Skip to content U.S. Department of Transportation/Federal Highway AdministrationU.S. Department of Transportation/Federal Highway Administration

Office of Planning, Environment, & Realty (HEP)
PlanningEnvironmentReal Estate

HEP Events Guidance Publications Awards Contacts

Carbon Sequestration Pilot Program

Estimated Land Available for Carbon Sequestration in the National Highway System

Final
May 2010

U.S. Department of Trasportation
Federal Highway Administration
Office of Planning, Envrionment and Realty

U.S. Department of Trasportation
Research and Innovative Technology Administration
John A. Volpe National Transportation Systems Center

Acknowledgements

This report was prepared by the U.S. Department of Transportation Federal Highway Administration (FHWA) and John A. Volpe National Transportation Systems Center (Volpe Center). The project team included Stephen Earsom of FHWA's Office of Project Development and Environmental Review, and Robert Hallett, Theresa Perrone, and Carson Poe of the Volpe Center's Multimodal Systems Research and Analysis Center of Innovation. Maggie Greenfield of MacroSys also contributed.

The project team wishes to thank the state Departments of Transportation (DOT) staff members, especially those at Minnesota and New Mexico DOTs, who graciously provided their time, knowledge, and guidance in conducting this pilot program. The DOTs in Alabama, Arizona, Delaware, Michigan, Montana, South Carolina, Utah, and Washington also provided valuable data. Their input enhanced FHWA's ability to explore the feasibility of sequestering carbon on highway rights-of-way.


Table of Contents


Executive Summary

The Federal Highway Administration (FHWA) established the Carbon Sequestration Pilot Program (CSPP) in 2008 to assess whether a roadside carbon sequestration effort through modified maintenance and management practices is appropriate and feasible for state Departments of Transportation (DOTs) when balanced against ecological and economic uncertainties. The purposes of the pilot were to:

CSPP findings are expected to inform DOTs that may be considering the implications of future climate change legislation or that independently want to evaluate the cost effectiveness of using National Highway System (NHS) right of way (ROW) to generate revenue from the sale of carbon credits, offset their own emissions, or meet statewide greenhouse gas emissions objectives.

The project team used data from Minnesota and several other states to estimate the amount of unpaved NHS ROW available for carbon sequestration-marking the first time that a rigorous study has been conducted to quantify the amount of state DOT-managed soft estate acreage. In the first of two analytical approaches used, ROW widths at random locations in nine states were manually measured on property maps to provide a distribution of common ROW dimensions and observed vegetation types. A subsequent geographic information system (GIS) analysis of 1,000 random locations nationwide provided insight into the types of land cover in close proximity to the NHS.

click on the figure to go to the tabular data

Results indicate that there are approximately 5.05 million acres in the NHS nationwide, with a likely range of 1.4 to 8.7 million acres. Roughly 68 percent, or 3.4 million acres, is unpaved. Evidence shows that the land cover has undergone little change since 1992.

The project team estimates the NHS ROW has approximately 91 million metric tons (MMT) of carbon currently sequestered in vegetation and is currently sequestering approximately 3.6 MMT of carbon per year, or 1.06 metric tons of carbon per acre per year. This equals the annual carbon dioxide emissions of approximately 2.6 million passenger cars. At its carbon equilibrium, the entire NHS ROW is estimated to be able to sequester between 425 and 680 MMT of carbon. Using a hypothetical carbon price of $20 per metric ton, this equates to a total potential value of $8.5 to $14 billion nationwide.

The availability of ROW property data was highly variable and, thus, was the major limiter in making these estimates. As discovered through this research, very few DOTs have had time, funding, or impetus to scan and geospatially reference their ROW property maps. The research here could have been vastly expedited had there been more DOTs with property maps in the needed electronic format and had their ROW data been easily accessible in a national GIS database.

In addition to this report, FHWA has developed a Highway Carbon Sequestration Estimator to help DOTs assess the return on investment for various carbon sequestration scenarios. The decision-support tool, which allows transportation officials to make estimates based on more state-specific considerations than possible here, is available upon request from carson.poe@dot.gov.

Even under the best scenarios, revenue generated from biological carbon sequestration will vary greatly from state to state based on carbon prices, management techniques, and ecological variability. However, considering the use of vegetation for living snow fences, landslide minimization, and other such human infrastructure protection may, in some cases, eventually be found to be more cost-effective than traditional engineering solutions, especially when all costs are included.


Introduction

The potential for land managers to generate revenue from biological carbon sequestration through sustainable forestry and replacing traditional ground cover with native grasses was the genesis of the Federal Highway Administration's (FHWA) Carbon Sequestration Pilot Program (CSPP). Federal statutes allow state Departments of Transportation (DOTs) to generate revenue from their land holdings. Since DOTs must retain unused buffers in their right-of-way (ROW) for safety, operations, and maintenance purposes, FHWA recognized that an opportunity might exist to shape the future of a burgeoning ecosystem service market.1

The National Highway System (NHS) is approximately 163,000 miles of roadway consisting of the Interstate Highway System and other roads important to the nation's economy, defense, and mobility.2 The NHS includes only 4 percent of the nation's roads but carries more than 40 percent of all highway traffic, 75 percent of heavy truck traffic, and 90 percent of tourist traffic. In 2007, approximately 69 percent of the NHS was classified as being located in rural areas. FHWA developed the CSPP to assess whether a roadside carbon sequestration effort on the NHS through modified maintenance and management practices is appropriate and feasible for DOTs when balanced against the economic and ecological uncertainties.3

The goals of the pilot were to:

  1. Develop estimates of the amount of revenue that DOTs could earn if they undertook such a effort using native vegetation;
  2. Determine the cost-effectiveness of a similar effort on a national scale; and,
  3. Create decision support tools that DOTs could use to determine the efficacy of programs in their states.

This paper addresses the first two purposes,4 refining a coarse estimate of the unpaved NHS ROW available for carbon sequestration that FHWA had made when establishing the merits of a pilot program. Results from the analysis include more accurate estimates of several variables for each state and for the nation as a whole, including:

An approximation of the carbon currently sequestered in NHS ROW is also presented. It should be noted that estimates here for the amount of land that could be converted to management for carbon sequestration constitute an upper bound. Net availability will undoubtedly be less, due to considerations for safety, operations, and maintenance.

