Report

Satellites reveal contrasting responses of regional climate to the widespread greening of Earth

See allHide authors and affiliations

Science  16 Jun 2017:
Vol. 356, Issue 6343, pp. 1180-1184
DOI: 10.1126/science.aal1727

The vegetation-climate loop

Just as terrestrial plant biomass is growing in response to increasing atmospheric CO2, climate change, and other anthropogenic influences, so is climate affected by those variations in vegetation. Forzieri et al. used satellite observations to analyze how changes in leaf area index (LAI), a measure of vegetation density, have influenced the terrestrial energy balance and local climates over the past several decades. An increase in LAI has helped to warm boreal zones through a reduction of surface albedo and to cool arid regions of the southern hemisphere by increasing surface evaporation. Furthermore, more densely vegetated areas displayed a greater capacity to mitigate the impact of rapid climate fluctuations on the surface energy budget.

Science, this issue p. 1180

Abstract

Changes in vegetation cover associated with the observed greening may affect several biophysical processes, whose net effects on climate are unclear. We analyzed remotely sensed dynamics in leaf area index (LAI) and energy fluxes in order to explore the associated variation in local climate. We show that the increasing trend in LAI contributed to the warming of boreal zones through a reduction of surface albedo and to an evaporation-driven cooling in arid regions. The interplay between LAI and surface biophysics is amplified up to five times under extreme warm-dry and cold-wet years. Altogether, these signals reveal that the recent dynamics in global vegetation have had relevant biophysical impacts on the local climates and should be considered in the design of local mitigation and adaptation plans.

A large part of the planet is greening in response to increasing atmospheric CO2, nitrogen deposition, global warming, and land-use change (1). This global and persistent increase in leaf area index (LAI; the amount of leaf area per unit of ground area) is enhancing the land carbon sink, leading to a negative feedback in the land-climate system that ultimately contributes to climate mitigation (2, 3). In addition, the associated variations in physiology, phenology, and structure of vegetation may regulate surface temperatures by affecting the water and energy exchanges between land and atmosphere (4, 5). The influence of plant biophysics on climate is increasingly recognized, given its potential role in enhancing or counteracting the climate benefits of carbon sequestration (69). However, these biophysical processes have not been considered in climate negotiations to date, given their uncertainty in sign and magnitude.

Although the effects of changes in vegetation on local climate have been explored through in situ measurements (such as FluxNet and field experiments) (10), it is unclear to what extent the results of such local-scale analyses can be extrapolated to larger areas. Repeated and consistent satellite observations have recently been used to assess the biophysical impacts of land-cover transitions on the global climate (1115). However, these studies focused primarily on the relatively small fraction of the global land that has experienced recent variation in forest cover (~3% in 2000–2012) (16), therefore overlooking the climate impacts of the widespread greening (1). Ultimately, the net effect of ongoing changes in vegetation density on the climate system is not well established, and assessments based on coupled land-atmosphere models have produced uncertain results (17, 18). The current availability of extensive Earth observations of land and climate variables may reduce this uncertainty and improve the representation of these effects in global circulation models.

We explored the interplay between local climate and temporal changes in LAI, a key structural parameter of vegetation that integrates human-induced and natural processes and largely controls land-climate interactions and feedbacks (3). In particular, we aimed to quantify the biophysical impacts associated with the greening of Earth, a global phenomenon of much larger extension than the direct land-use change. To this end, we analyzed satellite-retrieved dynamics of LAI, surface energy fluxes, and climate drivers at the global scale for the period 1982–2011 (supplementary text S1) (19, 20) by focusing on the interplay between LAI and the terms of the following energy balance equationLWoutRα – LE – H(1)where LWout is the outgoing longwave radiation and is therefore related to surface temperature, Rα is the sum of the downwelling longwave radiation and the absorbed shortwave radiation at the surface, and LE and H refer to the latent and sensible heat fluxes, respectively; the storage flux has been ignored because it is negligible at the annual time scale. We then explored at the pixel level and across climatological gradients the relationships between the relative year-to-year variations (Δ operator, dimensionless) of LAI and the terms of the surface energy budget reported in Eq. 1 (supplementary text S2).

