Earlier springs tip northern lands toward evaporative cooling, satellite records show

An earlier start of the growing season across northern lands is consistently linked to stronger latent heat fluxes, the energy the surface spends on evaporating water, while shifts at the season’s end produce weaker and far more scattered responses. That is the central result of a new analysis published in Biogeosciences, which quantifies how changes in the timing of spring green-up and autumn senescence reshape the surface energy balance across the Northern Hemisphere’s vegetated lands above 30 degrees north.

The study, led by Olivia Hau of the Max Planck Institute for Biogeochemistry in Jena, Germany, together with colleagues at Augsburg University, TUD Dresden University of Technology, the European Commission’s Joint Research Centre and other institutions, draws on Earth observation and reanalysis data covering 2001 to 2021. The team derived growing-season start and end dates from satellite-based leaf area index measurements and then asked how sensitive key surface energy variables, including albedo, latent heat, sensible heat and evaporative fraction, are to year-to-year variation in those dates.

Plant phenology exerts a first-order control on how land and atmosphere exchange water and energy. When foliage appears earlier in spring, canopy cover increases, which can raise canopy conductance and boost transpiration at the expense of sensible heat, while also altering how much sunlight the surface reflects. Increased leaf area also roughens the surface aerodynamically, generating more mechanical turbulence that can enhance both turbulent fluxes. Because these mechanisms operate simultaneously, the net effect of phenological shifts on near-surface climate has been difficult to pin down from observations alone.

To isolate the phenological signal, the researchers regressed spring and autumn means of each energy balance variable against the annual start-of-season and end-of-season dates, after removing long-term trends from the flux data to strip out confounding effects of the global warming trend. Spring and autumn windows were defined as twenty days either side of the mean transition dates. Phenological dates came from the GLASS leaf area index dataset, smoothed with a spline and thresholded at thirty percent of the annual maximum leaf area. The team also ran partial correlation analyses that statistically controlled for surface temperature, and they combined results from several independent data sources, including ERA5 reanalysis, the FLUXCOM up-scaled eddy covariance products and the GLEAM evaporation dataset, to gauge how robust each finding was across datasets.

The clearest signal emerged for latent heat. Across all vegetation types examined, an earlier start of season was associated with enhanced latent heat flux, with a spatial median of 0.47 watts per square metre per day of advance. The response ranged from 0.15 watts per square metre per day in croplands up to 0.68 watts per square metre per day in mixed forests, and the four datasets showed consistent large-scale patterns across most northern lands. The authors attribute this to increased energy supply through greater shortwave absorption and to evidence that expanding vegetation cover in spring enhances transpiration, regulated primarily by photosynthetic activity through stomatal conductance.

Albedo changes ran in the same direction in the regions where they were largest. In evergreen and deciduous needleleaf forests and in tundra, an earlier start of season was associated with albedo decreases of up to 0.003 per day, likely because spring green-up masks residual snow cover and darkens the surface. The researchers translated this into a first-order energy estimate: using the multi-decadal ERA5-Land average downward shortwave radiation of 257 watts per square metre for May and June over northern lands, an albedo decrease of 0.003 to 0.004 per day corresponds to roughly 0.77 to 1.01 watts per square metre per day of additional absorbed shortwave energy. That extra energy helps fuel the enhanced turbulent fluxes. Notably, mid-latitude regions in Europe, Central and Eastern Asia and eastern North America showed the opposite sign, with albedo increasing as the season started earlier, possibly because developing canopies shield darker soils or, in agricultural systems, because earlier sowing brings brighter crops to the surface.

Autumn told a more complicated story. Delays in the end of season were also associated with albedo decreases, though much smaller in magnitude, and with increases in latent heat flux of up to about 0.14 watts per square metre per day in mixed forests. But across other land-cover types the responses were weaker and more variable, and one dataset, GLEAM, even showed regionally negative sensitivities in forested areas. The authors suggest that soil moisture availability may be the limiting factor: an earlier spring start can progressively deplete soil moisture through extended summer transpiration, suppressing autumn evaporation independently of when the season ends. Where water is scarce at the onset of autumn, reduced stomatal conductance would dampen or even reverse the expected positive relationship between a delayed season end and latent heat.

