A Global Investigation of Outdoor Climatic Comfort

Abstract

In the era of climate crisis, the search for places that offer natural climatic comfort has become a crucial element in understanding the interaction between the environment and the overall quality of human life. Although indoor artificial climate control can provide comfort, it has significant environmental impacts and fosters a more artificial human experience. This study explores how climatic comfort varies worldwide, with a particular focus on outdoor environments where natural atmospheric factors directly influence human perception of comfort. We conducted a global survey, integrated with spatial climate databases, to model outdoor climatic comfort based on temperature, humidity, and natural lighting. The most comfortable locations were identified in tropical/equatorial regions at relatively high elevations. We discuss the results of the current global population distribution, along with past, present, and future demographic scenarios, thereby revealing a critical situation for countries in the Sahel and Middle Eastern deserts.

1. Introduction

The increasing trend of global warming, along with the consequent changes in climate patterns such as precipitation and wind [1], has heightened attention on the effects of climatic comfort on subjective well-being [2]. In this context, climatic comfort is a state in which individuals feel no need to alter their sensory interaction with the environment [3,4]. Climatic neutrality is an ideal state that enhances task performance [3,4,5], encompassing physiological, psychological, and environmental dimensions in both indoor and outdoor settings.

Thermal comfort is defined as a state of mental satisfaction with the thermal environment [6], and numerous studies have examined how temperature, relative humidity, and radiation affect thermal sensation [7]. However, in this paper, we address a broader concept of climatic comfort, one that includes not only thermal comfort but also the ways in which the perception of humidity and natural lighting directly affect human comfort. This is achieved by considering the feeling of humidity (or dryness) on the eyes, skin, and respiratory tract [8,9], as well as the effects of insufficient or excessive natural lighting on visual perception and on endorphin and vitamin D synthesis [10].

According to Wirz-Justice [10], humans exhibit neurobiological responses to seasonal changes in light and darkness, which affect circadian rhythms and cutaneous vitamin D synthesis. During winter, insufficient sunlight can contribute to vitamin D deficiency, which is associated with seasonal depression [11]. Sun exposure, as noted by Fell [12], may increase endorphin production. Molin et al. [13] and Rosenthal et al. [14] describe seasonal affective disorder, a mood condition marked by recurrent winter depression. Conversely, excessive light can cause eyestrain, fatigue, and headaches and may induce circadian disruption [15]. The adverse effects of both insufficient and excessive natural light can be partially mitigated through adaptation strategies—such as artificial lighting, shadowing, vitamin or hormonal supplementation, and the use of sunglasses—even though these measures may also produce environmental impacts when scaled to large populations.

Climatic comfort indices, initially designed for indoor environments, have since been expanded for application in urban planning at larger scales and outdoor settings [16,17,18]. Developing comfort indices based on microclimatic measurements presents significant challenges when applied at broader scales, such as the urban, regional, or global level. Some initiatives have interpolated weather station data to urban scales [19], but the recent availability of more comprehensive global climate databases has created new mapping opportunities. Although previous initiatives, such as those of Matzarakis and Amelung [20], were constrained by data availability at the time, the increased accessibility of detailed datasets now enables more refined analyses on a global scale. Di Napoli et al. [21] mapped the Universal Thermal Climate Index (UTCI) globally [22] using ERA5 reanalysis data, thereby identifying levels of physiological stress derived from laboratory experiments.

However, UTCI indicates physiological heat and cold stress based on laboratory-derived models but does not necessarily reflect the degree of subjective personal comfort. In addition, UTCI relies on standard assumptions regarding clothing, metabolism, and human activity and fails to consider cultural adaptations, acclimatization, and local subjective expectations of climate conditions in each location [23]. The Global Outdoor Comfort Index (GOCI) [24] was proposed to address these limitations, but its calibration currently covers only 29 cities and has yet to produce continuous global maps. To our knowledge, no studies have explicitly mapped global climatic comfort in relation to humidity and natural lighting.

Although indoor climatic comfort indices within environmental planning focus on maintaining pleasant levels of heat, humidity, and light, outdoor comfort assessment must account for constantly changing weather conditions. Consequently, planning studies have increasingly emphasized adaptation strategies to address climate variability [25]. For example, outdoor settings should offer shelter from solar radiation on hot days—preferably well-ventilated shelter—while also providing sun-exposed, wind-sheltered areas for colder days [26,27,28]. By the same logic, plans could incorporate a diverse mosaic of humid environments near water features alongside drier, elevated areas to accommodate periods of unpleasantly humid weather [29].

1.1. Climate, Well-Being, Mobility, and the Environment

The concept of climate migrants arises in a context where climate change and demographic pressures intensify the search for more comfortable environments. Projections of more frequent extreme climatic events, such as heat waves, storms, and droughts, are expected to further affect levels of climatic comfort [1]. Moreover, by 2050, around 200 million migrants could be displaced from regions rendered less viable by climate change [30], posing serious public health risks, particularly for economically disadvantaged communities. Migration choices reflect several factors, including safety and economic opportunities; however, changes in climatic comfort may also affect well-being before and after relocation. Furthermore, despite severe climate changes, wealthier individuals usually have more migration options, while poor populations may remain trapped [31]. Additional systems of privilege, including race, nationality, education, and health, can further influence individuals’ agency in residential and travel choices [32].

In land-use planning, increasing national and international mobility presents significant opportunities and challenges. Besides the direct impact of outdoor climatic comfort on well-being, the outdoor climate also acts as a transitional boundary layer influencing how people adapt to indoor climatic comfort. Settling in areas with higher outdoor climatic comfort can reduce the ecological footprint associated with air conditioning (heating, cooling, humidification, and dehumidification) and artificial lighting, whereas migration to less comfortable areas could increase this environmental impact. This issue is particularly relevant because air conditioning is the leading source of greenhouse gas emissions in urban areas of developed countries and the second largest source (after transportation) in cities of developing countries, with a short-term trend toward becoming the primary one [33].

