Local and Remote Climate Impacts from Expansion of Woody Biomass ...

Download PDF
1 downloads 1533 Views 7MB Size Report
Comment
Nov 1, 2012 - ... N. Murphy, RSMAS, Uni- versity of Miami, 4600 Rickenbacker Causeway, Miami, FL 33149. ..... in the spring and 3 W m. 22 in July. ..... 4. Discussion. We break new ground in analyzing the climate im- ..... NCAR Tech. Rep.
1 NOVEMBER 2012

MURPHY ET AL.

7643

Local and Remote Climate Impacts from Expansion of Woody Biomass for Bioenergy Feedstock in the Southeastern United States LISA N. MURPHY AND WILLIAM J. RILEY Climate Sciences Department, Earth Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California

WILLIAM D. COLLINS Climate Sciences Department, Earth Sciences Division, Lawrence Berkeley National Laboratory, and Department of Earth and Planetary Science, University of California, Berkeley, Berkeley, California (Manuscript received 22 September 2011, in final form 7 March 2012) ABSTRACT Many efforts have been taken to find energy alternatives to reduce anthropogenic influences on climate. Recent studies have shown that using land for bioenergy plantations may be more cost effective and provide a greater potential for CO2 abatement than using land for carbon sequestration. Native southern U.S. pines (i.e., loblolly) have excellent potential as bioenergy feedstocks. However, the land-cover change due to expansion of biofuels may impact climate through biophysical feedbacks. Here, the authors access the local and remote consequences of greater forest management and biofuel feedstock expansion on climate and hydrology using a global climate model, the NCAR Community Climate System Model, version 4 (CCSM4). The authors examine a plausible U.S. Department of Energy (DOE) biofuel feedstock goal by afforesting 50 million acres of C4 grasslands in the southeastern United States with an optimized loblolly plant functional type. Changes in sensible and latent heat fluxes are related to increased surface roughness, reduced bareground evaporation, and changes in stomatal conductance. In the coupled simulations, these mechanisms lead to a 18C cooling, higher atmospheric stability, and a more shallow planetary boundary layer over the southeastern United States during the summer; in winter, a cooling of up to 0.258C between 408 and 608N, a weakened Aleutian low, and a wetter Australia occurs. A weakened Aleutian low shifts the North Pacific storm track poleward in the future loblolly scenarios. These local and global impacts suggest that biophysical feedbacks need to be considered when evaluating the benefits of bioenergy feedstock production.

1. Introduction Dedicated energy crops, including forests and agricultural products, can be efficiently and sustainably converted into bioenergy to decrease our dependence on fossil fuels, mitigate climate change (Hall et al. 1991; Hedenus and Azar 2009), and provide energy security (DOE 2003). In this paper, we analyze the biophysical effects on climate of short-rotation woody crops, which are a potentially large source of bioenergy. Plantations forests grow faster, produce more biomass per unit area, and are more economically feasible with respect to harvesting and transporting biomass than unmanaged natural forests (Zhang and Polyakov 2010). Biomass

Corresponding author address: Lisa N. Murphy, RSMAS, University of Miami, 4600 Rickenbacker Causeway, Miami, FL 33149. E-mail: [email protected] DOI: 10.1175/JCLI-D-11-00535.1

harvested from trees, agricultural crops, wood residues, and other plants supplied roughly 55 000 GWh of electricity (EIA 2008) during 2006 and 6.5 billion gallons of ethanol (EIA 2012) during 2007 in the U.S. domestic energy market. Land conversion to perennial crops could become significant if the market for bioenergy increases. In 2005, the U.S. Department of Energy (DOE) estimated that the United States could sustain up to 1.3 billion tons per year of biomass production (Perlack et al. 2005). To produce this amount of biomass requires a considerable amount of land dedicated to growing perennial crops. We consider scenarios in which woody feedstock production is assumed to come from additional forests planted on marginal agricultural land released from crop rotation because of increased agricultural efficiency rather than on land occupied by extant forests. In the United States, under a moderate crop yield increase, 35

7644

JOURNAL OF CLIMATE

million acres could be available to grow perennial crops, while a high crop yield increase could leave 55 million acres available for perennials (Perlack et al. 2005). Over the past two centuries, the United States has undergone drastic transformation because of deforestation, drainage, agriculture, afforestation, and reforestation. In addition to altering the carbon cycle, land-use and land-cover change (LULCC) affect the biophysical properties of the earth’s surface and can thereby impact the land–atmosphere exchanges of energy and water at regional to global scales. For example, converting grasses to forests in midlatitudes decreases the surface albedo and increases absorption of solar radiation (Betts 2000; Feddema et al. 2005; Gibbard et al. 2005; Betts et al. 2007). However, Fall et al. (2010) examined the effects of converting grass and shrub to forest by contrasting observations and meteorological reanalysis and found that afforestation results in a small cooling. Increased roughness, which affects the land–atmosphere energy and mass exchange, and altered evapotranspiration are also predicted (Bonan 1997). Typically, forests have deeper rooting depths and higher evaporation rates per unit area compared to grasslands, and both of these properties tend to reduce sensible heat fluxes (SHFLX) and increase latent heat fluxes (LHFLX). Increased evapotranspiration can cool the local atmosphere through cloud and precipitation feedbacks (Bonan 2008). Afforestation in tropical and boreal regions has been shown to cause regional cooling (Bonan 2008) and warming (Bonan et al. 1992; Betts 2000; Betts et al. 2007), respectively. In midlatitudes, the climate effect of replacing grasslands with forest is less certain (Bonan 2008). While future climate change may impact forest productivity (Hargrove and Hoffman 2004; McCarthy et al. 2007; Eckert et al. 2010), forest management practices, including planting trees, fertilizing, suppressing fires, and draining land, have increased pine productivity in recent decades and may lessen the sensitivity to climate change. More relevant to our current analysis, several recent studies have examined the biogeophysical effects of bioenergy crop production. Higher leaf area index (LAI), albedo, transpiration, and rooting depth of perennial bioenergy crops (e.g., miscanthus) are expected to cause a regional cooling on the order of 18C compared to annual crops such as maize (Georgescu et al. 2009, 2011). Remote sensing data show that growing sugar cane crops on agricultural land cools the local land surface temperature by almost 18C because of enhanced evapotranspiration and higher albedo (Loarie et al. 2011). The goal of the current work is to quantify the biogeophysical effects of biofuel feedstock production on regional and hemispheric climate under a plausible twentyfirst-century deployment scenario in the southeastern

