Main

This work identifies regions of high cooling needs using 2,100 simulation runs of global mean surface temperature through the HadAM4 model1,2 across three global warming scenarios: historical (2006–2016), 1.5 °C and 2 °C. Rising extreme heat is already driving an unprecedented surge in cooling demand, with the energy required for cooling by 2050 predicted to be equivalent to the combined electricity capacity of the United States, European Union and Japan in 20163. But how much more cooling would be required if the Paris Agreement’s preferred 1.5 °C limit4 is overshot, and global mean temperature increases to 2.0 °C? The question is crucial, given the growing consensus that there is currently ‘no credible pathway to avoid warming to 1.5 °C’5.

Cooling degree days (CDDs) are a widely used indicator to examine warming and quantify cooling demand. CDDs measure how warm a given location is, by comparing the mean outdoor temperatures recorded each day with a standard temperature (usually 65 °F or 18 °C)3. For example, a day with a mean outdoor temperature of 30 °C has 12 CDDs. In this Article, we map annual CDDs and examine the most affected countries by warming from 1.5 °C to 2.0 °C projections. These are identified by absolute and relative cooling demand increases between these two scenarios. Absolute changes (abs-ΔCDD) show where human exposure to hotter weather will be severe. Relative changes (rel-ΔCDD) indicate large adaptation challenges in regions not traditionally prepared for increasing heat.

Previous work has mainly reported CDDs using historical data6,7. Model-based studies for specific areas of the world have also been reported8,9,10,11. Global model data, however, have only been analysed for specific years, leaving an important gap in predicting and preparing for cooling demand in fast approaching 1.5 °C and 2.0 °C scenarios. To calculate CDDs, we simulate 700 members per scenario using the citizen-science project climateprediction.net (CPDN), obtaining 6-hourly mean temperatures at a spatial resolution of 0.883° × 0.556°.

The findings of this study are summarized in Fig. 1 and Table 1. Figure 1a maps the difference in CDDs between the 1.5 °C and 2.0 °C scenarios, and Table 1a highlights the top ten countries with more than 5 million inhabitants that will experience, and subsequently need to respond to, the largest changes. Extended Data Table 1 includes the top 50 countries with a population of more than 2 million. A more extended list is provided in Supplementary Note 4. To examine variability, we map the standard deviation of results in Supplementary Note 3.

Fig. 1: Global CDD difference between 1.5 °C and 2 °C global warming scenarios.
figure 1

a, Absolute delta cooling degree days (abs-ΔCDD) from 1.5° to 2 °C global warming scenarios. b, Relative delta cooling degree days (rel-ΔCDD) from 1.5 °C to 2 °C global warming scenarios. Delta (Δ) refers to the incremental change in the variable. The absolute and relative delta from 1.5 °C to 2 °C scenarios were calculated using the mean annual CDDs per coordinate across ensemble members per scenario, involving 700 simulations each. Administrative boundaries were used from EuroGeographics.

Full size image
Table 1 Ranking of the top ten countries that will suffer the highest increase (absolute and relative) in area-weighted mean CDDs from 1.5 °C to 2.0 °C
Full size table

The results show that regions surrounding the Equator, particularly the Sub-Saharan region, will experience the largest increase in cooling demand (Fig. 1a). Table 1a shows that ten African countries are the nations with the largest change in CDDs, with important implications for their planning and building climate resilience. These countries align in a west–east band in central Africa. They mainly border Mauritania, Niger and Sudan, identified in ref. 6 to have the highest extreme heat historically. Mali and Chad were also previously reported to have the highest historical CDD6, and here we show that they will also experience a large increment in CDDs from a 1.5 °C to a 2.0 °C scenario. Indeed, the central African population not only had the highest requirements for cooling historically (2009–2018) but would also experience the highest surge in heat exposure and wide-ranging adaptation requirements.

Notably, the results of relative changes in CDDs (Fig. 1b and Table 1b) show that the Global North will experience dramatic relative increases in the number of days that require cooling. Table 1b is the first to rank the top ten most affected countries by their relative increases in CDDs globally. Eight of ten are European nations, which are traditionally unprepared for high temperatures and will require large-scale adaptation to heat resilience.