The findings can inform leadership at DOTs that are considering the implications of future climate change legislation and the transportation reauthorization bill or that might independently want to evaluate the cost effectiveness of using highway ROW for carbon sequestration and carbon offset trading.


Methods

The geographic focus of this study is the 48 contiguous United States. Two analytical approaches were developed to generate estimates of ROW acreage at a state scale and for the nation as a whole. The first method-the transect analysis-requires that states have fairly detailed datasets in the appropriate electronic format and involves significant manual manipulation by an analyst. The second method-the polygon area analysis-uses widely available data, and, while GIS modeling competence is needed, overall requires less time to generate results.

Transect Analysis

Using ESRI's ArcMap GIS application, Minnesota's NHS roadway network was overlaid with a one fourth-mile by one fourth-mile grid across the entire state.

At every instance where a road crossed the grid (over 55,000 occurrences), a point was made and assigned a unique identification number. Random numbers were generated and then sorted in ascending order. The first 40 random points in the sorted list were used as the sample for analysis.

The sample was overlaid in the GIS on computer-aided design (CAD) drawings and scanned, geospatially-referenced ROW maps available in Minnesota DOT's (Mn/DOT) Right of Way Mapping and Monitoring application.5 A transect line perpendicular to the highway and between property boundaries was then drawn at each random site (Figure 1). It was assumed that all land within the property boundaries was owned in fee simple, and no land was held in easement.

Figure 1. Example random site with transect and property boundary lines drawn

Figure 1 illustrates a right of way property map at a random location in Minnesota overlaid on an aerial photo of the same location. Lines have been manually drawn on the highway right of way property to delineate the property boundaries. A perpendicular line that transects the highway between the highway property boundaries has also been drawn to indicate where right of width measurements were manually taken.

Based on information in the underlying legal ROW property maps, the ROW widths were manually measured using the GIS application's embedded tool. Points were discarded in cases where the random site was positioned in the middle of a four-way intersection or where no ROW property map or CAD drawing was available to delineate property boundaries. In the analysis for Minnesota, this occurred twice.

The project team also used National Agriculture Imagery Program (NAIP) aerial imagery6 ,7 to discern visually unpaved and paved areas and to estimate transect widths for grass, shrubs, trees, grass/trees within the ROW at the random sites.

The grass/trees category describes land cover where the project team observed both grass and woody vegetation in roughly equal proportions. Under a passive restoration land management approach either vegetation type might be expected to be prevalent. Here again, widths for paved and vegetated portions of the ROW were measured manually for all of the random points using the GIS application's embedded ruler tool (Figure 2).

Figure2. Example random site with pavement and vegetation measurements shown

Figure 2 displays example right of way width measurements that were manually measured. The measurements, which are listed on top of an aerial photo of a random highway site, are specified for the paved and unpaved portions of the highway right of way. In this example, the right of way (from left to right) has the following widths: 70 feet of woody vegetation, 42 feet of pavement, 52 feet of grass, 42 feet of pavement, 28 feet of grass, and 50 feet of pavement. The total right of way width of this particular random location is 284 feet.

This exercise resulted in a distribution of ROW widths for each of the categories mentioned above. The average value of each of these categories was then multiplied by the total number of miles of NHS in the state, and appropriate conversions were made to arrive at acreage estimates.

Next, other state DOTs were contacted in random order to determine whether electronic and geospatially-referenced ROW property data similar to Mn/DOT's were available. The intent was to generate ROW acreage estimates for an additional 5-10 states, then extrapolate the results to all other states and the nation as a whole.

With the data from additional states in hand, the project team modified the methodology used for Minnesota by overlaying a 1-mile by 1-mile grid on the entire NHS network in the 48 contiguous United States. At every instance where a road crossed the grid, a point was made and assigned a unique identification number. Again, random numbers were generated, and the first 1,000 corresponding points were selected for inclusion in the sample. Points in states for which electronic and geospatially-referenced ROW property map data were available were used as locations where additional boundary to boundary measurements were made.

Using the available ROW and NAIP data, pavement and vegetation widths were manually made for all of the new sites. In total, 121 sites across 8 states were analyzed, in addition to those in Minnesota (Figure 3).

Figure3. Number and location of random sites

click on the figure to go to the tabular data

The resulting distribution of ROW widths was multiplied by the number of NHS miles in the U.S. to provide extrapolations of NHS ROW acreage. Specifically, the observed pooled average widths for all states' transect sites were multiplied by each individual state's NHS mileage to provide ROW estimates. In other words, no individual state's acreage estimate was derived solely from that state's own measurements:

ROW Acreage Estimate = (Observed Pooled Average Width * NHS mileage in ft by state)/43650 ft2

The standard deviation for the widths at all transect sites was also added to and subtracted from each state's average width value to provide ranges of expected ROW widths. The upper and lower width values for each state were then multiplied by the respective state's NHS mileage to provide estimated ROW ranges.

Polygon Area Analysis

Using ESRI's ArcMap GIS application, 300-foot diameter circle polygons were centered on each of the same random points included in the transect analysis (40 for Minnesota and 1,000 for the 48 contiguous states). Land cover data from the Multi-Resolution Land Characteristics Consortium's (MRLC) 2001 National Land Cover Database (NLCD) were extracted from each polygon providing percent cover for several land cover types in each polygon (Figure 4).8 The land cover distributions were compared with the land cover estimates from the transect analysis to determine the degree to which the polygon area analysis method is a suitable substitute to manual transect measurement in corroborating land cover types.

Figure4. Example random site with circle polygon drawn and land cover extracted

Figure 4 shows an example circle polygon generated in the geographic information system side-by-side with an aerial photo of the random location the circle polygon was generated. Circle polygons were used to extract land cover information for various random sites. In this example, the circle polygon and corresponding aerial image show that the land cover around the random location consisted of areas of open water and low-intensity development.

Land cover data from the MRLC's Retrofit Land Cover Change Data were also extracted from each polygon in order to discern any changes in land cover that might have occurred at the 1,000 random points between 1992 and 2001. Results from the polygon area analysis consist of distributions for (1) each land cover type found in 2001 and (2) cover changes that occurred from 1992 to 2001.


Results

There are approximately 163,000 miles of roadway in the NHS of the contiguous U.S., with Texas having the most mileage and Rhode Island having the least (Figure 5).

