enough period (usually 20 or 30 years) when global mean temperature reaches predefined anomalies (1.5 °C or 2 °C) compared to preindustrial levels. Then, for the RCM, the same GWL timing defined by the downscaled GCM is used to extract 30 years for analysis. The corresponding 30 years period is then extracted from the corresponding RCM simulations for analysis using 1971–2000 as a control (CTL) period, as discussed in Nikulin et al (2018). The time window 1861–1890 is taken as the preindustrial period, as it is a reasonable approximation for pre-industrial global mean temperature and it is available for all CMIP5 historical simulations, as underlined in Klutse et al (2018).

A brief synthesis is provided here, however detailed information can be found in Nikulin et al (2018). Information on the changes of the climate will be considered as robust when it fulfils the following two criteria: (a) more than 80% of considered model simulations agree on the sign of the supposed change; (b) the signal to noise ratio, i.e. the ratio of the mean to the standard deviation of the ensemble of climate change signals, is equal to or larger than one. While the first criterion considers only the agreement of the ensemble members, the second criterion is a metric of the strength of the climate change signal. The second criterion is necessary, as even if the first criterion is met, the change signal may be negligible. The responses to the 1.5 °C and 2 °C GWLs of extreme temperature are assessed through five (5) temperature-based indices from the Expert Team on Climate Change Detection and Indices (ETCCDI). These include TX10P, TX90P, TN10P, TN90P and HWDI for temperature (see table 1 for a full description of the indices).

An assessment of the daily temperature during the historical period (1982–2000) over Côte d'Ivoire was made using ground-based observations from SODEXAM meteorological stations, as shown in figure 2. In the Taylor diagram (Taylor 2001), the individual models interpolated to the nearest grid points around a station considered are represented in black dots, the ensemble mean in red and the meteorological observation in blue dots is used as reference data and labelled 'ref' in the figure 2. Because of differences in the calendars used by individual models (for example the simulations with HadGEM, use a 360 d calendar, while others are using the standard Gregorian calendar) the analysis with the Taylor diagram is made at a monthly scale.

Zoom In Zoom Out Reset image size

The analysis of the results shows that along the coastal area, in the southern climatic zone (stations of Abidjan, Adiaké, San Pedro, Sassandra and Tabou), most of the single members of the RCMs ensemble show lower standard deviation values close to the reference and relatively low values of RMSE. Also, the correlation values between these RCMs and these stations of the southern zone are between 0.7 and 0.9, except for Tabou where the values are between 0.6 and 0.8. Thus, the ensemble mean of the RCMs is located closer to the reference, suggesting a good replication of the observed temperature by the CORDEX- Africa RCMs over the southern part of Côte d'Ivoire, especially along the coast.

Over the central part of the country, for the stations of Bondoukou, Bouaké, and Yamoussoukro, the single members of the CORDEX-Africa RCMs display standard deviation values at both sides of the reference one. Thus, the ensemble mean has a standard deviation value mostly identical to the reference value. Also, across the stations of Daloa, Gagnoa, and Man, the values of standard deviation displayed by the single members and the ensemble mean are slightly higher than the reference. The values of the correlation coefficient, in this region, are varying from 0.6 to 0.8 for Yamoussoukro, 0.4–0.9 for Bondoukou, and from 0.4 to 0.8 in Bouaké, Daloa, Gagnoa, and Man. Also, the values of the RMSE displayed are relatively higher between these RCM members and the observation, suggesting an acceptable ability of these RCMs in reproducing the temperature patterns of the central part of Côte d'Ivoire.

In the northern climatic zone (stations of Korhogo and Odienné), the majority of the single ensemble members of the RCMs display standard deviation values close to the reference value. These northern stations also have smaller values of RMSE and a high correlation between these RCMs and the observations. Therefore, the ensemble mean of the CORDEX-Africa RCMs and the single members are located closer to the reference, suggesting a better performance of these RCMs over the northern part of Côte d'Ivoire compared to the other regions. This assessment of the simulated daily temperatures of CORDEX-Africa over the period 1982–2000 highlights an increasing performance of RCMs from the South towards the North of Côte d'Ivoire, in agreement with previous works which underlined that the skills of the RCMS varied from one location to another with higher performance away from the coastal area (Kouadio et al 2015).

