SEARCH

SEARCH BY CITATION

Keywords:

  • biofuel crops;
  • regional climate change;
  • land use change

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Model and Experiments Performed
  5. 3. Results and Discussion
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[1] Recent work has shown that current bio-energy policy directives may have harmful, indirect consequences, affecting both food security and the global climate system. An additional unintended but direct effect of large-scale biofuel production is the impact on local and regional climate resulting from changes in the energy and moisture balance of the surface upon conversion to biofuel crops. Using the latest version of the WRF modeling system we conducted twenty-four, midsummer, continental-wide, sensitivity experiments by imposing realistic biophysical parameter limits appropriate for bio-energy crops in the Corn Belt of the United States. In the absence of strain/crop-specific parameterizations, a primary goal of this work was to isolate the maximum regional climate impact, for a trio of individual July months, due to land-use change resulting from bio-energy crops and to identify the relative importance of each biophysical parameter in terms of its individual effect. Maximum, local changes in 2 m temperature of the order of 1°C occur for the full breadth of albedo (ALB), minimum canopy resistance (RCMIN), and rooting depth (ROOT) specifications, while the regionally (105°W–75°W and 35°N–50°N) and monthly averaged response of 2 m temperature was most pronounced for the ALB and RCMIN experiments, exceeding 0.2°C. The full range of albedo variability associated with biofuel crops may be sufficient to drive regional changes in summertime rainfall. Individual parameter effects on 2 m temperature are additive, highlight the cooling contribution of higher leaf area index (LAI) and ROOT for perennial grasses (e.g., Miscanthus) versus annual crops (e.g., maize), and underscore the necessity of improving location- and vegetation-specific representation of RCMIN and ALB.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Model and Experiments Performed
  5. 3. Results and Discussion
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[2] Many countries have aggressively promoted commercialization of technologies aimed at increasing domestic bio-energy production, as exemplified by the United States' Renewable Fuels Mandate [U.S. Congress, 2005], which is expected to consume 35% of the U.S. maize harvest by 2018 [U.S. Department of Agriculture, 2009]. In part, these efforts are justified on the grounds that displacing fossil fuels with plant-based biofuels will reduce greenhouse gas emissions thereby moderating anthropogenic influence on the climate system. However, recent studies have highlighted the effect of these policies on food prices [e.g., Naylor and Falcon, 2008] and emphasized that these price changes can lead to unintended but potentially large indirect effects on climate by motivating land use change around the world [Searchinger et al., 2008]. Another potentially important but unintended climate effect of large scale biofuel production is the direct effect on local and regional climates via changes in the energy and moisture balance of the surface upon conversion to biofuel crops. Such effects have been shown to be important when converting forests to grasslands, especially in high-latitude regions [Betts, 2000; Bonan, 2008] but the effects of changing from one type of grass to another are less clear.

[3] As an initial step toward quantifying the direct climate effects of land use and land cover change (LULCC) associated with the expansion of biofuel crops, we evaluated the potential effects of switching large regions from one type of grass to another. Although previous studies have suggested that climate effects of conversions from one grass species to another is relatively small [Gibbard et al., 2005], they typically considered only a relatively narrow range of biophysical attributes among grass species, and did not focus on regions of likely biofuel expansion. Here we perform a systematic sensitivity analysis across the full range of parameters that could reasonably be expected for four key bioenergy crops: maize, soy, Miscanthus, and switchgrass.

[4] We focus on the Corn Belt of the United States, an area dominated currently by maize and soybean, with possible future adoption of cellulosic ethanol feedstocks such as switchgrass and Miscanthus. Both observational [Carleton et al., 2001, 2008] and high-resolution modeling-based work [Weaver and Avissar, 2001; Adegoke et al., 2007] over the central U.S. have documented the climatic impact of contemporary agricultural land-use change through the effect on the surface energy budget. Further work has refined our understanding of crop-atmosphere interactions by considering the influence of key physiological characteristics [Alfieri et al., 2008].

[5] The goals of the current study are to (1) identify the potential magnitude of regional temperature and precipitation effects associated with bio-energy related LULCC, and (2) diagnose the relative climate importance of different parameters of bio-energy crops, as a way of informing future studies aimed at improving representations of particular crops.

2. Model and Experiments Performed

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Model and Experiments Performed
  5. 3. Results and Discussion
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[6] The latest version of the Weather Research and Forecasting (WRF version 3.1) modeling system [Skamarock et al., 2008] is used to conduct sensitivity experiments quantifying the maximum midsummer climate impact due to specification of the full range of biophysical parameters appropriate for bio-energy crops. A more detailed description of the model, options selected, and forcing data used is presented in Text S1 of the auxiliary material.