Figure 5. NHS mileage, by state*

click on the figure to go to the tabular data

Right-of-Way in the National Highway System

Transect widths were measured at 159 random sites across 9 states. The sample showed a total ROW range from 60 ft to 1,295 ft with an average of 257 ft. Unpaved ROW ranged from 0 ft to 1,047 ft with an average of 175 ft. A majority of the ROW was observed to be grass (Table 1 and Figure 6).

Table 1. Descriptive statistics for random site transect measurements*
AVERAGE TOTAL ROW257 ft 
Average paved 32%
Average unpaved 68%
Average grass 64%
Average trees 11%
Average grass/trees 14%
Average shrubs 11%
MEDIAN TOTAL ROW202 ft 
Median Paved70 ft 
Median Unpaved144 ft 

Figure 6. Frequency of observed widths at random sites **

click on the figure to go to the tabular data

**Note: The x-axes' scales differ for the two histograms.

It is estimated that there are approximately five million acres of NHS ROW nationwide. Adding and subtracting one standard deviation results in a range of 1.4 to 8.7 million acres. Approximately 68 percent of those acres (3.4 million) are estimated to be unpaved, with grasses expected to comprise the largest unpaved ROW portion (Table 2).

Table 2. Estimated Total NHS ROW acres*
NHS ROW AcreageU.S. Estimate (in 000s of acres)
Total1,400-8,700, likely ~ 5,000
Unpaved 400-6,400, likely ~ 3,400
Grassland200-2,800, likely ~ 2,200
Woody vegetation30-460, likely ~ 360
Grassland/woody vegetation mix36-600, likely ~ 470
Shrub30-500, likely ~ 390

*Total ROW and unpaved ROW acreage estimates for individual states are in Appendix A.

Land Cover in the National Highway System

Land cover data were collected at 1,000 random sites across the contiguous United States. The sample showed that the main land cover types in the vicinity of the NHS are "developed, open space" and "developed, low intensity." The most common land cover types not characterized as "developed" are "cultivated crops" followed by "deciduous forest" (Table 3). When disaggregated, the land cover results show "developed, open space" as the most prevalent land cover type near the NHS in 29 of 48 states.

Table 3. Polygon area analysis land cover percentages, nationwide
LAND COVER CLASSIFICATIONPERCENT
Developed, Open Space32.39%
Developed, Low Intensity25.92%
Developed, Medium Intensity14.22%
Cultivated Crops4.95%
Developed, High Intensity4.43%
Decidous Forest3.81%
Shrub/Scrub3.58%
Hay/Pasture3.33%
Herbaceuous2.55%
Evergreen Forest2.34%
Woody Wetlands1.06%
Emergent Herbaceuous Wetlands0.47%
Mixed Forest0.43%
Barren Land0.34%
Open Water0.18%

Less than four percent of the land cover in the polygons changed categories from 1992 to 2001 (Table 4). A majority of the change was from a vegetated cover type to the "Urban" cover type, which consists of primarily developed, open spaces and some low, medium, and high intensity lands.9 Forested portions of the polygons experienced a net loss of approximately 1.6 percent over the same time period. The grassland/shrub land cover segment of the polygons experienced a net loss of roughly 0.5 percent between 1992 and 2001.

Table 4. Land cover change at polygon area analysis random sites, 1992-2001
LAND COVER CHANGEPERCENT OF CIRCLE BUFFERS CHANGED
Forest to Urban1.545%
Agriculture to Urban1.183%
Grassland/Shrub to Urban0.479%
Wetlands to Urban0.150%
Forest to Agriculture0.120%
Grassland/Shrub to Agriculture0.090%
Agriculture to Forest0.067%
Agriculture to Grassland/Shrub0.056%
Forest to Grassland/Shrub0.050%
Agriculture to Wetlands0.037%
Grassland/Shrub to Forest0.015%
Agriculture to Barren0.013%
Agriculture to Open Water0.013%
Wetlands to Agriculture0.007%
Grassland/Shrub to Barren0.005%
Forest to Barren0.002%
Grassland/Shrub to Wetlands0.001%

Discussion

This discussion consists of three sections: ROW Estimates; Carbon Sequestration in the NHS; and, Conclusions. Recommendations for repeating or customizing the methods developed here are presented, as well as estimates of carbon sequestration rates and potential on NHS ROW.

ROW Estimates

The project team assumed that ROW acreage estimates for all 50 States could be generated from a multi-state distribution of ROW widths based on the known miles of NHS roads in each State; it was understood that acreage estimates for states where data were not collected would potentially be subject to more error if the sample of states was not sufficiently representative.

The transect analysis, which was used to develop the ROW distribution, was particularly valuable because it allowed the project team to use maps to quickly discern property boundaries, providing a means to take precise and accurate ROW measurements. It offered a degree of certainty likely not possible by any means other than taking surveys in the field. However, the reliance on electronic property maps was also the transect analysis principal drawback. The ability to perform the transect analysis was (and remains) contingent upon the availability of easily accessible, electronic and geospatially-referenced ROW property maps. It became apparent during the transect analysis that most state DOTs do not have maps in this format readily available, indicating an area where future federal funding might be directed. For the state DOTs that did have and were able to provide electronic property maps, there was an implicit assumption that all lands delineated as NHS ROW were owned in fee simple, rather than in easement. This, too, is a possible concern due to questions regarding how ownership rights to lands a state DOT manages might affect the process for selling carbon credits generated on those lands (FHWA 2009).

These realities, along with the fact that transects require time-intensive manual interpretation, led the project team to develop the polygon area analysis methodology to complement and potentially be a proxy for the transect analysis. The project team hypothesized that the percent land cover in the polygons would not be substantially different from that measured manually in the transect analysis. Thus, the project team used only the transect method for developing an estimate of acres of ROW across the US, and both the transect and polygon area methods for determining land cover types.

