3.1. Projected Number of Hot Days
The number of days with maximum AT above the AT thresholds in
Table 1 (defined as Hda1, Hda2, Hda3, Hda4 and Hda5 in
Section 2.3) were modelled and downscaled for Africa per year for 1961–2100 by CCAM. The average number of days per year for each Hda threshold in each specified time period were calculated; and the ensemble average of these average number of days are shown in this section.
Figure 1 below shows the average number of Hda2 per year for 1961–1990 (present day climate;
Figure 1a), and the average change in number of Hda2 per year for the times slices of 2011–2040 (
Figure 1b), 2041–2070 (
Figure 1c) and 2071–2100 (
Figure 1d). The change is displayed in
Figure 1b–d in order to easily highlight the projected impact that climate change will have on the number of hot days. Thus, in order to understand the absolute magnitude of Hda2 projected on average per year in 2011–2040, the change from
Figure 1b would need to be added to the number for the average number of days per year in
Figure 1a. For example, in present day climate, Johannesburg, South Africa is modelled to have 34.5 Hda2 per year on average for 1961–1990 (blue in
Figure 1a). In 2011–2040, Johannesburg is projected to have an increase of 35 Hda2 on average per year (yellow in
Figure 1b); this would give a total of 69.5 Hda2 on average per year in 2011–2040.
Figure 1.
CCAM model derived (A) average number of Hda2 per year in present climate; (B) projected change in average number of Hda2 per year in 2011–2040 compared to 1961–1990; (C) projected change in average number of Hda2 per year in 2041–2070 compared to 1961–1990; (D) projected change in average number of Hda2 per year in 2071–2100 compared to 1961–1990.
Figure 1.
CCAM model derived (A) average number of Hda2 per year in present climate; (B) projected change in average number of Hda2 per year in 2011–2040 compared to 1961–1990; (C) projected change in average number of Hda2 per year in 2041–2070 compared to 1961–1990; (D) projected change in average number of Hda2 per year in 2071–2100 compared to 1961–1990.
In general, there is an increase in Hda2 across the continent, which by using ATmax = 27 °C as the threshold for where heat may start impacting health, indicates that there are projected to be more days into the future where health may be impacted by high temperatures. The average number of Hda1 (ATmax < 27 °C) was also modelled (
Figure S1). Across the continent, the number of Hda1 decreased, which indicates that the number of days when health is less likely to be impacted by heat are projected to decrease across the continent. This clearly indicates that the potential risk to human health from high AT is projected to increase across the continent.
In
Figure 1a, a large part of equatorial Africa and coastal areas (in red) have many Hda2 already in the present day climate; thus it is not possible for there to be a large change in the number of Hda2. However, across the rest of Africa, there is an increase in the average Hda2 per year in each time slice, and the spatial trends in the projected increases are similar across time slices. For example, the areas with the largest projected increases in 2011–2041, such as the high-lying areas of the escarpment from Ethiopia through Tanzania (in orange and red in
Figure 1b), are also projected to experience the highest increases in the 2071–2100 time period (in red in
Figure 1d). In northern Africa, the increases in Hda2 do not have large spatial variability, while in southern Africa there is more spatial heterogeneity in the projected increases. High spatial resolution regional climate modelling focused on areas with high spatial heterogeneity may be useful in creating tailored projections in order to spatially resolve this heterogeneity.
Figure 2 displays the average number of Hda2 per year projected for 2071–2100. The color scale is the same as in
Figure 1a, which is the average number of Hda2 per year modelled for the current climate.
Figure 2 highlights that the spatial extent of Africa that will now have close to every day as Hda2 (in red) has extended from the current climate. In addition, the majority of Africa is projected to have, on average, over 5 months of the year as Hda2 (in yellow, orange and red). While much of South Africa and areas in the East African highlands are modelled to have fewer than 88 Hda2 days per year in the current climate (
Figure 1a), very few areas are projected to have fewer than 88 Hda2 per year (in blue) in 2071–2095.
Figure 2.
CCAM model outputs for number of Hda2 per year projected in 2071–2100.
Figure 2.
CCAM model outputs for number of Hda2 per year projected in 2071–2100.