Post-retirement migration to areas with pleasant climates is increasingly common, reflecting a pursuit of improved quality of life [34]. Remote work and international outsourcing strategies continue to expand, providing greater flexibility in residential locations and workplace choices [35]. These shifts influence human resource management in multinational firms [36], and studies highlight the impact of climatic comfort on productivity and well-being [37,38].

Due to space–time compression, global benefits increasingly hinge on socioeconomic status rather than geographic location [39]. Geospatial big data addresses challenges related to storage, management, analysis, and visualization, thereby facilitating regional discussions on migration and land-use planning [40]. This implies a significant linkage between efficient space use and the opportunities provided by geospatial big data for addressing critical issues related to sustainable urban development and environmental impacts.

It should also be acknowledged that sociotechnical and economic development tends to increase access to indoor climate control, thereby alleviating hardship in extreme climates. However, this progress also entails significant socioenvironmental impacts, including increased energy consumption and consequent pollution, greenhouse gas emissions, and depletion of natural resources. Moreover, shifting daily life toward climate-controlled indoor environments may foster a more artificial human experience and limit the enjoyment of outdoor activities in natural environments [41].

1.2. Objectives and Scientific Novelty

The main objective of this study is to map outdoor climatic comfort for human residence at the global scale, considering temperature, humidity, and natural lighting. Additionally, this study aims to assess how demographic dynamics intersect with these comfort patterns and influence human well-being.

The specific objectives of this article are to accomplish the following:

• Integrate outdoor comfort indicators based on relative humidity, solar radiation, and temperature into a unified comfort index;
• Evaluate the spatial global distribution of outdoor climatic comfort;
• Examine future demographic scenarios to identify alignments and misalignments between human dynamics and climatic comfort.

This work broadens prior research on climatic comfort by expanding the concept of comfort beyond thermal sensation, incorporating and integrating direct perceptions of humidity and natural lighting. Within this extended framework, this study’s main novelty is the development of a global climatic comfort database and the modeling of its relationship with spatial climate databases worldwide. The findings may be useful for policymakers, especially for adaptation to climate change at the international and regional levels, and may support urban planners and landscape architects at the local scale.

2. Materials and Methods

Figure 1 presents an overview of this study’s research methodology. In brief, an international network of researchers conducted a global survey regarding climatic comfort, followed by statistical modeling of how climatological variables influence participants’ reported comfort levels. We used the results of the models to construct indices of outdoor climatic comfort, then mapped these indices and compared them with demographic growth scenarios.

2.1. Survey Design and Disclosure

We designed a questionnaire to quantify outdoor climatic comfort, asking participants to report their perceptions of temperature, humidity, and natural lighting in outdoor settings. Respondents provided geographical location, age, and gender, identified the months of highest and lowest comfort (rated on a 7-point scale of comfort for each variable in each evaluated month), and indicated the importance of each climatic variables (scored from 0 to 10 for each variable—temperature, humidity, and natural lighting). The complete form is provided in the Supplementary Materials.

This study did not collect individual climatic comfort parameters, such as clothing, workplace, and metabolic correlates (e.g., activity level, diet, height, and weight), which are usually assessed in intra-urban or intra-building thermal comfort studies based on interviews. Our objective was not to obtain an “instant” on-site assessment (e.g., passersby on a specific street, in a park, or at a workplace) where location could be paired with each respondent’s immediate condition. Instead, the survey elicited climate satisfaction based on respondents’ subjective perception and memory of monthly patterns at their place of residence, which are assumed to incorporate the respondents’ adaptation to local climatic conditions. This methodological choice facilitated global, online dissemination of the survey.

We conducted the survey from 19 December 2021 to 1 February 2023 predominantly online, with supplementary in-person collection in Brazil and Italy, following ethical guidelines of informed and voluntarily consent approved by a research ethics committee. The dissemination strategy targeted participants from various geographic, climatic, and cultural contexts, while balancing gender and age ranges. We used social networks, online directories, and messaging apps to reach a diverse sample of participants. To ensure linguistic coverage, we engaged specific language-speaking groups in various countries, with survey versions in English, Portuguese, Spanish, Thai, French, and Chinese. In total, we made 7959 posts on distinct platforms: 4651 on Facebook, 2761 on Reddit, 210 on V Kontakte (Russia), 128 on LinkedIn, 78 on OK.ru (Russia), 30 on Groups.io, 20 on WhatsApp, 10 on Telegram, 10 on Wènjuàn (China), 10 on WeChat (China), 6 on Google Groups, 6 on Xing (Germany), 2 on Freelists, and 37 via email. Platform selection was based on their regional penetration and user demographics. Despite these efforts, some major platforms (e.g., Twitter/X, Instagram) were excluded due to limited access to group-based posting and brief content lifespans. Detailed dissemination procedures are provided in the Supplementary Materials.

2.2. Comfort Indices

To capture the respondents’ experiences, we developed a survey-based approach denominated mean sensation vote (MSV), which was modified from the theoretical predicted mean vote (PMV) model proposed by Fanger [42], originally designed for indoor environments using a Likert scale and later incorporated into the ASHRAE 55 standard [6]. This change is justified since our study includes a variety of non-steady contexts and multilingual surveys, whereas PMV is a model-based index calibrated for steady-state, mechanically conditioned indoor spaces with known metabolic rates, clothing, air speed, humidity, and radiant temperatures. The assumptions underlying PMV and its distinct thermal labels (cold, cool, warm, and hot) could introduce construct and translation biases, resulting in non-equivalent spacing in respondents’ interpretation when translated into different languages.

