Abstract

Satellite remote sensing data indicate that greenness has been increasing in the northern high latitudes, apparently in response to the warming of recent decades. To identify feedbacks of this land-cover change to the atmosphere, the authors employed the atmospheric general circulation model ARPEGE-CLIMAT, an adaptation of the Action de Recherche Petite Echelle Grande Echelle model for climate studies, to conduct a set of control and sensitivity modeling experiments. In the sensitivity experiments, they increased the greenness poleward of 60°N by 20% to mimic the manifestation of vegetation changes in the real world, and by 60% and 100% to represent potential aggressive vegetation change scenarios under global warming. In view of the direct exposure of vegetation to sunlight during the warm seasons, the authors focused their study on the results from late spring to early fall. The results revealed significant thermodynamic and hydrological impacts of the increased greenness in northern high latitudes, resulting in a warmer and wetter atmosphere. Surface and lower-tropospheric air temperature showed a marked increase, with a warming of 1°–2°C during much of the year when greenness is increased by 100%. Precipitation and evaporation also showed a notable increase of 10% during the summer. Snow cover decreased throughout the year, with a maximum reduction in the spring and early summer. The above changes are attributable to the following physical mechanisms: 1) increased net surface solar radiation due to a decreased surface albedo and enhanced snow–albedo feedback as a result of increased greenness; 2) intensified vegetative transpiration by the additional plant cover; and 3) reduced atmospheric stability leading to enhanced convective activity. The results imply that increased greenness is a potentially significant contributing factor to the amplified polar effects of global warming.

1. Introduction

Satellite remote sensing data show an increase of greenness in the northern high latitudes since the early 1980s (Myneni et al. 1997; Tucker et al. 2001; Zhou et al. 2001; Stow et al. 2004). Increased greenness is quantified by the satellite-derived normalized difference vegetation index (NDVI), calculated from measurements by the Advanced Very High Resolution Radiometer (AVHRR). From 1982 to 1999, the NDVI shows an overall increase of 10% in the Eurasian sector and 17% in the North American sector poleward of 60°N (Stow et al. 2004). The increased greenness is generally manifested by greater vegetation production, an earlier onset of greening, and a lengthening of the growing season.

This greenness increase, as discussed by Serreze et al. (2000), may be attributable to a number of contributing factors, including an increase of carbon dioxide in the atmosphere, enhancement of nutrients in the warmed soils, and possibly thawed permafrost, resulting from the amplified warming that has been observed in the northern high latitudes (ACIA 2004). Observations indicate that the strongest warming in the northern high latitudes has occurred over land (e.g., Houghton et al. 2001), resulting in a considerable response of the sensitive Arctic ecosystem (ACIA 2004, chapter 9), particularly in increases of shrubbiness (Sturm et al. 2001).

While global warming has significant biospheric consequences, leading to marked changes in greenness in northern high latitudes, these changes may in turn feed back to the climate system by influencing the atmosphere. The impacts of land-cover changes on local and global climate have been widely investigated (e.g., Dickinson and Henderson-Sellers 1988; Bonan 1999; Costa and Foley 2000; Voldoire and Royer 2004). These studies focused on deforestation, indicating that land-cover-induced changes in energy, moisture, and carbon fluxes at the land surface can significantly influence climate. Specifically, the impacts of high-latitude vegetation change on climate were also explored extensively. Bonan et al. (1992) investigated the impacts of the boreal forest vegetation on climate and indicated that the boreal forest can warm both winter and summer temperatures relative to the simulation with the replacement of forest with bare ground or tundra. Thomas and Rowntree (1992) used a parameterization of snow-covered land surface albedo to investigate the impacts of boreal deforestation and found that both temperature and precipitation decreased in the deforested regions. Chalita and Le Treut (1994) found an increase in snow cover when boreal forest is removed. Foley et al. (1994) investigated the effects of the northward extension of boreal forests on the mid-Holocene climate from palaeobotanical data and found that the northward extension of boreal forests can give rise to an additional warming of 4°C in spring and 1°C in the other seasons besides the warming from orbital variations alone. Levis et al. (1999) furthermore investigated the potential feedbacks of high-latitude vegetation on a doubled CO2 climate and suggested the feedbacks must be included in the assessments of anthropogenic climate change.

Recent observational evidence suggests that the land-cover changes exhibited in northern high latitudes are primarily characterized by increased greenness, in conjunction with warmer air and surface temperatures (Zhou et al. 2001; Stow et al. 2004; Walker et al. 2003). If global warming continues as projected by the most recent Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4) climate model simulations, this greenness increase may strengthen further. Previous studies (e.g., Bonan et al. 1992; Thomas and Rowntree 1992; Chalita and Le Treut 1994; Douville and Royer 1997), though most of them focused on the impacts of boreal deforestation, suggest that the Arctic climate is quite sensitive to changes in vegetation. Therefore, an important question naturally arises: How does increased greenness in northern high latitudes feed back to climate?

To answer this question, we employed the atmospheric general circulation model ARPEGE-CLIMAT, an adaptation of the Action de Recherche Petite Echelle Grande Echelle model for climate studies that was originally developed by Météo-France and further modified by the Bjerknes Center for Climate Research (Furevik et al. 2003), to conduct a set of simulations with both realistic and various aggressive degrees of greenness increase in northern high latitudes. Strong seasonality characterizes the land cover in the northern high latitudes. Only during warm seasons is low-lying vegetation exposed and allowed to exhibit explicit interactions with atmosphere. Owing to this, we decided to focus our attention on the thermodynamic and hydrological impacts of increased greenness in the high latitudes during the warm seasons. The model physics, configuration, and experimental design are described in section 2. Sections 3 and 4 describe the results of the experiments. Section 5 contains a discussion and summary of the results.

2. Model description and experimental design

The atmospheric general circulation model ARPEGE-CLIMAT is a spectral model, originally developed for numerical weather prediction by Météo-France and the European Centre for Medium-Range Weather Forecasts (ECMWF). It has been adapted by Deque et al. (1994) for use in climate simulations. Furevik et al. (2003) further extended it by coupling it with the Miami Isopycnic Coordinate Ocean Model (MICOM) to create the Bergen Climate Model (BCM). The version of the ARPEGE model used in this study is the same as that used to create the BCM.