The polygon area analysis relied on the 2001 NLCD and polygons centered on the roadway used to extract land cover data. The 2001 NLCD has 29 different land cover classifications, 15 of which occurred in some quantity in the polygon area analysis. Four of those land covers describe varying intensities of development: high, medium, low, and open space. Observation of a random subset of 200 circle polygons indicated that paved NHS areas were most commonly classified as "developed, low intensity," while "developed, medium intensity" typically referred to places where bridges or interchanges were located. "Developed, high intensity" NHS areas were usually in very urbanized areas near parking lots, dense building development, and other impervious surfaces. "Developed, open space"-the predominant land cover found in the polygon area analysisgenerally captured unpaved portions of NHS ROW. From a carbon sequestration standpoint, open space and low intensity developed areas are generally expected to have the most carbon sequestration potential. This should not, however, suggest that medium and high intensity development areas are not suitable for carbon sequestration. In fact, some of the widest ROWs the project team observed were at interchanges in locations predominately developed with high-intensity; the diameter of the circle polygon may have extracted only a portion of the land cover types identified at the same location in the transect analysis. In other words, areas of medium- and high-intensity development may also be able to accommodate vegetation as open- and low-intensity development areas would be expected to do so. A difference would be that alternative vegetation management practices in the former may require the balancing of more factors.

Paved NHS areas-those classified as "developed, low intensity" or "developed, medium intensity"-were predicted to account for roughly 35 percent of each circle polygon.10 This proportion corresponded to what the transect measurements showed. There, 32 percent of the observed ROW was paved. Furthermore, it was assumed the polygons would have a uniform bias toward vegetation that is closest to the road.

Assuming most woody vegetation is located further from the road, this might result in a disproportionately high percentage of grass relative to woody vegetation. However, the polygon area analysis indicated that grasses, forests, and shrubs respectively comprised 43.5 percent, 6.6 percent, and 3.5 percent of the unpaved area in the vicinity of the random points-nearly matching results from the transect analysis (43.9 percent grasses, 7.2 percent trees, and 7.7 percent shrub) (Table 5).

Table 5. Comparison of predicted land cover types from two analytical approaches
  Polygon Area
Analysis
Transect
Analysis
Paved areas34.832
Grasses43.543.9
Woody vegetation6.67.2
Shrubs3.57.7

Based on these results, the project team believes that that the polygon area analysis serves as a relatively robust surrogate to the transect method for determining land cover types. Similarly, it is expected that the land cover types the polygon area analysis identified can be used to refine the more coarse land cover categories of the transect analysis, thereby potentially providing more precision in state-level estimates for carbon sequestration potential. However, at this time the transect method remains our only available method for developing rigorous estimates of ROW acreage.

The following are recommended steps for state DOTs interested in repeating or customizing either of the analyses described:11

Transect Analysis

Polygon Area Analysis

Carbon Sequestration in the NHS

The project team assumed that ROW acreage estimates could be used to assess:

 

A variety of factors affect the long-term storage potential, or carbon stock equilibrium,12 of vegetation and soil. Considerable variations in sequestration rates have been demonstrated depending on geographic region, plant species (Stavins and Richard 2005, Birdsey 1992), management practices, and natural disturbance regimes such as fire.

The Chicago Climate Exchange (CCX)13 addresses this variation in part by crediting sequestration of grasslands between 0.4 and 1.0 metric tons of C/ac/yr, depending on location. Here, the project team used the average of these values, 0.7 metric tons of C/ac/yr, for grasses and shrubs. Representative carbon sequestration for reforestation activities in the U.S. have been estimated to be between 1.1 and 7.7 metric tons C/ac/yr over 120 years (Birdsey 1996) and up to 172.1 metric tons C/ac/yr for avoided deforestation14 (U.S. Government 2000). CCX has more refined sequestration rates for a variety of tree species. For these purposes, the project team averaged CCX sequestration rates for 21 to 25 year-old coniferous (2.21 metric tons C/ac/yr) and deciduous species (2.16 metric tons C/ac/yr)-the expected representative age of trees on the NHS, then applied them to calculate the estimates below. It should be noted that carbon sequestration rates for afforestation activities15 in the U.S. have been shown to be higher than reforestation sequestration rates-between 2.2 and 9.5 metric tons of C/ac/yr (ROW 1996). However, afforestation often requires significant inputs of labor, water, and fertilizer that would render the project cost prohibitive. For this and other reasons, FHWA strongly encourages native, self-sustaining vegetation.

Assuming trees can sequester carbon for 120-years and grasses up to 50 years (U.S. EPA) and portions of the NHS have been around for 50 years, the project team expected the oldest trees on the NHS to be nearly at their sequestration mid-points and grasses to be at their maximums.

Given that areas of the NHS remain under construction today, the project team further assumed the "average" NHS vegetation to be roughly 25 years old. Using the results from the transect and polygon area analyses and the sequestration rates described above, it is estimated that currently the total annual uptake of carbon on the NHS is approximately 3.6 million metric tons (MMT), or 1.06 metric tons of C/acre/year (Table 6). This equates to the annual carbon dioxide emissions of approximately 2.6 million passenger cars.16 The project team also estimates that NHS ROW has sequestered approximately 91 MMT of carbon over its existence.

Table 6. Estimated annual carbon uptake on the NHS
 Estimated
Acres
Carbon Sequestration
Rates (metric tons C/ac/yr)
Metric Tons
of C/year
Deciduous4778202.161032091
Coniferous294096 2.26664657
Mixed54608 2.21120684
Grasses2207596 0.701545317
Shrub389393 0.70272575
Total Unpaved3423513 3635324

It is worth noting that the project team assumed portions of the NHS ROW identified as "grass/trees" in the transect analysis could be managed toward being trees and thus treated them as trees in making these estimates. The project team recognizes this will not be ecologically appropriate in some cases. It also should be noted that these estimates assume that all unpaved NHS ROW could be used for carbon sequestration of appropriate vegetation type. For example, the clear zone17 would continue to be managed for grasses but might be mowed less frequently18 or converted from an introduced species such as annual rye grass to native perennial species that store more carbon underground (Cox et al. 2006).