To address these issues, we used consistent, systematic modifiers (e.g., “slightly” and “very”) to indicate intensity while maintaining a neutral midpoint (“comfortable”) and retained only the primary bipolar terms (e.g., “Hot” vs. “Cold”). In addition to reducing respondents’ cognitive load, this structure minimizes linguistic imprecision that may result from culturally specific interpretations of terms like “cool” or “warm” by providing a clearer, more parallel semantic gradient that is more robustly translated, as recommended by ISO 10551 [43] and Beaton et al. [44].

We structured our categories to the traditional 7-point sensation continuum to maintain comparability (

Table 1. Comparison of the Likert scales and respective labels for PMV and for MSV.

Likert Scale	PMV-ASHRAE [6,42]	Mean Sensation Vote (MSV)
		Thermal	Humidity	Natural Lighting
1	Cold	Very cold	Very humid	A lot of lack of clarity
2	Cool	Cold	Humid	Lack of clarity
3	Slightly cool	Slightly cold	Slightly humid	Slight lack of clarity
4	Neutral	Comfortable	Comfortable	Comfortable
5	Slightly warm	Slightly hot	Slightly dry	Slight excess of light
6	Warm	Hot	Dry	Moderate excess of light
7	Hot	Very hot	Very dry	Too much light

). We developed analogous humidity scales (from very humid to very dry) and natural lighting scales (from “a lot of lack of clarity” to “too much light”). To record comfort perceptions throughout the year, each participant responded to these questions twice: once for the month they preferred most and once for the month they disliked most. Subsequently, MSV was calculated as the empirical average of participants’ reported sensations, summarizing respondents’ overall sensation. In our statistical models, the scale was treated as ordinal.

We adjusted the original formula, based on the ISO 7730 [45] Predicted Percentage of Dissatisfied (PPD) model, to calculate the dissatisfaction rate (DR) for outdoor environments. Following Krüger et al.’s [46] adaptation of the ISO 7730 PPD model from indoor to outdoor thermal comfort, we increased the minimum dissatisfaction threshold from 5% to 12% while maintaining the same mathematical structure as in ISO 7730 [45] PPD equation. We then calculated the DR using Equation (1), classified MSV categories, and computed the DR for each month. The annual DR represents the averages of all monthly DR values, thus accounting for discomfort potentially caused by contrasting climatic extremes in different seasons. This metric was specifically chosen over a simple “annual average MSV” because it effectively captures and penalizes seasonal variability (e.g., a location with an uncomfortably hot summer and an uncomfortably cold winter), a critical nuance that simple annual averages would fail to represent.

$$DR=100-88x{e}^{-[\left(0.03353\times {MSV}^4\right)+\left(0.2179\times {MSV}^2\right)]}$$

(1)

2.3. Databases

Table 2. Climate global raster databases (1992–2021), with monthly time resolution, from ERA5 portal.

Database	Spatial Resolution *	Source
Air temperature, wind speed, solar radiation	0.1° (~11 km)	Era5-Land [47]
Mean radiant temperature (MRT), Universal Thermal Climate Index (UTCI)	0.25° (~28 km)	Era5 Heat [21]
Relative humidity	0.25° (~28 km)	Era5 Essential climate variables [48]
Cloud cover	0.25° (~28 km)	Era5 Single levels [49]

* Note: Resolution in km refers to the equatorial regions, gradually decreasing toward the poles.

provides a detailed overview of the climatic variables, their spatial resolutions, and respective data sources. We analyzed 30-year average climate data for 1992–2021. Details on the preparation of the raster datasets are provided in the Supplementary Materials.

To assess demographic scenarios, we used data from the Socioeconomic Data and Applications Center, drawing on shared socioeconomic pathways for demographic projections [50,51]. This study utilized SSP2 (Middle of the Road) demographic scenarios. We analyzed population projections for 2020 and 2100 at a 7.5-arc minute resolution (~14 km at the equator).

2.4. Statistical Analysis

We developed the statistical models in R (version. 4.5.2), and the code is provided in the Supplementary Materials. We processed climate raster data monthly for each climate variable and extracted the values to a points layer containing the locations and responses corresponding to each survey respondent.

We developed three distinct models for temperature, humidity, and natural lighting, producing MSV and DR maps and graphs for each variable. Ordinal generalized additive models with random effects from the mgcv package [52] (version. 1.9-4), in R inferred the relationship between reported climatic comfort levels and spatial climate data from ERA5. Additive models facilitate the assessment of nonlinear relationships among variables. Where monotonic relationships were theoretically expected (i.e., that the change in the explanatory variable always corresponds to a change in a specific direction in the explained variable, e.g., the average temperature at the “slightly hot” level is lower than the “very hot” level), we imposed monotonic smooths, constraining the number of knots in the splines to avoid overfitting and ensure model coherence. Each survey response received a weight to correct for gender and age imbalances, which was computed as the sampled proportion divided by the expected global demographic proportion [53].

Table 3. Demographic profile of sampled and target (global) population [53] and respective weights to compensate for model bias.

Age Group	Sex	Answers	Sample Proportion	Global Proportion (Target) [50]	Weight for Bias Compensation
0–29	Female	516	0.20	0.25	1.24
	Male	770	0.30	0.26	0.85
30–44	Female	308	0.12	0.10	0.86
	Male	415	0.16	0.10	0.62
45–60	Female	203	0.08	0.07	0.91
	Male	151	0.06	0.07	1.20
60+	Female	89	0.04	0.07	2.07
	Male	77	0.03	0.06	2.04

presents the sample and the estimated global demographic profile, alongside the weights used to compensate for imbalances in the models.

