The study, led by Pierre Etienne Banville with Alexander J. MacIsaac and Kirsten Zickfeld, tackles a long-standing blind spot in climate science. Forests influence climate in two distinct ways. Beyond storing carbon, they alter the physical properties of the land surface, changing how much sunlight is reflected, how much water evaporates, and how much heat is transferred to the air. These biogeophysical effects operate locally, at the site of the forest itself, and non-locally, in distant regions connected through atmospheric and oceanic transport. Previous work had established that large-scale forestation tends to produce non-local warming, particularly at high latitudes where dark forest canopies mask bright snow. But most studies focused on land-atmosphere interactions alone, leaving the ocean’s role largely unexplored, especially over the long timescales on which the ocean actually operates.

To isolate that role, the team ran paired multi-century simulations with the University of Victoria Earth System Climate Model, an intermediate-complexity model that couples a dynamic ocean, a thermodynamic-dynamic sea ice model, and a dynamic global vegetation model. The experimental design was elegant in its simplicity. The researchers first spun the model up for ten thousand years under pre-industrial conditions, then deforested the globe by allowing only grasses to grow, and finally let forests regrow on either 50 percent or 25 percent of land grid cells in a checkerboard pattern. This checkerboard arrangement is the key to separating local from non-local effects: over reforested cells, the temperature difference from the deforested baseline contains both local and non-local signals, while over cells that remained grassland, any difference must be entirely non-local. Crucially, one set of simulations used a fully dynamic ocean, while a second set prescribed fixed sea surface temperatures, effectively switching off ocean circulation and ocean-atmosphere and ocean-sea ice feedbacks.

The results were striking. In both sets of simulations, reforestation cooled the surface locally, most strongly in the tropics where broadleaf trees with large leaf areas dominate, and warmed it non-locally, most strongly at high latitudes. But when the dynamic ocean was included, the non-local warming became far larger in magnitude and covered a much greater geographic area. The mechanism begins over land. Forests darken the surface, increasing absorbed shortwave radiation, and alter the turbulent heat fluxes, raising the sensible heat flux that warms the overlying air. This warmer air is advected away from reforested regions, raising air temperatures globally. Through the Clausius-Clapeyron relationship, warmer air holds more water vapor, and that additional vapor strengthens the greenhouse effect, increasing incoming longwave radiation and warming sea surfaces everywhere.

From there, the ocean takes over and escalates the process. Warmer sea surface temperatures increase evaporation over the ocean, pumping even more water vapor into the atmosphere and reinforcing the greenhouse effect in a classic positive feedback. At high latitudes, the warmer ocean melts sea ice, exposing darker water that absorbs more sunlight, which warms the ocean further and melts more ice. The simulations also revealed a temperature-vegetation feedback: warmer conditions increase leaf area in both reforested and remaining grassland cells, darkening the surface and amplifying the warming. A surface energy balance decomposition showed that in the dynamic ocean runs, the increase in incoming longwave radiation driven by the water vapor feedback accounts for most of the non-local warming, whereas it plays only a modest role when sea surface temperatures are pinned in place.

Not every oceanic response pushes in the same direction. The warming of sea surfaces reduced the density of surface waters and, combined with freshwater input from melting sea ice in the Labrador Sea, weakened the Atlantic Meridional Overturning Circulation, the great conveyor belt that carries heat northward. During the initial phase of forest regrowth, this slowdown dampened high-latitude warming by reducing meridional heat transport, partially counteracting the water vapor and sea ice-albedo feedbacks. But over the following five centuries, the circulation recovered almost to its pre-reforestation strength, and the resumption of poleward heat transport contributed to the continued rise of sea surface temperatures at high latitudes, adding yet another layer to the amplification.

Perhaps the most consequential finding is temporal. By the time vegetation had regrown and stabilized, roughly five hundred years after reforestation began, the local effects had essentially settled. In the prescribed sea surface temperature runs, the non-local effects had stabilized too. In the dynamic ocean runs, however, the non-local warming kept strengthening for another five hundred years. The culprit is the ocean’s enormous thermal inertia: it takes centuries for the ocean to equilibrate with the atmosphere, as heat is slowly absorbed, mixed downward, and later resurfaced by currents. This sustained rise in sea surface temperature feeds the amplifying feedbacks, committing the planet to additional non-local warming long after the forests themselves have reached maturity. The authors draw a direct analogy to the committed warming from past carbon dioxide increases, which is likewise driven by oceanic processes.

