Current inequality and future potential of US urban tree cover for reducing heat-related health impacts

The study addresses how urban trees can mitigate heat-related health risks and energy use, and how these benefits are distributed across racial/ethnic groups in the United States. Against a backdrop of rising summer temperatures, more frequent and intense heat waves, and thousands of annual heat-related deaths, the authors investigate the current inequality in urban tree canopy cover and associated cooling benefits. Prior evidence shows lower-income and people-of-color (POC) neighborhoods typically have less tree cover and higher summer temperatures. The paper quantifies current disparities in protective benefits from trees, estimates the additional potential benefits of increased tree canopy under realistic planting constraints, and evaluates economic value and return-on-investment (ROI), with attention to equity and policy relevance for large-scale nature-based solutions and climate adaptation funding.

The literature indicates significant and growing heat-related mortality and morbidity exacerbated by climate change, though projections vary by region and scenario. Trees reduce urban heat through shading and evapotranspiration, with largely local effects. Numerous studies document unequal urban tree cover distribution, with POC and low-income neighborhoods experiencing less canopy and greater heat exposure. Heat increases electricity demand, and vegetation (including trees) can reduce building cooling loads; complementary strategies include cool roofs and energy efficiency. Urban trees also sequester carbon, forming part of natural climate solutions (NCS). Prior national estimates of urban reforestation potential (e.g., Fargione et al., Cook-Patton et al.) differ due to assumptions about plantable land and data resolution, with 30 m datasets potentially underestimating current canopy. Past research often focused on single or multiple cities; this study scales to thousands of municipalities using high-resolution canopy data and models heat-health and energy impacts while incorporating realistic constraints to planting.

Design: Four-phase analysis across 5723 municipalities within 100 US urbanized areas (>500 km²) housing 180 million people (2020). - Unit of analysis: US Census blocks (finest demographic unit). Population and race/ethnicity from 2020 Census (NHGIS). Population-weighted statistics used for aggregations. Data sources: - Tree canopy: 2 m NAIP imagery classified (forest/non-forest) using random forests in Google Earth Engine; validated against Urban Tree Canopy assessments (block-level R=0.97; median absolute error ~6%). - Impervious surface: NLCD 2019 (30 m continuous % impervious). - Land Surface Temperature (LST): Landsat 8 C2 LST, 100 m, mean JJA composites (2016–2020), quality-screened. - Air temperature: Global Historical Climatology Network summer (JJA) mean at 423 stations. Reforestation scenarios: - Constraints: (1) Physical: only non-impervious area considered plantable at block level; (2) Social/political/climatic: maximum feasible canopy set to the 90th percentile of observed tree cover for blocks within each urbanized area and impervious surface category (0–20%, 20–40%, 40–60%, 60–80%, 80–100%). - Scenarios: Incremental increases of canopy by 5% steps (5–40%) and an Ambitious Scenario (max possible per above constraints). Where nominal increases exceed block-level maximum, canopy capped at maximum. - Placement assumption: Additional canopy randomly overlays pre-existing non-tree covers proportionally to their pre-planting shares (e.g., fraction over impervious vs sparse vegetation). - Trees required: Calculated from average adult canopy area of 19.6 m² per stem; scenarios reflect adult canopy spacing and benefits. Temperature modeling (two-step regression): 1) LST model: Linear regressions per climate group (mesic vs arid) relate block-level LST to tree cover fraction and impervious surface fraction, including fixed effects for urbanized area. Sparse sample of 10,000 blocks used to reduce spatial autocorrelation. R²: mesic 0.76; arid 0.85. Example coefficients: mesic tree cover estimate −2.88 (SE 0.16); arid −8.06 (SE 0.55). Impervious cover positive in both. 2) Air temperature model: Linear regression of air temperature vs LST with biome-specific slopes and urbanized area fixed effects (N=423 stations). R²=0.49; slopes vary by biome (e.g., 0.25–0.46 °C air per 1 °C LST). - Propagation of uncertainty: Regression and parameter uncertainties propagated through estimates. Health impact estimation: - Mortality: Based on regional heat-mortality response curves (Bobb et al.) relating air temperature to mortality; applied to changes in modeled air temperature per block and aggregated. - Morbidity: Based on Gronlund et al. heat-related hospitalizations; scaled to ED and outpatient visits and associated productivity losses. Electricity and GHG impacts: - Electricity: Derived from Santamouris et al. (US studies) relating % increase in electricity use per 1 °C air temperature increase (2.9–8.5%); applied to summer quarter only and to utility-level residential consumption; avoided GHG from avoided kWh using EPA average grid carbon intensity. Carbon sequestration: - Net annual sequestration per m² canopy from Nowak et al. (0.226 kg C m⁻² yr⁻¹), accounting for mortality; note that emissions from planting/maintenance not included due to lack of national estimates. Economic valuation: - Avoided mortality: Value of a statistical life (age-weighted VSL), $5.7M (2015$, range $3.3–8.2M), converted to 2022$. - Morbidity: Cost-of-illness for hospitalizations, ED, outpatient visits and productivity. - Electricity: EIA utility-level usage and costs. - Carbon: EPA Social Cost of Carbon (SCC) $190/tCO₂ (2020$, 2% discount), applied to increased sequestration and avoided electricity emissions. Equity and ROI analyses: - Blocks classified as majority non-Hispanic white vs majority POC. Protective rate: deaths avoided per million people. ROI defined as avoided annual mortality per tree planted; high-ROI blocks designated as top 5%. Additional notes: - Analyses emphasize population-weighted block-level changes; regional fixed effects and biome/aridity stratifications capture spatial variability. All results reflect propagated statistical uncertainty.

