Citation: Smith, James E., and Linda S. Heath. 2006. Land use change and forestry and Annex
3.12. (Excerpted.) In: US EPA. Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990 2004. EPA 430-R-06-002. Washington, DC. Available at http://yosemite.epa.gov/oar/
globalwarming.nsf/content/ResourceCenterPublicationsGHGEmissionsUSEmissions
Inventory2006.html (18 August).
Note that EPA contractors or employees may have written or edited some text, or perhaps
formatted tables or redrew some figures for final EPA format.
EPA 430-R-06-002
INVENTORY OF GREENHOUSE GAS EMISSIONS AND SINKS:
1990-2004
April 15, 2006
U.S. Environmental Protection Agency
1200 Pennsylvania Ave., N.W.
Washington, DC 20460
U.S.A.
7. Land-Use Change and Forestry
7.1 Forest Land Remaining Forest Land
Changes in Forest Carbon Stocks (IPCC Source Category 5A1)
For estimating carbon (C) stocks or stock change (flux), C in forest ecosystems can be divided into the following
five storage pools (IPCC 2003):
•
Aboveground biomass, all living biomass above the soil including stem, stump, branches, bark, seeds, and
foliage. This category includes live understory.
•
Belowground biomass, all living biomass of coarse living roots greater than 2 mm diameter.
•
Dead wood, including all non-living woody biomass either standing, lying on the ground (but not including
litter), or in the soil.
•
Litter, including the litter, fumic, and humic layers, and all non-living biomass with a diameter less than 7.5
cm at transect intersection, lying on the ground.
•
Soil organic carbon (SOC), including all organic material in soil to a depth of 1 meter but excluding the
coarse roots of the above pools.
In addition, there are two harvested wood pools also necessary for estimating C flux, which are:
•
Harvested wood products in use.
•
Harvested wood products in landfills.
Carbon is continuously cycled among these storage pools and between forest ecosystems and the atmosphere as a
result of biological processes in forests (e.g., photosynthesis, growth, mortality, decomposition, and disturbances
such as fires or pest outbreaks) and anthropogenic activities (e.g., harvesting, thinning, clearing, and replanting). As
trees photosynthesize and grow, C is removed from the atmosphere and stored in living tree biomass. As trees age,
they continue to accumulate C until they reach maturity, at which point they store a relatively constant amount of C.
As trees die and otherwise deposit litter and debris on the forest floor, C is released to the atmosphere or transferred
to the soil by organisms that facilitate decomposition.
The net change in forest C is not equivalent to the net flux between forests and the atmosphere because timber
harvests do not cause an immediate flux of C to the atmosphere. Instead, harvesting transfers C to a "product pool."
Once in a product pool, the C is emitted over time as CO2 when the wood product combusts or decays. The rate of
emission varies considerably among different product pools. For example, if timber is harvested to produce energy,
combustion releases C immediately. Conversely, if timber is harvested and used as lumber in a house, it may be
many decades or even centuries before the lumber decays and C is released to the atmosphere. If wood products are
disposed of in landfills, the C contained in the wood may be released many years or decades later, or may be stored
almost permanently in landfills.
This section quantifies the net changes in C stocks in the five forest C pools and two harvested wood pools. The net
change in stocks for each pool is estimated, and then the changes in stocks are summed over all pools to estimate
total net flux. Thus, the focus on C implies that all C-based greenhouse gases are included, and the focus on stock
change suggests that specific ecosystem fluxes are not separately itemized in this report. Disturbances from forest
fires and pest outbreaks are implicitly included in the net changes. For instance, an inventory conducted after fire
counts only trees left. The change between inventories thus counts the carbon changes due to fires; however, it may
not be possible to attribute the changes to the disturbance specifically. The IPCC Good Practice Guidance for Land
Use, Land-Use Change, and Forestry (IPCC 2003) recommends reporting C stocks according to several land use
types and conversions, specifically Forest Land Remaining Forest Land and Land Converted to Forest Land.
Currently, consistent datasets are not available for the entire United States to allow results to be partitioned in this
way. Instead, net changes in all forest-related land, including non-forest land converted to forest and forests
converted to non-forest are reported here.