Unlike interval- or ratio-scale models, ordinal models effectively respect the ordering of categories without assuming equal distances between them, making them appropriate when the ordinal nature of the response variable is crucial to the analysis [54], as in our questionnaire. Because each respondent reported comfort levels for the most and least comfortable months, the models included a random effect per person (ID) and a binary indicator for whether the comfort level refers to the highest- or lowest-comfort month. The formulas of the models are provided in

Table 4. Models of climatic comfort for temperature, humidity, and natural lighting.

Predicted Variable	Mathematical Formula
Mean temperature sensation vote	s(temperature of the month) + s(annual mean temperature) + s(humidity of the month, k = 5) + s(annual mean humidity) + s(radiation of the month) + s(annual mean radiation) + s(wind speed of the month) + binary variable of worst or best month + s(random effect of respondent id)
Mean humidity sensation vote	s(humidity of the month) + s(annual mean humidity, k = 4) + s(temperature of the month) + binary variable of worst or best month + s(random effect of respondent id)
Mean natural lighting sensation vote	s(radiation of the month) + binary variable of worst or best month + s(random effect of respondent id)

Note: s() is the nonlinear function (spline) and k is the highest number of knots allowed in that spline.

. We assessed the goodness of fit of the models using Somers’ delta [55] via the DescTools package [56] (version. 0.99.60), in R, as recommended by Agresti and Tarantola [57] for ordinal models. Somers’ delta ranges from −1 (perfect negative association) to 0 (no association) to 1 (perfect positive association) and is accompanied by 95% confidence intervals [58].

2.5. Maps of Outdoor Climatic Comfort

Each model (temperature, humidity, and natural lighting) predicted the monthly climatic comfort as spatial rasters for MSV and DR. From these monthly spatial layers, we produced maps of the annual mean MSV and DR. In addition, we created a trivariate map to visualize DR jointly for temperature, humidity, and natural lighting. This method condenses the three variables into a single map using three-dimensional techniques [59]. Finally, a unified index map consolidated climatic comfort information based on participants’ weighted average of the importance assigned to temperature, humidity, and natural lighting. Its calculation is detailed in the Supplementary Materials.

The GOCI [24] model is a linear regression designed to predict thermal perception through the ASHRAE 55 [6] 7-point scale, facilitating comparison with the MSV used in the present study. Using ERA5 inputs, we mapped GOCI globally for each month with the empirically calibrated Equation (2) [24]. We then computed Pearson correlation coefficients between the annual DR raster layer pixels from GOCI and the model developed for temperature MSV, excluding Antarctica due to the lack of MRT data in the ERA5 database for that region. Finally, we converted the MSV scale to the DR scale for each month using Equation (2) [24].

GOCI = (0.053 × MRT) + (0.084 × temperature of the month) + (0.006 × humidity of the month) − (0.229 × wind speed of the month) − (0.056 × annual mean temperature) − (0.026 × annual maximum temperature) − (0.042 × annual minimum temperature) + (0.009 × latitude) − 0.908

(2)

We computed monthly Pearson correlation coefficients between UTCI rasters [21] and those from the temperature MSV, excluding Antarctica.

Bivariate maps, following Brown [60] and Prener [61], were used to compare climatic comfort with current population density and future demographic scenarios [48]. We discussed the results in the context of the IPCC’s [1] climate change trends.

3. Results and Discussion

We obtained a total of 2696 responses to the forms, primarily in English (1705), followed by Portuguese (667), Spanish (190), French (73), Thai (34), and Chinese (27). We excluded responses from participants with less than one year of residence in their country (157 responses), duplicate entries (4 responses), and 6 responses from small islands in Tonga due to the coarse spatial resolution of the climate databases. Therefore, a total of 2529 valid responses remained (94%). Of these responses, 36% were from the Americas, 30% from Europe, 17% from Asia, 13% from Africa, and 4% from Oceania, by continent. Responses collected directly online totaled 2379, while 317 responses were obtained through in-person interviews. After confirming that the response patterns from the two collection methods were not statistically different for the same locations, we combined the data from both the online and in-person surveys.

Respondents from 157 countries, 705 states and provinces, and 1235 cities participated in the study. Figure 2 maps the location of the 2529 valid responses on the various climate characteristics, overlaid on Köppen–Geiger [62] climate regions.

The generated databases and spatial analysis code in R are available as Supplementary Materials.

3.1. Models for Temperature, Humidity, and Natural Lighting

The formulas and Somers’ delta values used to assess statistical associations are reported in

Table 5. Sommers’ delta values for each model.

Predicted Variable	Sommers’ Delta
Mean temperature sensation vote	0.69 [95% CI = 0.68–0.71]
Mean humidity sensation vote	0.24 [95% CI = 0.22–0.26]
Mean natural lighting sensation vote	0.52 [95% CI = 0.50–0.54]

. All climatic explanatory variables presented statistically significant relationships with the predicted variables (p < 0.05). The temperature model was the most accurate, while the humidity model was less accurate, indicating greater variability in preferences and/or uncertainty about its influence on comfort. Natural lighting had intermediate precision and was simpler in terms of variables.

For temperature, an alternative model replaced radiation with MRT, but it yielded a lower Somers’ delta (0.68 [95% CI = 0.67–0.70]). Therefore, we retained the model with solar radiation as the predictor. Solar radiation is the main climatological driver in the MRT model in ERA5 [21]. Total cloud cover did not enhance any models’ predictions and was therefore excluded.

