Vegetation-based climate mitigation in a warmer and greener World

Earth system models project an increase in leaf area index (LAI) over much of the planet during the 21st century. Satellite observations over recent decades confirm widespread greening driven by elevated CO₂, nitrogen deposition, climate change, and land cover change. Vegetation changes alter land–atmosphere exchanges of water and energy, producing biophysical effects on surface temperature that can either amplify or counteract the climate benefits from carbon sequestration. The sign and magnitude of these biophysical effects depend strongly on background climate: greening tends to cool tropical and temperate regions via enhanced evapotranspiration but can warm snow-covered boreal regions by reducing albedo. Future warming, declining snow cover, and changing soil moisture and radiation will modulate these processes, potentially enhancing non-radiative cooling or reducing it via CO₂-driven stomatal closure and increased water-use efficiency. Robust observational constraints on climate-dependent biophysical sensitivities are lacking, and vegetation models remain uncertain. This study aims to quantify, in a robust and observation-driven way, how both biophysical and biochemical (carbon sequestration) mitigation potentials of vegetation evolve under future climate scenarios, considering concurrent changes in LAI and background climate.

Prior work indicates that increasing LAI has driven evaporation-driven cooling, especially in water-limited tropical regions, whereas in boreal areas increased LAI reduces albedo over snow, producing warming. The net effect in boreal regions remains debated, with discrepancies between observations and models. Studies agree that background climate modulates the relative importance of radiative (albedo) versus non-radiative (turbulent flux) processes. Projected declines in snow cover and soil moisture are expected to substantially influence land–atmosphere interactions. Elevated CO₂ increases water-use efficiency through partial stomatal closure, potentially damping the evaporative cooling benefits of greening. Large uncertainties persist in dynamic vegetation models regarding these processes and their future evolution.

The study combines Earth observations (2003–2014) with Earth system model projections to estimate how LAI changes affect surface air temperature (T) via biophysical pathways under future climates. The approach involves: 1) Estimating local biophysical temperature sensitivity to LAI (dT/dLAI) from observations. Air temperature is derived from MODIS day/night land surface temperature fused with in-situ data. LAI is from the GLASS product at 0.05° resolution. For each grid cell and month, temperature change between two years (ΔT) is decomposed into the component attributable to LAI change (ΔT_LAI) and a residual (ΔT_res) representing large-scale climate variability. ΔT_res is estimated from nearby (within 50 km) reference grid cells with stable LAI (<0.1 m²/m² change), using inverse-distance weighting. This is repeated for all 66 year pairs in 2003–2014. 2) Deriving climate-dependent sensitivities. Data are split into snow-covered (≥1% monthly snow cover) and snow-free conditions. For snow-covered cases (radiative dominance), dT/dLAI is modeled as a bivariate quadratic function of monthly snow cover fraction (SC, %) and downwelling shortwave radiation (SWdown, W/m²) using MODIS snow and ERA5 radiation, yielding an empirical equation (Eq. 3 in the paper). For snow-free cases (non-radiative dominance), the sensitivity is modeled as a bivariate quadratic function of evaporation (E, mm/day, from GLEAM) and SWdown, using dT/d(ln(LAI)) to account for lower sensitivity at high LAI (Eq. 4). 3) Validating sensitivities against CMIP6. The observation-derived sensitivity functions are applied to CMIP6-simulated historical drivers (LAI, SWdown, evaporation, snow cover; 18 models) for 2003–2014 to compare spatial patterns and zonal means of dT/dLAI with observations (r=0.78, RMSE=0.13), supporting use for future projections. 4) Projecting future biophysical impacts. Observation-based sensitivities are applied to CMIP6 future drivers under four SSPs (SSP126, SSP245, SSP370, SSP585) from 2015–2100 on a common 2×2° grid, yielding temperature changes induced by projected LAI trends and evolving climate drivers. Attribution separates: LAI effect (future LAI with baseline climate sensitivities), All effects (future LAI and future climate-modulated sensitivities), and Climate change effect (difference All effects − LAI effect). Contributions from individual drivers (solar radiation, evaporation, snow cover) are estimated by counterfactual runs holding each driver constant in turn. 5) Biochemical mitigation estimation. Using the linear relationship between cumulative atmospheric carbon and land temperature response (2.2±0.5 °C per 1 Tt C), the study estimates biochemical temperature mitigation from CMIP6-simulated increases in vegetation carbon stock (ΔB) via ΔT = 2.2 ΔB. 6) Ancillary details: Climate zones are from the Köppen–Geiger map aggregated to Equatorial, Arid, Temperate, Boreal (Polar excluded). Multiple LAI products were assessed; GLASS was selected for robustness and gap-free coverage, with sensitivity tests using Copernicus and GLOBMAP yielding similar results. All CMIP6 model fields were bilinearly interpolated to 2×2°. Seasonal analyses align hemispheres by shifting the southern hemisphere by six months to match the solar cycle.

