Urban areas are increasingly affected by climate extremes, and climate change is projected to increase their frequency and intensity. Global urbanization further concentrates populations and expands urban land, leading to a generally accepted notion that urbanization escalates population exposure to climate extremes. However, urban land patterns are human-designed and can be modified. This research investigates how climate change and urbanization, both individually and collectively, affect spatial population exposure to four climate extremes – hot days, cold days, heavy rainfalls, and severe thunderstorm environments (STEs) – across the continental U.S. at the end of the 21st century. The study emphasizes three aspects often overlooked: (i) the effects of urban land change on regional climate projections, particularly given significant projected urban expansion; (ii) continental-scale spatial patterns of population exposure, moving beyond individual city studies; and (iii) population exposure to climate extremes beyond temperature, including extreme precipitation and convective environments.

Existing literature primarily focuses on the amplifying effects of urbanization on climate extremes, particularly concerning temperature increases due to the urban heat island (UHI) effect. Studies have shown that urban lands can enhance climate extremes in areas of high population density. However, there is a gap in understanding how urban land patterns might moderate, rather than amplify, these exposures. While the modifying effects of urban lands on climate extremes are recognized, much of the research concentrates on individual cities, neglecting the continental-scale spatial patterns and the effects of urban land change on regional climate projections. Furthermore, the existing literature predominantly focuses on temperature extremes, overlooking the impact of changes in extreme precipitation and convective environments.

The study analyzes three future scenarios combining different climate, land use, and population conditions at the end of the 21st century (EOC; 2075–2100 summaries) for spatial population exposures to the four climate extremes at a 25-km resolution. These scenarios are: (i) high greenhouse gas induced warming with no population nor urban land change ("climate"); (ii) high greenhouse gas induced warming and high population growth with no urban land change ("climate+pop"); and (iii) high greenhouse gas induced warming, high population growth, and high urban land expansion ("climate+pop+landUse"). A historical scenario (BOC; 1980–2005 summaries) serves as a baseline. High scenarios are used to provide more data points on extreme cases for trend identification and pattern comparison. The effects of climate change, urban land expansion, and population change are captured as ratios between exposure counts (ECs) from different scenarios. These ratios effectively separate the effects of each variable. Population exposures are calculated at a 25-km resolution using products of average frequency maps of each weather extreme and spatial population data. Different aggregation strata (urban development densities, climate regions, and sizable urban centers) were applied to analyze the spatial patterns of exposures. Climate projections use published scenarios from previous work, generated using the Weather Research and Forecasting (WRF) model. Spatial urban land projections were produced using a novel data-driven urban land modeling framework (CLUBS-SELECT). Four weather extremes (hot days, cold days, heavy rainfall, and severe thunderstorm environments) were defined for each 25-km grid. The study uses ratio-based effect measures to isolate the effects of climate, land use, and population changes on population exposure. The robustness of these ratio-based measures is confirmed by analyzing various scenario pairs.

Over the 21st century, the U.S. average temperature increases by 4.4 °C, the national population becomes 2.3 times the BOC level, and the total amount of urban land 4.2 times. These changes lead to significant increases in population exposure to hot days (13.6 times BOC level), heavy rainfalls (7.5 times), and STEs (2.5 times), while cold days increase only slightly (1.1 times). While urban land effects on total national exposure are generally small, continued urbanization strongly affects the experience of climate extremes at the individual level. As urban land expands, exposures in non-urban areas decrease, while exposures in urban areas, especially those with mid-to-high development densities, increase. A significant shift occurs in the location of population exposure, with the majority shifting from non-urban to urban areas. Urbanization increases the average individual's exposure to climate extremes. However, surprisingly, urban land expansion can also decrease average exposures in some cases, particularly for heavy rainfalls in mid-to-high-density developments. This might be due to physical land-atmosphere mechanisms (e.g., UHI effects reducing extreme precipitation) and/or spatial population-urban-land dynamics. Climate effects are more prominent than urban land effects in influencing total population exposure counts, with the highest effects observed for hot days and the lowest for STEs. Population effects are consistently around the national total population change ratio (2.3), highlighting the limitations of current spatial population projection models. Analysis across climate regions reveals major changes in regional patterns for hot days and heavy rainfalls, while STEs and cold days show relatively stable patterns. The Southeast emerges as the most exposed region to three of the four climate extremes. Examination of 109 sizeable urban centers reveals that urban land effects can increase and decrease population exposures, and can even decrease exposures in urban centers where climate effects increase them. Urban land effects are generally smaller than climate effects and show no correlation with urban land total amounts or change amounts, suggesting that urban land patterns (spatial arrangements and land cover types) are key determinants. A loose association exists between urban land effects and climate trends, indicating the potential for urban land patterns to moderate population exposures independently of climate effects. For hot days, seven urban centers showed reduced exposure due to urban land changes in the simulated high urbanization scenario. These findings underscore the potential for urban land patterns to moderate population exposures to climate extremes at both the regional and city levels.

The findings challenge the conventional view that urbanization solely amplifies population exposure to climate extremes. The study demonstrates that urban land patterns can moderate, rather than intensify, the nation's future experiences of climate extremes. This emphasizes the importance of focusing on the spatial design of urban areas rather than simply on the total amount of urban land. Urban land designs cannot eliminate climate extremes, but they can significantly alter how individuals experience them. The study highlights the need for long-term planning in urban development, acknowledging that current urban systems often lack adequate climate considerations. To achieve long-term resilience, it is crucial to identify and implement adaptive urban design pathways, exploring potential long-term effects of alternative socioeconomic choices through large-scale spatiotemporal modeling. The potential to moderate climate extremes through thoughtful urban design presents an opportunity for societal growth and environmental justice.

This research demonstrates the potential for urban land patterns to moderate population exposures to climate extremes. Spatial patterns are more critical than total urban land area. This counters the common belief that urbanization solely amplifies these exposures. Future research should focus on identifying specific urban land patterns and associated mechanisms to effectively reduce exposure, emphasizing long-term planning in urban design and development. Developing better long-term socioeconomic models will further enhance these studies.

The study uses a single climate model, configuration, and forcing combination, limiting the assessment of model uncertainties. Current long-term spatial population models have limitations, necessitating further development in this area before in-depth investigations of spatial population patterns interacting with urban land and climate are feasible. These limitations suggest areas for future research to refine the findings and enhance the robustness of the conclusions.