Global forestation and deforestation affect remote climate via adjusted atmosphere and ocean circulation

The study investigates how global-scale changes in forest cover—both forestation and deforestation—modify Earth’s energy balance and consequently adjust large-scale atmospheric and ocean circulation. While prior debates have focused on carbon storage and local thermodynamic effects (albedo, evapotranspiration, roughness), the authors address the less-explored remote dynamical responses, including impacts on the intertropical convergence zone (ITCZ), midlatitude westerlies, and ocean circulation such as the Atlantic meridional overturning circulation (AMOC). Motivated by ongoing tropical deforestation and proposed massive tree planting campaigns, the paper quantifies potential side-effects of large-scale land-cover change on remote climate using idealized, fully coupled atmosphere–ocean simulations with fixed CO2 to isolate biogeophysical mechanisms.

Previous work has shown that biogeophysical impacts of land-use change depend strongly on location and that albedo changes can dominate global mean temperature responses. Studies with slab oceans or prescribed SSTs suggested that extratropical forestation/deforestation can shift the ITCZ and cross-equatorial heat transport, and that tropical deforestation can weaken tropical convection and alter extratropical circulation via Rossby wave sources. LUMIP/CMIP6 experiments indicated substantial near-surface ocean temperature responses to global deforestation and a strong sensitivity of ocean thermohaline circulation to large-scale land-use change. However, the fully coupled atmospheric and oceanic circulation responses under global forestation/deforestation remained unclear, motivating the present CESM2 coupled simulations to resolve remote effects on weather and climate patterns.

Model: Community Earth System Model (CESM) v2.1.2 in fully coupled mode: CAM6 atmosphere (32 vertical levels), CLM5 land, POP2 ocean (60 vertical levels), CICE5 sea ice, MOSART river routing, at ~1° horizontal resolution. Forcing: Preindustrial conditions constant throughout. Experimental design: A preindustrial control (control) uses historical preindustrial land cover (~30% forest). Two idealized experiments branch from the equilibrated control: (1) forest: globally convert non-forest plant functional types (grassland, cropland, shrubs, urban) to forest, retaining bare soil fractions; result ~80% forest cover. Distribution of tree functional types follows preindustrial composition or zonal mean where preindustrial forest is zero. (2) grass: convert the same areas to grassland, reducing forest cover to 0% and replacing with grass. CO2 mixing ratio is fixed across all simulations. Integration length: Control spun up for 200 years; then all three (control, forest, grass) run for 300 years. Analysis period excludes the first 50 years after branching (years 50–300) to avoid initial adjustment; some diagnostics examine early decades and long-term evolution. Statistical testing: Two-sided Wilcoxon rank-sum tests on annual or seasonal means (years 50–300), with Benjamini–Hochberg false discovery rate control (α_fdr = 0.05); only significant differences shown/quoted. Radiative forcing equivalence: Top-of-atmosphere net radiative imbalance (first 5 years and last 5 years) relative to control is translated to equivalent CO2 forcing using ΔF = 5.35 ln(C/C0). Diagnostics: Meridional heat transport partitioned into atmospheric (dry and moist static energy) and ocean components; AMOC computed in height and potential density coordinates with a maximum index between 20–70°N; near-surface variables (2 m temperature, 10 m wind, cloud fractions, precipitation); identification of jet streams via vertically averaged wind speed between 400–100 hPa (a_vel ≥ 30 m s−1) and classification into deep (Δu,rel < 0.4) and shallow (Δu,rel > 0.4) jets; E-vector diagnostics for eddy momentum flux convergence; Hadley cell via meridional mass stream function analyzed for boreal winter (Oct–Mar) and summer (Apr–Sep).