- By 2100 under the high-emission SSP585 scenario, greening is projected to mitigate global land warming by 0.71 ± 0.40 °C in total. Of this, approximately 83% (0.59 ± 0.41 °C) is attributable to biochemical effects from increased vegetation carbon sequestration, while 0.12 ± 0.05 °C arises from biophysical land–atmosphere interactions. - The relative contribution of vegetation-based mitigation to overall land warming is larger under stringent mitigation scenarios: about 35 ± 20% under very stringent scenarios (e.g., SSP126) versus 11 ± 6% under SSP585. - Biophysical mitigation increases in absolute magnitude with warmer scenarios (higher CO₂ concentrations) but declines in relative terms compared to overall warming. - Roughly half of the biophysical cooling stems from the greening itself (LAI increases), and half from concurrent changes in background climate that amplify cooling. Two mechanisms drive this amplification: (i) reduced snow cover lowers radiative warming from winter/spring greening in Boreal, Temperate, and cold Arid zones, and (ii) enhanced non-radiative cooling via stronger coupling between LAI and evaporation under increased temperature and vapor pressure deficit. At the global scale, future changes in evaporation and snow cover contribute about equally to enhanced biophysical cooling. - Spatial and seasonal patterns: Largest biophysical cooling occurs over drylands (e.g., African savanna) due to high dT/dLAI under high radiation and low evaporation; limited changes over humid tropical forests due to low sensitivity in well-watered canopies. In SSP585, boreal summer (JJA) cooling reaches up to 0.55 ± 0.1 °C at northern high latitudes, partially offset by slight winter warming from radiative effects of winter greening. - Sensitivity trends: dT/dLAI decreases notably in boreal zones due to declining snow cover; regions with increasing evaporation often show increased sensitivity (stronger cooling response). - Model–observation consistency: Observation-derived sensitivities applied to CMIP6 historical drivers reproduce observed spatial and zonal patterns (r = 0.78, RMSE = 0.13). - Overall, biochemical mitigation related to greening is about five times larger than biophysical mitigation, but biophysical effects peak during the warmest months and in arid regions, where societal vulnerability to heat is high.

The study demonstrates that vegetation can mitigate future land warming through both biophysical and biochemical pathways, with absolute mitigation increasing under warmer, higher-CO₂ scenarios but representing a smaller fraction of total warming under those scenarios. By constraining temperature sensitivities to LAI with Earth observations and embedding their dependence on key climate drivers (snow cover, radiation, evaporation), the analysis addresses the uncertainty surrounding the evolution of vegetation biophysical effects under climate change. It shows that background climate change itself enhances the biophysical cooling potential, via both reduced radiative warming in snow-affected regions and stronger evaporative coupling as atmospheric demand rises. The results underscore the heterogeneous nature of biophysical mitigation: strongest in drylands and during warm seasons, limited in humid tropics, and seasonally compensating in boreal regions (summer cooling vs winter warming). These findings highlight the relevance of vegetation-based climate actions (afforestation, reforestation, restoration) as components of mitigation portfolios, particularly for achieving stringent climate targets where their relative contribution is higher.

By integrating observation-derived climate-dependent sensitivities with CMIP6 projections, the study quantifies future vegetation-driven climate mitigation. It finds that greening will mitigate land warming by about 0.71 °C under SSP585 by 2100, predominantly through biochemical carbon uptake, with a smaller but societally relevant biophysical cooling that is amplified by future climate changes (reduced snow cover and enhanced evaporative demand). Biophysical mitigation is strongest in arid regions and during the warm season. While absolute mitigation grows with warming, relative importance is greater under stringent mitigation pathways. These insights support the inclusion of vegetation-based solutions in climate strategies and call for enhanced modeling and observation to reduce uncertainties. Future research should better capture non-local atmospheric teleconnections, species shifts under climate change, the magnitude and persistence of CO₂ fertilization and water-use efficiency effects, and improve satellite-based temperature and LAI retrievals in high-latitude, low-light conditions.

- Non-local biophysical effects (teleconnections via large-scale circulation affecting clouds, precipitation, radiation, snow cover, evaporation) are likely underrepresented; the short observational baseline (2003–2014) with modest LAI changes may not trigger or capture such effects. - Passive satellite observations at high latitudes in winter are uncertain due to low daylight and high solar zenith angles. - Grid cells with high variability in dT/dLAI were excluded, yet substantial spread remains at high latitudes. - CMIP6 models do not simulate natural plant species shifts under climate change, introducing uncertainty in projected LAI and biophysical responses. - Future LAI increases may be overstated in ESMs; if so, projected mitigation (especially medians) may be overestimated. - The air temperature product (satellite-LST fused with stations) carries uncertainties, particularly over elevated terrain, potentially affecting sensitivity estimates. - The separation of driver contributions (solar radiation, evaporation, snow) is approximate; real-world interactions complicate precise attribution. - The sensitivity computations assume a uniform 1 m²/m² LAI change across months, whereas in some regions (e.g., boreal) greening is seasonal, potentially biasing annual signals.