3.1.1. Temperature Comfort Model

The conditional plot in Figure 3 shows the classified monthly temperature comfort levels, adjusted for auxiliary covariates. The vertical axis of the conditional plot represents a latent variable without direct empirical interpretation, while the breaks (dashed red lines) indicate threshold cut-points from the cumulative-link ordinal GAM, calibrated by the model to represent the underlying intervals for each class of comfort level, as a data-driven approach. The spacing between the thresholds reflects where adjacent categories of the predicted ordinal variable best separate after covariate adjustment. The results partially align with WHO [64] recommendations of ~18–24 °C for comfortable temperatures (20–24 °C for infants and the elderly). In Figure 3, the slope between −5 °C and 15 °C is less steep, indicating that temperatures below and above this range exert a stronger effect on reported human comfort. A comparison between the temperature cut-points for each comfort level in MSV (this study), UTCI, and Physiological Equivalent Temperature (PET) is provided in the Supplementary Materials.

The conditional plot in Figure 4 (mean annual temperature) shows a pattern opposite to that in Figure 3 (monthly temperature), reflecting how mean annual temperature shapes behavioral adaptation and subjective expectations in regions that are generally hotter or colder year-round. A similar trend is observed in the GOCI model [24], which also accounts for global-scale adaptation to thermal comfort. Consequently, this adaptation effect partially offsets the differences observed in regions with more extreme temperatures. Consistent with Figure 4, many studies in tropical and equatorial countries—although primarily focused on indoor settings—report greater tolerance to higher temperatures. Reported comfort ranges include 24–29 °C in Indonesia [65], 26–28 °C in Lagos (Nigeria) [66], 26–32.45 °C in Hyderabad (India) [67], and 25.9–33.8 °C in Jaipur (India) [68]. By contrast, ASHRAE [6] recommends indoor comfort ranges of 19.44–27.7 °C in the United States. According to Figure 4, adaptation effects on thermal comfort are most pronounced in regions where annual temperatures fall below 5 °C and exceed 24 °C.

The conditional plots for the remaining explanatory variables affecting temperature comfort are provided in the Supplementary Materials. Increases in monthly solar radiation and humidity were directly associated with warmer comfort levels, similar to the effects observed for monthly temperature, whereas annual mean solar radiation exhibited an inverse association—analogous to the pattern seen with annual mean temperature—reflecting adaptation processes. Higher wind speeds partially mitigated perceived warmth, exhibiting an inverse association.

Figure 5 and Figure 6 present maps of the 30-year mean annual temperature for MSV and DR. In Figure 5, the thermal MSV spans from very cold to hot; the warmest temperature class (very hot) is absent because no locations fall into that category on an annual-mean basis. It is important to note that even in regions with seemingly pleasant annual mean temperatures, seasonal extremes may occur during winter and summer.

Temperate zones exhibit more comfortable mean annual temperatures on the MSV scale (Figure 5), whereas equatorial regions display lower DR values (Figure 6). This indicates that, although annual mean temperature may appear comfortable in temperate zones, seasonal variability is high—often resulting in uncomfortably hot summers and uncomfortably cold winters. Accordingly, particularly in temperate regions, environmental planning should account for this weather variability by providing a variety of outdoor microclimatic refuges that enable adaptative behavior among residents, as recommended by Lenzholzer [26] and Zölch et al. [28].

By contrast, equatorial regions exhibit relatively consistent temperatures throughout the year. Considering the DR map (Figure 6), together with the adaptation offset associated with mean annual temperature (Figure 4), adaptation to thermal conditions appears more effective in equatorial regions than in temperate zones or polar regions.

The most comfortable temperatures occur at high elevations in equatorial regions, including the Andes and the high plateaus of Central Africa. These regions are notable as proposed cradles of the human species [69,70], although their climates may have changed since then. Future work could assess whether current patterns align with their respective paleoclimatic conditions.

3.1.2. Humidity Comfort Model

The conditional plot in Figure 7 illustrates the effect of monthly relative humidity on the corresponding comfort level. Peak comfort occurs between 60% and 78% relative humidity (Figure 7), contrasting with WHO [8] and EPA [9] indoor recommendations of 30–50%, which aim to prevent mold growth and air contamination. This suggests that indoor, health-oriented humidity standards may not always align with outdoor climatic comfort. A pragmatic compromise of ~55% may better balance comfort with health considerations. Similarly to the temperature model, the mean annual values (conditional plot shown in the Supplementary Materials) display an inverse pattern relative to the monthly values, reflecting adaptation strategies and subjective expectations in climates that are generally drier or wetter throughout the year.

Figure 8 and Figure 9 present MSV and DR maps related to humidity. The very dry regions of Greenland and Antarctica (Figure 8) appear as the least comfortable (Figure 9), followed by the arid areas of the Tibetan Plateau, Mongolia, the Sahara, and the Atacama Desert (Chile), as well as northeastern Russia and northern Canada. Highly humid equatorial regions—such as the Amazon and Central Africa (Figure 8)—also exhibit relatively lower comfort in Figure 9.

3.1.3. Natural Lighting Comfort Model

The conditional plot in Figure 10 illustrates the effect of the only predictor, monthly downward surface solar radiation (SSR), on natural lighting comfort MSV. It demonstrates how SSR can serve as a physically based proxy for the luminous–radiative environment, which is associated with comfort- and health-related impacts of lighting, while noting that it is an energy measure rather than one of photopic illuminance. Notably, “natural lighting MSV,” as designed in the survey, also reflects the perceived sufficiency or excess of sunlight exposure over a month rather than instantaneous glare, which aligns with the monthly resolution of SSR. The pattern aligns with Kent et al. [71], who reported that monthly radiation below ~10,000 kJ/m² is associated with reduced vitamin D synthesis, which affects serotonin levels and cognitive function, particularly in patients with depression, thereby triggering seasonal psychological disorders. Subsequent studies using the same dataset corroborated this association [72] and highlighted the negative impacts of vitamin D deficiency on elevated blood pressure and cholesterol levels [73,74], thereby increasing cardiovascular risks. Conversely, the model indicates that monthly solar radiation above ~18,000 kJ/m² becomes increasingly excessive for outdoor climatic comfort.