VOLUME 25

United States (SUS). Southern pine plantations are currently among the most intensively managed forests in the world. Intensively managed forests will have to increase in the future to meet the increasing demand for wood and fiber (Sedjo and Botkin 1997). These forests also preserve carbon sequestration because forests are not replaced with agriculture or urban areas. There are currently ;45 million acres of plantation forest in the southern United States, of which 83% are loblolly pine (LP; Pinus taeda L.; 75%) and slash pine (Pinus ellioti; 25%) (Smith et al. 2009; Zhang and Polyakov 2010). In the SUS, loblolly pine forests cover 55 million acres and slash pine cover 13 million acres. Slash pine forests have declined since the 1950s because of conversion to faster growing loblolly pine (Smith et al. 2001). We chose to parameterize for loblolly only assuming increasing loss and conversion of slash to loblolly. Here, we define the SUS as the land area between 258 and 358N and between 758 and 988W, which encompasses the following states: Oklahoma, Louisiana, Mississippi, Arkansas, Alabama, Florida, Georgia, South Carolina, North Carolina, and eastern Texas. LP is a native species to the SUS, has a broad geographic distribution, is cost effective for conversion to cellulosic ethanol (Huang et al. 2004; Frederick et al. 2008), and has been chosen as the prime candidate for plantation bioenergy in the SUS (Kline and Coleman 2010). Unlike some hardwood species (e.g., poplar, eucalyptus) and grasses (e.g., switchgrass, miscanthus), LP can be easily established on marginal lands to reduce its competition for fertile agricultural land. LP grows more slowly than deciduous trees with annual leaf turnover. However, LP has a relatively rapid 18-month needle turnover rate for evergreen species that yields a seasonally varying LAI (Zhang and Allen 1996). This study investigates the local and remote climate impacts of land-cover change in the SUS using a coupled land–atmosphere global climate model [Community Climate System Model, version 4 (CCSM4)]. We apply eddy covariance data to parameterize the land surface model [the Community Land Model (CLM4)] representation of LP and perform a series of three simulations to characterize interactions with the atmosphere for current and potential future vegetation distributions consistent with an extreme expansion of forest biofuel feedstock production. In our current scenario, we replace all NET with LP, which presents a substantial increase in current LP forests. Although this land-use change is drastic, it allows us to test the sensitivity of our model as well as place upper bounds on the climate change possible because of expanding forest management and biofeedstock production in the United States. Our future scenario represents an additional contribution from expanding biofeedstock production in the United States.

1 NOVEMBER 2012

MURPHY ET AL.

TABLE 1. Total one-sided LAI (m2 m22) for January and June for NET and LP. Total LAI (m2 m22) Month

NET

LP

January June

4 4.5

1.5 6

Local energy and water budgets, local atmospheric temperature changes, and remote atmospheric responses in both the Northern and Southern Hemispheres associated with this LULCC are discussed. Methods are presented in section 3, local and far-field responses in section 4, additional context for interpretation of the simulations in section 5, and conclusions in section 6.

2. Methods a. Model description Simulations in this study are performed with the National Center for Atmospheric Research (NCAR) CCSM4 (Gent et al. 2011). Our global interactive land–atmosphere simulations are based upon the Community Atmosphere Model, version 4 (CAM4) (Neale et al. 2010). The land component of CCSM4, the CLM4, represents the major components of the earth’s biogeophysical and biogeochemical (BGC) cycles. Vegetation is categorized into 16 plant functional types (PFTs) that are characterized by the geometric, physical, and physiological attributes that affect energy and water fluxes. Plant physiological processes include photosynthesis, respiration, transpiration, and phenology. Each PFT has its own LAI, stem area index (SAI), canopy height, root distribution, and optical properties.

b. Model modifications CLM4 has a single PFT that represents temperate needleleaf evergreen trees (NET). In CLM4, the default temperate needleleaf tree PFT (NET) has a flat seasonal LAI with almost no difference between winter and summer values. In contrast, observations of LP from the Duke loblolly forest site [section 3b(1)] show a strong seasonal cycle in LAI (Table 1). Because of the distinguishing characteristics of LP relative to other NET, and since LP and slash pine are the two dominant forest types in the SUS, it is necessary to represent the specific biophysical properties of LP in our simulations. We therefore use an inversion technique to develop a new PFT parameter set to represent LP [section 3b(2)].

1) DATA Eddy covariance flux tower observations have the potential to reduce uncertainties in model parameterizations.

7645

To constrain the PFT parameters, we optimize simulations of latent and sensible fluxes against measured time series of these fluxes from the Ameriflux (http://public. ornl.gov/ameriflux/) site (358N, 798W) in the Duke loblolly forest (Stoy et al. 2006; Novick et al. 2009). Continuous half-hourly observations of ecosystem level exchanges of energy, momentum, and water are available from 1998 to 2009. Since this site experienced a strong drought from 2001 to 2002 and data is limited before this period, we use data from 2003 to 2005 in our inversion method. Observed monthly LAI at the Duke Forest loblolly Ameriflux site is taken from the Global Inventory Modeling and Mapping Studies (GIMMS) dataset, which is a normalized difference vegetation index (NDVI) product that covers the period from 1981 to 2006 (Tucker et al. 2004). The GIMMS dataset is derived from the Advanced Very High Resolution Radiometer (AVHRR) instrument imagery. In the following simulations, LP trees have a prescribed LAI that is set to the observed monthly-mean Duke loblolly forest values averaged between 2003 and 2005, unless otherwise noted. In the absence of observations, LP SAI is set to the default NET values.

2) OPTIMIZATION METHOD AND IMPLEMENTATION

We generate ensembles of simulations by running CLM4 in offline mode with the new LP PFT forced with observed meteorology and other surface conditions appropriate to this site. The members of the ensemble are distinguished by perturbed parameters in the PFT. The PFT parameters are then optimized by minimizing the difference between simulated and observed fluxes. From diurnal to seasonal time scales, vegetation exerts a biophysical effect on climate through the partitioning of net radiation into sensible and latent heat fluxes. Better representations of the seasonal changes in photosynthesis and stomatal conductance are critical for modeling energy fluxes of ecosystems (Xu and Baldocchi 2003). Since we are interested in the biophysical climate feedbacks from LP establishment, we focus our optimization on predicted sensible and latent heat fluxes. We create a 36-member ensemble by changing two parameters that influence stomatal conductance through the photosynthesis calculation: m (mp), the slope of conductance to photosynthesis relationship (default m 5 6), and f (flnr), the fraction of leaf N in the Rubisco enzyme (default f 5 0.05). For the inversion, we choose six values of each to span the reasonable parameter space: m 5 5, 6, 7, 8, 9, and 10 and f 5 0.05, 0.06, 0.07, 0.08, 0.09, and 1. We run CLM in site mode with atmospheric forcing provided from the Duke loblolly forest flux tower. Each