Pasture, rangeland, and agricultural land that is reserved for conservation purposes store carbon at equilibrium levels ranging from 73 to 159 metric tons of C/acre and average 113 metric tons. Mature, never harvested forests have higher equilibrium levels per acre, varying from 286 to 1,179 metric tons of C/acre and averaging 465 metric tons (Birdsey 1992 and CBO 2007). While harvesting forests can decrease the equilibrium level of carbon (Ruben et al. 2005), it was assumed that trees are not harvested on highway ROW-though harvesting timber presumably could be a DOT land management activity. Keeping these figures in mind, the point of carbon saturation on the NHS ROW is expected to be between 425 and 680 MMT (Appendix C). At current sequestration rates, carbon saturation on the NHS is not expected to occur for at least 75 years, and perhaps longer for areas of woody vegetation. That said, sequestration rates are expected to decline over time, and the actual carbon saturation point may be sooner if NHS vegetation is older than assumed. Using a hypothetical carbon price of $20 per metric ton, the estimated carbon volume equates to a total potential value of $8.5 to $14 billion nationwide. It is not unreasonable to conceive of even higher carbon prices (Benítez et al. 2006 assumes $50 per metric ton), as some modeling studies, consistent with attaining certain emissions goals during this century, show carbon prices rising to as high as $80 per ton of CO2-equivalent by 2030 and $155/ton of CO2-equivalent by 2050 (IPCC 2007).

The project team's estimated carbon sequestration maximum is based on the assumption that grasses are not converted to a different land cover. However, U.S. Forest Service data on both total historical forested land and total grassland pasture and rangeland suggest that many states could sequester additional volumes of carbon if alternative land management activities were undertaken. For example, from 1953 to 2002, forested land in Minnesota decreased by 6 percent. Presumably, some of the previously forested land is located the NHS ROW and could be shifted to a forest management strategy, increasing the estimated volume of carbon potentially sequestered in the state. This does not include other possible gains from restoring grassy areas to native grassland communities. Although some states may be more forested now than at other times during the past century, it is assumed that land along the NHS has been cleared, and thus the potential for additional carbon sequestration elsewhere is likely similar to that described in this example.

It should also be noted that these values represent calculations from aggregated data. Specific numbers will vary widely from state to state, and states are strongly encouraged to use FHWA's Highway Carbon Sequestration Estimator (or other appropriate tool), which allows transportation officials to assess the return on investment for various carbon sequestration scenarios based on more state-specific considerations than possible here. The tool is available upon request from carson.poe@dot.gov.

Conclusions

In the current debate regarding national climate change legislation, the U.S. Congress is placing great emphasis on minimizing the cost of any cap-and-trade system on the economy and consumers. Allowing the sale of carbon offsets opens up a potential revenue stream for those who wish to adopt carbon sequestration as a land management strategy.

In the highway context, there are considerable ecological, economic, and political uncertainties related to whether highway land management practices for carbon sequestration can offer a practical source of revenue. This research only examined a few of these uncertainties, particularly those relating to the amount of NHS land that might be available for carbon sequestration and, in turn, the magnitude of revenue possible should a cap-and-trade system be established. Even under the best scenarios, revenue generated from biological carbon sequestration will vary greatly from state to state based on carbon prices, management techniques, and ecological variability. However, development of a carbon market is one step toward a more complete valuation of ecosystem services. Furthermore, considering the use of vegetation for living snow fences, landslide minimization, and other such human infrastructure protection may, in some cases, eventually be found to be more cost-effective than traditional engineering solutions, especially when all costs are included.


APPENDICES

Appendix A. Estimated NHS ROW Acres, by state

  Estimated
Total NHS
Acres
Estimated Total
Acres Range
Estimated
Unpaved
NHS Acres
Estimated Unpaved
Acres Range
  LOWER UPPER   LOWER UPPER
Rhode Island8,5912,37314,8095,85069111,005
Delaware10,7642,97418,5547,32986613,788
Vermont22,5456,22838,86215,3521,81428,879
New Hampshire25,2336,97143,49517,1822,03032,322
Connecticut29,8758,25351,49620,3432,40438,268
Maine40,02611,05868,99527,2553,22151,272
Maryland45,10712,46277,75330,7153,63057,781
West Virginia57,54915,89999,19939,1874,63173,717
Massachusetts60,51916,719104,31941,2094,87077,522
New Jersey63,33117,496109,16743,1255,09681,125
Nevada68,06318,803117,32346,3465,47787,186
Utah68,61618,956118,27646,7235,52187,894
Idaho74,52320,588128,45750,7455,99795,460
South Carolina83,40023,041143,76056,7906,711106,832
North Dakota83,59923,095144,10256,9256,727107,086
Louisiana84,63523,382145,88957,6316,810108,414
Arizona86,70923,955149,46459,0436,977111,071
Mississippi87,68524,224151,14659,7087,056112,321
South Dakota90,73925,068156,41061,7877,301116,233
Indiana91,38325,246157,52062,2267,353117,057
New Mexico92,97025,684160,25563,3067,481119,090
Arkansas93,02625,700160,35263,3447,485119,162
Nebraska93,52525,838161,21363,6847,526119,802
Kentucky93,72125,892161,55063,8187,541120,052
Wyoming94,28426,047162,52064,2017,587120,773
Iowa101,03627,913174,15968,7998,130129,422
Washington101,63428,078175,19069,2068,178130,189
Oklahoma104,42528,849180,00271,1078,403133,764
Tennessee105,05929,024181,09471,5388,454134,576
Virginia107,76629,772185,76073,3828,671138,043
Colorado110,15630,432189,88075,0098,864141,105
Kansas116,99732,322201,67379,6679,414149,869
Oregon117,12032,356201,88379,7519,424150,025
Montana120,61433,321207,90782,1309,705154,501
Alabama122,34233,799210,88583,3079,844156,715
Minnesota122,77233,918211,62783,6009,879157,266
Wisconsin131,64436,369226,91989,64110,593168,630
North Carolina134,15537,062231,24791,35010,795171,846
Florida134,89737,267232,52691,85610,855172,797
Ohio139,50738,541240,47394,99511,225178,702
Georgia140,38338,783241,98395,59211,296179,825
Missouri143,20939,563246,85497,51611,523183,444
Michigan149,50041,302257,698101,80012,030191,503
New York159,80444,148275,459108,81612,859204,701
Pennsylvania170,47947,097293,861116,08513,718218,376
Illinois179,16649,497308,835122,00014,417229,504
California233,89964,618403,180159,27018,821299,614
Texas432,339119,440745,238294,39434,788553,807