Forest C storage pools, and the flows between them via emissions, sequestration, and transfers, are shown in Figure
7-1. In the figure, boxes represent forest C storage pools and arrows represent flows between storage pools or
between storage pools and the atmosphere. Note that the boxes are not identical to the storage pools identified in
2
this chapter. The storage pools identified in this chapter have been altered in this graphic to better illustrate the
processes that result in transfers of C from one pool to another, and emissions to the atmosphere as well as uptake
from the atmosphere.
Figure 7-1: Forest Sector Carbon Pools and Flows.
Approximately 33 percent (303 million hectares) of the U.S. land area is forested. Approximately 250 million
hectares are located in the conterminous 48 states and form the basis for the estimates provided in this chapter.
Seventy-nine percent of the 250 million hectares are classified as timberland, meaning they meet minimum levels of
productivity and are available for timber harvest. Historically, the timberlands in the conterminous 48 states have
been more frequently or intensively surveyed than other forestlands. Of the remaining 51 million hectares, 16
million hectares are reserved forestlands (withdrawn from management for production of wood products) and 35
million hectares are lower productivity forestlands (Smith et al. 2004b). From the early 1970s to the early 1980s,
forest land declined by approximately 2.4 million hectares. During the 1980s and 1990s, forest area increased by
about 3.7 million hectares. These net changes in forest area represent average annual fluctuations of only about 0.1
percent. Given the low rate of change in U.S. forest land area, the major influences on the current net C flux from
forest land are management activities and the ongoing impacts of previous land-use changes. These activities affect
the net flux of C by altering the amount of C stored in forest ecosystems. For example, intensified management of
forests can increase both the rate of growth and the eventual biomass density 1 of the forest, thereby increasing the
uptake of C. Harvesting forests removes much of the aboveground C, but trees can grow on this area again and
sequester C. The reversion of cropland to forest land increases C storage in biomass, forest floor, and soils. The net
effects of forest management and the effects of land-use change involving forest lands are captured in the estimates
of C stocks and fluxes presented in this chapter.
In the United States, improved forest management practices, the regeneration of previously cleared forest areas, as
well as timber harvesting and use have resulted in net uptake (i.e., net sequestration) of C each year from 1990
through 2004. Due to improvements in U.S. agricultural productivity, the rate of forest clearing for crop cultivation
1 The term “biomass density” refers to the mass of vegetation per unit area. It is usually measured on a dry-weight basis. Dry
biomass is about 50 percent carbon by weight.
3
and pasture slowed in the late 19th century, and by 1920 this practice had all but ceased. As farming expanded in
the Midwest and West, large areas of previously cultivated land in the East were taken out of crop production,
primarily between 1920 and 1950, and were allowed to revert to forests or were actively reforested. The impacts of
these land-use changes still affect C fluxes from forests in the East. In addition, C fluxes from eastern forests have
been affected by a trend toward managed growth on private land. Collectively, these changes have nearly doubled
the biomass density in eastern forests since the early 1950s. More recently, the 1970s and 1980s saw a resurgence
of federally-sponsored forest management programs (e.g., the Forestry Incentive Program) and soil conservation
programs (e.g., the Conservation Reserve Program), which have focused on tree planting, improving timber
management activities, combating soil erosion, and converting marginal cropland to forests. In addition to forest
regeneration and management, forest harvests have also affected net C fluxes. Because most of the timber harvested
from U.S. forests is used in wood products, and many discarded wood products are disposed of in landfills rather
than by incineration, significant quantities of C in harvested wood are transferred to long-term storage pools rather
than being released rapidly to the atmosphere (Skog and Nicholson 1998). The size of these long-term C storage
pools has increased during the last century.
Changes in C stocks in U.S. forests and harvested wood were estimated to account for an average annual net
sequestration of 627 Tg CO2 Eq. (171 Tg C) over the period 1990 through 2004 (Table 7-5, 7-6, and Figure 7-2). In
addition to the net accumulation of C in harvested wood pools, sequestration is a reflection of net forest growth and
increasing forest area over this period, particularly before 1997. The increase in forest sequestration is due more to
an increasing C density per area than to the increase in area of forestland. Forestland in the conterminous United
States was approximately 246, 250, and 251 million hectares for 1987, 1997, and 2002, respectively, only a 2
percent increase over the period (Smith et al. 2004b). Continuous, regular annual surveys are not available over the
period for each state; therefore, estimates for non-survey years were derived by interpolation between known data
points. Survey years vary from state to state. National estimates are a composite of individual state surveys. Total
sequestration declined by 21 percent between 1990 and 2004. Estimated sequestration in the litter carbon pool had
the greatest effect on total change; the net rate of accumulation in litter decreased by 56 Tg CO2 Eq. Aboveground
biomass and soil carbon had the next largest effects on total change; the net rate of accumulation in these pools
decreased by 28 and 24 Tg CO2 Eq., respectively.