In order to understand the number of days projected across different apparent temperature thresholds, Hda3, Hda4 and Hda5 were analyzed. Applying different thresholds is helpful to understand the potential severity of health impacts, as higher temperatures lead to an increase in mortality. Additionally, the thresholds can be used to understand at what AT threshold different regions are projected to begin to experience increases in “hot days”. For example, as stated above, for the highlands of East Africa, this increase is projected to begin at the Hda2 threshold.
Figure 3 displays the number of days on average per year for the present day climate and the projected change in the average number of days per year in the time slice 2071–2100 for Hda3, Hda4 and Hda5 (all time slices for Hda3, Hda4 and Hda5 are displayed in
Figures S2–S4 of supplementary data).
Figure 3.
CCAM model derived (A) average number of Hda3 per year in present climate (1961–1990); (B) change in average number of Hda3 per year in 2071–2100 compared to 1961–1990; (C) average number of Hda4 per year in present climate (1961–1990); (D) change in average number of Hda4 per year in 2071–2100 compared to 1961–1990; (E) average number of Hda5 per year in present climate (1961–1990); (F) change in average number of Hda5 per year in 2071–2100 compared to 1961–1990.
Figure 3.
CCAM model derived (A) average number of Hda3 per year in present climate (1961–1990); (B) change in average number of Hda3 per year in 2071–2100 compared to 1961–1990; (C) average number of Hda4 per year in present climate (1961–1990); (D) change in average number of Hda4 per year in 2071–2100 compared to 1961–1990; (E) average number of Hda5 per year in present climate (1961–1990); (F) change in average number of Hda5 per year in 2071–2100 compared to 1961–1990.
There was not a projected decrease in number of days for any of the three thresholds. For Hda3, equatorial Africa, and the western part in particular, there are already numerous days above ATmax = 32 °C, and, as in Hda2 (
Figure 1), large increases are not possible. The largest projected increases in Hda3 are seen in a band from Angola, across northern Zambia, southern Democratic Republic of Congo and Tanzania, up across Uganda, Kenya, and Ethiopia (
Figure 3b in red). Additionally, there are pockets of projected increases in Hda3 in western Africa (e.g., in Cameroon and Nigeria). In general, these areas are on the border of the areas that already have a large number of Hda3 in the present climate. This suggests that several of these areas displaying large increases in
Figure 3b may have many days very close to the threshold in the current climate, and thus the projected increasing temperature will result in shifting a large number of days from just below the Hda3 threshold to above it. This is similar to the increases seen in the high-lying areas of the escarpment in Hda2 in
Figure 1.
The spatial extent of areas in Africa with large number of Hda4 days in the present climate is much smaller than Hda2 and Hda3 (
Figure 3c). It is at the Hda4 threshold where western equatorial Africa, in particular, is projected to see large increases in number of days. In southern Africa, the northern and eastern Botswana borders with Angola and Zambia, much of Mozambique, and some of the eastern coastal areas are projected to see increases (in yellow and orange in
Figure 3d); the rest of the region does not experience many Hda4 in the current climate, and is not projected to see comparatively large increases in number of days by the end of the century.
There are very few Hda5 days in the present climate anywhere in Africa (
Figure 3e, note the small scale on the color bar). The spatial extent of areas in Africa that are projected to see increases in Hda5 by 2100 is narrow, with increases in the Democratic Republic of Congo, which is most likely driven by changes in relative humidity, and in the eastern Sahara desert. Previous research has found that the eastern Sahara is one region that is projected to see some of the largest increases in temperature on the continent [
67]. Additionally, both of these areas may have many days already close to this threshold, and thus the projected increase moves many days over the AT threshold.
3.2. Rate of Increase of Projected Number of Hot Days
In order to begin to understand what areas of Africa are projected to see the largest rates of increase in AT, a time series of the 11-year moving average for the number of hot days for all thresholds (
i.e., Hda2, Hda3,
etc.) was calculated for the full model domain (
i.e., Africa). The time series for 12 selected cities are displayed in
Figure 4,
Figure 5 and
Figure 6 below in order to highlight the projected impact on large African cities, as well as to highlight the variability in the projected increases and rate of increases (magnitude and shape) projected for the continent. The 12 cities were selected due to their large and growing populations (
Table 2), and for a representative geographical spread of the areas highlighted. Those cities that were analyzed are Lagos, Nigeria; Cairo, Egypt; Kinshasa-Brazzaville conurbation, Democratic Republic of the Congo and Republic of the Congo; Johannesburg, South Africa; Mogadishu, Somalia; Khartoum, Sudan; Dar es Salaam, Tanzania; Casablanca, Morocco; Nairobi, Kenya; Luanda, Angola; Addis Ababa, Ethiopia; and Dakar, Senegal. Not all cities are shown in each figure to ease the viewing of these figures; those cities that were projected to either see no change or a very little change are not shown below, but are included in the
supplementary material (Figures S5–S7).