Other studies have reported significant associations between seasonal and geographic variation in solar radiation—and their respective effects on the synthesis of vitamin D, serotonin, melatonin, and cortisol—and outcomes such as suicide [75,76], depression [13,77,78,79], sleep quality [80,81,82], arthritis [83], cancer risk [84], mood of social media posts [85], and internet searches related to mental disorders [86]. However, the heterogeneity in measurement units, geographic scope, and statistical methods still makes it challenging to directly compare the results of these studies with our indicator of climatic comfort derived from global climate databases for solar radiation.

Figure 11 and Figure 12 present MSV and DR maps of natural lighting comfort. Deserts—such as the Sahara (Africa), Atacama (Chile), and parts of the Middle East—exhibit greater discomfort due to excessive natural light. DR gradually increases toward the poles, as natural lighting falls below comfortable levels. Similarly to the MSV and DR maps for temperature comfort (Figure 5 and Figure 6), temperate regions exhibit comfortable annual mean values for natural lighting comfort but experience high summer and low winter extremes, whereas equatorial areas achieve better DR due to stable—albeit slightly excessive—natural lighting throughout the year.

3.2. Integrated Maps of Climatic Comfort

The trivariate map (Figure 13) visualizes DRs for temperature (red), humidity (green), and natural lighting (blue) using additive color mixing. Lighter areas indicate more comfortable climatic conditions, whereas darker areas indicate less favorable conditions. When temperature (red) and humidity (green) show higher comfort, the map displays yellow; a combination of comfortable humidity (green) and natural lighting (blue) produces cyan. The combination of the most comfortable natural lighting (blue) and temperature (red) yields magenta, while the most comfortable combination of all three variables results in white.

The most comfortable areas primarily correspond to high-elevation equatorial regions, where the climate remains relatively more stable throughout the year. By contrast, darker areas—such as Greenland, Antarctica, northeastern Russia, and northern Canada—are uncomfortable across all considered climatic dimensions. Magenta areas, particularly in equatorial and tropical regions, exhibit a favorable combination of natural lighting and temperature, but not humidity. Green regions, for example, much of Europe, the United States, Australia, and parts of Africa, exhibit comfortable humidity but less comfortable natural lighting and temperature. Yellow regions, such as Mexico and East Africa, have acceptable temperatures and humidity but excessive natural lighting.

The map of the unified index of outdoor climatic comfort (Figure 14), expressed as the average DR, integrates data on temperature, humidity, and natural lighting, weighted according to their relative importance. Mean importance scores were 8.09 for temperature, 7.32 for natural lighting, and 6.88 for humidity. The map depicts higher comfort at high elevations in equatorial regions and lower comfort in desert and cold regions.

3.3. Comparison with GOCI and UTCI Indices

Figure 15 presents the mean annual DR map for GOCI. The Pearson correlation between the GOCI model’s DR and our modeled thermal comfort DR is 0.72. However, the mean annual DR estimated by the GOCI (43.03%) is substantially lower than that of our model (67.16%).

It is worth noting that these values are area-weighted rather than population-weighted: some regions with extreme conditions (e.g., deserts and polar regions) may exhibit high DR but low population. Additionally, Golasi et al. [24] found R² = 0.379 for GOCI, not considering cities with extreme weather patterns, whereas our study achieved Somers’ delta = 0.69 across a broader climatic diversity. Since GOCI was empirically calibrated only for 29 cities, its performance can vary significantly when applied to regions beyond the original calibration set.

Figure 16 reports the Pearson correlation for each month between UTCI and the modeled temperature comfort in this study. The lowest correlation occurred in June (0.295), and the highest in March (0.965), with a 12-month mean of 0.754. The UTCI diverged most from modeled thermal comfort in June, particularly in the Southern Hemisphere during winter.

One hypothesis is that UTCI was primarily calibrated to Northern Hemisphere inhabitants and does not insufficiently account for local adaptation capabilities, leading to different comfort patterns in the Southern Hemisphere under colder conditions. Krüger et al. [23] similarly reported this pattern and argued for the recalibration of UTCI for Southern Hemisphere cities, possibly reflecting prolonged exposure to cold and limited heating infrastructure in many buildings.

3.4. Spatial Relationships Between Climatic Comfort and Demographic Patterns

Figure 17 displays two variables: population density in 2020 and the dissatisfaction rate of the unified outdoor climatic comfort index. Regions with darker purple shades indicate higher population density, whereas darker blues indicate higher dissatisfaction. Overlapping purple and blue areas highlight regions with both high population density and high dissatisfaction. Critical areas include the Sahel (south of the Sahara Desert), as well as some parts of the Middle East and Central Asia (Saudi Arabia, Iran, Pakistan, Afghanistan, northern China, and Kyrgyzstan).

Figure 18 shows projected population change from 2020 to 2100 alongside the percentage of the population experiencing dissatisfaction. It maps population growth trends (2020–2100) and highlights areas with substantial population growth (dark purple), high dissatisfaction (dark blue), and low growth (light gray). These patterns are especially pronounced in the Sahel and parts of the Middle East and Asia, where rapid population growth may become a future concern due to less favorable climatic conditions.