Globally, Switzerland and the United Kingdom will see the largest relative variation in cooling demand (30%). This is relevant, as current cooling studies for Switzerland and the United Kingdom are, at best, limited. For Switzerland, only two studies in 2006 and 2021 were found12,13, which warned of the accelerating demand for cooling (compared with heating demand). In the case of the United Kingdom, the country with the second-largest relative increase in CDDs, only one 2009 predictive study is found14. The latter aligns with the large relative change of our results (but for different temperature increases), reporting that the energy (and emissions) from air conditioners almost doubles from 2004 to 2030 in London. However, these 2009 study results were not set in the global context we provide. Additional statistics on the relative (and absolute) increase in CDDs in countries with more than 2 million inhabitants are provided in Supplementary Note 5, this time exclusively considering urban areas. This urban area-weighted analysis identifies Ireland, the United Kingdom and Finland as the top three most affected countries—foreshadowing important questions about prioritizing sustainable cooling access and heat resilience strategies in their cities.

Our results enhance and complement the existing literature. A previous study examining predictions of CDDs in Europe9 reports changes in CDDs between Representative Concentration Pathways (RCP4.5 and RCP8.5) in different years from historical (1981–2010) to the period 1981–2100. It models temperature at different years rather than forcing specific global warming scenarios, as in our analysis. While the results are analogous regarding the highest absolute increase in Europe to be in Mediterranean countries, no relative changes are reported. Another study reports European CDDs (that is, Mediterranean) in a 2.0 °C scenario (with spatial resolution >200 km2)15, showing that the further south, the more the absolute change of CDD increases.

In our study, other large regions of high CDD relative increase are found in the mountain ranges of the Andes in South America, crossing the continent from North to South, and the Himalayas in Central Asia, which extend into the Southwest of China. This brings additional insight for sustainability planning as previous CDD predictions16,17 for China under different RCP scenarios did not highlight this region for its relative increase in cooling demand. Further research on changing climate in these regions is needed as no additional studies have been found.

Supported by these results, we argue how immediate and unprecedented climate adaptation interventions are required worldwide to be prepared for a hotter world. An increasing number of stocktake studies4,5 make clear that limiting a surge in global mean temperature to 1.5 °C is increasingly out of reach. We show that moving from a 1.5 °C to 2.0 °C warmer planet would dramatically exacerbate heat exposure and energy demand for cooling. There has already been an increase in global surface temperature of 1.09 °C above pre-industrial levels between 2011 and 20204,18. The total difference in cooling demand from today to a 2.0 °C warmer planet would be greater than our analysis maps, requiring a key focus on an issue that has traditionally been a blind spot for sustainability debates19.

For this study, the differences in CDDs reported are built on the largest ensemble of 700 simulations for each scenario to ensure internal climate variability and at the current highest available temporal resolution of temperatures. The 6-hourly mean temperature predictions result in high granularity of cooling demand variations. The geographical resolution of 0.833° × 0.556° allows examination of the whole planet under one lens while managing the computational intensity of large datasets.

The absolute change in CDDs values shows that African countries will experience the highest increase in cooling demand. These conditions will pose further stress to the continent’s socio-economic development and energy networks, and their implications for equitable access to cooling, issues that require much additional research given the limited studies of this rising threat in the African context20. Further, the results on relative changes indicate that countries that will experience the most drastic increases in CDDs are traditionally prepared for heating, not cooling. These countries will require acute and long-term adaptation to make their populations and the built environment more heat resilient, including broad cooling access through sustainable pathways21. Much can be shared and learnt from countries across the world as they tackle this global challenge.

Overall, CDDs are a valuable indicator of normalized temperature exposure, and are useful to enable a top-down comparison of global warming scenarios between regions. As research grows, additional socio-economic, technical and environmental variables, such as humidity, solar irradiance and wind speed, are needed for more precise cooling demand estimations. It should also be noted that individual thermal comfort expectations differ across communities and countries, depending on conventions, physiology and cultural norms, among others3.