Table 2.
Population per city as reported by the United Nations Human Settlements Programme [
68].
Table 2.
Population per city as reported by the United Nations Human Settlements Programme [68].
City, Country | Population in 2010 | Projected Population in 2020 | Projected Population in 2025 |
---|
Cairo, Egypt | 11,031,000 | 13,254,000 | 14,740,000 |
Lagos, Nigeria | 10,788,000 | 15,825,000 | 18,857,000 |
Kinshasa-Brazzaville conurbation, Democratic Republic of the Congo and Republic of the Congo | 9,972,000 | 14,396,000 | 16,899,000 |
Luanda, Angola | 4,790,000 | 7,555,000 | 8,924,000 |
Khartoum, Sudan | 4,516,000 | 6,018,000 | 7,090,000 |
Johannesburg, South Africa | 3,763,000 | 4,421,000 | 4,732,000 |
Nairobi, Kenya | 3,237,000 | 4,939,000 | 6,143,000 |
Dar es Salaam, Tanzania | 3,415,000 | 5,677,000 | 7,276,000 |
Casablanca, Morocco | 3,009,000 | 3,580,000 | 3,911,000 |
Addis Ababa, Ethiopia | 2,919,000 | 3,881,000 | 4,705,000 |
Dakar, Senegal | 2,926,000 | 4,227,000 | 5,064,000 |
Mogadishu, Somalia | 1,426,000 | 2,693,000 | 3,309,000 |
Figure 4 displays the 11-year moving average of Hda2 per year,
Figure 5 displays the 11-year moving average of Hda3 per year, and
Figure 6 displays the 11-year moving average of Hda4 per year for selected African cities. In each graph, three model analyses were plotted for each threshold as a function of time; the ensemble 10th percentile (blue), the ensemble median (black), and the ensemble 90th percentile of the number of days per year (red).
Figure 4,
Figure 5 and
Figure 6 show that for each city the three outputs show the same trends, however, there is a difference in magnitude and this difference varies depending on the city.
In
Figure 4, the plots for Dar es Salaam and Khartoum highlight examples of cities that are projected to have a small change in the number of Hda2, as those cities already experience a large number of Hda2 in today’s climate. Addis Ababa and Nairobi see a non-linear rate of increase in the projected number of days, with the rate increasing in the latter part of the century. Both of these cities are in the area in red in
Figure 1, which indicates that they are projected to see the largest increases in Hda2;
Figure 4 also highlights the projected magnitude of the rate of change and how it is projected to change in the time period modelled. The other cities highlighted, namely Johannesburg, Dakar, Cairo and Casablanca, also see increases in Hda2, however, the shape of the time series is closer to linear than the time series for Nairobi and Addis Ababa.
In
Figure 5, Casablanca, Nairobi and Addis Ababa are not shown as they are projected to see very small increases in Hda3 compared to the other cities studied, which lead to few Hda3 projected in 2095 (the ensemble 90th percentile value of Hda3 < 33 days in 2095). Thus, these cities are projected to see increases in days with ATmax ≥ 27 °C (
i.e., Hda2) but not in days with the threshold of ATmax ≥ 32 °C (
i.e., Hda3); this indicates that the largest projected increases are occurring in the AT range of 27 °C ≤ ATmax ≤ 32 °C.
Figure 4.
Eleven-year moving average of the number of Hda2 per year in selected cities in Africa. The ensemble 10th percentile (blue), the ensemble median (black) and the ensemble 90th percentile (red) of number of days per year are shown.
Figure 4.
Eleven-year moving average of the number of Hda2 per year in selected cities in Africa. The ensemble 10th percentile (blue), the ensemble median (black) and the ensemble 90th percentile (red) of number of days per year are shown.
Johannesburg and Dakar are both projected to have increases in Hda3 from very few days in the current climate, with Dakar projected to see the increases earlier in the century than Johannesburg. The other cities are projected to see steady increases in Hda3, and all, except for Cairo, are projected to see increases in Hda3 until almost every day is projected to be an Hda3 by the end of the century.