Sensible heat fluxes, by contrast, showed no coherent large-scale pattern and diverged strongly between datasets, particularly in temperate and boreal Eurasia, where FLUXCOM generally indicated more negative or neutral sensitivities than ERA5 and GLEAM. Deciduous needleleaf forests showed sensitivities of about -0.24 watts per square metre per day, while other vegetation types displayed weaker and often mixed responses. In high northern latitudes, however, a pattern of increased sensible heat alongside increased latent heat was emerging, consistent with the larger shortwave absorption from reduced albedo. Evaporative fraction, the ratio of latent to total turbulent heat, largely mirrored the latent heat patterns in spring, with sensitivities reaching about -0.002 per day in deciduous needleleaf forests, indicating a shift of the energy partition toward evaporation as the season began earlier.

The authors are careful about what these observational patterns can and cannot establish. Because surface temperature both influences and responds to phenology through coupled energy-balance processes, observational analyses alone cannot disentangle the feedbacks or provide quantitative causal attribution. Still, the concurrent increases in latent heat and evaporative fraction, together with negative partial correlations of latent heat and evaporative fraction with start of season after controlling for temperature, lead the team to hypothesize that phenological shifts may partially offset greenhouse-gas-induced surface warming through enhanced evaporation, a potential negative feedback on surface warming. This interpretation is consistent with independent modelling evidence showing that earlier season starts enhance evaporative cooling across northern vegetated regions, and with large-scale observational analyses demonstrating that leaf area increases reduce the Bowen ratio through enhanced transpiration.

The study also quantified how robust its estimates were to methodological choices, testing alternative leaf area thresholds and analysis window widths. The reference settings generally fell near the centre of the distributions obtained from the alternatives, and the sign of the response was preserved in almost all cases. Sensible heat proved the most robust variable, while autumn surface temperature was the most sensitive to the choice of analysis setting. Mixed forests showed the greatest sensitivity to methodological choices, whereas croplands, perhaps reflecting their managed nature, produced the most compact and consistent distributions. Importantly, the variability across vegetation types and energy variables substantially exceeded the variability introduced by the tested settings, suggesting the observed responses reflect underlying ecosystem characteristics rather than analytical choices. The authors flag two main limitations: satellite-based phenology reflects canopy greenness rather than its underlying drivers, and the analysis does not explicitly account for interactions between concurrent start and end of season shifts, such as the possibility that an earlier spring increases the likelihood of summer soil moisture depletion that then shapes autumn responses. They call for improving the explicit representation of dynamic vegetation-controlled land surface dynamics in climate reanalysis efforts, such as the successors of ECMWF’s ERA5, and for future work applying comparable sensitivity analyses to land surface models, quantifying legacy effects of phenological shifts, and integrating site-level observations of phenology and energy fluxes to understand stand-scale mechanisms.

Subject of Research: Sensitivity of surface energy balance fluxes to shifts in growing season start and end dates across northern lands

Article Title: Phenology and surface energy balance changes across northern lands

Article References: Hau, O., Forkel, M., Buermann, W., Kranz, J., Migliavacca, M., Weber, U., Panwar, A., & Winkler, A. J. (2026). Phenology and surface energy balance changes across northern lands. Biogeosciences, 23(21), 7705-7725. https://doi.org/10.5194/bg-23-7705-2026

Image Credits: AI Generated

DOI: 10.5194/bg-23-7705-2026

Keywords: plant phenology, surface energy balance, latent heat flux, albedo, evaporative fraction, northern lands, start of season, end of season, satellite remote sensing, evaporative cooling, climate feedback, soil moisture