7646

JOURNAL OF CLIMATE

ensemble run is integrated for 11 yr with forcing data from 1998 to 2008. We disregard the first 5 yr as a spinup period and use the daytime [1000–1600 local time (LT)] data from 2003 to 2005 to optimize the model parameters. We choose the daytime period because it contributes the dominant diurnal variability in energy fluxes to the atmosphere. There are several methods to minimize the difference between observations and modeled output (Hasselmann 1998; Herbei et al. 2008; Murphy et al. 2004; Ricciuto et al. 2008). Here, we use a Bayesian inference method to characterize the uncertainty in climate model parameters. According to the Bayes theorem, the posterior probability is proportional to the likelihood of the data given the parameters multiplied by the prior probability of the parameters. For more details on the methodology, the reader can refer to Goes et al. (2010). We draw 10 000 samples of the posterior probability distribution through a Markov chain Monte Carlo (MCMC) method. We estimate two model parameters (m and f) given the observations of sensible and latent heat flux using the climatological-mean average daytime energy flux observations based on the period 2003–05. As an integral part of our normal multivariate likelihood function, we estimate three additional statistical parameters: the bias coefficient, correlation length, and standard deviation. In the absence of a clear alternative, we choose uniform priors for the model parameters and for the bias and correlation length statistical parameters. This method accounts for cross correlation between the (measured– modeled) residual time series and for autocorrelation in each of the residual series. Neglecting autocorrelation in the residuals can result in biased and overconfident parameter estimates. For the cross-covariance matrix, we use an inverse Wishart prior following Goes et al. (2010). The fact that we find a correlation value close to 1 and an autocorrelation value near zero implies that the residuals can be treated independently without the loss of much information. We jointly optimize the model and statistical parameters to calculate the joint posterior probability density function (PDF). The final parameter set chosen is based on median of the joint posterior probability values. The inversion results for the parameters give the following values: m 5 10 and f 5 0.05 (see Table 2 for the properties of the statistical distributions). Results are described in section 4a below.

c. Experimental design We perform coupled simulations using prescribed sea surface temperatures (SSTs) to investigate the impact of land-cover change in the SUS with atmospheric feedbacks. All simulations described in this study use prescribed PFT coverage and omit N cycling impacts. We

VOLUME 25

TABLE 2. Properties of the statistical distributions [median and 95% credible interval (CI)] of the parameter for each considered source of information (LHFLX and SHFLX) and the posterior (joint distribution considering all sources of information). Parameters m

f

Obs

Median

95% CI

Median

95% CI

LHFLX SHFLX Posterior

10 7 10

1.93 0.04 2.89

0.05 0.06 0.05

0.96 0.03 0.01

use the standard 28 CCSM4 resolution (1.98 in latitude and 2.58 in longitude) and 26 levels in the vertical. We develop three present-day (PD) climate scenarios and one future scenario that are all forced with presentday concentrations of greenhouse gases. The scenarios are 1) a control simulation that uses the default NET tree PFT (control); 2) replacement of all NET in the SUS (135 million acres; 33% of SUS) with the new optimized LP PFT (using optimized m and f values) with the default LAI values for NET (PD loblolly old LAI); 3) same as scenario 2 but utilizing the monthly Duke Forest loblolly LAI values at all grid cells with LP (PD loblolly); and 4) same as scenario 3, with an additional conversion of 50 million acres (12% of SUS) of C4 grasslands to LP trees throughout the SUS (future loblolly). Afforestation of C4 grasslands occurs along the southern and western edges of our domain. While our PD loblolly scenarios have more loblolly pine than the most recent estimates, it was our decision, naturally subjective, to produce experiments that are aligned with the model baseline. Each simulation was 60 yr long. We discard the first 20 yr of our simulations as a spinup period and average over the last 40 yr to compute the climatology and statistical significance. We characterize the statistical significance of differences between our simulations using the Student’s t test with the null hypothesis that the differences are indistinguishable from unforced internal variability in the same fields. The spatial distribution and magnitude of the internal variability are derived from the control simulations. The significance is then evaluated at each grid point by setting an upper bound p on the probability that unforced variability could generate the time-mean differences obtained at that point. In our figures, statistically significant differences with p # 0.05 are stippled. This method does not take into account how many grid cells would be considered significant 1 out of 20 times (5%) by random chance (Livezey and Chen 1983). A more rigorous statistical test would account for autocorrelation within the time series, which can reduce the number of false positives (Zwiers and von Storch 1995; von Storch

1 NOVEMBER 2012

MURPHY ET AL.

7647

and Zwiers 1999). Yet this test still would not account for spatial correlation within fields (von Storch and Zwiers 1999). Geopotential height, in particular, has long correlation length scales (Findell et al. 2009) indicating that height anomalies in adjacent grid cells may not be independent. Rather than employ Bayesian methodology to test field significance (Livezey and Chen 1983), one can more easily use fewer degrees of freedom (Van den Dool and Chervin 1986; Wang and Shen 1999; Findell et al. 2009). Here we examine whether our estimates of statistical significance depend on the sampling period by comparing the statistically significant results at different time periods within the last 40 yr of each simulation (similar to Subin et al. 2012). This allows us to discuss only the significant changes that were consistent among all time slices.

3. Results a. Loblolly pine parameterization The generic NET parameterization used in CLM substantially overestimates sensible heat flux and underestimates latent heat flux at diurnal, monthly, and seasonal time scales. We calculated the average daytime (1000– 1600 LT) observations of energy fluxes averaged over the years 2003–05. These values are constrained by our ensemble spread (Fig. 1). The new PFT parameter choices lead to a substantial improvement in the seasonal sensible and latent heat energy budgets (Fig. 1). Compared to the default NET (control), LP trees decrease sensible and increase latent heat flux. We performed a global CLM offline run forced by National Centers for Environmental Prediction (NCEP) reanalysis data and replaced NET in the SUS with our new optimized LP [parameter choices are described in section 2b(2)]. Greater evapotranspiration reduces the near surface atmospheric temperature. Averaged across the SUS region, we predict a cooling of 0.28–0.758C due solely to these parameter changes (m and f). Greater evapotranspiration within the canopy substantially reduces monthly-mean ground evaporation by up to 8 W m22 and consequently decreases soil moisture during summer.

FIG. 1. The average daytime (1000–1600 LT) (top) SHFLX and (bottom) LHFLX (W m22) averaged over the period 2003–05. Individual ensemble members are shown in gray, the observations are shown in red, a run with default CLM parameters is shown in blue, and the best ensemble member predicted from the Bayesian inversion method is shown in green.

b. Local changes 1) NET ENERGY BUDGET We examine the difference in the energy budget terms (net radiation, sensible heat flux, latent heat flux, and ground heat flux) averaged over the SUS for our future loblolly, control, and PD loblolly scenarios (Fig. 2). Substituting the current distribution of LP for NET in the SUS does not substantially alter the annual net radiation but does lead to monthly differences as large as 1.5 W m22. Replacing grasslands with LP plantations

(future loblolly) leads to increases in net radiation as large as 3–4 W m22 in the spring and 3 W m22 in July. Afforestation replaces bright grassland with darker forests that absorb more incoming radiation. The SUS annualmean net radiation is 118.7, 118.4, and 120.6 W m22 in the control, PD loblolly, and future loblolly scenarios, respectively. A change in radiant heating at the surface can alter sensible and latent heat partitioning, which can in turn