Table 7-5. Net Annual Changes in Carbon Stocks (Tg CO2/yr) in Forest and Harvested Wood Pools
Carbon Pool
1990
1998
1999
2000
2001
2002
2003
Forest
(563.3)
(412.7) (423.2) (420.2) (420.2) (420.2) (420.2)
Aboveground Biomass
(338.5)
(287.5) (306.6) (310.3) (310.3) (310.3) (310.3)
Belowground Biomass
(64.8)
(55.1) (59.5) (60.3) (60.3) (60.3) (60.3)
Dead Wood
(43.5)
(41.6) (35.5) (33.2) (33.2) (33.2) (33.2)
Litter
(82.9)
(12.4) (24.9) (26.6) (26.6) (26.6) (26.6)
Soil Organic Carbon
(33.6)
(16.0)
3.2
10.1
10.1
10.1
10.1
(206.1 (214.7) (210.8) (213.8) (214.4) (215.6)
Harvested Wood
(210.1)
Wood Products
(47.6)
(51.9) (61.5) (58.7) (59.0) (59.2) (60.4)
Landfilled Wood
(162.4)
(154.2) (153.1) (152.1) (152.8) (155.3) (155.1)
Total Net Flux
(773.4)
(618.8) (637.9) (631.0) (634.0) (634.6) (635.8)
2004
(420.2)
(310.3)
(60.3)
(33.2)
(26.6)
10.1
(217.0)
(60.8)
(156.2)
(637.2)
Note: Parentheses indicate net C sequestration (i.e., a net removal of C from the atmosphere). Total net flux is an estimate of the actual net flux
between the total forest C pool and the atmosphere. Forest estimates are based on interpolation and extrapolation of inventory data as described
in the text and in Annex 3.12. Harvested wood estimates are based on results from annual surveys and models. Totals may not sum due to
independent rounding.
Table 7-6. Net Annual Changes in Carbon Stocks (Tg C/yr) in Forest and Harvested Wood Pools
Carbon Pool
1990
1998
1999
2000
2001
2002
2003
2004
Forest
(153.6)
(112.6) (115.4) (114.6) (114.6) (114.6) (114.6) (114.6)
Aboveground Biomass
(92.3)
(78.4) (83.6) (84.6) (84.6) (84.6) (84.6) (84.6)
Belowground Biomass
(17.7)
(15.0) (16.2) (16.4) (16.4) (16.4) (16.4) (16.4)
Dead Wood
(11.9)
(11.4)
(9.7)
(9.1)
(9.1)
(9.1)
(9.1)
(9.1)
Litter
(22.6)
(3.4)
(6.8)
(7.2)
(7.2)
(7.2)
(7.2)
(7.2)
Soil Organic Carbon
(9.2)
(4.4)
0.9
2.8
2.8
2.8
2.8
2.8
(56.2) (58.5) (57.5) (58.3) (58.5) (58.8) (59.2)
Harvested Wood
(57.3)
Wood Products
(13.0)
(14.2) (16.8) (16.0) (16.1) (16.1) (16.5) (16.6)
Landfilled Wood
(44.3)
(42.1) (41.8) (41.5) (42.2) (42.3) (42.3) (42.6)
Total Net Flux
(210.9)
(168.8) (174.0) (172.1) (172.9) (173.1) (173.4) (173.8)
4
Note: Parentheses indicate net C sequestration (i.e., a net removal of C from the atmosphere). Total net flux is an estimate of the actual net flux
between the total forest C pool and the atmosphere. Forest estimates are based on interpolation and extrapolation of inventory data as described
in the text and in Annex 3.12. Harvested wood estimates are based on results from annual surveys and models. Totals may not sum due to
independent rounding.