[7] We use a single coarse grid covering the contiguous U.S., northern Mexico, and southern Canada (Figure 1). The domain is discretized by 160 and 100 points in the east-west and north-south directions, respectively, with a horizontal grid spacing of 32-km. The vertical dimension contains 29 levels, with 9 within the first 1.5-km of the surface so as to better resolve planetary boundary layer (PBL) processes.

image

Figure 1. Domain extent and default landscape representation used for all experiments. For illustrative purposes, we aggregate all cropland grid cells, consisting of 5 distinct classes, into one single representation, denoted as “Cropland”.

Download figure to PowerPoint

[8] We focus on the month of July, given that the land surface exerts its greatest hydroclimatic control in this region during the summer [Koster et al., 2004]. Three CONTROL simulations are produced for July 1995, 1996, and 1997 (as they represent what may be considered July-averaged climate conditions) to first test the model's ability to appropriately simulate summer conditions. We then selected lower and upper limits for four key parameters in the land surface model – albedo (ALB), leaf area index (LAI), minimum canopy resistance (RCMIN) and rooting depth (ROOT) – based on literature and available field data appropriate for bio-energy crops (Table 1). For example, Hatfield and Carlson [1979] measured the albedo of three different Midwest maize varieties as ranging from 0.16 to 0.24, and soybean exhibits a comparable albedo range [Costa et al., 2007]. With limited information for switchgrass or Miscanthus, we therefore incorporated these values as boundary ALB limits appropriate for all bio-energy crops.

Table 1. Limits of Biophysical Parameters Appropriate for Bioenergy Crops Used for Sensitivity Testsa
Biophysical Parameters Altered and Naming Convention UsedMINMAX
AlbedoALBMIN = 0.16b[maize]ALBMAX = 0.24b[maize]
Leaf Area IndexLAIMIN = 3c[maize,soy]LAIMAX = 8d[Miscanthus]
Minimum Canopy Resistance (sm−1)RCMIN(min) = 30e,f[maize,soy]RCMIN(max) = 100g,h[maize,soy,switchgrass]
Rooting Depth (cm)ROOTMIN = 20i,j[soy]ROOTMAX = 200k,l[switchgrass,Miscanthus]

[9] Ranges for the other three parameters were similarly defined (Table 1), with ranges made wide enough to reflect the tendency of perennial biomass crops such as Miscanthus toward higher LAI and rooting depths. The range in RCMIN variability was explained by contributions from annual (i.e., maize and soybean) and perennial crops (i.e., Miscanthus). Although values more extreme than considered here have been documented (e.g., Miscanthus LAI has been measured to be in excess of 10 [Heaton et al., 2008]) we believe these values represent a reasonable range in expected bio-energy crop biophysical parameter variability and are appropriate for identifying the potential magnitude of climate impacts and the relative role of different parameters.

[10] Eight simulations were conducted for each of three years, using the lower and upper bounds for each of the aforementioned four parameters. In each simulation, the parameters were changed only for distinct agricultural grid cells within a specified geographic region (102.5°W–82.5°W and 37°N–48.5°N). The twenty-four default USGS land use categories were first interpolated to the model grid prior to simulation initialization. Parameter values for cells classified in the USGS system as dryland cropland and pasture, irrigated cropland and pasture, mixed dryland/irrigated cropland and pasture, or cropland/grassland mosaic were then replaced with the relevant values from Table 1.

[11] Including the trio of CONTROL experiments (which used the NOAH LSM default biogeophysical representation), a total of 27 simulations were performed. All experiments were initialized on June 24, 12 Z, and were continued through July 31, 18 Z, using a 120-second timestep for both the atmospheric and land-surface model components. The initial simulation week (corresponding to June) was treated as a spinup and was consequently discarded, leaving the month of July, with results written every 6 h, for analysis.

3. Results and Discussion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Model and Experiments Performed
  5. 3. Results and Discussion
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[12] A detailed description of the model's ability to reproduce the July climate, evaluated against suitable temperature and precipitation observations, is presented in Text S2 and Figures S1S3.

[13] Although no tuning of the model was carried out in order to improve correspondence with observations, we did spectrally nudge wave numbers 0, 1, and 2 (i.e. greater than about 2500-km wavelength) [Miguez-Macho et al., 2004, 2005] thus constraining the model solution of the longest atmospheric wavelengths (above the PBL only) to that of the driving boundary fields. Usage of this technique leads to more reliable sensitivity results [Miguez-Macho et al., 2004, 2005].