IPCC [1] projects rising global mean temperatures, altered precipitation patterns, and more frequent extreme weather events. These trends have direct implications for vulnerable regions such as the Sahel and the Middle East, where growing population density places pressure on scarce resources, including water and food (see Figure 17 and Figure 18). Climate projections also indicate escalating impacts on natural and human systems, including risks to biodiversity and human health, and heighten the potential for social conflict. Migration driven by climate impacts and other conflicts may also direct large numbers of people to areas that are not necessarily more climatically comfortable, thereby increasing environmental burdens from artificial adaptations, such as indoor climate control (temperature and humidity) and lighting. The correlation between population growth and less favorable climatic conditions underscores the need for robust adaptation strategies, which must be tailored to cultural and economic contexts, as well as the practical constraints faced by policymakers and urban planners.

For a more detailed assessment of climate change impacts on outdoor climatic comfort, future studies could apply the models developed in this study to projected scenarios from global and regional climate models. Nevertheless, IPCC [1] findings already indicate several relevant trends, including the following:

• In temperate regions, the climate is likely to become less comfortable due to more frequent heat waves and hotter summers, partially offset by warmer—and thus less uncomfortable—winters.
• Global warming may render some currently comfortable, densely populated tropical and equatorial areas less hospitable—for example, Indonesia and coastal areas of Brazil and Eastern Africa. However, this assessment should consider the local adaptation capacity, as modeled in this study (Figure 4).
• Extremely cold regions are expected to become more thermally comfortable due to global warming. Nonetheless, the low and declining population density in these regions limits the global significance of this benefit.
• Regarding humidity, densely populated monsoon regions—such as South and Southeast Asia, West Africa, and Southeast Brazil—are likely to experience more pronounced seasonal humidity extremes, becoming uncomfortably wetter during rainy seasons and uncomfortably drier during dry seasons.
• Changes in climatic comfort are expected to be more pronounced in continental interiors, particularly Central Asia, whereas coastal areas—which are typically more densely populated—may benefit from the moderating influence of proximity to the sea.
• Climate projections indicate a northward shift in the Intertropical Convergence Zone (ITCZ), altering the extent of uncomfortably wet climates, alongside a northward expansion of the subtropical high-pressure belts (deserts and semi-arid zones). These shifts will necessitate adaptation strategies for populations experiencing changing climatic conditions.

Evidence suggests that most contemporary migration choices are primarily driven by socioeconomic inequalities, with individuals and groups relocating in search of better employment and livelihood opportunities, as supported by several economic studies on migration [87,88,89]. These dynamics can be further exacerbated by war, social conflict, increasingly frequent extreme climate events, and sea level change, with the latter two being linked to climate change [90]. In this context, several factors, such as political stability, governance capacity, cultural ties, and infrastructure resilience, interact with climate-related migration [91]. From this perspective, climatic comfort is rarely the primary criterion influencing migratory choices and is likely relevant only for specific groups, such as retirees, professionals engaged in remote or international work, or seasonal tourists. Without challenging this view, we emphasize that although climatic comfort may be secondary in migration decisions, its effects on quality of life are often underappreciated. These effects include persistent exposure to uncomfortable climate conditions, increased artificialization of environments in attempts to adapt, and the environmental impacts associated with artificial climate control. Advances in communication, transport, and services may expand opportunities for more comfortable and sustainable human settlement worldwide, even when climatic comfort patterns are misaligned with current demographic dynamics.

3.5. Policy Implications

To guide high-level policy and resource allocation, policymakers can use the comfort indices developed in this study as a basic tool to identify and prioritize regions vulnerable to extreme climate stress. These indices can also be incorporated into public health monitoring systems to protect the most susceptible populations by enabling targeted alerts and efficient allocation of resources during extreme climate events, such as heat waves, cold waves, droughts, or prolonged periods of precipitation.

Concrete, local-scale initiatives can then be linked to these global patterns. Policymakers in “high-risk” areas identified by our global maps may first utilize ground-level monitoring to determine the precise morphological or environmental factors contributing to climatic discomfort in their most vulnerable districts, following multi-scalar approaches [92]. However, most literature on urban climate adaptation primarily addresses thermal comfort, whereas the findings of this study indicate that a monitoring framework integrating a broader range of climate characteristics can provide a more comprehensive understanding of climatic comfort. Once local-scale climatic comfort issues are identified, authorities can implement climate-appropriate interventions. For example, increasing urban tree cover not only provides shade during hot periods but also reduces exposure to excessive sunlight and can improve outdoor humidity through evapotranspiration. Conversely, open areas designed to increase solar radiation exposure can enhance comfort during colder periods, particularly in high-latitude regions where insufficient natural light becomes a limiting factor for comfort. Except in highly humid equatorial rainforest regions, our results suggest that most cities can improve outdoor humidity outdoor comfort by incorporating waterscapes. In regions receiving intense solar radiation, constructing green spaces with integrated photovoltaics (BGIPV) can optimize energy production while simultaneously providing thermally comfortable, shaded, and humidity-moderated outdoor environments [93].

3.6. Limitations, Uncertainties, and Prospects for Future Studies

Whereas prior studies considered humidity and solar radiation primarily in terms of their influence on thermal comfort, this article offers a broader perspective on climatic comfort that incorporates the combined and distinct effects of these variables. However, considering the variables analyzed and the inherent subjective diversity in human–climate interaction, several simplifications and assumptions were necessary.

This research focused on individuals’ perceptions of climatic comfort, which may be partially related to work productivity. However, climate’s effect on productivity depends on task type (manual vs. cognitive), setting (indoors vs. outdoors), task-specific lighting requirements, and corresponding adaptation strategies.

The modeling and mapping results have clear limitations. Our adaptation of the PMV scale based on ASHRAE 55 [6] standard, while justified to enhance consistency across multilingual forms, may affect the correlations among subjective assessments, thereby limiting the comparability with datasets using the unmodified scale [94].