7648

JOURNAL OF CLIMATE

VOLUME 25

FIG. 2. The average monthly difference between the future loblolly scenario and our control (red) and the PD loblolly and control (blue) for (a) net radiation, (b) SHFLX, (c) LHFLX, and (d) ground heat flux.

impact boundary layer energetics and structure. Slightly lower levels of net radiation decreases the sensible heat fluxes in the PD loblolly scenario. However, increased annual transpiration in LP compared to NET increases the latent heat flux from 63.3 W m22 in the control scenario to 66.4 W m22 in the PD loblolly scenario and 66.2 W m22 in the future loblolly scenario averaged across the SUS. The largest difference, in both the PD and future loblolly scenarios, occurs during the summer when the LAI reaches its maximum value. The future loblolly scenario results in an annually averaged increase in latent heat flux of ;3 W m22 and a decrease in sensible heat flux of ;1 W m22. The PD loblolly simulation increases (decreases) latent (sensible) annual heat fluxes by ;3 W m22. The reduction in sensible heat flux is larger in the PD loblolly scenario than in the future

loblolly scenario since the LAI is much lower for grass than for forest. Our future expansion scenario replaces grasslands with forests, which increases the LAI and surface roughness and therefore increases canopy evapotranspiration. The ground heat flux remains effectively unchanged in all simulations.

2) EFFECT OF LAI CHANGES VERSUS PFT PHYSIOLOGY CHANGES

In this section, we examine the impact of the imposed LAI on climate by comparing the anomalies with respect to the control. The LAI of LP is lower during winter and higher by almost a factor of 2 during the summer compared to NET. The local differences in energy fluxes between the PD loblolly (old LAI) and PD loblolly and the control are largely insignificant (Figs. 3a,b, 4a,b versus

1 NOVEMBER 2012

MURPHY ET AL.

7649

FIG. 3. The mean DJF (left) sensible and (right) LHFLX anomalies (W m22) for (a),(b) PD loblolly (old LAI) minus control, (c),(d) PD loblolly minus control, and (e),(f) future loblolly minus control. Shaded grid cells represent statistically significant cells (using a Student’s t test with p 5 0.05).

Figs. 3c,d, 4c,d), indicating that changes in the energy fluxes are due to the PFT physiology changes (LP appropriate m and f values) and not the LAI changes. Although summertime [June–August (JJA)] differences in SHFLX (Figs. 4a,c) and LHFLX (Figs. 4b,d) are similar in magnitude, wintertime [December–February (DJF)] differences are larger over the SUS with the imposed LP LAI (PD loblolly) (Figs. 3a,c). LHFLX changes over the oceanic regions north of Australia and the North Pacific show similar responses in both the PD loblolly (old LAI) and PD loblolly simulations (Figs. 3b,d, 4b,d). The differences (with respect to the control) in humidity (Fig. 5), clouds (Fig. 6), and 2-m air temperature (Fig. 7) show regions where the atmospheric response is affected by changes in LAI (Figs. 5a,b, 6a,b, 7a,b versus Figs. 5c,d, 6c,d, 7c,d). In particular, atmospheric properties in the high-latitude regions of both the Northern and Southern Hemisphere respond differently to the new LP LAI. However, these regions typically have very

high interannual variability in temperature and cloud cover, and therefore the difference in responses may not be statistically significant. The PD loblolly (old LAI) scenario results in lower 2-m air temperature and 2-m relative humidity during winter over the SUS (Figs. 5a, 7a). During the summer, there is higher humidity over western Eurasia and the Middle East and a 0.58C increase in the 2-m air temperature over northern Canada (Figs. 5b, 7b). These changes are not observed in our PD loblolly scenario (Figs. 5c,d, 7c,d).

3) LAND–ATMOSPHERE RESPONSE In this section, we examine the atmospheric response to replacing NET with LP (PD loblolly) and afforesting C4 grasslands in the SUS (future loblolly). Our PD loblolly simulation has the largest impact on the SUS humidity, clouds, and temperature during the summer compared to the control (Figs. 5d, 6d, 7d) because the

7650

JOURNAL OF CLIMATE

VOLUME 25

FIG. 4. As in Fig. 3, but for the mean JJA.

PFT physiology changes have the largest impact on summer energy fluxes (Figs. 3d, 4d) and because the difference in LAI between LP and NET is largest in the summer. These differences result in substantial changes in the temperature and humidity of the boundary layer. In JJA, afforestation and higher LAI (due to the grassland conversion) in the future loblolly scenario increases the interception of precipitation by the canopy by 0.08 mm day21. Most of this water evaporates from the leaf surfaces before reaching the ground, resulting in lower ground evaporation, runoff, and water storage. Summertime canopy transpiration increases over the control in both the PD (11 W m22) and future (15 W m22) loblolly scenarios (not shown). A statistically significant increase in the 2-m relative humidity occurs during summer in all three scenarios over the SUS with up to 5% higher humidity over Georgia and South Carolina (Fig. 5). Humidity increases together with evapotranspiration. The summertime PD loblolly scenario shows a broader area of increased humidity

extending eastward from the Rocky Mountains compared to the future loblolly scenario in which the area of increased humidity is confined to the SUS (Figs. 5d,f). In the future loblolly scenario, anticyclonic rotation over the SUS brings moist air from the Gulf Stream inland, whereas in the PD loblolly scenario more moisture is directed from the Gulf of Mexico into the central plains and results in increased humidity in the interior United States. Replacing grasslands with forest in the future loblolly scenario reduces the surface albedo and therefore contributes to an increase in net radiation at the surface (Fig. 2a). This effect is reduced in part due to an increase in low-level cloud cover that effectively limits radiation at the surface. More moisture in the boundary layer increases the summertime low-level cloud cover over the SUS by ;2%–3%: that is, a 25% relative change in the regional cloud cover (Fig. 6). The largest increase over the SUS is predicted in the future loblolly scenario. Increased low-level clouds reduce solar heating at the surface, and this lower insolation in turn cools the lower

1 NOVEMBER 2012

MURPHY ET AL.