Figure 5.
Eleven-year moving average of the number of Hda3 per year in selected cities in Africa. The ensemble 10th percentile (blue), the ensemble median (black) and the ensemble 90th percentile (red) of number of days per year are shown.
Figure 5.
Eleven-year moving average of the number of Hda3 per year in selected cities in Africa. The ensemble 10th percentile (blue), the ensemble median (black) and the ensemble 90th percentile (red) of number of days per year are shown.
In
Figure 6, Kinshasa-Brazzaville is shown for the first time, as at the Hda2 and Hda3 thresholds the conurbation already had a large number of days in the current climate and thus could not see an increase into the future (
Figures S5 and S6). For Hda4, this conurbation is projected to experience a large non-linear increase. Cairo and Khartoum have a steady increase in the number of days. It should be noted that across
Figure 4,
Figure 5 and
Figure 6, for Cairo, and across
Figure 5 and
Figure 6 for Khartoum, the shape of the time series was very similar showing a very steady increase. Many of the other cities showed a variable rate of increasing days at one of the thresholds (
i.e., Nairobi in
Figure 5, Kinshasa-Brazzaville in
Figure 6); however, neither Cairo nor Khartoum are projected to see this type of increase. This difference in the shape of the time series in different cities, which has implications for the rate of increase of the number of hot days, highlights the variability of the projected impact of climate change on apparent temperature across the continent. This reiterates the need for high resolution regional climate projections that provide projections on smaller areas with a finer spatial scale which can provide finer details of spatial differences in temperature and the drivers of the heterogeneity. Further investigation of the meteorological reasons driving the differences in the rate of change for these cities fall beyond the scope of this paper.
Figure 6.
Eleven-year moving average of the number of Hda4 per year in selected cities in Africa. The ensemble 10th percentile (blue), the ensemble median (black) and the ensemble 90th percentile (red) of number of days per year are shown.
Figure 6.
Eleven-year moving average of the number of Hda4 per year in selected cities in Africa. The ensemble 10th percentile (blue), the ensemble median (black) and the ensemble 90th percentile (red) of number of days per year are shown.
A Monte Carlo analysis was performed for the time series for Hda2, Hda3 and Hda4 from the median of the ensemble members (e.g., black lines in
Figure 4,
Figure 5 and
Figure 6) in order to test for the significance of the increasing trends seen. Across the full model domain, the increasing trend was significant at the 99% significance level, except for areas that saw no change in the number of days either because the area already has 365 days per year at or above that threshold in the current climate (e.g., Lagos in Hda2) or that the area does not currently have any days at or above a threshold and is not projected to have any increase at that level in the future (e.g., East African Highlands in Hda4). Thus, at all areas when an increase is projected, the increase was determined to be statistically significant.
In order to analyze the rate of increase across Africa, the average increase in days per year over the full timescale of the calculated 11-year moving averages (
i.e., 1966–2095) was calculated for Hda2, Hda3 and Hda4 and is shown in
Figure 7. An average increase was selected for this analysis in order to provide a comparable analysis of the rate of increase across the entire domain. In
Figure 7a, the average rate of increase for Hda2 is displayed. In this figure, the highlands in East Africa are projected to see the largest average rate of increase (red). Nairobi is projected to see an average rate of increase of 1.98 Hda2 per year, and Bujumbura, Burundi is projected to see one of the higher projected average rates of increase at 2.36 Hda2 per year. In comparison, Johannesburg, South Africa, which also is projected to see large increases in the number of Hda2 (
Figure 1) is projected to see an average rate of increase of 1.10 Hda2 per year. While the ability of populations to acclimatize to new climatic regimes is not well-understood, it may be assumed that a higher rate of temperature increase can lead to a higher risk of negative health impacts. This analysis highlights that while both the East African highlands and areas in southern Africa are projected to see large increases in Hda2, the average rate of increase is projected to be higher in the highlands, and thus this area may be a “hot spot” with an increased risk of negative health impacts from heat.