• Global mean temperature response: Forestation warms global 2 m temperature by ~+0.5 °C (most >+1 °C over northern extratropical land and the Sahel; some regions >+2 °C). Deforestation cools by −1.6 °C, with particularly strong high-latitude cooling (<−4 °C) due to snow- and sea-ice–albedo feedbacks. Global mean clear and full-sky albedo decrease by 0.01 in forest; increase by 0.025 and 0.026 in grass, respectively. Heat days (>30 °C) increase by >15 to >30 days yr−1 over many midlatitude/tropical land in forest; decrease by >30 days in parts of tropical Africa due to higher evapotranspiration and roughness. - Radiative forcing equivalence: Initial global mean TOA forcing +0.70 W m−2 (≈ +40 ppm CO2-equivalent) in forest and −0.85 W m−2 (≈ −42 ppm) in grass—about one-third of historical CO2 increase since 1850 (126 ppm). Years 5–10: +0.57 and −0.91 W m−2; end of simulations (years 296–300): +0.41 and −0.35 W m−2, indicating radiative equilibrium not yet reached. - Ocean–atmosphere meridional heat transport: In forest, total poleward heat transport reduces in the Northern Hemisphere (−5% peak) and increases slightly in the Southern Hemisphere (+2%). The NH reduction is dominated by ocean heat transport decreases of ~10–25% between ~10°S and 60°N; atmospheric dry transport decreases but is mostly offset by increased latent transport. In grass, NH total transport increases by ~10% (peak) with ocean heat transport increases of ~20–50% between 10–60°N; SH total transport decreases by ~3%. - AMOC: Forestation weakens and shallows the AMOC; deforestation strengthens and deepens it across ~30°S–60°N. AMOC maximum (20–70°N) changes: −22% (forest) and +49% (grass). Changes implicate altered water mass transformation in the subpolar North Atlantic; a North Atlantic “warming hole” appears in forest akin to GHG warming scenarios. - Sea surface and ocean heat content: Forestation yields a pronounced North Atlantic cooling patch (warming hole) and increased heat content in mid/upper ocean (except warming-hole region), with decreased deep ocean heat content consistent with a shallower, weaker AMOC; grass shows opposite-signed changes. - Near-surface winds and clouds: 10 m winds decrease over most land in forest (increase in grass), with regional exceptions over strongly forested areas (e.g., central North America, Sahel, N India, N Australia). Global mean cloud cover decreases in forest and increases in grass; over northern extratropical land, cloud cover decreases >2.5% (forest) and increases up to 10% (grass), with low clouds contributing strongly. Oceanic cloud changes indicate coupled ocean–atmosphere adjustments (e.g., increased clouds over E North Atlantic in forest; increased clouds over E North Pacific and W North Atlantic in grass). - Precipitation: Global mean precipitation increases by +0.8% (forest) and decreases by −3.0% (grass). Sensitivity ~1.6% K−1 (forest) and ~1.90% K−1 (grass), smaller than CMIP6 CO2-driven responses (2.1–3.1% K−1). Euro-Mediterranean annual mean precipitation decreases by >5% in forest and increases by ≥10% in grass. Tropical rainfall exhibits robust latitudinal shifts: northward in forest; southward in grass, with some seasonal/regional nuances. - Midlatitude circulation and jets: Forestation warms extratropics through the troposphere, weakening and shifting NH westerlies poleward (30–50°N), with poleward migration of deep eddy-driven jets in both hemispheres, reduced E-vector divergence over the western Gulf Stream region indicating weaker eddy momentum flux convergence and weaker westerlies; storm tracks shift poleward, consistent with warming scenarios. Deforestation strengthens NH westerlies (20–60°N), increases eddy kinetic energy and deep jet frequency, shifts SH jets equatorward, and does not shift NH jet latitude, with stronger jets especially over the W North Atlantic in boreal winter and across NH midlatitudes in summer. - Hadley cell and ITCZ: In forest, NH Hadley cell weakens in boreal winter (~2–4%); SH cell modestly intensifies in boreal summer. In grass, boreal winter NH cell strengthens and broadens southward (10–90% increase) while SH cell narrows northward, shifting the ITCZ south (confirmed by energy-flux-equator). In boreal summer, the SH branch weakens due to reduced inter-hemispheric temperature contrast. These Hadley changes are consistent with the observed tropical precipitation shifts. - Process chain: Initial land warming (forestation) increases sensible heat over NH land and warms near-surface air, reducing ocean–atmosphere heat fluxes over adjacent oceans (weaker winds and reduced air–sea temperature contrasts), decreasing atmospheric energy input over oceans and reducing atmospheric meridional heat transport. Over decades, AMOC slowdown drives a strong ocean heat transport response and the North Atlantic cooling, eventually returning atmospheric heat transport toward baseline.

The simulations show that idealized global forestation and deforestation substantially adjust large-scale circulation, thereby modulating remote climate beyond local thermodynamic effects. Forestation warms the globe and particularly NH extratropical land, weakens and shifts NH midlatitude westerlies poleward, and slows and shallows the AMOC, producing a North Atlantic warming hole and poleward-shifted storm tracks. Deforestation induces broadly opposite but stronger responses due to amplified snow–ice–albedo feedbacks: global cooling, stronger NH westerlies, equatorward SH jet shifts, a strengthened and deepened AMOC, and a southward-shifted ITCZ with compensating Hadley cell changes. These circulation adjustments explain widespread changes in precipitation, clouds, winds, and temperature patterns, including Euro-Mediterranean drying under forestation and tropical rainfall latitudinal shifts. The results underscore that large-scale land-cover interventions can trigger ocean circulation responses, especially in the Atlantic, which in turn feedback on atmospheric circulation. Consequently, planning of forestation strategies must account for remote impacts mediated by atmosphere–ocean coupling, not solely local carbon and energy-balance effects.

The study demonstrates that large, idealized changes in global forest cover can substantially reorganize atmospheric and ocean circulation: forestation yields global warming, weaker and poleward-shifted NH midlatitude circulation, and a slowed AMOC, while deforestation produces opposite, stronger responses and a southward-shifted ITCZ with seasonally varying Hadley cell adjustments. These dynamics lead to significant remote impacts on precipitation, cloudiness, and winds across ocean basins and continents. Given proposals for massive tree planting, such biogeophysical and dynamical consequences must be integrated into policy and project design. Future research should assess these mechanisms with more realistic, regional and global forestation scenarios, transient future climates, inclusion of biogeochemical feedbacks, multi-model ensembles to quantify inter-model spread, and refined process studies of AMOC sensitivity (e.g., deep-water formation regions, air–sea fluxes, mesoscale storms, and gyre dynamics).

• Idealized scenarios (near-complete global forestation or deforestation) are not realistic deployment pathways but used to isolate mechanisms; extrapolation to partial or regional forestation requires caution. - CO2 is held fixed; biogeochemical feedbacks from carbon cycle changes are not included, so total climate impacts (thermodynamic plus biogeochemical) are not represented. - Despite 300-year integrations and excluding the first 50 years, full equilibrium is not reached (residual radiative imbalances), and deep-ocean adjustments continue. - Results are model- and parameterization-dependent; prior multi-model studies show substantial inter-model spread in temperature responses to land-use change, implying potential variability in circulation responses. - Spatial resolution and specific land-surface parameterizations may influence regional circulation, cloud, and air–sea flux responses. - Statistical significance is assessed on annual/seasonal means; some diagnostics (e.g., E-vector divergence) exhibit patchiness and limited significance in places.