7651

FIG. 5. (left) DFJ and (right) JJA 2-m relative humidity anomalies (percent) for (a),(b) PD loblolly (old LAI) minus control, (c),(d) PD loblolly minus control, and (e),(f) future loblolly minus control. Shaded grid cells represent statistically significant cells (using a Student’s t test with p 5 0.05).

atmosphere by 0.58C between the surface and 600 hPa (not shown). Over the SUS in all three scenarios, the local 2-m air temperature is significantly reduced by up to 18C during the summer (Figs. 7d,f) and by 0.58C in the annual average. The maximum daily temperature is lower in the summer for both the PD and future loblolly scenarios, while the minimum daily temperature remains the same in the PD loblolly and decreases by 0.58C in the future scenario. Cooling stabilizes the lower atmosphere and reduces the height of the planetary boundary layer by 50 m (not shown). Since cooler air holds less water vapor, we would expect an increase in precipitation rate over the SUS in the PD and future loblolly scenarios (Fig. 8b). As expected, precipitation increases by 0.75 mm day21 over the central plains and by larger amounts over the Gulf of Mexico in summer (Figs. 8b,d) in both scenarios. Increased LHFLX also contributes to

increased precipitation over the central plains in summer. In winter, there is a significant increase in precipitation over the eastern United States in the future loblolly scenario associated with a small increase in SUS LHFLX at the expense of SHFLX.

c. Remote changes 1) NORTHERN HEMISPHERE CHANGES In winter, the zonal-mean 2-m air temperature is reduced by 0.258C between 408 and 608N (Figs. 7c,e). Cooling is strongest from Eastern Europe through Russia. During the summer, cooling occurs over western Canada and is largest in the future loblolly scenario (Fig. 7f). This cooling is associated with greater lower-level cloud cover, which reduces the shortwave radiation at the surface. Higher geopotential heights (not shown) and increased sea level pressure (;2 hPa) are predicted over

7652

JOURNAL OF CLIMATE

VOLUME 25

FIG. 6. As in Fig. 5, but for low-level cloud cover (percent).

the Gulf of Alaska and Bering Sea during the winter in our PD and future loblolly scenarios (Fig. 9). The Aleutian and Icelandic lows are both semipermanent lows through which synoptic-scale low pressure systems frequently pass and intensify (Zhu et al. 2007). These centers have significant effects on the atmospheric circulation in the Northern Hemisphere. We predict a relatively weakened Aleutian low. A slight northward shift in the storm track is inferred from an increase in the eddy kinetic energy (EKE) (a diagnostic for storminess) poleward of 608N (Figs. 10b,c). Typically, cyclonic rotation carries moist air from the Gulf of Alaska inland, where it ascends at the southern edge of the mountains resulting in cloud formation. A reversal in the circulation results in less low-level cloud cover over Alaska and western Canada and more low-level cloud cover over the Gulf of Alaska (Figs. 6c,e). In winter, the future loblolly scenario shows increased sea level pressure near the Azores and decreased sea

level pressure near Iceland. This pattern is indicative of the positive phase of the North Atlantic Oscillation (NAO), which is a dominant pattern of large-scale variability in the Northern Hemisphere. An increase in precipitation over the western Sahel and a small decrease over the Saharan desert occurs in both our PD and future loblolly scenarios (Figs. 8b,d) and results from changes in atmospheric circulation. This feature is consistent with greater upper-level divergence seen in the velocity potential field, a measure of atmospheric divergence (Fig. 11). The velocity potential, which is proportional to the irrotational component of the wind field, is a measure of divergence and convergence of the large-scale atmospheric circulation. Negative (positive) velocity potential anomalies correspond to regions of enhanced (suppressed) convection. In summer, upper-level atmospheric divergence induces greater vertical velocity over the western Sahel (Figs. 11b,d).

1 NOVEMBER 2012

MURPHY ET AL.

7653

FIG. 7. As in Fig. 5, but for 2-m air temperature (kelvins). The zonally averaged 2-m temperature anomaly is to the right of each plot.

2) SOUTHERN HEMISPHERE CHANGES Vorticity transport by a perturbed atmospheric divergent circulation can result in teleconnections between tropics and extratropical regions (Sardeshmukh and Hoskins 1987). The PD and future scenarios result in increased velocity potential over the eastern Indian Ocean in winter, indicating suppressed convection (Figs. 11a,c). There is a notable shift in precipitation from the tropics to the extratropics in the eastern Indian Ocean and western Pacific. A decrease of 1 mm day21 occurs over the eastern Indian Ocean in both PD [PD loblolly and PD loblolly (old LAI)] and future loblolly simulations (Fig. 8). The winter precipitation rate increases by similar magnitude over Australia. A decrease in the 2-m air temperature occurs over southern Australia (Figs. 7c,e) and coincides with lower SHFLX (Figs. 3c,e), higher LHFLX (Figs. 4c,e), and increased relative humidity over Australia (Figs. 5c,e).

4. Discussion We break new ground in analyzing the climate impacts of bioenergy crops by parameterizing for an important bioenergy tree crop in the SUS and implementing it in a state-of-art GCM. Our main focus is on the biogeophysical feedbacks resulting from the implementation of new trees and conversion of grasslands to forests. Typically in middle to high latitudes afforestation has a warming effect on climate due to increased solar absorption resulting from albedo changes (Gibbard et al. 2005; Bonan 2008). However, as shown in this study, LP trees have higher evapotranspiration than NET trees and grasslands, which causes regional cooling. Additional climate feedbacks could arise from BGC processes, such as changes in carbon fluxes, with sizeable climate impacts on regional and global scales. For instance, LP trees are more productive than other forest types, and therefore their establishment may act as an important

7654

JOURNAL OF CLIMATE

VOLUME 25

FIG. 8. (left) DJF and (right) JJA total precipitation (convection and large scale) anomalies (mm day21) for (a),(b) PD loblolly minus control and (c),(d) future loblolly minus control. Shaded grid cells represent statistically significant cells (using a Student’s t test with p 5 0.05).

regional sink of atmospheric carbon (Hedenus and Azar 2009). In opposition, harvest would decrease carbon sequestration rates and may also act as a source of carbon to the atmosphere (Depro et al. 2008). The net effect of the life cycle of carbon in loblolly trees is dependent on several factors, namely, site quality, rotation length, management practices, and physiographic region. While there are several studies on this subject, the inclusion of biomass harvest for fossil fuel offset has yet to be considered. Therefore, additional research is needed to characterize the full life cycle. The resulting spatial and temporal heterogeneity in stand height, albedo, and LAI would alter the climate through changes in the energy budget, hydrology, and atmospheric circulation and indirectly affect the BGC cycle. We did not account for elevated CO2 in our future loblolly scenario. However, increased carbon uptake by new LP plantations may reduce some of the expected warming. Arora and Montenegro (2011) found global cooling in their partial and complete afforestation scenarios in which croplands were converted to forests in conjunction with a continuous increase in emissions. While we did not account for these nontrivial anthropogenic influences, we believe this study provides avenues for future research.

other forest types (see Sun et al. 2010, their Table 5). A higher albedo results in more reflection of incoming radiation and possibly more surface cooling. Although albedo is prognosed in the model, the actual cooling from the introduction of LP plantations would be larger than what is predicted in our scenarios since we have retained the default NET leaf properties for LP. This albedo effect may offset more anthropogenic warming over the eastern United States. In our simulations, replacing grasslands with LP plantations (future loblolly scenario) reduces the annual-mean albedo of the SUS from 0.12 to 0.11.

b. Local changes Our predicted local decrease in SHFLX and increases in LHFLX associated with conversion of C4 grasses to LP are consistent with observations contrasting European grasslands and forests (Teuling and Seneviratne 2011). Differences in available energy, canopy roughness, and the timing of physiological functioning explain the observed differences in SHFLX and LHFLX of the contrasting vegetation surfaces. The higher summer LAI and greater evapotranspiration that are characteristic of LP act to cool the eastern United States.

a. Albedo

c. Remote changes

Depending on the age of the stand, LP albedo ranges from 0.25 to 0.31, which is larger than typical albedos of

Statistically significant changes in atmospheric circulation throughout the Northern Hemisphere suggest that

1 NOVEMBER 2012

MURPHY ET AL.