Stock estimates for forest and harvested wood C storage pools are presented in Table 7-7. Together, the
aboveground live and forest soil pools account for a large proportion of total forest C stocks. C stocks in all non-soil
pools increased over time. Therefore, C sequestration was greater than C emissions from forests, as discussed
above. Figure 7-3 shows the carbon-average densities for live trees on forest land, including both above- and
belowground biomass.
Table 7-7. Carbon Stocks (Tg C) in Forest and Harvested Wood Pools
Carbon Pool
1990
1998
1999
2000
2001
Forest
39,508
40,812 40,529 40,645 40,760
Aboveground Biomass
14,334
14,938 15,016 15,100 15,184
Belowground Biomass
2,853
2,967
2,982
2,998
3,014
Dead Wood
2,409
2,488
2,499
2,509
2,518
Litter
4,492
4,565
4,569
4,575
4,583
Soil Organic Carbon
15,420
15,460 15,464 15,463 15,460
2,365
2,421
2,480
2,537
Harvested Wood
1,915
Wood Products
1,134
1,248
1,262
1,279
1,295
Landfilled Wood
781
1,117
1,159
1,200
1,242
Total Carbon Stock
41,423
42,782 42,951 43,125 43,297
2002
40,874
15,269
3,031
2,627
4,590
15,458
2,595
1,311
1,284
43,470
2003
40,989
15,499
3,074
2,638
4,636
15,455
2,654
1,327
1,327
43,643
2004
41,103
15,438
3,064
2,545
4,604
15,452
2,713
1,344
1,369
43,816
1998
41,218
15,723
3,080
2,554
4,612
15,449
2,772
1,360
1,411
43,990
Note: Forest C stocks do not include forest stocks in Alaska, Hawaii, or U.S. territories, or trees on non-forest land (e.g., urban
trees). Wood product stocks include exports, even if the logs are processed in other countries, and exclude imports. Forest
estimates are based on interpolation and extrapolation of inventory data as described in the text and in Annex 3.12. Harvested
wood estimates are based on results from annual surveys and models. Totals may not sum due to independent rounding.
Inventories are assumed to represent stocks as of January 1 of the inventory year. Flux is the net annual change in stock. Thus,
an estimate of flux for 2004 requires estimates of C stocks for 2004 and 2005.
Figure 7-2: Estimates of Net Annual Changes in Carbon Stocks for Major Carbon Pools (Tg C yr-1)
Note: Estimates for harvested wood are based on the same methodology and data as the previous U.S. Inventory (EPA 2005).
Estimates for all pools are based on measured forest inventory data as described in the text. Total Net includes all forest pools:
biomass, dead wood, litter, forest soils, wood products, and landfilled wood.
Methodology
The methodology described herein is consistent with Good Practice Guidance for Land Use, Land Use Change, and
Forestry (IPCC 2003) and the Revised 1996 IPCC Guidelines for National Greenhouse Gas Inventories
(IPCC/UNEP/OECD/IEA 1997). Estimates of net C flux from Land-Use Change and Forestry, including all pools
5
except harvested wood, were derived from periodic and annualized inventories of forest stocks. Net changes in C
stocks were interpolated between survey years. Carbon emissions from harvested wood were determined by
accounting for the variable rate of decay of harvested wood according to its disposition (e.g., product pool, landfill,
combustion). 2 Different data sources were used to estimate the C stocks and stock change in (1) forests
(aboveground and belowground biomass, dead wood, and litter), (2) forest soils, and (3) harvested wood products.
Therefore, these pools are described separately below.
Figure 7-3: Average Carbon Density in the Forest Tree Pool in the Conterminous U.S. During 2005.
Live Biomass, Dead Wood, and Litter Carbon
The estimates of non-soil forest C stocks are based on data derived from forest inventory surveys. Forest survey
data were obtained from the USDA Forest Service, Forest Inventory and Analysis (FIA) program (Frayer and
Furnival 1999, Smith et al. 2001). Surveys provide estimates of the merchantable volume of wood and other
variables that are used to estimate C stocks. Estimates of temporal change such as growth, mortality, harvests, or
area change are derived from repeated surveys, which were conducted every 5 to 14 years, depending on the state.