The model uses truncated spherical harmonics to represent fields in the horizontal and a hybrid sigma-pressure coordinate in the vertical. In this study, we use a triangular truncation at wavenumber 63 and a linear grid with a roughly 2.8° resolution in the horizontal. There are 31 vertical levels, ranging from the surface to 10 hPa. The time step is 30 min. Physical parameterizations used include Morcrette’s (1991) radiative scheme, Bougeault’s (1985) convective scheme, Ricard and Royer’s (1993) statistical cloud scheme, and the Louis et al. (1981) turbulence scheme. The lower boundary conditions over land points are provided by the Interaction between Soil, Biosphere, and Atmosphere (ISBA) land surface scheme (Noilhan and Planton 1989; Mahfouf et al. 1995). More information about the details of the model’s physical processes can be found in Deque et al. (1994).

In the present study, the ARPEGE-CLIMAT model was first integrated for a 10-yr simulation of the present-day climate to serve as the control experiment (CTL). The simulation began on 1 January, forced by climatological sea surface temperatures, sea ice, and a present-day vegetation map over land with initial and boundary conditions taken from National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalyses (Kalnay et al. 1996). Sensitivity simulations with the same integration time span were then performed, in which the vegetation coverage and leaf area index (LAI) poleward of 60°N were increased. Vegetation properties at grid cells outside this region remained the same as in CTL. In this way, the climatic response to the high-latitude greenness increase can be identified. The reason for increasing both vegetation coverage and LAI is that these two vegetation parameters are closely correlated with greenness. Vegetation coverage measures the horizontal spread of vegetation and LAI measures its vertical density. Vegetation coverage ranges from 0% to 100% and LAI from 0 to 6 in m2 m−2. The most significant impacts of increasing vegetation coverage and LAI in the land surface model ISBA include a reduction to surface albedo and the enhancement of evapotranspiration. In ISBA, vegetative evapotranspiration (Eυ), including transpiration (Etr) and direct evaporation (Er), is described by

 
formula

where veg is the vegetation coverage, ρa is the air density, qsat(Ts) is the saturation specific humidity for a given surface temperature Ts, qa is the specific humidity of air at altitude Za (the lowest model layer), δ = (Wr/Wr max)2/3 serves as a weighting function to suppress Etr in favor of Er as the vegetation surface becomes increasingly wet, Wr is the water intercepted by vegetation, Wrmax is the maximum water capacity intercepted by vegetation, Ra and Rs are the air and surface resistances respectively and are represented by

 
formula
 
formula

where CH is the exchange coefficient, a function of the thermal stability of the atmosphere, Va is the wind speed at height Za, Rsmin is the minimal stomatal resistance, a function of the vegetation type, and LAI is the leaf area index. In addition, F1, F2, F3 and F4 are functions measuring the influence of the solar radiation, hydric tension, deficit to saturation and air temperature, respectively, on the stomatal regulation. Detailed descriptions of the parameterizations for these four functions can be found in Noilhan and Planton (1989).

From Eq. (1), a linear relationship exists between the vegetation coverage (veg) and the evapotranspiration (Eυ). Furthermore, LAI exerts an influence on the transpiration (Etr) via surface resistance (Rs) as seen in Eq. (3). When LAI increases, there will be a reduction in the surface resistance (Rs), and vegetative transpiration will accordingly be enhanced. It should be noted that evaporation also occurs from the nonvegetated portions of each grid cell, which of course decreases as the vegetation coverage increases. The total evaporation is a weighted linear combination of evaporation (evapotranspiration) from the bare and vegetated fractions of each grid cell. Thus, under the conditions of increased greenness, an increase in the total evaporation contributed by the increased vegetative evapotranspiration can be offset to some extent by a decreased nonvegetative evaporation.

Surface albedo is a linear weighted combination of vegetation albedo and bare soil albedo in the ISBA [see Eq. (5)]. Because vegetation has a smaller albedo than the soil, surface albedo is reduced as the vegetation is increased. Surface emissivity and surface roughness length are both determined by the dominant vegetation type and have not been changed in this study. When snow is present, however, the impacts of the increased greenness are included. The snow parameterization used in the ISBA model was included by Douville et al. (1995) and was also 1 of the 21 models for snow simulation comparisons by Slater et al. (2001). As indicated by Slater et al. (2001), the most important properties of snow as it relates to the climate system are its high albedo, low thermal conductivity and roughness length, and ability to store water within the hydrological cycle. In the ISBA model, surface albedo and roughness length vary according to the fraction of snow coverage, which is determined by

 
formula

where Pn is the total snow fraction, Png and Pnv are the snow fraction covering bare soil and vegetation, respectively, Wn is the snow liquid water equivalent, z0 is the roughness length of the topography, z0υ is the vegetation roughness length determined by the vegetation type, and ρs is the snow density.

The total surface albedo is then given by

 
formula

where αg is the bare soil albedo, αs is the snow albedo as a function of temperature and snow age, and αv is the vegetation albedo.

The surface roughness length in the ISBA model is given by a quadratic formula:

 
formula

where z0cr= 0.001, and z0f = (z20υ + z20)1/2 is the background roughness length.

The snow fraction is calculated with an empirical parameterization, but is also a critical variable in determining the snow-covered surface properties and plays an important role in snow melting (Slater et al. 2001). The masking effect of the vegetation is taken into account via the vegetation coverage veg in Eq. (4), which thereby leads to an impact on Pn by increased greenness.

In this study, we conducted three sensitivity experiments, in which the vegetation coverage and LAI have 20%, 60%, and 100% increase with an upper limit of 100% for vegetation coverage and 6 m2 m−2 for LAI, and named them as G20, G60, and G100, respectively (Table 1). These experiments represent a realistic increase of greenness (20%), as has been demonstrated via remote sensing observations, as well as two potential aggressive increases in the future. A 15-yr on-site experiment at Toolik Lake, Alaska, mimicking the effects of faster decomposition of organic matter in a warmer climate, was performed by Shaver et al. (2001). Their results demonstrated that the primary production in fertilized plots was increased by 2.5 times and the leaf area in fertilized plots was twice that of controls, which suggests that a more aggressive greenness increase (e.g., 100%) is possible with the continuation of global warming. The maximum fractional vegetation coverage in the experiment CTL is 80%–90% over most of Alaska, northwestern Canada, and southern Eurasia and 40%–70% over northeastern Canada and northern Eurasia for land poleward of 60°N (Fig. 1a). Accordingly, small changes (10%–20%) are imposed on the areas with 80%–90% coverage and larger changes (30%–50%) on the areas with 40%–70% coverage in the experiment G100 (Fig. 1c). The maximum LAI in the experiment CTL shows a similar pattern except for a small LAI (<0.5) over eastern Russia (Fig. 1b). The initial LAI over northeastern Canada, northern Europe, and eastern Russia is relatively small (∼1), and so the changes imposed on these areas in the experiment G100 are also very small (Fig. 1d).