Appendix B. Polygon Area Analysis Land Cover Percentages, by state

  Devel.
Open
Space
Devel.
Low
Intensity
Devel.
Med.
Intensity
Devel.
High
Intensity
De-
cidous
Forest
Mixed
Forest
Ever-
green
Forest
Cul-
tivated
Crops
Emergent
Herbaceuous
Wetlands
Hay/
Pasture
Her-
baceuous
Open
Water
Shrub/
Scrub
Woody
Wetlands
Barren
Land
ND62.6%28.7%0.3% 1.5%  2.3%  1.3%3.3%   
NE58.8%20.5%5.6%1.8%   10.8%  2.5%    
NV51.3%13.1%7.7%   1.5%   2.3% 22.8% 1.3%
CO51.0%17.5%17.0%5.6% 0.7% 0.8%0.9%0.5%4.9% 1.0%  
SC50.2%18.3%9.5%6.8% 0.7%7.0%  6.3%  1.1%  
ME49.2%28.0%14.4%  8.4%         
CT48.2%41.4%10.4%            
OR47.3%24.0%9.1%5.1% 0.7%4.8%1.8%3.6%  1.1%2.2%0.3% 
SD45.8%26.1%2.5%0.2%  1.5%9.6%0.6%6.6%7.2%    
AR45.1%26.5%10.5%2.2%4.1%1.6%3.2%5.9% 0.7%     
NM44.9%21.0%17.0%2.6%      2.2% 12.3%  
WY42.1%29.5%1.7%   1.1%3.6%  3.8% 18.1%  
OK41.6%17.5%21.1%3.0%   1.4%0.2%8.0%7.2%    
GA41.5%30.6%9.9%2.5%4.5%1.7%6.7%0.4% 1.5%0.7%    
MT40.7%18.9%7.0%    13.4% 1.1%9.2% 7.4%1.8% 
MS40.4%29.7%6.8%1.1%4.1%1.7%4.1%  4.6% 0.6%2.3%1.5%3.1%
KS39.8%30.3%7.2%3.5%1.7%0.6% 3.0% 4.5%9.4%    
TN39.7%33.0%10.5%0.6%8.8%1.8% 1.8% 2.7%   1.1% 
IA38.9%18.2%21.6%3.1%2.6%  8.3% 4.5%2.8%    
NH36.8%28.4%24.1%0.7%6.2% 2.3%  1.5%     
AL36.4%33.0%6.9%0.9%6.5%0.5%0.2%1.7% 6.9%1.1%0.6%2.0%2.9%0.5%
PA36.3%30.5%10.0%0.5%10.7%0.4%0.8%2.9%0.5%5.8%   1.7% 
NY34.4%28.1%11.5%7.1%5.3% 1.8%3.3% 6.2%  0.2%2.0% 
NC34.3%22.7%5.0%0.7%7.4%2.0%5.0%3.4%0.9%7.1%2.7%  8.8% 
TX33.9%23.4%12.5%6.7%0.6% 1.5%3.4% 3.4%5.9% 8.5% 0.1%
MA33.2%22.1%20.8%7.0%9.2% 1.8%  0.6%   4.2%1.2%
MN32.0%19.6%15.7%10.3%6.0% 2.3%6.2%0.3%4.9%0.9%  1.8% 
ID31.7%27.4%4.9%0.5%0.5% 20.6%1.4% 1.1%4.1% 6.6%1.2% 
RI28.1%28.8%22.2% 20.9%          
MO28.0%31.2%16.7%3.7%5.9%  3.0% 10.9%0.5%    
IN27.7%33.9%19.7%1.6%4.7%  10.6%0.9% 0.7%    
AZ27.2%24.8%9.5%0.5%   5.7%  3.1% 29.2%  
FL26.1%29.1%15.8%9.6% 0.9%2.4%1.6%4.3%0.6%4.2% 1.0%4.3% 
WA25.5%31.7%21.3%2.3% 0.6%2.7%3.7% 2.2%0.7% 5.0%0.5%3.8%
OH24.7%31.5%23.4%3.6%0.7%  15.6% 0.5%     
WV22.1%11.2%24.7%2.3%34.0%    5.6%     
WI21.9%29.2%16.1%3.9%3.5% 2.3%16.3%1.0%4.6%  0.2%1.0% 
CA21.8%18.1%18.7%4.7%  6.9%7.7% 1.1%7.9% 11.5% 1.6%
VA20.9%33.2%19.8%5.5%10.1% 0.5%2.5% 7.2%0.3%    
KY19.2%19.2%16.3%4.5%22.0% 3.0%8.5% 1.2%1.9%3.9% 0.3% 
IL16.6%29.5%21.5%12.2%1.8%  13.9%0.3%4.1%   0.1% 
VT15.6%23.9%4.5% 13.1%7.6%4.2%17.0% 4.5%8.1% 1.5%  
UT14.8%11.8%22.9%19.2%  0.8%     29.9%0.7% 
NJ14.6%14.6%33.1%11.5%10.6%0.8% 5.2%5.8%0.5%   1.8%1.5%
MD13.5%28.0%32.4%4.6%4.6%  8.0% 8.9%     
MI13.2%30.1%30.5%7.6%4.7%1.0%0.7%5.3%0.1%2.6%0.5%  3.8% 
LA11.8%51.3%3.9%6.3%  5.8%8.1%1.8%4.9%  3.3%2.9% 
DE4.9%26.2%16.8%4.9%   47.2%       

Appendix C. Carbon Sequestered on NHS, by state*

  Unpaved
Acres
Carbon
Sequestered
(metric tons/acre/yr)
Carbon Equilibrium
(Metric Tons of Carbon)
      Low Estimate High Estimate
RI58489392500,0861,157,842
DE732711767626,5451,450,630
VT15347246471,312,3153,038,384
NH17176275851,468,7683,400,616
CT20336326601,738,9624,026,193
ME27246437582,329,8655,394,303
MD30705493122,625,6156,079,048
WV39174629133,349,7997,755,741
MA41196661613,522,6908,156,033
NJ43110692353,686,4068,535,083
NV46331744083,961,8259,172,756
UT46708750133,994,0209,247,297
ID50728814704,337,82310,043,298
SC56771911754,854,57311,239,723
ND56906913924,866,11711,266,450
LA57612925254,926,45811,406,157
AZ59024947935,047,18911,685,683
MS59688958595,103,98611,817,184
SD61767991985,281,75012,228,759
IN62205999025,319,22512,315,524
NM632851016375,411,59212,529,381
AR633241016985,414,85812,536,942
NE636641022445,443,92412,604,238
KY637971024575,455,30212,630,582
WY641801030735,488,07312,706,457
IO687761104555,881,10413,616,435
WA691831111085,915,92113,697,046
OK710831141606,078,40214,073,235
TN715151148536,115,27514,158,608
VA733571178126,272,85414,523,449
CO749841204256,411,96314,845,526
KS796411279046,810,19715,767,551
OR797251280386,817,31515,784,031
MT821031318587,020,72016,254,973
AL832801337477,121,30116,487,845
MN835731342177,146,35316,545,847
WI896111439167,662,74317,741,438
NC913211466617,808,89318,079,817
FL918261474727,852,09018,179,831
OH949641525128,120,41818,801,087
GA955601534708,171,43218,919,199
MO974841565598,335,90519,300,000
MI1017661634378,702,10820,147,865
NY1087801747019,301,86721,536,478
PA1160471863719,923,26422,975,190
IL12196019586910,428,93724,145,967
CA15921825570313,614,81831,522,191
TX29429847264225,165,62658,265,610