Monthly temporal resolution hides daily and even hourly variability in climate variables. We used 30-year means because this is a standard practice in climatology for characterizing stable patterns and average interannual differences; however, under ongoing global warming, the assumption of climate stationarity is limited [95]. The maps presented depict annual means and therefore do not explicitly show monthly variations or monthly extremes. Another limitation is the uncertainty of spatialized data, particularly in regions with sparse surveying and meteorological monitoring. Regarding humidity, we used relative humidity due to data availability and acknowledge its inverse relation to temperature; other variables, such as absolute humidity (amount of precipitable vapor in the air), may also be relevant for future climatic comfort assessments.

Due to the spatial resolution of the input dataset, the general climate index is better suited for use by international agencies and national or state planners to prioritize regions and seasons for adaptation and to anticipate energy demands and well-being pressures. It is not intended for block-level urban design without appropriate local calibration. Nevertheless, the conditional plots, which illustrate how comfort responds to variations in climate variables, can be applied at the local planning level when combined with detailed, site-specific measurements.

This study expands the analysis of climatic comfort by incorporating humidity and solar radiation alongside additional factors that influence thermal perception. However, despite including these variables, several simplifications and assumptions were necessary. Human responses to the climate exhibit substantial subjective diversity in the psychological, cultural, and physiological dimensions [96], which vary across demographic characteristics such as gender, age, skin pigmentation, height, weight, and clothing practices [97,98,99]. For example, women generally have lower metabolic rates than men, and cultural norms shape socially acceptable clothing choices—both of which can influence adaptability to local climatic conditions [100,101]. A key challenge in mapping climatic comfort—especially at the global scale—is the lack of continuous spatial datasets for these factors at spatial and time resolutions compatible with climate data.

In this context, it would be pertinent to further investigate how gender, age, and duration of residence affect climatic comfort using the dataset compiled in this study. The literature offers testable hypotheses—for example, that women would feel more comfortable under slightly warmer conditions, and that both women and older adults are more sensitive to climate variability [102,103]—but these claims still require validation through global survey data and have not been evaluated regarding responses to humidity and natural lighting. Furthermore, access to means for adapting to climatic discomfort—within and across countries—is also influenced by social privileges related to income, race, education, and gender [104].

The survey is convenience-based and exhibits heterogeneity across regions; however, we partially mitigate this limitation through demographic post-stratification weights and a mixed-effects modeling structure with respondent-level random intercepts. Expanding survey data collection—particularly in underrepresented regions with distinctive climates, such as hot deserts and extremely cold areas—would further enhance the robustness of this study. The limited number of survey languages constrained spatial coverage, reflecting the research team’s language capabilities and origins, and also restricted in-person data collection. Nevertheless, to our knowledge, no prior survey has achieved a comparable spatial distribution across climatic regions, as illustrated in Figure 2, nor has any survey covered the full global range of temperature, humidity, and solar radiation as extensively (Figure 3, Figure 4, Figure 7 and Figure 10).

Another promising avenue for future research is modeling past climatic comfort using paleoclimatic spatial datasets (e.g., PaleoClim [105]) and comparing them with archeological spatial databases (e.g., ROCCEH [106]) to investigate the relationship between climatic comfort and human emergence and migration patterns. Finally, it would be valuable to simulate future climatic comfort under projected climate change scenarios—including adaptation to global warming and urban heat islands—integrated with demographic and migration projections.

4. Conclusions

The index introduced in this study enhances understanding of climatic comfort by extending its assessment beyond thermal sensation to also incorporate humidity and natural lighting. When considered together, the generated maps reveal both the potential and constraints of different regions in terms of overall outdoor comfort. Our analysis highlights a notable mismatch between areas of highest climatic comfort and regions with the greatest population density and growth. Among the spatial patterns observed, the high plateaus of equatorial Central–Western Africa demonstrate particularly favorable climatic conditions, consistent with proposed hypotheses regarding the origins of the human species.

A key finding is that perceived outdoor comfort is strongly influenced by human adaptation. Our results provide empirical support for adaptive thermal comfort theory [107], demonstrating that people in warmer climates tolerate higher temperatures than standard predictive models suggest—a factor that helps explain observed divergences from UTCI. The nonlinear relationships identified indicate that comfort indices based on linear formulations, such as GOCI [24] and PET [108], while convenient and practical, may offer lower precision, especially in regions with extreme climatic conditions.

The findings have several important policy implications. First, the climatic comfort maps help urban planners anticipate future energy demands for lighting and air conditioning, which are major contributors to greenhouse gas emissions in rapidly urbanizing regions with less comfortable climates. Second, identifying areas with high climatic comfort—such as the Andean highlands—can guide long-term strategies for developing new economic centers or remote-work hubs, potentially alleviating population pressures on less comfortable coastal megacities. Finally, these results provide landscape architects and urban planners with actionable insights to design urban microclimates—through parks, squares, and green corridors—that enhance public health, foster restorative outdoor experiences, and mitigate local climatic stress [28].

Our findings highlight a crucial consideration for urban planners designing and managing public spaces: people report preferring higher outdoor humidity levels (60–78%) than those typically recommended for indoor health (30–50%). Moreover, discomfort arising from both excessive and insufficient natural lighting underscores the need for cities to provide a balance of shaded and sunlit areas [26], while also recognizing the neurobiological importance of light for mood regulation and vitamin D synthesis [11,69].

This study provides a quantitative framework to understand the interplay between well-being, human settlement, and climate. It emphasizes that climatic comfort encompasses not only physical conditions but also psychological and social dimensions. In this context, anticipating future demographic shifts, planning sustainable urban development, and mitigating the environmental impacts of adaptation—such as increased energy use for indoor climate control—require knowledge of where how, and to what extent climates are comfortable or not worldwide. We hope these results stimulate reflection and inform policy discussions on migration, sustainable development, space–time compression, and territorial planning at a global scale.