Table 1.

Summary of experiments: 1) CTL uses climatological vegetation coverage (Vegc) and leaf area index (LAIc), 2) G20 increases Vegc and LAIc by 20%, 3) G60 increases Vegc and LAIc by 60%, and 4) G100 increases Vegc and LAIc by 100%, all increases occur only over land poleward of 60°N.

Summary of experiments: 1) CTL uses climatological vegetation coverage (Vegc) and leaf area index (LAIc), 2) G20 increases Vegc and LAIc by 20%, 3) G60 increases Vegc and LAIc by 60%, and 4) G100 increases Vegc and LAIc by 100%, all increases occur only over land poleward of 60°N.
Summary of experiments: 1) CTL uses climatological vegetation coverage (Vegc) and leaf area index (LAIc), 2) G20 increases Vegc and LAIc by 20%, 3) G60 increases Vegc and LAIc by 60%, and 4) G100 increases Vegc and LAIc by 100%, all increases occur only over land poleward of 60°N.
Fig. 1.

(a) Vegetation coverage and (b) LAI in the experiment CTL, and the differences of (c) vegetation coverage and (d) LAI between the experiments G100 and CTL.

Fig. 1.

(a) Vegetation coverage and (b) LAI in the experiment CTL, and the differences of (c) vegetation coverage and (d) LAI between the experiments G100 and CTL.

Since the model takes some time to reach quasi equilibrium in response to the altered vegetation, we exclude the first year for all experiments in the following analysis. In addition, only vegetated areas are included in the analysis. Glaciated areas, such as most of Greenland and a very small part of southern Alaska, have been excluded as have areas with snow depths greater than 0.5 m in August.

3. Thermodynamic impacts of greenness increase

a. Surface energy budget

The surface energy budget consists primarily of net solar radiation (SW), net longwave radiation (LW), sensible heat flux (SH), and latent heat flux (LH). Downward SW and upward LW represent a significant heat source and sink, respectively. Figure 2 illustrates the seasonal cycles of area-averaged monthly mean surface energy budgets for SW, LW, SH, and LH over land poleward of 60°N in the experiment CTL, where downward fluxes are defined as positive and upward fluxes as negative. The LW and LH are negative throughout the year, contributing to heat loss and surface cooling. SH is positive during winter from October through March, becoming negative from late spring through early autumn. The large positive SW plays a dominant role in surface heating from April to September.

Fig. 2.

Seasonal cycles of area-averaged monthly mean surface energy budgets over land poleward of 60°N in the experiment CTL.

Fig. 2.

Seasonal cycles of area-averaged monthly mean surface energy budgets over land poleward of 60°N in the experiment CTL.

The SW absorbed by the surface is determined by the surface albedo. Since vegetation has a lower albedo than bare soil, increased vegetation coverage accordingly results in an overall decrease in the surface albedo and, in turn, alters surface heat budgets through an increase in the surface absorption of solar radiation in the sensitivity experiments G20, G60, and G100 (Fig. 3a). The decrease of the surface albedo can be amplified especially when solar radiation is strong and snow cover is present, since darker vegetation can effectively mask the brighter snow-covered ground. Differences in SW between the sensitivity experiments and CTL are heterogeneous over the course of the year. The maximum increase seen in late spring and early summer occurs systematically in the three sensitivity experiments, while little increase appears in December and January when the incoming SW is very small. The strong increase during late spring and early summer in the sensitivity experiments can be accounted for by the positive feedback between snow cover and SW. When SW is increased, snow melts more rapidly, which, in turn, results in a decrease in surface albedo and thus a further increase in SW. More information about snow cover changes will be given in section 3b. An interesting result is that there is little change in SW in the sensitivity experiments during July. This is due to an increase in cloudiness associated with changes in precipitation, which blocks downward solar radiation and offsets the effects of decreased albedo caused by increased greenness. Generally the fractional cloud cover decreases slightly for most of the year and total cloud water increases during the summer months in the sensitivity experiments (Fig. 4). Increased cloud water, implying an increase in thick and low clouds in the sensitivity experiments, reflects more solar radiation. Meanwhile a slight increase in fractional cloud cover occurs in July, which, together with the increased cloud water, prevents more solar radiation from reaching the surface in the sensitivity experiments. Overall, changes in SW are positively correlated with the greenness increase—the more greenness is increased, the stronger the change that occurs in SW. The increase in SW reaches a maximum of about 20 W m−2 in May in the sensitivity experiment G100, while the largest SW increase is about 8 W m−2 in G20. The geographical distribution of the SW difference between G100 and CTL in May is displayed in order to show the maximum impacts of vegetation increase on the surface radiation. There is a broad increase in SW absorption throughout in the landmass surrounding the Arctic Ocean where a greenness increase was prescribed (Fig. 5). A Student’s t test was used to assess the statistical significance of the differences between the sensitivity experiments and CTL. Only the differences that are statistically significant at the 99% level are shaded in Fig. 5. The broad increase in SW over the land surrounding the Arctic Ocean in the sensitivity experiments (G60 and G20 are not shown) is statistically significant at the 99% level.

Fig. 3.

Seasonal cycles of the differences of surface energy budgets (sensitivity experiments minus CTL) for (a) surface net solar radiation, (b) surface net longwave radiation, (c) sensible heat flux, and (d) latent heat flux.

Fig. 3.

Seasonal cycles of the differences of surface energy budgets (sensitivity experiments minus CTL) for (a) surface net solar radiation, (b) surface net longwave radiation, (c) sensible heat flux, and (d) latent heat flux.

Fig. 4.

Seasonal cycles of the differences of area-averaged monthly mean (a) cloud cover and (b) cloud water over land poleward of 60°N between the sensitivity experiments and CTL.

Fig. 4.

Seasonal cycles of the differences of area-averaged monthly mean (a) cloud cover and (b) cloud water over land poleward of 60°N between the sensitivity experiments and CTL.

Fig. 5.

Differences of monthly mean surface net solar radiation between G100 and CTL in May; only the differences significant at 99% level are indicated with yellow shadow.

Fig. 5.

Differences of monthly mean surface net solar radiation between G100 and CTL in May; only the differences significant at 99% level are indicated with yellow shadow.