Current inequality and protective benefits: - Majority white neighborhoods vs majority POC in 5723 municipalities: tree canopy 35% vs 24%; impervious surface 42% vs 56%. - Population-weighted median air temperature reduction due to existing trees: 1.01 ± 0.03 °C (white) vs 0.82 ± 0.03 °C (POC); differential cooling ~0.19 ± 0.05 °C. - Annual avoided impacts due to current trees (totals across study areas): • Mortality: 632 ± 100 (white) vs 442 ± 97 (POC); difference 190 ± 139 more deaths avoided in white neighborhoods. • Morbidity: 114,936 ± 7444 (white) vs 84,805 ± 7271 (POC); difference 30,131 ± 10,406 more avoided cases in white neighborhoods. • Electricity: 6.2 ± 0.3 TWh (white) vs 4.8 ± 0.3 TWh (POC); difference 1.4 ± 0.5 TWh. - Protective rate (deaths avoided per million population) is generally lower in majority POC neighborhoods, with the greatest inequality in the US Northeast Corridor. Ambitious reforestation potential (feasible maximum under constraints): - Additional cooling: population-weighted mean summer air temperature reduction 0.38 ± 0.014 °C (up to 1.8 °C locally). - Trees required: ~1.2 billion additional trees. - Additional annual avoided impacts: mortality 464 ± 83 people; morbidity 80,785 ± 6110 cases; electricity 4.3 ± 0.2 TWh. - Climate benefits: increased net sequestration 23.7 ± 0.2 MtCO₂e yr⁻¹; avoided electricity-related emissions 2.1 ± 0.1 MtCO₂e yr⁻¹. - Economic value: total annual benefits ~$9.62 ± 0.45 billion (largest shares from carbon sequestration and avoided mortality). 5% canopy increase scenario yields ~$2.54 ± 0.12 billion annually. ROI and equity: - ROI (avoided mortality per tree) is higher on average in majority POC neighborhoods for all planting ambition levels due to higher densities and lower baseline canopy. - Targeting high-ROI blocks (top 5%) can yield benefits that meet or exceed planting and maintenance costs (e.g., 5% target in high-ROI POC blocks: $32 ± 4M benefits vs $29 ± 7M costs; ROI ~1.12). Spatial patterns: - Greatest additional health protection per capita from planting occurs in dense urban cores (often majority POC); greatest planting capacity and carbon gains occur in lower-density suburban/exurban areas (often majority white). - Among urbanized areas, large, dense, naturally forested regions (e.g., Northeast, parts of West) show the largest increases in protective rate under ambitious planting.