[14] We begin assessment of biophysical parameter specification distinct to bio-energy crops by evaluating the impact on 2 m temperature (Figure 2). The maximum impact occurs for the ALB and RCMIN simulations, while the effect due to bracketing of LAI and ROOT is largely confined to the geographical region where the perturbation was applied. Maximum absolute differences in 2 m temperature for the ALB simulations approach 1°C locally. The sensitivity to RCMIN specification illustrates maximum absolute differences in 2 m temperature in excess of 1°C locally, in good agreement with modeling results conducted by Alfieri et al. [2008] over the southern Great Plains, who simulated a 1°C–1.5°C response to nearly the same range in RCMIN as we specified here. For both the ALB and RCMIN experiments, the landscape imposed impact on near-surface temperature is comparable to the effect of historical land cover change over the Midwest/Great Plains. In both instances the local effects are advected downstream by the mean wind, thereby impacting locales removed from the geographical area of biogeophysical modification, although the magnitude is reduced. For example, although the biophysical representation over Oklahoma remained unaltered among all experiments, both pairs of sensitivity tests (i.e. ALB and RCMIN simulations) produced 2 m absolute temperature differences exceeding 0.5°C. Effects are also apparent, although reduced in magnitude, over the lower Ohio River Valley, Kentucky River Valley, Tennessee River Valley, and northeastern U.S. Further, we note that although the impact illustrated by the ROOT experiments remains largely localized (similar to the LAI experiments), maximum absolute differences in 2 m temperatures exceed 1°C locally. The magnitude and pattern of influence shown for July 1997 is similar for July 1995 (not shown) and 1996 (not shown), suggesting that notwithstanding large-scale dynamical variability among the trio of years, the imposed landscape forcing assumes a primary role in altering the near surface climate.

image

Figure 2. WRF simulated July 1997 2-m temperature [°C] difference for (a) ALBMIN − ALBMAX, (b) LAIMIN − LAIMAX, (c) RCminMIN − RCminMAX, and (d) ROOTMIN − ROOTMAX. Black rectangle in plots delineate portion of region used in the spatially averaged calculations presented in Table 1.

Download figure to PowerPoint

[15] We next define an area, bounded by 105°W–75°W and 35°N–50°N (the box outlined in Figure 2) which defines the geographical region wherein the imposed landscape perturbations exhibit their influence. Table 2 shows the simulated response, averaged over this sub-domain across all sensitivity experiments. The 2 m temperature response to the full range of albedo specification is similar among the different July months investigated (see also Table S1 for individual July results). This response is second, ranking behind only the response for the RCMIN simulations. Although the full range of LAI and ROOT experiments produce responses limited in geographical extent, they remain non-negligible.

Table 2. Mean Response Across Sub-domain Displayed in Figure 2 (105°W–75°W and 35°N–50°N) to Full Range of Biophysical Parameter Specification for All Three July Monthsa
 2-m Temperature [°C]2-m Dew-Point [°C]Precipitation [mm day−1]
  • a

    Bold denotes values depicting the same sign for all three July months.

ALBMIN − ALBMAX0.2200.20
LAIMIN − LAIMAX0.13−0.16−0.02
RCMIN(min) − RCMIN(max)−0.310.380.07
ROOTMIN − ROOTMAX0.18−0.27−0.06

[16] The response of simulated 2 m dew-point temperature is least for the ALB simulations although the response in accumulated rainfall is greatest, exceeding 0.1 mm day−1 for all 3 July months. The monthly mean differences in latent and sensible heat flux between ALBMIN and ALBMAX (not shown) reveal an increase (for the ALBMIN experiments) in the sub-domain region of influence (due to increased absorption of solar radiation). During the course of the month, the ALBMIN experiments undergo an increase in net radiation, serving to destabilize the lower atmosphere through both moisture enhancement (due to increased evapotranspiration) and low-level heating (via turbulent heating). Enhanced instability assists in lifting the added moisture higher into the PBL where convective systems may preferentially take advantage of both the extra water and dynamical lift in enhancing simulated rainfall. All other pairs of experiments produce an increase in one radiation balance term, but a nearly commensurate decrease in the other. For example, despite the substantial increase in low-level moisture (Table 2) due to a lowering of canopy-level resistance (i.e., the RCMIN experiments) the sensible heating reduction strongly limits the vertical lift required to condense the additional water, resulting in reduced precipitation relative to the ALB experiments.