Moreover, Gbobaniyi et al (2014) noticed substantial variable biases in both the magnitude and spatial extent from model to model in the CORDEX-Africa RCMs. These biases in the RCMs were related to several sources, including boundary condition, internal dynamic physical schemes convective parameterization and horizontal resolution (Crétat et al 2014, Klutse et al 2015).

Across Côte d'Ivoire, the percentage of days when the daily minimum temperature is below the 10th percentile (TN10P) decreases by up to −12% under GWL 1.5 °C and −15% under GWL 2 °C (figures 3(a)–(d)). These decreases suggest a reduction of the occurrence of cold night events, especially along the coastal region and the northern part of the country of up to 12% (GWL 1.5 °C) and 15% (GWL 2 °C) from about 20% of the time during the reference period (1971–2000). This reduction is at a minimum over the central parts of the country, less than 5% with GWL 1.5 °C and approximately 5% for GWL 2 °C. Compared to GWL 1.5 °C, the extra warming to GWL 2 °C could bring an additional reduction of the occurrence of cold nights by 2% – 4%, with maxima mainly located in the western, the southern and northern part of the country.

Zoom In Zoom Out Reset image size

An increase in GWL to both 1.5 °C and 2 °C will increase the highest values of the minimum temperature (TN90P), from 23% to 25% during the reference period, to about 36% and 54% respectively (figures 3(e)–(h)). This shows that exceeding the GWL threshold of both 1.5 °C and 2 °C will result in an increase in warm nights compared to the historical period. The projected increase in warm nights shows a north–south gradient with maximum values towards the coast. Exceeding the GWL 1.5 °C and reaching the 2 °C threshold is projected to cause an increase in warm nights, characterised by a north–south gradient, with 5% in the north to 14% towards the coastal part of the country.

The spatial distribution of the annual mean changes in TX10P due to the GWLs at 1.5 °C and 2 °C, over Côte d'Ivoire (figures 3(i)–(l)), suggests a decrease, from about 19% during the historical period, up to 9% and 12% respectively. The spatial distribution pattern of these changes shows a south–north gradient with highest reductions of TX10P in the south (compared to the north) at both GWLs at 1.5 °C and 2 °C. These changes show that the percentage of cold days will reduce across the country as the global temperature rises above the level of 1.5 °C and 2 °C. The occurrence of cold days decreases by approximately 3.2% at GWL 2 °C compared to GWL 1.5 °C, I with a maximum reduction located over the coastal zones and along the eastern border with Ghana.

The projected changes in the spatial distribution of the annual mean of TX90P over Côte d'Ivoire show an increase under GWLs at 1.5 °C and 2 °C (figures 3(m)–(p)). The percentages of occurrence of TX90P increase from 28% during the historical period by about 19% and 27% in the context of GWLs at 1.5 °C and 2 °C respectively compared to the historical period.

Thus, in the context of GWL at 1.5 °C and 2 °C, the warm days events are projected to be 12% – 18% and 15% – 30% respectively more frequent compared to the reference period, which recorded an occurrence of about 20%. The maximum values of these changes are projected in the coastal zone under GWL at 1.5 °C and spread across the country in the context of GWL at 2 °C.

The enhanced GWL at 2 °C (compared to GWL 1.5 °C) will result in an additional increase of up to 8% in the occurrence frequency of warm day events.

An enhancement of the GWL to the 1.5 °C and 2 °C thresholds may lead to changes in the temperature pattern across Côte d'Ivoire similar to the West African region discussed by Kumi and Abiodun (2018) and Klutse et al (2018).