In conjunction with the SW changes, LW, SH, and LH also show noticeable changes in the sensitivity experiments, though the magnitudes of the effects are smaller than those seen in SW (Figs. 3b–d). In general, LW, SH, and LH decrease (increase upward) in the sensitivity experiments, indicating that the increased greenness causes these three terms of the energy budget to transfer more energy from the surface to the atmosphere. The increased heat loss from these three components of the energy budget partially offsets the increased SW. Seasonally, these three energy budget components have their maximum decreases (increases upward) of LW and SH during spring, while the minimum decreases are found in winter. There is a slight positive anomaly of LW in July and the decrease (increase upward) in SH in July is also very small, which can be attributed to the increased clouds as discussed above.

b. Temperature increase

The surface air temperatures in areas of enhanced greenness increase markedly in the sensitivity experiments relative to CTL throughout the year (Fig. 6), although the temperature increases are not spatially homogeneous over the land areas poleward of 60°N. Nevertheless, the broad spatial scale of the warming suggests that the greenness increase contributes to polar amplification of global warming by increasing the temperature in the terrestrial portion of the northern high latitudes. The increased SW due to the greenness increase provides more energy for the turbulent fluxes. The sensible heat flux is also notably increased and thereby warms the lower atmosphere.

Fig. 6.

(a) Seasonal cycles of area-averaged monthly surface temperatures over land poleward of 60°N in the experiment CTL, and (b) the differences between sensitivity experiments and CTL.

Fig. 6.

(a) Seasonal cycles of area-averaged monthly surface temperatures over land poleward of 60°N in the experiment CTL, and (b) the differences between sensitivity experiments and CTL.

Climatologically, the area-averaged monthly mean surface air temperature over land poleward of 60°N reaches a maximum in July and a minimum in January. Superimposed on the seasonal cycle, there is a larger temperature increase in winter and spring when greenness is increased, with a smaller increase seen in the summer. Maxima occur in June and in December and January, with a temperature increase of about 2°C for each. The smaller increase in the surface air temperatures during July and August is primarily due to the small changes in heat budgets during these two months (shown in Fig. 3), which can be explained by the offsetting effects of increased clouds. The significant increase in spring temperature, seen especially in G100, can be accounted for by the positive feedback noted earlier between snow cover and SW. The warm anomaly in winter is not simply caused by the local atmosphere–land surface interactions. Rather it is complicated by the dynamic processes associated with atmospheric general circulation.

The warmings of 1°–2°C obtained here from the vegetative parameter changes may be compared with the greenhouse warming of northern land areas in greenhouse gas scenario experiments. In the recent IPCC simulations done in support of the Fourth Assessment Report, the annual mean warmings poleward of 60°N averaged over all (14) models contributing to the IPCC model archive are 2°–4°C (B1 scenario), 2°–5°C (A1B scenario), and 3°–6°C (B2 scenario) for the late twenty-first century (Chapman and Walsh 2006). In winter, the warmings are generally a degree or two larger than the annual means, while in spring they are slightly smaller. The IPCC model simulations were generally performed with fixed (seasonally varying) vegetative cover. Thus the 1°–2°C warmings from the vegetative changes are smaller than, but as much as 25%–50% as large as, the CO2-incuded warming in the absence of vegetative changes.

A rise in temperature caused by increased greenness is not constrained to the near-surface layer—it extends to the midtroposphere. The differences in the vertical profiles of area-averaged annual mean air temperature over land poleward of 60°N between the sensitivity experiments and CTL (Fig. 7) show warmings of about 0.9°, 0.8°, and 0.75°C in G100, G60, and G20, respectively, at 1000 hPa. Although the magnitude of these temperature changes decreases with height, the increase in temperature is still greater than 0.5°C below 900 hPa and between 0.1° and 0.3°C at 700 hPa.

Fig. 7.

(a) Profiles of area-averaged annual mean air temperature over land poleward of 60°N in the experiment CTL, and (b) the differences between sensitivity experiments and CTL.

Fig. 7.

(a) Profiles of area-averaged annual mean air temperature over land poleward of 60°N in the experiment CTL, and (b) the differences between sensitivity experiments and CTL.

c. Snow cover reduction

Seasonal cycles of area-averaged monthly mean snow cover over land poleward of 60°N in CTL, as well as the differences between the sensitivity experiments and CTL, are shown in Fig. 8. Climatologically, the CTL snow cover reaches its peak (∼180 mm) in April and minimum (almost zero) in August. The changes in snow cover in the sensitivity experiments show a reduction, with the largest reduction in May and June. The reduction in the experiment G100 is much larger than that in G60 and G20.

Fig. 8.

(a) Seasonal cycles of area-averaged monthly mean snow cover over land poleward of 60°N in the experiment CTL, and (b) the differences between sensitivity experiments and CTL.

Fig. 8.

(a) Seasonal cycles of area-averaged monthly mean snow cover over land poleward of 60°N in the experiment CTL, and (b) the differences between sensitivity experiments and CTL.

The large reduction in snow cover in May and June in the sensitivity experiments is attributed to rapid snowmelt in late spring, when solar radiative forcing predominates. As indicated in sections 3a and 3b, the strongest increase in absorbed SW in the sensitivity experiments occurs in spring, increasing the rate of snowmelt (Fig. 9) and leading to a faster snow removal than in CTL. As this process continues, the snow–albedo feedback intensifies the speed of the snowmelt.

Fig. 9.

Seasonal cycles (February–August) of area-averaged monthly mean snowmelt over land poleward of 60°N in the experiments CTL, G100, G60, and G20.

Fig. 9.

Seasonal cycles (February–August) of area-averaged monthly mean snowmelt over land poleward of 60°N in the experiments CTL, G100, G60, and G20.

In addition, the overall decrease in snow cover throughout the year is caused by a stronger snowmelt as well as a weaker snow accumulation in the greenness experiments. Annual snowfall north of 60°N decreases by about 5 mm in G20, 7 mm in G60, and 12 mm in G100 (Table 2). The snowfall is basically balanced by the snowmelt and snow sublimation in all experiments, which results in an almost zero snow cover during summer as shown in Fig. 8a. The reduction in snow accumulation could be driven by the shift of some snowfall to rainfall, associated with the increased surface and tropospheric air temperatures when the greenness is increased (Fig. 6 and 7). We found that there was little change in the accumulated precipitation during winter, so if the increased temperature resulted in the transfer of some snow to rain, there would be less precipitation in the form of snow, which would necessarily reduce snow accumulation and hence snow cover. These changes in snow cover under the increased greenness scenario are consistent with the observed snow cover changes associated with the warming trend in the northern high latitudes (Groisman et al. 1994; Serreze et al. 2000; ACIA 2004).

Table 2.