Our results show that temporal variations in LAI are closely related to changes in the surface energy budget and that background climate conditions play a key role in modulating these interactions (Fig. 1). In cold-temperate and boreal regions (mean annual air temperature Ta < 280 K), increases in LAI are associated with a boost in absorbed radiation (Rα) (Fig. 1C, top left corner), mostly because of the reduction of surface albedo (fig. S1C), which is followed by surface warming and a consequent increase in LWout (Fig. 1A). By contrast, in warm regions (Ta > 290 K) increasing LAI is associated with cooling and hence a reduction in LWout (Fig. 1A, top right corner), owing to the enhancement of LE (Fig. 1E) that is particularly evident at moderate and low precipitation levels (P < 1400 mm) (Fig. 1, B and F). As expected, LE and H show opposite patterns because they represent alternative pathways for the release of energy from the land surface (Fig. 1, G and H). The complex relationship between LAI and the surface energy balance highlighted in Fig. 1 is therefore dominated by radiative terms (Rα) in cold climates and by turbulent energy fluxes (H and LE) in warm climates. This dichotomy highlights the need to consider all the terms of the energy budget in order to quantify biophysical land-climate interactions (3).

Fig. 1 Relative interannual variations of LAI and the components of the surface energy balance across climatological gradients.

(A to H) Global relative variations in annual longwave outgoing (ΔLWout), absorbed radiation (ΔRα), latent heat flux (ΔLE), and sensible heat flux (ΔH) are illustrated against the interannual relative variations in leaf area index (ΔLAI, y axis) and the climatological median of air temperature [Ta, x axis, in (A), (C), (E), (G), respectively] and precipitation [P, x axis, in (B), (D), (F), (H), respectively] (supplementary text S2, materials and methods).

We further assessed the interplay between LAI and surface biophysics under extreme climate conditions produced by opposite anomalies in precipitation and temperature (for example, exceptionally warm-dry or cold-wet years). For this purpose, we quantified climatic anomalies through empirical cumulative probability distributions (F) of the concurrent interannual variations in precipitation (ΔP) and air temperature (ΔTa) (supplementary text S2). Results show that large negative relationships between LWout (and thus surface temperature) and LAI in extreme warm-dry and cold-wet years are five times larger in magnitude than those observed under average climate conditions and are closely linked to the corresponding variations in latent and sensible heat fluxes (Fig. 2, A to D, signal during climate anomalies in the top left and bottom right corners to be compared with the average climate conditions shown in the centers). We argue that the observed amplification of land biophysical impacts under climate extremes is due to the coupling between soil moisture, LAI, and surface temperature (20, 21). In fact, high levels of precipitation lead to moist soils that stimulate vegetation greening (fig. S2E, bottom right corner), which in turn promotes high rates of latent fluxes (fig. S2C, bottom right corner) and leads to an enhanced cooling of the surface (fig. S2A, bottom right corner). Conversely, low levels of precipitation induce water stress and a reduction in LAI (fig. S2E, top left corner), which feeds back to climate through the increase in sensible heat fluxes (fig. S2D, top left corner) and, potentially, reduced atmospheric moisture and cloud formation. These processes may ultimately lead to a further increase in incoming solar radiation, resulting in larger heat fluxes and rising temperatures (fig. S2A, top left corner). Remarkable differences between climate zones emerge over the dominant warm/dry–cold/wet gradient (fig. S3, A to D). Larger covariation of climate and LAI, particularly with regard to latent fluxes, is found in arid regions compared with other climate zones because of the stronger coupling of vegetation dynamics with soil moisture (22).

Fig. 2 Climate anomalies and relationship between surface energy balance terms and LAI.

(A to D) Ratios of interannual variations in energy fluxes [(A) ΔLWout, (B) ΔRα, (C) ΔLE, and (D) ΔH] to ΔLAI expressed as a function of the empirical cumulative probabilities of the anomalies in precipitation [FP), x axis] and air temperature [FTa), y axis]. F close to zero indicates extremes with low values of the climate variables, equivalent to cold-dry anomalies, and F close to one indicates extremes with high values, equivalent to warm-wet anomalies.

The complex interplay between vegetation density and surface energy balance highlighted above led us to investigate the global climate impacts of the recent greening. To this end, we combined the observed LAI trends with the sensitivity of biophysical processes to variations in LAI in order to estimate LAI-related decadal changes in the mean value and daily range of surface temperatures. Concerning leaf area, it has been shown that over the past three decades, most global vegetated areas have experienced a widespread greening, with a global median of 0.025 ± (1 × 10−4) m2 m−2 decade−1 (Fig. 3, A to C) and a statistically significant positive trend on 46% of the global vegetated area.