The spatial distribution of the annual mean of HWDI indices (figures 3(q)–(t)) revealed increases of these indices for both 1.5 °C and 2 °C of GWLs. These increases in HWDI of up to 25 and 55 d respectively for GWLs 1.5 °C and 2 °C indicate that warm spells are expected to lengthen over Côte d'Ivoire. Also, the spatial distribution shows that the maxima are located over the coastal region in the southern part of Côte d'Ivoire. These warm spells are expected to be more pronounced along the coast and in the northern half of the country. Moving from GWL 1.5 °C to 2 °C lengthens the warm spells by up to 27 d, with maximum values located along the coast, in the middle east, and the north–west of the country.

The previous section discussed the annual mean and spatial distribution of the changes in temperature extremes over Côte d'Ivoire. It reported an increase in warm days, nights, and spells associated with a decrease in cold days and nights in the context of GWL of both 1.5 °C and 2 °C thresholds. The current section focuses on the seasonal mean changes in TN10P, TN90P, TX10P, TX90P and HWDI under GWLs at 1.5 °C and 2 °C, across Côte d'Ivoire and for each of its climatic zones as defined in figure 1. Following Yapo et al (2019), four seasons, namely January–February–March (JFM), April–May–June (AMJ), July–August–September (JAS) and October–November–December (OND) have been used for each of the three climatic zones.

The assessment of the projected changes in temperature extremes over Cote d'Ivoire and its climatic regions projects a decrease in the occurrence frequency of cold nights and days, associated with an increase in the occurrence frequency of warm nights and days. These expected changes may also result in an increase in the length of the maximum warm spell per year at 1.5 °C and 2 °C GWLs. This implies that the expected changes in temperature extremes would affect the diurnal cycle of temperature over the entire country and its climatic regions, during all of the seasons. Thus, these changes in temperature will potentially impact agriculture, crop production, and food security as well as human health and energy.

The maximum seasonal changes in the frequency of warm nights is projected to occur during JAS under GWL at 1.5 °C at the country level and in all of the climatic zones. In the context of GWL at 2 °C, the maximum rise in warm night events occurs during the season OND in the southern and central part of the country and during the season AMJ for the northern zone and at country level (table 2). The minimum changes in warm night occurrence are mainly expected during JFM. The maximum decrease, in cold night events, under GWL at 1.5 °C is expected during JAS in southern and northern Cote d'Ivoire. In the context of GWL at 2 °C, the maximum decrease in the occurrence of cold nights are projected in AMJ across the south and centre and during JAS in the north of Côte d'Ivoire (table 2). The minimum decrease in cold night events varies between seasons and regions at GWL 1.5 °C and mainly in OND under GWL 2 °C. The maximum increase in warm day occurrences is expected during JFM in all climatic zones, while the rainy seasons are relatively less affected. The maximum decrease in cold days, under GWL at 1.5 °C, is expected in JFM in the south and north and OND in the centre. In the context of GWL at 2 °C, the maximum decrease in cold nights is expected in OND. This suggests more frequent warm nights during the WAM period, when the parts with bimodal rainfall patterns (southern and central) experience the little dry season, meaning that the JAS season will experience more warmer nights. Warm days are projected to increase during JFM in a period that already records frequent warm events. The increase of 0.5 °C from the 1.5 °C threshold to 2 °C may not only impact the amplitude of temperature extreme events such as cold days/nights and warm days/nights, but also induce a slight shift of these amplitudes between seasons.