7655

FIG. 9. Northern Hemisphere polar stereographic projection of (left) DJF and (right) JJA sea level pressure anomalies (hPa) for (a),(b) PD loblolly minus control and (c),(d) future loblolly minus control. Shaded grid cells represent statistically significant cells (using a Student’s t test with p 5 0.05).

expanding loblolly forests in the SUS will affect remote climate through atmospheric teleconnections. In our simulations, replacing NET with LP leads to a zonally averaged cooling of 0.258C between 408 and 608N and an increase in the winter surface pressure over the Aleutian low. A change in the position and intensity of the Aleutian low has implications for the storm track over the North Pacific (Salathe´ 2006). The increase in the zonally averaged EKE, a diagnostic for storminess (Lau 1988; Chang et al. 2002) in the poleward direction, implies a corresponding poleward shift in the North Pacific storm track in our simulations.

Our results suggest regional land-use change in the SUS may impact the dominant modes of variability in the Northern Hemisphere, such as the NAO (Wallace and Gutzler 1981; Hurrell et al. 2003). During the winter, our future loblolly simulation produces an increase in surface pressure near the Azores and a decrease in pressure near Iceland. This pattern suggests an expansion of LP plantations in the SUS might favor more positive phase NAO events. While our predicted wind and temperature fields are in agreement with observed changes associated with positive NAO events (Hurrell 1995; Hurrell and Deser 2009), we do not predict a wetter Europe.

7656

JOURNAL OF CLIMATE

VOLUME 25

The suppression of precipitation over the Indian Ocean may be related to dynamical reductions in the synoptic upper-level atmospheric divergence. However, the increased precipitation over Australia is apparently due to increased LHFLX flux at the expense of SHFLX flux over Australia. Increased evaporational cooling reduces the 2-m air temperature and is accompanied by higher 2-m humidity. Although low-level cooling could stabilize the atmospheric boundary layer and inhibit convection, our results show the elevated humidity enhances cloud growth and convective precipitation. Regional-scale perturbations over the SUS may influence remote locations through the advection of vorticity by the perturbations in the large-scale circulation. We hypothesize that this mechanism may be responsible for the change in energy fluxes over Australia.

d. Potential for storm damage One billion tons of biomass is required annually in order to reach the goal of replacing 30% of our current petroleum consumption with biofuels by 2030 (Perlack et al. 2005). Under a high crop yield scenario, 55 million acres could be available from cropland, idle cropland, and pastureland to grow perennial crops. In our future loblolly scenario, we assume 50 million acres of grassland in the SUS are converted to LP plantations. This change represents 90% of the total land area available to grow perennial crops. However, many mechanisms could impact the viability and productivity of such a change, including changes in climate and extreme climate events. There is appreciable risk of forest disturbance in the SUS due to a variety of mechanisms including a particularly high susceptibility to hurricanes. In 2005, Hurricane Katrina resulted in a loss of between 43.9 6 8.4 Tg C (Negro´nJua´rez et al. 2010) and 105 Tg C (Chambers et al. 2007) because of forest destruction.

e. Caveats

FIG. 10. EKE (m2 s22) in DJF. The changes shown are for zonaland time-mean DJF kinetic energy for (a) the control, (b) PD loblolly minus control, and (c) future loblolly minus control. Shaded grid cells represent statistically significant cells (using a Student’s t test with p 5 0.05).

Our predicted climate impact of expanding bioenergy production in the SUS may be altered because of the particulars of our simulations. First, the imposed leaf properties of the LP may result in a land surface that is too dark and a weaker predicted cooling. Second, our simulations use fixed sea surface temperatures and therefore constrain the effects of land-cover change. Using a slab ocean model rather than prescribed SSTs would allow the mixed layer temperature to adjust to changes in predicted fluxes, rather than require large changes in air–sea fluxes to hold SST constant. Future work will include fully coupled simulations to examine this effect. In disagreement with recent estimates based on forest inventory (Smith et al. 2009), CLM has an unrealistically

1 NOVEMBER 2012

MURPHY ET AL.

7657

FIG. 11. As in Fig. 8, but for the velocity potential (m2 s21).

high amount of needleleaf evergreen trees at the expense of broadleaf deciduous trees in the SUS. It was our decision, naturally subjective, to produce experiments that are aligned with the model baseline. In this idealistic scenario, we replace all needleleaf evergreen trees with loblolly pines, therefore overestimating the number loblolly pines in the SUS. Additionally, constraints from disturbance, soil and land resources, and larger-scale climate changes may also limit the total areal conversion expected. Although these scenarios are somewhat extreme, they provide an upper limit to the potential climate changes arising from the increased forest management and continued growth of biofuels. We predict a 0.758 and 18C cooling in our respective PD and future loblolly scenarios. In our scenarios, including the future loblolly, greenhouse gas concentrations remain fixed at year 2000 levels. The A1B scenario (Christensen 2007) predicts a 18C warming over much of the SUS due to anthropogenic warming associated with increasing greenhouse gas concentrations by 2029. Therefore, our expected cooling from LP plantations may be overcome by anthropogenic warming within the next two decades. It is important to note that the southeastern United States is a biogenic volatile organic compound (BVOC) emission hot spot. BVOC emissions from forests produce harmful tropospheric ozone that forms a cooling haze in the summer (Goldstein et al. 2009). Afforestation may increase BVOC emission rates and exacerbate regional cooling in the future.

5. Conclusions We have developed a new CLM4 plant functional type for LP, an important biofuel feedstock crop and also the most planted forest species in the United States, in order to study the climatic effects from large-scale introduction of these trees. With our new PFT parameterization, we examine the future climate impacts of expanding woody crops, specifically LP plantations, in the SUS. We find that a plausible, regional introduction of biofuels has both local and remote climate impacts. Vegetation can influence climate through biophysical effects. LP trees influence atmospheric processes differently than the default NET through its effect on partitioning net radiation into LHFLX and SHFLX. Our results show that LP trees increase LHFLX and decrease SHFLX compared to the default NET. Increased LHFLX is largely due to increased canopy transpiration. Greater evapotranspiration reduces the near-surface (2 m) air temperature by more than 18C in our PD and future loblolly simulations. This finding suggests that there may be a regional bias in CCSM4, since the default PFT in its CLM4 land component does not accurately represent these dominant trees. Our results show that future afforestation with LP under an extreme scenario of expansion cools the Northern Hemisphere midlatitudes, weakens the Aleutian low, shifts the North Pacific storm track poleward, and has implications on Australian hydroclimate via atmospheric