Historically, the FIA program did not conduct detailed surveys of all forest land, but instead focused on land capable
of supporting timber production (timberland) 3 . Over time, however, individual state surveys gradually started to
2 The wood product stock and flux estimates presented here use the production approach, meaning that they do not account for C
stored in imported wood products, but do include C stored in exports, even if the logs are processed in other countries. This
approach is used because it follows the precedent established in previous reports (Heath et al. 1996).
3 Forest land in the United States includes land that is at least 10 percent stocked with trees of any size. Timberland is the most
productive type of forest land, which is on unreserved land and is producing or capable of producing crops of industrial wood.
Productivity is at a minimum rate of 20 cubic feet of industrial wood per acre per year. The remaining portion of forest land is
6
include reserved and less productive forest lands. The C stock estimates provided here include all forest land, see
Annex 3.12 for discussion of how past data gaps on these lands were filled.
Temporal and spatial gaps in surveys were addressed with the new national plot design and annualized sampling
(Alerich et al. 2005), which were recently introduced by FIA. Annualized sampling means that a portion of plots
throughout each state is sampled each year, with the goal of measuring all plots once each 5 years. Sampling is
designed such that partial inventory cycles provide usable, unbiased samples of forest inventory. Thus, many states
have relatively recent partial inventories, yet not all states are currently surveyed this way. All annualized surveys
initiated since 1998 have followed the new national plot design for all forestlands, including reserved and less
productive lands.
For each periodic or annualized inventory in each state, each C pool was estimated using coefficients from the
FORCARB2 model (Birdsey and Heath 1995, Birdsey and Heath 2001, Heath et al. 2003, Smith et al. 2004a).
Estimates of C stocks made by the FORCARB2 coefficients at the plot level are organized somewhat differently
than the standard IPCC pools reported in Table 7-7. However, the estimators are compatible with reorganizing the
pools following IPCC LULUCF Good Practice Guidance (2003). For example, the biomass pools here include the
FORCARB2 pools of live trees and understory vegetation, each of which are divided into aboveground versus
belowground portions. Calculations for the tree portion of the aboveground biomass C pool were made using
individual-tree or volume-to-biomass conversion factors for different types of forests, depending on the data
available for each survey (Jenkins et al. 2003, Smith et al. 2003). Biomass was converted to C mass by dividing by
two because dry biomass is approximately 50 percent C (IPCC/UNEP/OECD/IEA 1997). The other portion of
aboveground biomass, live understory C, was estimated from inventory data using tables presented in Birdsey
(1996). Litter C was estimated from inventory data using the equations presented in Smith and Heath (2002). Down
dead wood was estimated using a FORCARB2 simulation and U.S. forest statistics (Smith et al. 2001).
Forest Soil Carbon
Estimates of soil organic carbon stocks are based solely on forest area and on average soil C density for each broad
forest type group. Thus, any changes in soil C stocks are due to changes in total forest area or the distribution of
forest types within that area. Estimates of the organic C content of soils are based on the national STATSGO spatial
database (USDA 1991) and follow methods of Amichev and Galbraith (2004). These data were overlaid with FIA
survey data to estimate soil C on forest lands by broad forest type group.
Forest Carbon Stocks and Fluxes
The overall approach for determining forest C stock change was to estimate forest C stocks based on data from two
forest surveys conducted several years apart. Carbon stocks were calculated separately for each state based on
inventories available since 1990 and for the most recent inventory prior to 1990. For each pool in each state in each
year, C stocks were estimated by linear interpolation between survey years. Similarly, fluxes were estimated for
each pool in each state by dividing the difference between two successive stocks by the number of intervening years
between surveys. Stocks and fluxes since the most recent survey were based on extrapolating estimates from the last
two surveys. C stock and flux estimates for each pool were summed over all states to form estimates for the
conterminous United States. Data sources and methods for estimating individual C pools are described more fully in
Annex 3.12.
