Population growth and global warming are significantly increasing global energy demand, particularly in the building sector. While heating remains a major energy consumer in many developed countries, cooling demand is rapidly growing, especially in developing nations located in warmer climates. This surge is driven by rising populations, increased ownership of cooling equipment, and the escalating effects of climate change. In IEA member countries, space cooling already accounts for up to 20% of total energy consumption and up to 30% of total electricity use in residential buildings, with this share expected to increase significantly in the coming years. Many large metropolitan areas are situated in developing countries with warm to hot climates, presenting a considerable challenge for future energy supply and its capacity to meet the rising demand. This study focuses on the global trend of cooling energy demand, utilizing Cooling Degree Days (CDDs) as a key indicator to quantify the relationship between temperature and cooling energy needs. CDDs provide a simple but effective measure of the cumulative temperature exceeding a base temperature threshold, reflecting the overall cooling requirement. The study also investigates the temporal clustering of CDDs to assess whether periods of intense cooling demand occur randomly or tend to cluster, thereby having implications for energy storage and peak production capacity planning. Previous research has shown the significance of population weighting in CDDs to better represent energy load and the importance of considering humidity in accurately reflecting human thermal stress and consequent cooling demand. This study aims to provide a comprehensive global assessment of national-level CDD trends and, for the first time, a thorough analysis of their temporal clustering, using a new dataset of energy-related climate indicators generated by the International Energy Agency (IEA) and the Euro-Mediterranean Center on Climate Change (CMCC). This dataset, based on ERA5 reanalysis, provides various climate indicators relevant for analyzing both energy supply and demand at both grid and national levels. The study focuses on the trends in national CDDs and their temporal clustering, particularly highlighting emerging tropical countries highly vulnerable to thermal stress and with limited adaptive capacity.

Existing literature has documented the rising global energy consumption, driven primarily by increased demand across sectors, including buildings. Studies have shown that residential energy use for heating is substantial in IEA member countries but the share of cooling is rapidly increasing. Research has also highlighted the correlation between electricity consumption for cooling and ambient air temperature. Degree days, particularly CDDs, have been employed as metrics for predicting future building energy demand. However, existing studies often lack the inclusion of critical factors such as population weighting and humidity in their CDD calculations. Population weighting is crucial for accurately reflecting energy load distribution, as sparsely populated areas contribute less to the overall energy demand. Similarly, incorporating humidity, alongside temperature, is essential for a more accurate representation of human thermal comfort and consequent cooling needs, particularly in tropical regions. Existing research has also investigated the spatial distribution of HDDs and CDDs, but less attention has been given to temporal clustering of these indices. Understanding temporal patterns is crucial for effective energy supply planning and for optimizing energy storage and peak production capacity. Studies on temporal clustering in climate indices have shown evidence of extreme events clustering in alternating quiet and active periods. While the sensitivity of electricity consumption to temperature depends on various factors, a first-order relationship with outdoor air temperature is widely accepted.

This research leverages a newly developed IEA-CMCC dataset based on ERA5 reanalysis, covering 2000-2021. The dataset includes numerous energy-related climate indicators, available at both grid and national levels. The study employs CDDs and humidity-corrected CDDs (CDDhum), calculated using various base temperatures (primarily 21°C). National-level CDDs are population-weighted to accurately represent energy consumption profiles. The warm season is defined as May-October in the Northern Hemisphere and November-April in the Southern Hemisphere, based on the analysis of different six-month periods' averages. The study employs linear regression to analyze the annual trends of total cooling demand (CDDhum21 per year and CDD21 per year). To examine the temporal clustering of intense CDDs (defined as exceeding the 90th percentile of daily values), a binary time series (1 for intense days, 0 otherwise) is constructed. The coefficient of variation (CV) of inter-event times (time in days between consecutive intense days) is calculated. CV values exceeding 1 indicate temporal clustering. In addition, the study defines clusters as groups of at least two consecutive intense days and analyzes the number of clusters (N_C) and the size of the largest cluster (M_C) to further quantify clustering. To detect significant changes in clustering between the two decades 2000-2009 and 2011-2020, a bootstrapping method is used. Furthermore, Ripley's K-function is employed as another metric for temporal clustering, comparing the observed K-function to a simulated homogeneous Poisson process. To assess the impact of spatial aggregation, the analysis is also performed at the grid point level, with indicators averaged using population weights. This allows for a comparison of results obtained from aggregated versus disaggregated data.