The study demonstrates that urban trees already deliver substantial heat-risk reduction but that these benefits are inequitably distributed, with majority POC neighborhoods receiving less cooling and lower protective rates due to lower canopy and higher imperviousness. By modeling realistic, constraint-aware reforestation scenarios, the authors show significant potential to further reduce heat-related mortality, morbidity, and electricity use, while providing meaningful climate mitigation through carbon sequestration and avoided grid emissions. The findings directly address the research questions by quantifying current disparities, estimating feasible planting outcomes, and evaluating where and how planting yields the largest protective benefits per tree. Equity emerges as central: prioritizing dense, low-canopy neighborhoods—often majority POC—maximizes health ROI and helps redress disparities. While total costs can exceed monetized benefits when applied uniformly, targeted high-ROI strategies align benefits and costs more closely, supporting effective, equitable deployment of nature-based adaptation aligned with funding mechanisms such as the Inflation Reduction Act. Comparisons with European analyses highlight how differing city densities and target definitions affect estimated benefits, reinforcing the value of locally constrained, block-level modeling for US contexts.

This work provides a high-resolution, national assessment of current inequities in urban tree canopy and their consequences for heat-related health and energy impacts, and it quantifies the additional, feasible benefits of reforestation under realistic constraints. Key contributions include: (1) documenting that majority POC neighborhoods receive less cooling and protection from existing trees; (2) estimating an ambitious, yet plausible, reforestation need (~1.2 billion trees) and associated national benefits (avoiding ~464 additional deaths annually, reducing morbidity and electricity use, and delivering ~23.7 MtCO₂e yr⁻¹ of sequestration); (3) demonstrating that ROI for mortality reduction is higher in POC neighborhoods, and that targeted high-ROI investments can make benefits comparable to costs. Future research should: refine plantable area definitions using finer local data and planning constraints; incorporate lifecycle GHG emissions from planting/maintenance; improve city-specific model parameterization (e.g., building stock, microclimate, AC prevalence); evaluate additional co-benefits (air quality, mental health, stormwater); and test implementation strategies that combine tree planting with complementary heat mitigation measures (e.g., cool roofs, depaving) for maximum equitable impact.

- Data resolution and temporal mismatches: 2 m canopy, 30 m impervious, 100 m LST, and differing census boundary vintages (2010 urbanized area boundaries, 2016 canopy, 2020 demographics). Block-level aggregation mitigates but does not eliminate errors. - Plantable area and target assumptions: Physical constraint implemented at block level; maximum feasible canopy set at the 90th percentile within impervious categories—ambitious but not a true maximum. Does not fully account for local policies, land-use conflicts, or potential infrastructure changes (e.g., depaving, green roofs). - Temperature modeling: Two-step regression with moderate explanatory power for air temperature (R²=0.49); parameters allowed to vary regionally/biome-wise but not within cities, potentially under- or over-estimating local effects. - Health and energy models: Use regional/national relationships that may not capture building-level or neighborhood-specific variability (e.g., AC access, building characteristics). - Carbon accounting: Net sequestration excludes GHG emissions from planting and maintenance due to lack of representative national data; lifecycle emissions could reduce net benefits. - Costs: Planting/maintenance costs based on a sample of cities and municipal operations; costs may vary locally and may differ for non-municipal or volunteer programs. - Equity correlations: Race/ethnicity correlates with imperviousness and aridity, which influence modeled temperature impacts; while analyzed explicitly, causal attribution is limited by observational design.