Figure 7b highlights the average rate of increase in Hda3 over the time period studied. For this threshold, many areas surrounding the East African Highlands, as well as parts of Angola and Nigeria and Cameroon are projected to experience the highest rates of increase, with smaller average increases projected for southern Africa (in yellow) and northern Africa (in light blue). In
Figure 7c, much of western equatorial Africa is projected to experience the largest average rate of increase in Hda4. The rate of increase in this area according to
Figure 7a,b was low because in the current climate there already are many Hda2 and Hda3, and thus there is little potential for increase and thus there is a small increase in the projected number of days. Conversely, the rate of increase in, for example, the highlands of East Africa in Hda4 (
Figure 7c) is low because there are very few or no days in this threshold in the current climate, and none projected during the timescale analyzed in this study. Thus, it is critical to analyze
Figure 7 together with the
Figure 1 and
Figure 3 in order to interpret the results of the average rate of increase.
Figure 7.
The average rate of increase of the 11-year moving average of (A) Hda2; (B) Hda3; (C) Hda4 for the median ensemble member for 1966–2095.
Figure 7.
The average rate of increase of the 11-year moving average of (A) Hda2; (B) Hda3; (C) Hda4 for the median ensemble member for 1966–2095.
3.3. Symptom Bands
The assessment of AT thresholds is helpful for understanding the trends in increasing temperature across Africa with respect to potential temperature-related health impacts. In order to better understand in which AT range the largest increases are seen, the number of days within each symptom band was analyzed.
Figure 8 shows the ensemble average of the average number of days per year within each Symptom Band (see
Table 1) for 1961–1990 (present day climate (left)) and the change (increasing or decreasing) in average days per year for 2070–2100 (right) (all time slices are shown in
Figures S8–S10). The ensemble average of the average number of days per year in the 30 year period is shown in
Figure 8.
In the projections of number of days within each Symptoms Band in
Figure 8, there are areas of Africa that will see decreases in days across all Symptom Bands. The decreases in days in the Symptom Bands are not because there are fewer “hot days”, but rather that days are moving up to more severe Symptom Bands.
Figure 8.
CCAM model derived (A) average number of Symptom Band I days per year in present climate (1961–1990); (B) change in average number of Symptom Band I days per year in 2071–2100 compared to 1961–1990; (C) average number of Symptom Band II days per year in present climate (1961–1990); (D) change in average number of Symptom Band II days per year in 2071–2100 compared to 1961–1990; (E) average number of Symptom Band III days per year in present climate (1961–1990); (F) change in average number of Symptom Band III days per year in 2071–2100 compared to 1961–1990.
Figure 8.
CCAM model derived (A) average number of Symptom Band I days per year in present climate (1961–1990); (B) change in average number of Symptom Band I days per year in 2071–2100 compared to 1961–1990; (C) average number of Symptom Band II days per year in present climate (1961–1990); (D) change in average number of Symptom Band II days per year in 2071–2100 compared to 1961–1990; (E) average number of Symptom Band III days per year in present climate (1961–1990); (F) change in average number of Symptom Band III days per year in 2071–2100 compared to 1961–1990.
For Symptom Band I (
Figure 8a,b), the majority of Africa is projected to see decreases in the average number of days per year in this band (blue and green in
Figure 8b); the highlands in East Africa, and parts of northern and southern Africa are the only areas projected to see increases (yellow, orange and red in
Figure 8b). For Symptom Band I, it appears that there is a difference in the areas that have few of these days in the current climate (blue in
Figure 8a) due to having either generally warmer temperatures (
i.e., tropical Africa) or generally cooler temperatures (
i.e., parts in southern Africa and the highlands in East Africa). The tropical African areas are projected to see decreases in the number of days in this Symptom Band, while the latter areas are projected to see increases.
In
Figure 8c, it can be seen that much of the area in equatorial Africa that was in blue in
Figure 8a is now in red; the red in
Figure 8c indicates a large number of days in Symptom Band II in the current climate. Many parts of southern Africa and the highlands of East Africa have few days in Symptom Band II in the current climate.
Figure 8d shows that while large parts of southern Africa are projected to see increases in days in Symptom Band II (yellow, orange and red), the majority of northern and equatorial Africa is projected to see decreases (blue and green).
In the current climate, only a small spatial extent of the continent has many days in Symptom Band III (
Figure 8e), and the greatest increases are projected in equatorial Africa (
Figure 8f). Only a small part of the Western Sahara is expected to see decreases in Symptom Band III (39 °C ≤ ATmax < 51 °C). This is also a region projected to see increases in Hda5 (ATmax ≥ 51 °C,
Figure 3f) and thus this decrease in Symptom Band III appears to be due to a shifting of the days to a higher temperature and a more severe Symptom Band.