Fig. 3 Long-term trends in LAI and associated changes in surface temperature.

(A) Spatial map of the satellite-based LAI trends (δLAI; 1982–2011). Areas labeled with black dots indicate trends that are statistically significant (Mann-Kendall test; P < 0.05) (supplemenatry text S5, materials and methods). (B) Zonal median of LAI trends at 5° latitudinal resolution and corresponding interquartile range shown as a black line and gray shaded band, respectively. (C) Trends in LAI binned as a function of climatological medians of precipitation (P, x axis) and air temperature (Ta, y axis). (D to F) Same as (A) to (C), but for the sensitivity of mean daily surface temperature to LAI (Embedded Image). Latitudinal profiles of daily, nighttime, and daytime surface temperatures are shown in (E) in black, blue, and red, respectively. (G to I) Same as (D) to (F), but for the trends in daily surface temperature (Embedded Image) related to variations in LAI, as computed with Eq. 2.

Variations in surface temperature (TS) associated with long-term variations in LAI have been computed asEmbedded Image(2)where δLAI is the long-term trend in annual mean LAI and Embedded Image is the sensitivity of TS to LAI. The sensitivity term has been derived as the partial derivative in a multiple regression of surface temperature against LAI, precipitation, and incoming shortwave radiation (Fig. 3, D to F). The derived signal Embedded Image integrates the bidirectional interactions between LAI and TS. All predictors in Eq. 2 are quantified for each pixel over a centered 9 by 9 (2.25°) spatial window (a sensitivity analysis to different levels of spatial aggregation is detailed in supplementary text S3). Details on regression analysis, statistical significance, and uncertainty propagation of the trends are reported in supplementary texts S4, S5, and S6, respectively.

Results show a large spatial variability of temperature sensitivity to changes in LAI (Fig. 3, D to F) that ultimately translates into latitudinal and climatological gradients of LAI-driven variations in TS (Fig. 3, G to I). A localized and statistically significant LAI-related warming up to ~0.4 K decade−1 occurs in the cold and wet climates of boreal regions, with hot spots in northern Canada and central Europe (Fig. 3, G to I). This warming effect results from the combination of a prevalent greening signal (Fig. 3A) and a positive sensitivity of TS to LAI (Fig. 3, D to F), which is connected to the reduction in albedo and the consequent increase in available radiative energy in areas with extended snow cover (Fig. 1C). Such a pattern emerges distinctly over tundra, where the progressive shrubification (23) may have amplified the phenomena [zonal median 0.014 ± (1 × 10−3) K decade−1] (table S1). Browning in northeastern America and Eurasia (Fig. 3A), mostly attributable to forest disturbances (24), has led to an opposite mild cooling that is sufficient to offset the LAI-related warming of the tundra when averaged over the whole boreal domain [–0.001 ± (6 × 10−4) K decade−1] (table S2). The observed patterns are largely consistent with the reported warming effects of northern afforestation and reforestation (11, 14).

In contrast, arid and semi-arid regions of the Southern Hemisphere (South Africa, southeastern America, and Australia) show a LAI-related cooling trend [regional cooling up to ~–0.4 K decade−1 and zonal median –0.036 ± (1 × 10−3) K decade−1] (Fig. 3, D to F, and table S2). This pattern stems from moderate greening rates (Fig. 3A) and the high negative sensitivity of TS to LAI (Fig. 3, D to F), which is associated to the close link between vegetation cover and latent heat fluxes in water-limited regions (Fig. 1, E to H). In arid regions, the interplay between LAI and TS is typically larger during daytime, when the large share of transpiration occurs, whereas in the boreal zone, we observed comparable relations of LAI on TS during daytime and nighttime (table S2).

The impacts of the recent greening on global temperature are limited because of the compensation of opposite local effects across different climate regions (Fig. 3I). On average, we estimate a global biophysical cooling of –0.007 ± (3 × 10−4) K decade−1 related to long-term changes in LAI, which outweighs the recent estimates of climate warming driven by deforestation (15). A decrease in the diurnal variation of surface temperature is also related to greening, as the result of a stronger global daytime cooling as compared with the nighttime variation [–0.015 ± (4 × 10−4) K decade−1 and –0.0002 ± (2 × 10−4) K decade−1, respectively]. When the effect of covarying climate drivers is accounted (by computing in Eq. 2 the sensitivity of TS to LAI as total derivative instead of partial derivative) (supplementary text S4), a spatially consistent doubling of the signals is emerging (fig. S4). This suggests that positive feedbacks in the land-climate system may amplify the biophysical impacts of variations in LAI on surface energy fluxes.