*These values represent calculations from aggregated data. Observed pooled average pavement and vegetation widths for all states' transect sites were multiplied by each individual state's NHS mileage to provide ROW estimates. The project team then used average sequestration rates for grasses and coniferous and deciduous tree species to calculate the carbon sequestration estimates below. Specific numbers will vary widely from state to state, and states are strongly encouraged to use FHWA's Highway Carbon Sequestration Estimator (or other appropriate tool) to assess the return on investment for various carbon sequestration scenarios based on more state-specific considerations than possible here. The tool is available for download at www.climate.dot.gov.


Sources Consulted

Anderson, J., E. Hardy, J. Roach, and R. Witmer. 1976. A Land Use and Land Cover
Classification System for Use with Remote Sensor Data, Geological Survey Professional Paper 964. 28p. landcover.usgs.gov/pdf/anderson.pdf. Accessed 5/12/2010

American Association of State Highway and Transportation Officials. 2002. Roadside
Design Guide, 3rd Edition.

Benítez, Pablo, Ian McCallum, Michal Obersteiner, and Yoshiki Yamagata. 2006. Global
potential for carbon sequestration: geographical distribution, country risk and policy
implications.  Ecological Economics 60: 572-583.

Birdsey, Richard A. 1992. Carbon Storage and Accumulation in United States Forest
Ecosystems, General Technical Report W0-59. U.S. Department of Agriculture,
Forest Service. http://www.nrs.fs.fed.us/pubs/gtr/gtr_wo059.pdf. Accessed 5/12/2010

Birdsey, Richard A. 1996. Regional Estimates of Timber Volume and Forest Carbon for
Fully Stocked Timberland, Average Management After Final Clearcut Harvest. Forests and Global Change: Vol. 2, Forest Management Opportunities for Mitigating Carbon Emissions, R.N. Sampson and D. Hair (eds.), pp. 309-334, American Forests, Washington, DC.

Birdsey, Richard A., Pregitzer, K., Lucier, A., 2006. Forest carbon management in the
United States: 1600-2100. Journal of Environmental Quality 35, 1461-1469.
jeq.scijournals.org/cgi/reprint/35/4/1461. Accessed 5/12/2010

Congressional Budget Office. September 2007. The Potential for Carbon Sequestration in
the United States.
http://www.cbo.gov/publication/19138. Accessed 5/12/2010

Cox, Thomas S. et al. 2006. Prospects for Developing Perennial Grain Crops. BioScience.
Vol. 56 No.8. August 2006. www.landinstitute.org/pages/Bioscience_PerennialGrains.pdf. Accessed 5/12/2010

Neeff, Till, et al. 2009. The Forest Carbon Offsetting Survey 2009. EcoSecurities.
www.ecosecurities.com/Registered/ECOForestrySurvey2009.pdf. Accessed 5/12/2010

FHWA. 2009. CSPP Implementation and Next Steps Progress Report.
climate.dot.gov/documents/FINAL_C-Seq_Report_021109.pdf. Accessed 5/12/2010

Forman, Richard et al. 2003.  Road Ecology: Science and Solutions. Island Press. 
Washington, D.C.

Hatton, D.B. 1982. A Proposal of Forest Management Alternatives to the Present Method of
Vegetation Control of a Section of Interstate 95, Bangor to Newport, Maine. College
of Forest Resources, University of Main, Orono, and Maine Department of Transportation.

Homer, C. C. Huang, L. Yang, B. Wylie and M. Coan. 2004. Development of a 2001 National
Landcover Database for the United States. Photogrammetric Engineering and Remote Sensing, Vol. 70, No. 7, July 2004, pp. 829-840.

IPCC. 2007. Summary for Policymakers. In: Climate Change 2007: Mitigation. Contribution
of Working Group III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [B. Metz, O.R. Davidson, P.R. Bosch, R. Dave, L.A. Meyer (eds)], Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. www.ipcc.ch/pdf/assessment-report/ar4/wg3/ar4-wg3-spm.pdf. Accessed 5/12/2010

Row, C. 1996. Effects of Selected Forest Management Options on Carbon Storage. Forests
and Global Change, vol. 2: Forest Management Opportunities for Mitigating Carbon Emissions N. Sampson and D. Hair (eds.). American Forests, Washington, DC, pp. 59-90.

Ruben N. Lubowski, Andrew J. Plantinga, and Robert N. Stavins. January 2005. Land-Use
Change and Carbon Sinks: Econometric Estimation of the Carbon Sequestration Supply Function. Journal of Environmental Economics and Management, vol. 51, no. 2, pp. 135-152, Appendix C. www.rff.org/documents/RFF-DP-05-04.pdf. Accessed 5/12/2010

Sohngen, Brent and Robert H. Beach. 2006. Avoided Deforestation as a Greenhouse Gas
Mitigation Tool: Economic Issues for Consideration. Ohio State University and RTI International.
www-agecon.ag.ohio-state.edu/people/sohngen.1/forests/AvoidedDeforestation_v5post.pdf. Accessed 5/12/2010

Stavins, Robert and Kenneth Richards. 2005. The Cost of U.S. Forest-based Carbon
Sequestration. PEW Center on Global Climate Change. www.pewclimate.org/docUploads/Sequest_Final.pdf. Accessed 5/12/2010