Area-averaged annual snowfall (F) and the summation of annual snowmelt (M) and annual snow sublimation (S) over land poleward of 60°N in the experiments CTL, G20, G60 and G100 (unit: mm m−2 yr−1). Changes of snowfall in G20, G60 and G100 relative to CTL are in parentheses (unit: %).

Area-averaged annual snowfall (F) and the summation of annual snowmelt (M) and annual snow sublimation (S) over land poleward of 60°N in the experiments CTL, G20, G60 and G100 (unit: mm m−2 yr−1). Changes of snowfall in G20, G60 and G100 relative to CTL are in parentheses (unit: %).
Area-averaged annual snowfall (F) and the summation of annual snowmelt (M) and annual snow sublimation (S) over land poleward of 60°N in the experiments CTL, G20, G60 and G100 (unit: mm m−2 yr−1). Changes of snowfall in G20, G60 and G100 relative to CTL are in parentheses (unit: %).

4. Hydrological impacts of greenness increase

a. Enhancement of evapotranspiration and increase in surface and atmospheric moisture

Increased greenness enhances vegetative transpiration during the warm season, as shown by the differences in vegetation transpiration between the experiments G100 and CTL (Fig. 10). The transpiration increase in G100 is almost homogeneously distributed over the landmass north of 60°N, with large centers located in the north-central Eurasian and American continents. The Student’s t test was again used to assess the statistical significance of the differences between the sensitivity experiments and CTL. Only the differences that are statistically significant at the 99% level are shaded in Fig. 10. The increases in vegetative transpiration in the sensitivity experiments (G60 and G20 are not shown) are statistically significant at the 99% level.

Fig. 10.

Differences of seasonal mean vegetation transpiration between G100 and CTL during June–August; only the differences significant at 99% level are indicated with yellow shadow.

Fig. 10.

Differences of seasonal mean vegetation transpiration between G100 and CTL during June–August; only the differences significant at 99% level are indicated with yellow shadow.

Vegetative transpiration, functioning as the hydraulic conductance between soil and atmosphere via vegetation roots and stomates, transfers soil moisture from root layers in the deep soil to the atmosphere. Note that vegetative transpiration is distinguished from bare soil evaporation, which is measured as the moisture loss from surface soil. Moisture loss from the wet canopy surface to the atmosphere is measured by vegetation evaporation. The sum of vegetation transpiration and vegetation evaporation is defined as vegetation evapotranspiration. The total evaporation from the land surface to the atmosphere is the summation of vegetation evapotranspiration and bare soil evaporation.

The area-averaged monthly mean total evaporation (including the three terms mentioned above) for the experiments CTL, G20, G60, and G100 over land poleward of 60°N is shown in Fig. 11. Dramatic increases of vegetation evapotranspiration occur relative to CTL from May through September in all three sensitivity experiments, with a maximum in June–July. Vegetation evapotranspiration in the experiment G100 is nearly double that of the experiment CTL in June and July, with the largest difference reaching ∼1 kg m−2 per day. On the other hand, due to less bare soil being exposed to the atmosphere as a result of the increased vegetation coverage, there is a reduction in bare soil evaporation in the sensitivity experiments. Vegetative evapotranspiration has a much stronger seasonal variability than bare soil evaporation. The enhanced vegetative evapotranspiration makes a critical contribution to the increase in total evaporation from the land surface to the atmosphere during May–July. The increases of total evaporation during May–July in all three sensitivity experiments pass the Student’s t test at the 99% significant level.

Fig. 11.

Seasonal cycles of area-averaged monthly mean evaporation (including bare soil evaporation, vegetation evaporation, and vegetation transpiration) over land poleward of 60°N in the experiments CTL, G20, G60, and G100.

Fig. 11.

Seasonal cycles of area-averaged monthly mean evaporation (including bare soil evaporation, vegetation evaporation, and vegetation transpiration) over land poleward of 60°N in the experiments CTL, G20, G60, and G100.

The partitioning of changes in total evaporation in the sensitivity experiments (strengthened vegetation evapotranspiration and weakened bare soil evaporation) in turn affects the soil moisture. Surface soil moisture is increased in the sensitivity experiments due to the reduction in direct evaporation from surface soil (Fig. 12a). As mentioned above, transpiration transfers soil moisture from the deep soil layers to the atmosphere and has little impact on the surface soil, so the enhanced transpiration does not reduce surface soil moisture. However the total soil moisture is decreased during June–September in the sensitivity experiments, especially G100 and G60, due to the enhanced vegetation transpiration (Fig. 12b). A significant increase in total soil moisture occurs in the sensitivity experiments during April and May due to faster snowmelt. There is also a slight increase in total soil moisture in the sensitivity experiments during October, which is likely caused by the larger rainfall (Fig. 15, later) and weakened transpiration (Fig. 11) in the sensitivity experiments for that month. The decrease in total soil moisture during the winter months is consistent with the larger underground runoff due to warmer soil temperature in the sensitivity experiments.

Fig. 12.

Seasonal cycles of (a) area-averaged monthly mean surface soil moisture and (b) total soil moisture over land poleward of 60°N in the experiments CTL, G100, G60, and G20.

Fig. 12.

Seasonal cycles of (a) area-averaged monthly mean surface soil moisture and (b) total soil moisture over land poleward of 60°N in the experiments CTL, G100, G60, and G20.

Fig. 15.

Seasonal cycles of area-averaged monthly mean total rainfall, convective rainfall, and large-scale rainfall over land poleward of 60°N in the experiments CTL and G100.

Fig. 15.

Seasonal cycles of area-averaged monthly mean total rainfall, convective rainfall, and large-scale rainfall over land poleward of 60°N in the experiments CTL and G100.

Increased total evaporation from land results in an increase in the atmospheric moisture during the summer months (Fig. 13). The differences in the profiles of area-averaged summer (June–August) mean specific humidity over land poleward of 60°N between the sensitivity experiments and CTL shows a considerable increase in moisture in the lower troposphere, with a maximum value near the surface. As with air temperature, the change in moisture grows as the greenness is increased.

Fig. 13.

(a) Profiles of area-averaged seasonal mean specific humidity over land poleward of 60°N in the experiment CTL and (b) the differences between sensitivity experiments and CTL.

Fig. 13.

(a) Profiles of area-averaged seasonal mean specific humidity over land poleward of 60°N in the experiment CTL and (b) the differences between sensitivity experiments and CTL.