These general increases in warm day/night and decreases in cold day/night occurrences, associated with the occurrence of longer warm spells expected over Côte d'Ivoire, are in line with Klutse et al (2018), who show a rise in temperature across West Africa at both 1.5 °C and 2 °C GWL using CORDEX data under the RCP 8.5. Moreover, Nikulin et al (2018) show that the West African region has already experienced a 1.5 °C warming since 2004 and the temperature is expected to continue to increase through 2049. They further noticed that some models indicated that the temperature rise has already crossed the 2 °C GWL since the year 2012 and would continue to rise through 2066. Furthermore, the current results are in agreement with a similar analysis in temperature extreme over Côte d'Ivoire (Yapo et al 2019) that projected an increase, during the periods 2031–2060 and 2071–2100, in the frequency of warm spell days, very warm days, and the warm nights across the country, under both RCP4.5 and RCP8.5. Using the low-warming projections of NCAR-CESM to assess the differences in extreme temperature events under the GWLs at 1.5 °C and 2.0 °C, relative to the recent climate taken from 1976 to 2005 over Africa, Iyakaremye et al (2021) found that the frequency of occurrence of hot days (TX90P) is projected to increase with global warming across the African continent, while the frequency of cold day events (TN10P) is expected to decrease, especially across West Africa. Moreover, Mba et al (2018) using CORDEX projection under RCP8.5 indicated that despite large uncertainties associated with projections at 1.5 °C and 2 °C, the 0.5 °C increase in global temperature is associated with a larger regional warming response in Africa. The intercomparisons between the changes due to GWL at 1.5 °C and 2 °C reveal generally marginal differences, in agreement with Klutse et al (2018). Most of the CORDEX-Africa models (generally more than 90%) used in the present work, across Côte d'Ivoire and the respective climatic zones, agree on the signs of the changes in temperature extreme. This indicates robust results for the increase in warm days/nights and the decreases in cold days/nights, associated with the occurrence of longer warm spells. Our results clearly indicate that these changes in temperature extremes, which would occur over all the climatic regions and during all of the seasons in Côte d'Ivoire, may certainly influence heatwave intensity and occurrence and may also contribute to increased disaster risks. Additionally, previous works showed that across Africa, an increase in hot nights, long and persistent heatwaves, and disaster risks are expected to increase under GWL at 1.5 °C and will be amplified in the 2 °C level (Dosio et al 2018, Weber et al 2018, Djalante 2019).

Several areas of the country are projected to experience warmer hot extremes which will necessarily impact agriculture, crop production, and food security in a country like Côte d'Ivoire, whose economy is mainly based on rain-fed agriculture, since surface temperature may affect plant growth, development and crop yield. The current agricultural system will need to be restructured in light of the present results, by developing and integrating species more tolerant to this temperature increase, to ensure food security. Concerning the energy sector, these longer warm spells, at the country level will probably increase temperature discomfort, which will induce more energy demand and consumption for cooling systems in both households and offices (Akara et al 2021). It will also affect solar energy production (Danso et al 2022) which remains one of the major sources of low carbon energy. Moreover, since our results highlighted that a difference of 0.5 °C may not only impact extreme temperature amplitudes but may induce some seasonal shifts, changes will also be expected in the current pattern of energy demand. In a country where electricity production and access is still climate-dependent, as seen recently (in 2021) with widespread power shortage due to climatic hazards (low precipitations, high temperatures), it will be necessary and useful for decision-makers to put in place more sustainable energy systems, such as renewable energy, integrating the findings of this work in the infrastructure designs. These projected changes in temperature in Africa, can be an opportunity for African countries to put in place common policies, to better develop tools and use the continent solar energy reserves which are estimated to be about 60 000 000 TWh yr⁻¹ (representing 40% of the global total solar energy potential), making Africa the most sun-rich continent in the world (Liu 2015).

As underlined by Nikulin et al (2018), the change in temperature over Cote d'Ivoire shown in the present results may be biased by some specific RCMs of the ensemble considered. Similar to their finding, the CORDEX ensemble used in this study is at least biased towards ten members of one RCM. Although in such a case, it is advisable to carefully analyse and understand the individual models considered, though this has not been analysed here. Therefore, our results only provide an overview of temperature expectations under GWLs at 1.5 °C and 2 °C over Côte d'Ivoire and its sub-climatic zones.