7658

JOURNAL OF CLIMATE

teleconnections. It is clear that the biophysical responses to our imposed land-cover change can have large impacts on local and global climate and that these effects need to be considered in analyses of biofuel feedstock production. Acknowledgments. This research was supported by the Director, Office of Science, Office of Biological and Environmental Research of the U.S. Department of Energy under Contract DE-AC02-05CH11231 as part of their Climate and Earth System Modeling Program. REFERENCES Arora, V. K., and A. Montenegro, 2011: Small temperature benefits provided by realistic afforestation efforts. Nat. Geosci., 4, 514–518. Betts, R. A., 2000: Offset of the potential carbon sink from boreal forestation by decreases in surface albedo. Nature, 408, 187– 190. ——, P. D. Falloon, K. K. Goldewijk, and N. Ramankutty, 2007: Biogeophysical effects of land use on climate: Model simulations of radiative forcing and large-scale temperature change. Agric. For. Meteor., 142, 216–233. Bonan, G. B., 1997: Effects of land use on the climate of the United States. Climatic Change, 37, 449–486. ——, 2008: Forests and climate change: Forcings, feedbacks, and the climate benefits of forests. Science, 320, 1444–1449. ——, D. Pollard, and S. L. Thompson, 1992: Effects of boreal forest vegetation on global climate. Nature, 359, 716–718. Chambers, J. Q., J. I. Fisher, H. C. Zeng, E. L. Chapman, D. B. Baker, and G. C. Hurtt, 2007: Hurricane Katrina’s carbon footprint on U. S. Gulf Coast forests. Science, 318, 1107–1107. Chang, E. K. M., S. Y. Lee, and K. L. Swanson, 2002: Storm track dynamics. J. Climate, 15, 2163–2183. Christensen, J. H., and Coauthors, 2007: Regional climate projections. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 847–940. Depro, B. M., B. C. Murray, R. J. Alig, and A. Shanks, 2008: Public land, timber harvests, and climate mitigation: Quantifying carbon sequestration potential on U.S. public timberlands. For. Ecol. Manage., 255, 1122–1134. DOE, 2003: Roadmap for agricultural biomass feedstock supply in the United States. U.S. Department of Energy Document DOE/NE-ID-11129, 100 pp. [Available online at http://www. inl.gov/technicalpublications/Documents/3323197.pdf.] Eckert, A. J., J. van Heerwaarden, J. L. Wegrzyn, C. D. Nelson, J. Ross-Ibarra, S. C. Gonzalez-Martinez, and D. B. Neale, 2010: Patterns of population structure and environmental associations to aridity across the range of loblolly pine (Pinus taeda L., Pinaceae). Genetics, 185, 969–982. EIA, 2008: Renewable energy annual 2006—Renewable energy trends in consumption and electricity. U.S. DOE Energy Information Administration Rep., 76 pp. ——, cited 2012: U.S. oxygenate production. U.S. DOE Energy Information Administration. [Available online at http://www. eia.gov/totalenergy/data/annual/showtext.cfm?t5ptb1003.] Fall, S., D. Niyogi, A. Gluhovsky, R. A. Pielke, E. Kalnay, and G. Rochon, 2010: Impacts of land use land cover on temperature trends over the continental United States: Assessment using the North American Regional Reanalysis. Int. J. Climatol., 30, 1980–1993.

VOLUME 25

Feddema, J. J., K. W. Oleson, G. B. Bonan, L. O. Mearns, L. E. Buja, G. A. Meehl, and W. M. Washington, 2005: The importance of land-cover change in simulating future climates. Science, 310, 1674–1678. Findell, K. L., A. J. Pitman, M. H. England, and P. J. Pegion, 2009: Regional and global impacts of land cover change and sea surface temperature anomalies. J. Climate, 22, 3248– 3269. Frederick, W. J., Jr., S. J. Lien, C. E. Courchene, N. A. DeMartini, A. J. Ragauskas, and K. Iisa, 2008: Production of ethanol from carbohydrates from loblolly pine: A technical and economic assessment. Bioresour. Technol., 99, 5051–5057. Gent, P. R., and Coauthors, 2011: The Community Climate System Model version 4. J. Climate, 24, 4973–4991. Georgescu, M., D. B. Lobell, and C. B. Field, 2009: Potential impact of US biofuels on regional climate. Geophys. Res. Lett., 36, L21806, doi:10.1029/2009GL040477. ——, ——, and ——, 2011: Direct climate effects of perennial bioenergy crops in the United States. Proc. Natl. Acad. Sci. USA, 108, 4307–4312. Gibbard, S., K. Caldeira, G. Bala, T. J. Phillips, and M. Wickett, 2005: Climate effects of global land cover change. Geophys. Res. Lett., 32, L23705, doi:10.1029/2005GL024550. Goes, M., N. M. Urban, R. Tonkonojenkov, M. Haran, A. Schmittner, and K. Keller, 2010: What is the skill of ocean tracers in reducing uncertainties about ocean diapycnal mixing and projections of the Atlantic meridional overturning circulation? J. Geophys. Res., 115, C12006, doi:10.1029/2010JC006407. Goldstein, A. H., C. D. Koven, C. L. Heald, and I. Fung, 2009: Biogenic carbon and anthropogenic pollutants combine to form a cooling haze over the southeastern US. Proc. Natl. Acad. Sci. USA, 106, 8835–8840. Hall, D. O., H. E. Mynick, and R. H. Williams, 1991: Cooling the greenhouse with bioenergy. Nature, 353, 11–12. Hargrove, W. W., and F. M. Hoffman, 2004: The potential of multivariate quantitative methods for delineation and visualization of ecoregions. Environ. Manage., 34, S39–S60. Hasselmann, K., 1998: Conventional and Bayesian approach to climate-change detection and attribution. Quart. J. Roy. Meteor. Soc., 124, 2541–2565. Hedenus, F., and C. Azar, 2009: Bioenergy plantations or longterm carbon sinks?—A model based analysis. Biomass Bioenergy, 33, 1693–1702. Herbei, R., I. W. McKeague, and K. G. Speer, 2008: Gyres and jets: Inversion of tracer data for ocean circulation structure. J. Phys. Oceanogr., 38, 1180–1202. Huang, C. H., G. D. Kronrad, and S. L. Cheng, 2004: Economic analyses of sequestering carbon in loblolly pine, cherrybark oak, and northern red oak in the United States. Environ. Manage., 33, S187–S199. Hurrell, J. W., 1995: Decadal trends in the North Atlantic Oscillation, regional temperatures and precipitation. Science, 269, 676–679. ——, and C. Deser, 2009: North Atlantic climate variability: The role of the North Atlantic Oscillation. J. Mar. Syst., 78, 28–41. ——, Y. Kushnir, G. Ottersen, and M. Visbeckand, 2003: The North Atlantic Oscillation: Climatic Significance and Environmental Impact. Geophys. Monogr., Vol. 134, Amer. Geophys. Union, 279 pp. Kline, K. L., and M. D. Coleman, 2010: Woody energy crops in the southeastern United States: Two centuries of practitioner experience. Biomass Bioenergy, 34, 1655–1666.