Synchronized increases in air temperature and humidity occur in the high latitudes of the Northern Hemisphere in the context of global warming effects (e.g., Kattsov and Walsh 2000). Increased greenness brings about similar changes, suggesting positive correlations and feedbacks in changes to these two parameters in the polar region. Additionally, because water vapor is a significant greenhouse gas, an increase in specific humidity will enhance the greenhouse effect, which will in turn favor a further amplification of the temperature increase.

b. Intensification of summer convective precipitation

In conjunction with the changes to land surface properties and atmospheric temperature and humidity, considerable changes to precipitation occur as well. Seasonal cycles of area-averaged monthly mean precipitation over land poleward of 60°N in CTL and the three sensitivity experiments show that precipitation primarily occurs from May to November, reaching a peak from June to August (Fig. 14). Relative to CTL, there is increased precipitation in the sensitivity experiments during the peak season; the increases in G100 and G60 are about 10% during the summer. The precipitation increases during May–July in all three sensitivity experiments pass the Student’s t test at the 99% significant level.

Fig. 14.

Seasonal cycles of area-averaged monthly mean precipitation over land poleward of 60°N in the experiments CTL, G100, G60, and G20.

Fig. 14.

Seasonal cycles of area-averaged monthly mean precipitation over land poleward of 60°N in the experiments CTL, G100, G60, and G20.

The increased total precipitation during summer is primarily the result of greater convective rainfall (Fig. 15), which indicates that increased greenness tends to reduce atmospheric stability and intensify atmospheric convective activity. Climatologically, convective rainfall is about 3 times greater than the large-scale rainfall during summer over land areas poleward of 60°N. Convective rainfall peaks in July, while large-scale rainfall reaches its maximum in September. A comparison between G100 and CTL indicates that the increase in convective rainfall begins in spring and reaches a maximum in June. Similar to the total precipitation, the convective rainfall differences during May–July between the sensitivity experiments and CTL are statistically significant at the 99% level.

To elucidate the strengthening of convective activity, we computed atmospheric moist static energy over land poleward of 60°N during June–August in all sensitivity and control experiments using the monthly mean data. The profile of the area-averaged moist static energy in CTL shows a positive vertical gradient (Fig. 16a), indicating that the mean state of the atmosphere over the study domain is stable. The differences in the area-averaged moist static energy show an increase in the sensitivity experiments relative to CTL, especially at the lower atmospheric levels (Fig. 16b). The increase in the moist static energy can be accounted for by the increases in both temperature and humidity in response to the increased greenness. Consequently, the vertical gradient of moist static energy decreases in the sensitivity experiments, indicating a reduction in the atmospheric stability. Decreased atmospheric stability provides a thermodynamic environment, which favors an increase in convective activity and associated convective precipitation. The increases seen in convective activity and convective precipitation can result from an increase in either its frequency or its intensity.

Fig. 16.

(a) Profiles of area-averaged seasonal mean moist static energy over land poleward of 60°N, and (b) the differences between sensitivity experiments and CTL during June–August.

Fig. 16.

(a) Profiles of area-averaged seasonal mean moist static energy over land poleward of 60°N, and (b) the differences between sensitivity experiments and CTL during June–August.

c. Changes in overall precipitation minus evaporation (P − E)

Both precipitation and evaporation increase under the scenarios of increased greenness. The difference between precipitation and evaporation (PE) is a critical parameter in the hydrological balance because it corresponds closely to runoff and river discharge over annual time scales. The differences between the sensitivity experiments and CTL in area-averaged annual mean PE over land poleward of 60°N show an increase in G20 (∼3 mm) and G60 (∼7 mm), but a decrease in G100 (∼−4 mm) (Table 3). This suggests that a complex nonlinearity exists in vegetation–atmosphere interactions. Small and intermediate increases in greenness result in an increase in PE due to a stronger increase in precipitation than in evaporation, while larger increases of greenness in the experiment G100 gives rise to a reduction of PE. The precipitation increase in G100 is almost the same as that in G60 (Fig. 14), but the evaporation increase in G100 is larger than that seen in G60 (Fig. 11). There is the same amount of precipitation reaching the canopy surface in the experiments G100 and G60; however, more precipitation will be intercepted by the canopy in G100 and subsequently evaporate back to the atmosphere, which results in a decrease in PE in G100. The PE differences between the sensitivity experiments and the control experiment are significant at the 90% level for each of G100-CTL, G60-CTL, and G20-CTL.

Table 3.

Area-averaged annual precipitation (P) minus evaporation (E) over land poleward of 60°N in the experiments CTL, G20, G60, and G100.

Area-averaged annual precipitation (P) minus evaporation (E) over land poleward of 60°N in the experiments CTL, G20, G60, and G100.
Area-averaged annual precipitation (P) minus evaporation (E) over land poleward of 60°N in the experiments CTL, G20, G60, and G100.

5. Summary and discussions

Significant impacts of vegetation on climate have been recognized. The high-latitude vegetation feedbacks on climate have been investigated by various numerical modeling studies (e.g., Bonan et al. 1992; Thomas and Rowntree 1992; Chalita and Le Treut 1994; Foley et al. 1994; Douville and Royer 1997; Levis et al. 1999), though most of them focused on the boreal deforestation. Models also indicate that a change from forest to cropland alters the energy and water budgets of the Mississippi River basin (Twine et al. 2004). However, recent observational data demonstrate that the northern high latitudes have instead been greening up. The satellite-derived NDVI show a greenness increase of about 10% in the Eurasian sector and 17% in the American sector poleward of 60°N (Stow et al. 2004). This remarkable greenness increase is probably attributable to the effects of the air temperature increase observed over northern terrestrial regions (Serreze et al. 2000; ACIA 2004). Motivated by this recent observational finding of increased greenness, together with the likelihood of future warming as projected by climate modeling studies (ACIA 2004; Houghton et al. 2001), which may further enhance the greenness increase, we have conducted a set of modeling experiments to identify the impacts of high-latitude vegetative changes on the atmosphere.

The land cover in the northern high latitudes has obvious seasonality. Vegetation is fully exposed and exhibits explicit interactions with the atmosphere primarily during the warm seasons. Therefore, the focus of this study was on the results from the warm seasons. Comparisons between the control and sensitivity experiments with increased vegetation demonstrate that greenness increase has significant thermodynamic and hydrological consequences in the northern high latitudes. Generally, a greenness increase contributes to both a warmer and wetter atmosphere and land surface. Specifically, the surface air temperature markedly increases in the sensitivity experiments, with a maximum increase of 2°C (for doubled greenness). Such warming is not negligible in comparison with greenhouse-driven warmings simulated by models with fixed vegetation. The increase in air temperature extends to the midtroposphere, decreasing in magnitude with height. Both the precipitation and evaporation are increased in the sensitivity experiments; the precipitation is enhanced by about 10% in the season of peak rainfall (June through August) in G60 and G100. In addition, snow cover decreases noticeably throughout the year, with the largest reduction seen in May and June. These results are consistent with previous modeling studies. Bonan et al. (1992), for example, showed that the boreal forest can result in a warmer temperature than the bare ground or tundra. Thomas and Rowntree (1992) reported a cool and dry climate for the boreal deforested regions. Snow cover increased when Chalita and Le Treut (1994) removed boreal forest in their numerical studies.