The reconstructed time series of LAI-related changes in land surface temperature (Embedded Image) further elucidate the contrasting regional climate responses in a scenario of global warming and widespread greening (supplementary text S7). In cold and humid regions (Ta < 280 K and P > 800 mm), greening is leading to an amplification of surface warming owing to the dominant effect of radiative processes (variations in albedo) (Fig. 4A). Conversely, in warm regions (Ta > 290 K) the increasing LAI leads to climate mitigation driven by the strength of plant-mediated evaporative cooling in arid environments (Fig. 4B). Altogether, the recent greening has therefore reduced the spatial variability of temperatures across Earth. Ultimately, for ~60% of the global vegetated area, greening has buffered the dominating warming signal, with a local mitigation effect of ~14% (supplementary text S8 and fig. S5). For the remaining areas, mostly located in the boreal zone, LAI trends have amplified the rise in air temperatures, leading to an additional warming of ~10%.

Fig. 4 Biophysical effects of the global greening on recent temperature trends.

(A) Variation in LAI, LAI-related surface temperature (Embedded Image), overall land surface temperature (TS), and air temperature (Ta) expressed with respect to the first observational year (1982) and spatially aggregated over cold-wet regions (annual Ta < 280 K and P > 800 mm). Regression lines are overlaid for each variable, and corresponding coefficients of determination are reported in the label (supplementary text S7, materials and methods). (B) As (A), but for warm regions (annual Ta > 290 K). (C) Relations between the long-term trend in air temperature (δTa, on the x axis) and the LAI-related trend in surface temperature (Embedded Image, on the y axis) spatially aggregated for different biomes. Upward- and downward-pointing triangles indicate positive and negative sensitivity of TS to LAI (Embedded Image), respectively. The size of the triangle refers to absolute value of sensitivity. Spatial domains of biomes are shown in fig. S6.

When aggregated at biome level, the trend in air temperature (δTa) and the contribution of LAI to this trend (Embedded ImageEmbedded Image are positively correlated (R2 = 0.53, P = 0.003) (Fig. 4C). Therefore, biomes that are experiencing larger climate warming are those where LAI effects are contributing to a substantial amplification of the climate signal, and biomes that are experiencing milder climate warming are where LAI effects are contributing phenomenon is of particular concern for cold biomes (such as tundra and boreal forest/taiga), where the rapid greening in combination with the positive sensitivity of surface temperature to LAI is contributing to the accelerated warming of ecosystems that are particularly vulnerable to climate change.

Our analysis reveals that the increase in vegetation density has had substantial climate impacts in recent decades. The sign and magnitude of these impacts largely depend on the background climatic condition and the extent of greening. Our observation-driven assessment provides a comprehensive description of the global relationships between vegetation cover and surface energy fluxes that may serve as a benchmark for global climate models. Future changes in environmental conditions (climate or atmospheric CO2) could somehow alter the dominant mechanisms observed in today’s climate (25, 26). Considering the projected exacerbation of climate variability over most of the globe (27) and our findings on the importance of the LAI-climate feedbacks during extreme climate conditions (Fig. 2), the future trajectories of the vegetation-atmosphere system are likely to coevolve even more closely than at present. Understanding how biophysical feedbacks will develop under future scenarios is key to improving projections of the future Earth’s climate. Overall, our findings emphasize the relevance of biophysical land-climate feedbacks and may help the development of more integrated and effective climate mitigation and adaptation strategies.

Supplementary Materials

www.sciencemag.org/content/356/6343/1180/suppl/DC1

Materials and Methods

Figs. S1 to S11

Tables S1 and S2

References (2852)

References and Notes

  1. Acknowledgments: Data availability and locations may be found in the supplementary materials. The study was funded by the Seventh Framework Program (FP7) LUC4C project (grant 603542). D.G.M. acknowledges support from the European Research Council (ERC) under grant agreement 715254 (DRY–2–DRY). The authors declare no competing financial interests. G.F. and A.C. conceived and designed the study; D.G.M. provided gap-filled net radiation, climate, and evapotranspiration data; G.F. analyzed the data; and G.F. and A.C. interpreted the results and wrote the manuscript, with contributions from R.A. and D.G.M.
View Abstract

Navigate This Article