1 NOVEMBER 2012

MURPHY ET AL.

Lau, N. C., 1988: Variability of the observed midlatitude storm tracks in relation to low-frequency changes in the circulation pattern. J. Atmos. Sci., 45, 2718–2743. Livezey, R. E., and W. Y. Chen, 1983: Statistical field significance and its determination by Monte-Carlo techniques. Mon. Wea. Rev., 111, 46–59. Loarie, S. R., D. B. Lobell, G. P. Asner, Q. Mu, and C. B. Field, 2011: Direct impacts on local climate of sugar-cane expansion in Brazil. Nat. Climate Change, 1, 105–109, doi:10.1038/ nclimate1067. McCarthy, H. R., R. Oren, A. C. Finzi, D. S. Ellsworth, H.-S. Kim, K. H. Johnsen, and B. Millar, 2007: Temporal dynamics and spatial variability in the enhancement of canopy leaf area under elevated atmospheric CO2. Global Change Biol., 13, 2479–2497. Murphy, J. M., D. M. H. Sexton, D. N. Barnett, G. S. Jones, M. J. Webb, and M. Collins, 2004: Quantification of modelling uncertainties in a large ensemble of climate change simulations. Nature, 430, 768–772. Neale, R. B., and Coauthors, 2010: Description of the NCAR Community Atmosphere Model (CAM4). NCAR Tech. Rep. NCAR/TN-4851STR, 224 pp. [Available online at http:// www.cesm.ucar.edu/models/ccsm4.0/cam/docs/description/ cam4_desc.pdf.] Negro´n-Jua´rez, R., D. B. Baker, H. C. Zeng, T. K. Henkel, and J. Q. Chambers, 2010: Assessing hurricane-induced tree mortality in U.S. Gulf Coast forest ecosystems. J. Geophys. Res., 115, G04030, doi:10.1029/2009JG001221. Novick, K., R. Oren, P. Stoy, M. Siqueira, and G. G. Katul, 2009: Nocturnal evapotranspiration in eddy-covariance records from three co-located ecosystems in the southeastern U.S.: Implications for annual fluxes. Agric. For. Meteor., 149, 1491– 1504. Perlack, R. D., L. L. Wright, A. F. Turhollow, R. L. Grahan, B. J. Stokes, and D. C. Erbach, 2005: Biomass as a feedstock for a bioenergy and bioproducts industry: The technical feasibility of a billion-ton annual supply. U.S. DOE Oak Ridge National Laboratory Rep. DOE/GO-102005-2135, 78 pp. Ricciuto, D. M., K. J. Davis, and K. Keller, 2008: A Bayesian calibration of a simple carbon cycle model: The role of observations in estimating and reducing uncertainty. Global Biogeochem. Cycles, 22, GB2030, doi:10.1029/2006GB002908. Salathe´, E. P., Jr., 2006: Influences of a shift in North Pacific storm track on North American precipitation under global warming. Geophys. Res. Lett., 33, L19820, doi:10.1029/2006GL026882. Sardeshmukh, P. D., and B. J. Hoskins, 1987: The generation of global rotational flow by steady idealized tropical divergence. J. Atmos. Sci., 45, 1228–1251. Sedjo, R. A., and D. Botkin, 1997: Using forest plantations to spare natural forests. Environment, 39, 14–20. Smith, W. B., J. S. Vissage, D. R. Darr, and R. M. Sheffield, 2001: Forest resources of the United States, 1997. U.S. Department

7659

of Agriculture Forest Service North Central Research Station General Tech. Rep. NC-219, 198 pp. ——, P. D. Miles, C. H. Perry, and S. A. Pugh, 2009: Forest resources of the United States, 2007. U.S. Department of Agriculture Forest Service Washington Office General Tech. Rep. WO-78, 336 pp. Stoy, P. C., and Coauthors, 2006: Separating the effects of climate and vegetation on evapotranspiration along a successional chronosequence in the southeastern U.S. Global Change Biol., 12, 2115–2134. Subin, Z. M., L. N. Murphy, F. Li, C. Bonfils, and W. J. Riley, 2012: Boreal lakes moderate seasonal and diurnal temperature variation and perturb atmospheric circulation. Tellus, 64A, 15639, doi:10.3402/tellusa.v64i0.15639. Sun, G., and Coauthors, 2010: Energy and water balance of two contrasting loblolly pine plantations on the lower coastal plain of North Carolina, USA. For. Ecol. Manage., 259, 1299–1310. Teuling, A. J., and S. I. Seneviratne, 2011: The contrasting role of forests and grasslands on land-climate interaction during recent European heat waves. Proc. Topical Conf. Earth Observation for Land-Atmosphere Interaction Science, Rome, Italy, ESA, 4 pp. Tucker, C. J., J. E. Pinzon, and M. E. Brown, 2004: Global inventory modeling and mapping studies. University of Maryland, College Park Global Land Cover Facility Dataset NA94apr15b. n11-VIg, 2.0. Van den Dool, H. M., and R. M. Chervin, 1986: A comparison of month-to-month persistence of anomalies in a general-circulation model and in the earth’s atmosphere. J. Atmos. Sci., 43, 1454– 1466. von Storch, H., and F. W. Zwiers, 1999: Statistical Analysis in Climate Research. Cambridge University Press, 484 pp. Wallace, J. M., and D. S. Gutzler, 1981: Teleconnections in the geopotential height field during the Northern Hemisphere winter. Mon. Wea. Rev., 109, 784–812. Wang, X. C., and S. S. Shen, 1999: Estimation of spatial degrees of freedom of a climate field. J. Climate, 12, 1280–1291. Xu, L., and D. D. Baldocchi, 2003: Seasonal trends in photosynthetic parameters and stomatal conductance of blue oak (Quercus douglasii) under prolonged summer drought and high temperature. Tree Physiol., 23, 865–877. Zhang, D. W., and M. Polyakov, 2010: The geographical distribution of plantation forests and land resources potentially available for pine plantations in the U.S. South. Biomass Bioenergy, 34, 1643–1654. Zhang, S. S., and H. L. Allen, 1996: Foliar nutrient dynamics of 11-year-old loblolly pine (Pinus taeda) following nitrogen fertilization. Can. J. For. Res., 26, 1426–1439. Zhu, X., J. Sun, Z. Liu, Q. Liu, and J. E. Martin, 2007: A synoptic analysis of the interannual variability of winter cyclone activity in the Aleutian low region. J. Climate, 20, 1523–1538. Zwiers, F. W., and H. von Storch, 1995: Taking serial correlation into account in tests of the mean. J. Climate, 8, 336–351.