Through quantitative investigation of the associated thermodynamic and hydrological impacts of increased greenness in the model experiments, the following physical processes were identified as the major factors responsible for the thermal and hydrologic changes:

  1. Solar radiation absorbed by the land surface increases when greenness increases due to the decreased surface albedo, which contributes significantly to the temperature increase. In addition, the increased greenness amplifies the positive snow–albedo feedback during the melting season, which leads to faster snowmelt and a stronger increase in surface net solar radiation.

  2. Vegetative evapotranspiration is significantly enhanced by increased greenness during summer, when more root-level soil moisture is transferred to the atmosphere. Direct evaporation from surface soil is reduced due to the greater soil coverage by vegetation, which reduces the exposure of bare soil to the atmosphere. Thus vegetation evapotranspiration makes an important contribution to the increase in total evaporation from the land surface to the atmosphere without decreasing soil moisture at the surface, though it does reduce the total soil moisture.

  3. Changes to surface and low-tropospheric thermodynamic and hydrological conditions, brought about by the greenness increase, reduce atmospheric stability, leading to an increase in convective activity and associated convective rainfall in the summer. The increased convective activity, in turn, increases cloud water, which blocks more downward solar radiation and offsets to some degree the increased radiative forcing at the surface.

The net PE is a crucial parameter for closing the hydrological budget. Both precipitation and evaporation increase with the increase in greenness. However, the relationship between net PE and greenness shows nonlinearity. PE increases when greenness is increased by 20% and 60%, while it decreases when the greenness is increased even further by 100%, pointing to a shift from a dominance of increased P to a dominance of increased E as the greenness becomes progressively larger.

This study has potentially important implications for the interpretation of future warming over Arctic terrestrial regions. The aforementioned results suggest that increased greenness can contribute to the polar amplification of global warming, due to the resultant albedo decrease and associated increase of SW absorption at the surface. Simultaneously, hydrological impacts of increased greenness result in increased atmospheric humidity, indicating that the changes to air temperature and humidity should be positively correlated at high latitudes. The increased water vapor can be expected to enhance the greenhouse effect, further contributing to the surface and tropospheric air temperature increase, as well as to the polar amplification of global warming. Critical to this interpretation of the polar amplification are changes to vegetation in middle latitudes, where many models project a summer drying under enhanced greenhouse forcing (Houghton et al. 2001).

This study represents a first attempt to quantify the role of increased greenness of high-latitude vegetation in climate change. Given that the magnitude of the vegetative effects obtained here are of the same magnitude as (although smaller than) direct greenhouse-driven changes, the climatic feedbacks involving increased high-latitude greenness appears to merit further investigation, both observationally and with models, in order that projections of climate change can be placed on a sounder physical basis. In particular, we did not include an interactive Arctic sea ice and ocean in this study. As Bonan et al. (1992) indicated, feedbacks from vegetation change could be strengthened with interactive sea ice and ocean compared to the prescribed sea ice and sea surface temperature in model simulations. Thus it would be worthwhile to further investigate the impacts of greenness increase in the context of fully coupled atmosphere–sea ice–ocean models.

Acknowledgments

This study was supported in part by a grant of High Performance Computing (HPC) resources from the Arctic Region Supercomputing Center (ARSC) at the University of Alaska Fairbanks as part of the Department of Defense HPC Modernization Program and by the National Science Foundation through Grant OPP-0327664 to the International Arctic Research Center. We thank Oyvind Byrkjedal for providing the climate model ARPEGE-CLIMAT and Jeremy Krieger for manuscript editing. We are grateful to Dr. Christa Peters-Lidard and the anoymous reviewers, whose constructive comments and suggestions substantially improved the content and presentation of this paper. The computing resources were provided by ARSC.