The study reveals a significant increase in cooling degree days (CDDs) globally over the past two decades, indicating a rise in cooling energy demand. This increase is more pronounced when considering humidity-corrected CDDs (CDDhum), highlighting the substantial impact of humidity on cooling requirements, particularly in tropical regions. Analysis of the top 23 countries with the highest cooling demand reveals a strong tendency for intense CDD events to cluster in time. This temporal clustering is more evident when humidity is incorporated into the analysis. The number of intense days (N), the number of clustered events (N_C), and the maximum cluster size (M_C) all significantly increased in the second decade (2011-2020) compared to the first (2000-2009) for a majority of countries, particularly in South America, Southeast Asia, and Africa. Several countries, such as India, Thailand, and Vietnam, experienced a substantial increase in the frequency and duration of intense CDD periods. Countries like Bahrain, Qatar, and Iraq show high variability in CDD values, suggesting a greater tendency for clustering. The analysis using Ripley's K-function also supports the findings of increased temporal clustering. The comparison between aggregated and disaggregated data shows a general agreement in the key indicators (N, N_C, M_C), demonstrating that the pre-aggregated IEA-CMCC dataset is suitable for temporal analysis. However, differences exist in M_C for some countries, potentially related to population distribution and warranting further investigation. The study also analyzes dry CDDs (CDD21), which exhibit similar trends, but the magnitude and clustering are generally less pronounced than those of CDDhum21. The linear regression analysis highlights countries with the most pronounced increase in annual accumulated CDDs (CDDhum21 and CDD21). Bangladesh, in particular, showed a significant rise in CDDhum21 per year, reflecting a substantial increase in the intensity and duration of cooling needs. The geographical pattern of intense CDDs and CDDhum21 shows the highest values for tropical countries in the Northern Hemisphere.

The findings of this study directly address the research question concerning the evolution of country-level energy demand for cooling. The observed increases in CDDs and the increased temporal clustering of intense events strongly support the hypothesis of a significant rise in cooling energy demand globally. The inclusion of humidity in the analysis enhances the accuracy of the results, particularly for tropical and humid regions, where humidity significantly influences human thermal comfort and, thus, cooling needs. The significance of the results lies in their implications for energy planning and policymaking. The increased cooling demand, particularly the clustering of intense periods, poses a considerable challenge for energy providers. The results emphasize the need to account for non-stationarity in climate when planning for peak energy demand. Traditional approaches that assume a stationary climate may underestimate future cooling demand due to neglecting the increased clustering of intense events. The observed trends, particularly in emerging economies, highlight the urgency of developing robust energy strategies to address the future energy demands related to cooling. The results also highlight the importance of considering humidity in the analysis of cooling energy demand and its temporal clustering.

This study provides a comprehensive global assessment of trends in national-level cooling degree days (CDDs) and their temporal clustering, revealing a significant rise in cooling energy demand over the past two decades. The inclusion of humidity proves to be crucial for accurately reflecting cooling demand, especially in tropical regions. The increased temporal clustering of intense CDD events presents a significant challenge for energy planning and requires strategies to address potential future energy shortages. Further research should focus on specific regions with the most pronounced increases in CDD and temporal clustering, particularly those in emerging economies, and should include analyses focusing on other factors, including urban sprawl. Investigating the influence of various factors on the temporal clustering of CDDs would also enhance our understanding of this phenomenon and allow for better energy planning.

This study uses a relatively short time series (2000-2021). While this captures recent trends, it may not fully capture longer-term climate variability or potential future changes. The analysis relies on population-weighted CDDs, which may not fully capture the spatial distribution of cooling energy consumption within countries. Furthermore, the study focuses on cooling energy demand from the building sector, neglecting other energy-intensive cooling applications. Finally, the study's ability to fully capture the complex interaction between climate change, population growth, economic development, and cooling energy demand could be limited due to the scope and data availability.