The researchers stress that these results are not an artifact of an extreme, planet-scale experiment. A sensitivity analysis with reforestation covering 25 percent of land grid cells, closer to the extent of historical deforestation, reproduced the same geographic and temporal patterns at roughly half the magnitude, with ocean dynamics driving a similar proportion of the non-local warming. The team also acknowledges limitations. The University of Victoria model uses a simplified single-layer atmosphere that cannot represent cloud feedbacks or atmospheric circulation changes, and its land surface scheme produces a decrease in evapotranspiration with reforestation that runs counter to most other models. Still, comparisons with the CMIP6 model ensemble under the Land Use Model Intercomparison Project suggest that including clouds would likely weaken tropical non-local warming while strengthening it at high latitudes, leaving the ocean’s amplifying role fundamentally intact.

The policy implications are uncomfortable but important. Net-zero frameworks typically treat a tonne of carbon stored in trees as equivalent to a tonne of carbon not emitted from fossil fuels, yet the full Earth system response to forestation unfolds over timescales far longer than most policy horizons. Large-scale forestation projects that appear beneficial when judged on carbon sequestration and local biogeophysical effects may carry a hidden, growing debit in the form of committed non-local warming that intensifies for centuries, particularly in the Arctic. The authors argue that decision-making frameworks must consider the complete Earth system response over sufficiently long timeframes, and they suggest that geological net zero, in which emitted carbon is permanently stored in geological reservoirs, would sidestep these biogeophysical complications entirely. Forests still deliver enormous benefits for biodiversity, ecosystem services, and carbon storage, but this study makes clear that counting on them to offset fossil fuel emissions requires accounting for an ocean that remembers, amplifies, and answers back on its own slow schedule.

Subject of Research: Ocean-mediated non-local biogeophysical warming effects of large-scale reforestation over multi-century timescales

Article Title: Ocean dynamics amplify remote warming effects of reforestation

Article References: Ocean dynamics amplify remote warming effects of reforestation. (n.d.). https://doi.org/10.5194/esd-17-1315-2026

Image Credits: AI Generated

DOI: 10.5194/esd-17-1315-2026

Keywords: reforestation, ocean dynamics, climate modeling, biogeophysical effects, sea surface temperature, AMOC, water vapor feedback, sea ice-albedo feedback, net-zero policy, carbon sequestration, Earth System Dynamics, thermal inertia

Surface albedo, the fraction of incoming solar radiation that a surface reflects back toward the sky, is a recognized driver of climate change in the physical science literature. Snow-covered boreal forests, dark conifer canopies, bright croplands, and grey urban rooftops each interact with sunlight in profoundly different ways. When land cover changes, so does the amount of solar energy absorbed at the surface, producing a radiative forcing that can either warm or cool the planet independently of any carbon flux. The Intergovernmental Panel on Climate Change acknowledges these biogeophysical effects, and studies stretching back to Robert Betts’ landmark 2000 paper in Nature have shown that afforestation in snowy high latitudes can offset, or even reverse, the carbon benefits of new trees because forests are darker than snow. Yet despite this well-established physics, albedo impacts are rarely quantified alongside greenhouse gas emissions in the assessment frameworks that actually inform decisions.

The barrier, the authors argue, has been data. To calculate the albedo change caused by converting one land cover type to another, an analyst needs albedo values for both the original and the new land cover at the same location, under the same atmospheric conditions, and at the same time of year. Existing simplified methods typically rely on single, static albedo values per land cover type, values that do not capture how albedo fluctuates through the day as the sun’s angle changes, nor how it varies with real atmospheric conditions such as cloud cover and aerosol loading. Albedo is not a fixed property of a surface: it depends on solar zenith angle, on whether the sky is clear or cloudy, on snow cover, and on the structure of vegetation canopies. Averaging all of that complexity into one number per land cover type introduces systematic error, and until now there has been no dataset designed specifically to support the direct calculation of albedo change between land cover types.

Loog and her colleagues set out to close that gap with a methodology that estimates daily-mean blue-sky albedo values for multiple land cover types at the same location. The term blue-sky is significant: it refers to albedo under realistic, mixed atmospheric conditions rather than the idealized clear-sky or entirely overcast scenarios that many remote sensing products assume. The approach draws on MODIS satellite data, accessed through Google Earth Engine, including the MODIS BRDF and albedo model parameters and land cover products, combined with ERA5 reanalysis data from the Copernicus Climate Data Store to characterize atmospheric conditions. By anchoring the estimates to real, spatially specific observations and weighting them toward realistic skies, the methodology produces albedo values that reflect what the atmosphere actually encounters, not a laboratory idealization.

The result is a global climatological dataset of surface albedo values for 22 distinct land cover types, each resolved in space and time. A crucial validation step was statistical: the researchers tested whether the albedo values assigned to each land cover type were genuinely distinguishable from one another, and found that all 22 types were statistically distinct. That matters because the entire premise of land-cover-based albedo accounting rests on the assumption that different land covers have meaningfully different reflectivities. If the distributions overlapped heavily, converting between two types would produce noise rather than signal. The statistical separation confirms that the dataset carries real, physically meaningful information about how land cover shapes the surface energy balance.