REFERENCES

REFERENCES
ACIA
,
2004
:
Impacts of a Warming Arctic: Arctic Climate Impact Assessment.
Cambridge University Press, 140 pp
.
Bonan
,
G. B.
,
1999
:
Frost followed the plow: Impacts of deforestation on the climate of the United States.
Ecol. Appl.
,
9
,
1305
1315
.
Bonan
,
G. B.
,
D.
Pollard
, and
S. L.
Thompson
,
1992
:
Effects of boreal forest vegetation on global climate.
Nature
,
359
,
716
718
.
Bougeault
,
P.
,
1985
:
A simple parameterization of the large-scale effects of cumulus convection.
Mon. Wea. Rev.
,
113
,
2108
2121
.
Chalita
,
S.
, and
H.
Le Treut
,
1994
:
The albedo of temperate and boreal forest and the Northern Hemisphere climate: A sensitivity experiment using the LMD GCM.
Climate Dyn.
,
10
,
231
240
.
Chapman
,
W. L.
, and
J. E.
Walsh
,
2006
:
Simulations of Arctic temperature and pressure by global coupled models.
J. Climate
,
in press
.
Costa
,
M. H.
, and
J.
Foley
,
2000
:
Combined effects of deforestation and doubled atmospheric CO2 concentrations on the climate of Amazonia.
J. Climate
,
13
,
18
34
.
Deque
,
M.
,
C.
Dreveton
,
A.
Braun
, and
D.
Cariolle
,
1994
:
The ARPEGE/IFS atmosphere model: A contribution to the French community climate modeling.
Climate Dyn.
,
10
,
249
266
.
Dickinson
,
R. E.
, and
A.
Henderson-Sellers
,
1988
:
Modeling tropical deforestation: A study of GCM land-surface parameterizations.
Quart. J. Roy. Meteor. Soc.
,
114
,
439
462
.
Douville
,
H.
, and
J. F.
Royer
,
1997
:
Influence of the temperate and boreal forests on the Northern Hemisphere climate in the Météo-France climate model.
Climate Dyn.
,
13
,
57
74
.
Douville
,
H.
,
J. F.
Royer
, and
J. F.
Mahfouf
,
1995
:
A new snow parameterization for the Meteo-France Climate model. Part I: Validation in stand-alone experiments.
Climate Dyn.
,
12
,
21
35
.
Foley
,
J. A.
,
J. E.
Kutzbach
,
M. T.
Coe
, and
S.
Levis
,
1994
:
Feedbacks between climate and boreal forests during the Holocene epoch.
Nature
,
371
,
52
54
.
Furevik
,
T.
,
M.
Bentsen
,
H.
Drange
,
I. K. T.
Kindem
,
N. G.
Kvamstø
, and
A.
Sorteberg
,
2003
:
Description and validation of the Bergen Climate Model: ARPEGE coupled with MICOM.
Climate Dyn.
,
21
,
27
51
.
Groisman
,
P. A.
,
T. R.
Karl
, and
R. W.
Knight
,
1994
:
Observed impact of snow cover on the heat balance and rise of continental spring temperatures.
Science
,
263
,
198
200
.
Houghton
,
J. T.
,
Y.
Ding
,
D. J.
Griggs
,
M.
Noguer
,
P. J.
van der Linden
,
X.
Dai
,
K.
Maskell
, and
C. A.
Johnson
,
2001
:
Climate Change 2001: The Scientific Basis.
Cambridge University Press, 881 pp
.
Kalnay
,
E.
, and
Coauthors
,
1996
:
The NCEP/NCAR 40-Year Reanalysis Project.
Bull. Amer. Meteor. Soc.
,
77
,
437
471
.
Kattsov
,
V. M.
, and
J. E.
Walsh
,
2000
:
Twentieth-century trends of Arctic precipitation from observational data and a climate model simulation.
J. Climate
,
13
,
1362
1370
.
Levis
,
S.
,
J. A.
Foley
, and
D.
Pollard
,
1999
:
Potential high-latitude vegetation feedbacks on CO2-induced climate change.
Geophys. Res. Lett.
,
26
,
747
750
.
Louis
,
J. F.
,
M.
Tiedtke
, and
J-F.
Geleyn
,
1981
:
A short history of the PBL parameterization at ECMWF. Proc. ECMWF Workshop on Planetary Boundary Layer Parameterization, Reading, United Kingdom, ECMWF, 59–80
.
Mahfouf
,
J. F.
,
A. O.
Manzi
,
J.
Noilhan
,
H.
Giordani
, and
M.
Deque
,
1995
:
The land surface scheme ISBA within the Meteo-France climate model ARPEGE. Part I: Implementation and preliminary results.
J. Climate
,
8
,
2039
2057
.
Morcrette
,
J-J.
,
1991
:
Radiation and cloud radiative properties in the ECMWF operational weather forecast model.
J. Geophys. Res.
,
96
,
9121
9132
.
Myneni
,
R. B.
,
C. D.
Keeling
,
C. J.
Tucker
,
A.
Asrar
, and
R. R.
Nemani
,
1997
:
Increased plant growth in the northern high latitudes from 1981 to 1991.
Nature
,
386
,
698
702
.
Nicholls
,
N.
,
G. V.
Gruza
,
J.
Jouzel
,
T. R.
Karl
,
L. A.
Ogallo
, and
D. E.
Parker
,
1996
:
Observed climate variability and change.
Climate Change 1995: The Science of Climate Change, J. T. Houghton et al., Eds., Cambridge University Press, 133– 192
.
Noilhan
,
J.
, and
S.
Planton
,
1989
:
A simple parameterization of land surface processes for meteorological models.
Mon. Wea. Rev.
,
117
,
536
549
.
Ricard
,
J. L.
, and
J. F.
Royer
,
1993
:
A statistical cloud scheme for use in an AGCM.
Ann. Geophys.
,
11
,
1095
1115
.
Serreze
,
M. C.
, and
Coauthors
,
2000
:
Observational evidence of recent change in the northern high-latitude environment.
Climatic Change
,
46
,
159
207
.
Shaver
,
G. R.
,
M. S.
Bret-Harte
,
M. H.
Jones
,
J.
Johnstone
,
L.
Gough
,
J.
Laundre
, and
F. S.
Chapin
III
,
2001
:
Species composition interacts with fertilizer to control long-term change in tundra productivity.
Ecology
,
82
,
3163
3181
.
Slater
,
A. G.
, and
Coauthors
,
2001
:
The representation of snow in land-surface schemes: Results from PILPS 2(d).
J. Hydrometeor.
,
2
,
7
25
.
Stow
,
D. A.
, and
Coauthors
,
2004
:
Remote sensing of vegetation and land-cover change in arctic tundra ecosystems.
Remote Sens. Environ.
,
89
,
281
308
.
Sturm
,
M.
,
C.
Racine
, and
K.
Tape
,
2001
:
Climate change: Increasing shrub abundance in the Arctic.
Nature
,
411
,
546
547
.
Thomas
,
G.
, and
P. R.
Rowntree
,
1992
:
The boreal forests and climate.
Quart. J. Roy. Meteor. Soc.
,
118
,
469
497
.
Tucker
,
C.
,
D.
Slayback
,
J.
Pinzon
,
S.
Los
,
R.
Myneni
, and
M.
Taylor
,
2001
:
Higher Northern Latitude NDVI and growing season trends from 1982 to 1999.
Int. J. Biometeor.
,
45
,
184
190
.
Twine
,
T. E.
,
C. J.
Kucharik
, and
J. A.
Foley
,
2004
:
Effects of land cover change on the energy and water balance of the Mississippi River basin.
J. Hydrometeor.
,
5
,
640
655
.
Voldoire
,
A.
, and
J. F.
Royer
,
2004
:
Tropical deforestation and climate variability.
Climate Dyn.
,
22
,
857
874
.
Walker
,
D. A.
, and
Coauthors
,
2003
:
Phytomass, LAI, and NDVI in northern Alaska: Relationships to summer warmth, soil pH, plant functional types and extrapolation to the circumpolar Arctic.
J. Geophys. Res.
,
108
.
8169, doi:10.1029/2001JD000986
.
Zhou
,
L.
,
C. J.
Tucker
,
R. K.
Kaufimann
,
D.
Slayback
,
N. V.
Shabanov
, and
R. B.
Myneni
,
2001
:
Variations in northern vegetation activity inferred from satellite data of vegetation index during 1981 to 1999.
J. Geophys. Res.
,
106
,
20069
20083
.

Footnotes

Corresponding author address: Jing Zhang, Geophysical Institute, University of Alaska Fairbanks, P.O. Box 757320, Fairbanks, AK 99775-7320. Email: jing@gi.alaska.edu