[17] The full extent of albedo variability present among bioenergy crops, from a minimally imposed value to the maximum, promotes a consistent decrease in simulated precipitation of 5% for July 1997 and July 1995, and 6% for July 1996. Modeling studies investigating the effect of deforestation due to soybean expansion versus pasture expansion in Amazonia found a precipitation signal exhibiting the same sign although of considerably larger magnitude [Costa et al., 2007]. While our values may not appear extreme, differences in local rainfall accumulation may be substantial with important implications for local hydroclimatology. Dynamic patterns of surface heterogeneity (e.g., soil moisture changes due to previous day's enhancement or reduction in rainfall) may affect subsequent storm cell development [Weaver, 2004], may exhibit pronounced persistence (monthly or longer scale), and may therefore impart further hydroclimatic influence [Georgescu et al., 2003].

4. Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Model and Experiments Performed
  5. 3. Results and Discussion
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[18] Our results show that large scale conversion of land cover from one crop type to another, as might be expected with biofuels, has the potential to significantly affect local and regional climate in the central U.S. Although differences in experimental setup preclude a direct comparison, the magnitude of the impacts presented in this research is bracketed by other regional climate modeling studies that have shown both a slight warming [Copeland et al., 1996] and cooling [Bonan, 2001; Bounoua et al., 2002; Baidya-Roy et al., 2003] effect due to historical LULCC over the central U.S. While all biophysical parameters investigated as part of this work have a non-negligible impact on near-surface climate, the impact of the ALB and RCMIN experiments are of greatest significance (Table 2).

[19] Understanding the precise impacts of specific biofuel crops will require more detailed knowledge of crop and location specific biophysical parameters, and in part the goal of this study was to identify which parameters are most important to constrain with further work. However, some general implications for conversion to cellulosic biofuel crops such as Miscanthus or switchgrass are apparent. Specifically, the tendency of perennials like Miscanthus to exhibit higher LAI and rooting depth suggests a cooling contribution of ∼1–2°C locally in July, notwithstanding any possible differences in ALB or RCMIN. Phenological contrasts between perennial versus annual bio-energy crops may reveal additional climate impacts that are the subject of ongoing work.

[20] While we consider these sensitivity results robust, we recognize the importance of a model intercomparison employing the methodology used here to incorporate a range of LSM choices (to account, for example, for differences in landscape representation or evapotranspiration calculations) and parameterization selection (e.g., different means of approximating subgrid convection).

[21] Future work should also focus on other regions with current or likely future bio-energy driven LULCC (e.g., Brazil). It is likely that only a subset of crops and regions will affect climate in meaningful ways, but identifying these cases will be critical to informing biofuel policy design that considers not only greenhouse gas impacts but also the direct climatic effect resulting from land use change.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Model and Experiments Performed
  5. 3. Results and Discussion
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Model and Experiments Performed
  5. 3. Results and Discussion
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

Auxiliary material for this article contains two text files, three figures, and a table.

Auxiliary material files may require downloading to a local drive depending on platform, browser, configuration, and size. To open auxiliary materials in a browser, click on the label. To download, Right-click and select “Save Target As…” (PC) or CTRL-click and select “Download Link to Disk” (Mac).

See Plugins for a list of applications and supported file formats.

Additional file information is provided in the readme.txt.

FilenameFormatSizeDescription
grl26473-sup-0001-readme.txtplain text document2KReadme.txt
grl26473-sup-0002-txts01.pdfPDF document737KText S1. Details of WRF as used in this paper, including parameterizations and initial and boundary forcing data used.
grl26473-sup-0003-txts02.pdfPDF document734KText S2. Evaluation of the WRF Control simulations against suitable observational datasets of gridded temperature and precipitation.
grl26473-sup-0004-fs01.pdfPDF document849KFigure S1. WRF-simulated temperature and precipitation against suitable observational datasets of gridded temperature and precipitation for July 1997.
grl26473-sup-0005-fs02.pdfPDF document849KFigure S2. WRF-simulated temperature and precipitation against suitable observational datasets of gridded temperature and precipitation for July 1996.
grl26473-sup-0006-fs03.pdfPDF document846KFigure S3. WRF-simulated temperature and precipitation against suitable observational datasets of gridded temperature and precipitation for July 1995.
grl26473-sup-0007-ts01.pdfPDF document26KTable S1. Mean response across the sub-domain displayed in Figure 2 to full range of biophysical parameter specification for each individual July.
grl26473-sup-0008-t01.txtplain text document1KTab#x2010;delimited Table 1.
grl26473-sup-0009-t02.txtplain text document0KTab#x2010;delimited Table 2.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.