The comparison with existing simplified methods is where the study delivers its most striking numbers. When the new dataset was set against values derived from conventional approaches that ignore diurnal fluctuations and realistic atmospheric conditions, the simplified values underestimated surface albedo by 2.7 to 5.8 percent on average. That may sound modest, but the distribution of errors is wide: in some cases the discrepancies reached more than plus or minus 50 percent. For an analyst calculating the climate impact of converting cropland to forest, or of an urban expansion project, an error of that magnitude in the underlying albedo values could flip the sign of the estimated radiative forcing, turning a projected warming into a projected cooling or vice versa. In other words, the simplified numbers are not just imprecise; in edge cases they can be qualitatively wrong.

Why does the diurnal cycle matter so much? Albedo generally increases as the sun sits low in the sky, because light strikes surfaces at grazing angles and interacts with canopy structure and roughness in ways that enhance reflection. A daily-mean albedo that properly weights this variation, hour by hour, differs systematically from a snapshot taken when the satellite happens to pass overhead, typically in mid-morning when the sun is high and albedo is near its daily minimum. Atmospheric conditions compound the effect, since clouds and aerosols alter both the intensity and the angular distribution of incoming radiation. By integrating these effects, the new methodology produces values that better represent the true daily exchange of energy between the surface and the atmosphere, which is precisely the quantity that radiative forcing calculations need.

The practical implications reach into several influential arenas. In life cycle assessment, the standard tool for evaluating the environmental footprint of products and services, land use impacts have long been dominated by biogeochemical considerations such as carbon stock changes, while biogeophysical effects like albedo have remained optional or absent. Researchers including R. M. Bright and collaborators have proposed metrics for converting albedo-induced radiative forcing into carbon dioxide equivalence, and case studies on biofuels, forestry, pavements, and greenhouse agriculture have demonstrated that albedo can rival carbon in importance for certain land-intensive systems. A reliable, globally consistent albedo dataset removes one of the main technical excuses for leaving these effects out. The same logic applies to national and corporate carbon accounting under frameworks such as the Greenhouse Gas Protocol, whose land sector guidance has been grappling with how to treat biogeophysical impacts, and to the evaluation of nature-based climate solutions, where recent studies have warned that tree planting at northern high latitudes may be far less climate-positive than carbon accounting alone suggests.

The dataset and the tooling around it are openly available. The albedo dataset itself is deposited in a Zenodo repository, all input datasets are public, and the authors have released their scripts on GitHub, including a Python-based tool that lets users retrieve albedo values at a 500-meter spatial scale, matching the resolution of the MODIS land cover products. Mann-Kendall trend test results and proxy comparison outputs are available in a separate repository. This openness lowers the barrier for practitioners in life cycle assessment, carbon accounting, and policy analysis to begin experimenting with albedo-inclusive assessments, and it invites scrutiny and improvement of the methodology by the wider research community.

There remain challenges on the road from dataset to decision. Translating albedo change into a CO2-equivalent metric involves methodological choices about time horizons and the decay of radiative forcing that are still actively debated in the literature. The dataset is climatological, describing typical conditions rather than year-to-year variability, and land cover classifications at 500-meter resolution inevitably simplify the mosaic of real landscapes. Nonetheless, the study provides what has arguably been the missing ingredient: a defensible, spatially and temporally specific, physically realistic set of albedo values that can be dropped directly into impact assessment calculations. As governments and corporations increasingly lean on land-based climate strategies, from reforestation to regenerative agriculture to urban greening, the energy reflected back to space by those landscapes will no longer be an afterthought. With this dataset, the mirror-like half of the land-climate equation finally has numbers that accountants, assessors, and policymakers can work with.

Subject of Research: Surface albedo effects of land cover change for climate impact assessment

Article Title: Enhancing climate impact assessment with surface albedo effects of land cover changes: methodology and dataset

Article References: Loog, K., Bjørn, A., & Margni, M. (2026). Enhancing climate impact assessment with surface albedo effects of land cover changes: methodology and dataset. Journal of Industrial Ecology, 30(4), 1869-1884. https://doi.org/10.1007/s44498-026-00127-8

Image Credits: AI Generated

DOI: 10.1007/s44498-026-00127-8

Keywords: surface albedo, land cover change, climate impact assessment, life cycle assessment, radiative forcing, MODIS, remote sensing, carbon accounting, land use, climate policy, biogeophysical effects, environmental impact assessment