Woodbury, Peter B., James E. Smith, and Linda S. Heath. 2006. Carbon Sequestration in
the U.S. Forest Sector from 1990 to 2010. USDA Forest Service. Forest Ecology and Management 241 (2007) 14-27. www.ncrs.fs.fed.us/pubs/jrnl/2007/nrs_2007_woodbury_001.pdf.
Accessed May 12, 2010

U.S. Department of Agriculture. National Agriculture Imagery Program imagery.
www.fsa.usda.gov/FSA/apfoapp?area=home&subject=prog&topic=nai. Accessed 5/12/2010

U.S. Environmental Protection Agency. 2005. Greenhouse Gas Mitigation Potential in U.S.
Forestry and Agriculture. www.epa.gov/sequestration/pdf/greenhousegas2005.pdf. Inactive 5/7/2012

U.S. Environmental Protection Agency. 2010. Motor Vehicle Emission Simulator 2010
(MOVES). www.epa.gov/otaq/models/moves/index.htm. Accessed 5/12/2010

U.S. Government. 2000. United States Submission on Land-Use, Land-Use Change, and
Forestry. www.state.gov/www/global/global_issues/climate/000801_unfccc1_subm.pdf. Accessed 5/12/2010


Endnotes

1 Ecosystem services are defined as inherent functions of natural ecosystem's that benefit human populations at little or no additional cost.  These functions include flood storage, water quality treatment, carbon sequestration, provision of wildlife habitat, genetic diversity, and landscape diversity.  Human alteration of the natural environment often eliminates or disrupts natural ecosystem functions, and requires human intervention and investment (economic cost) to replace lost functions.  Some functions, such as genetic diversity and landscape pattern diversity, are difficult, if not impossible, to replace. 

2The NHS includes the Eisenhower Interstate System; other principal arterials; the Strategic Highway Network; major strategic highway network connectors; and intermodal connectors that provide highway access between major intermodal facilities and the other subsystems.

3After assessing all 50 states, Washington D.C., and Puerto Rico against a number of criteria, New Mexico DOT (NMDOT) and Minnesota DOT (Mn/DOT) were selected to participate in the pilot program. Details, methods, results, and lessons learned from the NMDOT pilot are documented in FHWA's February 2009 CSPP Implementation and Next Steps Progress Report and not reported here.

4Hatton (1982) studied a section of highway in Maine to estimate the state's total highway ROW acreage that could potentially be used for forestry purposes. Forman (2002) offers an approach for making a more rigorous estimate of total road surface and roadside areas. Other literature providing unpaved highway ROW acreage estimates was unavailable.

5Mn/DOT's Right of Way Mapping and Monitoring application: www.dot.state.mn.us/maps/gisweb/row/

6 The NAIP acquires aerial imagery during the agricultural growing seasons in the continental United States. Imagery is available for all lower 48 states. Beginning in 2003, NAIP was acquired on a 5-year cycle. 2008 was a transition year, and a three-year cycle began in 2009: www.fsa.usda.gov/FSA/apfoapp?area=home&subject=prog&topic=nai.

7 Minnesota's Department of Natural Resources offered more detailed Gap Analysis Program land cover imagery, which allows for identification down to the species level. These data are paid for by square mile and were not consulted due to the fact that this analysis required only distinguishing between forested and non-forested land covers.

8For NLCD land cover class definitions, see www.mrlc.gov/nlcd_definitions.php.

9NLCD 1992/2001 Retrofit Change Data definitions: www.mrlc.gov/changeproduct_definitions.php

10Equation: 70,650 sq ft polygon area / 24,600 sq ft predicted NHS pavement area [300' road length per polygon x 82' average pavement width from transect analysis] = 34.8%

11 The spreadsheets used to record the manual transect measurements, as well as more detailed instructions for repeating the polygon area analysis, are available upon request. Contact Carson Poe at carson.poe@dot.gov for more information

12Congressional Budget Office. The Potential for Carbon Sequestration in the United States. 2007. www.cbo.gov/publication/19138

13CCX operates North America's only cap and trade system for all six greenhouse gases. GHG reductions achieved through CCX are the only reductions made in North America through a legally binding compliance regime, providing independent, third party verification.

14Avoided deforestation (AD) refers to the protection of existing forests by reducing deforestation and forest degradation rates. Carbon sequestration rates for AD are higher than those for reforestation because its alternative - deforestation - creates significant emissions itself; it has been estimated that tropical deforestation accounts for as much as 25 percent of global human-caused GHG emissions (Houghton 2005). Sedjo and Sohngen (2006) note "[n]ot only is carbon lost to the atmosphere from net reductions in forest cover, but newly afforested or reforested lands store far less carbon per hectare (currently) than mature stands being deforested. In addition, the geographical variation in forest cover trends has important implications for carbon emissions because of the large differences in carbon stock per hectare across regions. In general, the tropical areas experiencing net deforestation have higher carbon stocks in forest biomass per hectare than temperate regions experiencing net afforestation." Recent data have indicated that a market for AD carbon offsets is highly desirable among industry sectors: www.ecosecurities.com/Registered/ECOForestrySurvey2009.pdf

15The IPCC defines afforestation as the "planting of new forests on lands that, historically, have not contained forests. EPA defines afforestation more broadly as "the establishment of trees on lands that were without trees for some period of time." According to EPA, [d]iffering interpretations of this time period will dictate whether the establishment of forest cover is considered to represent afforestation or reforestation." www.epa.gov/sequestration/pdf/ghg_part3.pdf

16This calculation assumes average passenger car emissions are 5.0292 metric tons of CO2-eq per car per year (U.S. EPA 2010), or 1.371463 metric tons of carbon/car/year. 3.6 MMT estimated annual sequestration potential of NHS / 1.371463 MT/C/car/year = 2.6 million passenger cars.

17The American Association of State and Highway Transportation Official's Roadside Design Guide, 3rd Edition defines a "clear zone" as the total roadside border area, starting at the edge of the traveled way, available for safe use by errant vehicles. The desired minimum width is dependent upon traffic volumes and speeds and on the roadside geometry. Simply stated, it is an unobstructed, relatively flat area beyond the edge of the traveled way that allows a driver to stop safely or regain control of a vehicle that leaves the traveled way.

18Less frequent mowing may not affect carbon sequestration rates of grasses but can reduce maintenance costs and carbon emissions.

Updated: 10/31/2014
HEP Home Planning Environment Real Estate
Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000