Several important policy implications stem from these results. First, this work clearly indicates that every small increase in global warming will affect heat exposure and cooling demand worldwide, driving the need for immediate, unprecedented and localized adaptation. Second, it is in the national interest of all Global North and South countries to work towards the 1.5 °C target, given that they will be the most affected by the relative and absolute change in CDDs, respectively. Current planning and implementation of energy and climate policies across countries must be designed to be prepared for and build resilience to a hotter local climate. It is important to recognize that the dramatic, and often inequitable, rise in cooling demand can no longer be ignored but rather be addressed through socio-technical levers of change19, which support holistic sustainable solutions.

Methods

Ensembles of 2,100 global climate simulations for mean temperature for three scenarios were generated using the HadAM4 Atmosphere-only General Circulation Model1,2 from the UK Met Office Hadley Centre. The scenarios followed the half-a-degree additional warming prognosis and projected impacts experiment design protocol22, specifically: historical (2006–2016), 1.5 °C and 2 °C above pre-industrial levels. Thus, the model was forced to achieve the increase in temperature for scenarios 1.5 °C and 2.0 °C, regardless of when this occurs. The simulations output 6-hourly mean temperatures at a horizontal resolution of 0.833 longitude and 0.556 latitude, where each scenario involves 70 individual members for a 10 year period (700 runs per scenario), aiming to ensure internal variability. This simulation experiment ran within the CPDN climate simulation environment23. CPDN uses the Berkeley Open Infrastructure for Network Computing24 framework, tasking more than 30,000 globally distributed volunteer members of the public.

Biases in simulated temperature were identified and corrected using a quantile mapping approach. The bias correction was performed in the entire ensemble using reference temperature data from ERA5 for the same timeframe of the historical scenario (2006–2016). Biases are calculated for each percentile in the cumulative distribution function from the historical scenario compared with ERA5 observations. Then, the calculated biases are added to the simulations of the historical, 1.5 °C and 2 °C scenarios to correct the biases of each percentile, assuming that the bias is unchanging between scenarios. This ensures the preservation of the ensemble’s internal variability, and the cumulative distribution of the ensemble aligns with the cumulative distribution of the observations. Further details and validation of the climate model are provided in Supplementary Note 1.

CDDs were used to compare global warming scenarios. CDDs are a widely used indicator to measure temperature exposure and cooling demand through dry bulb temperature. Annual CDDs were calculated for the ensemble members per scenario (700 simulated years) in all coordinates according to equation (1):

$$\begin{array}{rcl}{{\mathrm{CDD}}}&=&\frac{{\mathop{\sum }\limits_{t=0}^{t=m}}\left({T}_{t}-{T}_{{{\mathrm{base}}}}\right)}{n},{T}_{t} > {T}_{{{\mathrm{threshold}}}}\end{array}$$
(1)

where \(t\) is the time step, \(m\) is the last time step of the year, \(n\) is the number of time steps in one day (n = 4, given 6-hourly data), \({T}_{t}\) is the mean outdoor temperature at time t, Tbase is the reference temperature used to calculate the temperature difference, and Tthreshold is the outdoor temperature value above which the temperature differences are calculated. Tthreshold and the baseline temperature, Tbase, was defined as 18 °C, following the most widespread approach in previous studies to enable comparison3. However, this methodology can have several modifications depending on available data, context and application (Supplementary Note 2). It should be noted that since we are evaluating the absolute and relative change between scenarios, the modification of CDD calculation criteria has few implications in the findings.

Then, mean annual CDDs and standard deviation per coordinate across ensemble members (700 simulations) were obtained for the 1.5 °C and 2 °C scenarios, and deltas were computed. Finally, the area-weighted statistics per country were calculated using QGIS geographic information system. Supplementary Note 4 lists the top 100 countries with more than 2 M population. Additionally, Supplementary Note 5 also introduces the top 100 countries by considering only urban area-weighted statistics per country to consider the dimension of urban contexts. This last ranking should be considered carefully since 44% of the population still lives in rural areas25.

This study has the following limitations. CDDs were calculated using the dry bulb temperature following the standard approach, which does not account for the influence of humidity or other environmental variables on perceived thermal comfort. CDDs may also be underestimated in urban areas since the urban heat island effect was ignored.

Supplementary Information provides additional details of the methods and results associated with the climate model (Supplementary Note 1), CDDs (Supplementary Note 2), additional statistical results (Supplementary Note 3) and a more extended ranking of countries according to different criteria (Supplementary Notes 4 and 5).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.