Harvested Wood Carbon
Estimates of C stock changes in wood products and wood discarded in landfills were based on the methods
described by Skog and Nicholson (1998). Carbon stocks in wood products in use and wood products stored in
landfills were estimated from 1910 onward based on historical data from the USDA Forest Service (USDA 1964,
Ulrich 1989, Howard 2001), and historical data as implemented in the framework underlying the North American
classified as either reserved forest land, which is forest land withdrawn from timber use by statute or regulation, or other forest
land, which includes less productive forests on which timber is growing at a rate less than 20 cubic feet per acre per year. In
2002, there were about 199 million hectares of timberland in the conterminous United States, which represented 79 percent of all
forest lands over the same area (Smith et al. 2004b).
7
Pulp and Paper (NAPAP, Ince 1994) and the Timber Assessment Market and the Aggregate Timberland Assessment
System Timber Inventory models (TAMM/ATLAS, Haynes 2003, Mills and Kincaid 1992). Beginning with data on
annual wood and paper production, the fate of C in harvested wood was tracked for each year from 1910 through
2004, and included the change in C stocks in wood products, the change in C in landfills, and the amount of C
emitted to the atmosphere (CO2 and CH4) both with and without energy recovery. To account for imports and
exports, the production approach was used, meaning that C in exported wood was counted as if it remained in the
United States, and C in imported wood was not counted.
Uncertainty
The forest survey data that underlie the forest C estimates are based on a statistical sample designed to represent the
wide variety of growth conditions present over large territories. However, forest survey data that are currently
available generally exclude timber stocks on most forest land in Alaska, Hawaii, and U.S. territories. For this
reason, estimates have been developed only for the conterminous United States. Within the conterminous United
States, the USDA Forest Service mandates that forest area data are accurate within 3 percent at the 67 percent
confidence level (one standard error) per 405,000 ha (106 acres) of timberland (Alerich et al. 2005). For larger
areas, the uncertainty in area is concomitantly smaller. For growing stock volume on timberlands, the accuracy is
targeted to be 5 or 10 percent for each 28.3 million m3 (109 cubic feet) at the same confidence level. An analysis of
uncertainty in growing stock volume data for timber producing lands in the Southeast by Phillips et al. (2000) found
that nearly all of the uncertainty in their analysis was due to sampling rather than the regression equations used to
estimate volume from tree height and diameter. Standard errors for growing stock volume ranged from 1 to 2
percent for individual states and less than 1 percent for the 5-state region. However, the total standard error for the
change in growing stock volume was estimated to be 12 to 139 percent for individual states, and 20 percent for the
5-state region. The high relative uncertainty for growing stock volume change in some states was due to small net
changes in growing stock volume. However, the uncertainty in volume change may be smaller than was found in
this study because estimates from samples taken at different times on permanent survey plots are correlated, and
such correlation reduces the uncertainty in estimates of changes in volume or C over time (Smith and Heath 2000).
In addition to uncertainty in data summarized for inventory surveys, there is uncertainty associated with the
estimates of specific C stocks in those forest ecosystems. Estimates for these pools are derived from extrapolations
of site-specific studies to all forest land since survey data on these pools are not generally available. Such
extrapolation introduces uncertainty because available studies may not adequately represent regional or national
averages. Uncertainty may also arise due to: (1) modeling errors, (e.g., relying on coefficients or relationships that
are not well known); and (2) errors in converting estimates from one reporting unit to another (Birdsey and Heath
1995). An important source of uncertainty is that there is little consensus from available data sets on the effect of
land use change and forest management activities (such as harvest) on soil C stocks. For example, while Johnson
and Curtis (2001) found little or no net change in soil C following harvest, on average, across a number of studies,
many of the individual studies did exhibit differences. Heath and Smith (2000a) noted that the experimental design
in a number of soil studies limited their usefulness for determining effects of harvesting on soil C. Because soil C
stocks are large, estimates need to be very precise, since even small relative changes in soil C sum to large
differences when integrated over large areas. The soil C stock and stock change estimates presented herein are
based on the assumption that soil C density for each broad forest type group stays constant over time. As more
information becomes available, the effects of land use and of changes in land use and forest management will be
better accounted for in estimates of soil C (see “Planned Improvements,” below).
A quantitative uncertainty analysis was developed for the estimates of C stock and flux presented here. The analysis
incorporated the information discussed above as well as preliminary uncertainty analyses of previous C estimates
developed according to the same or similar methodologies as applied here (Heath and Smith 2000b, Smith and
Heath 2000, Skog et al. 2004). Some additional details on the analysis are provided in Annex 3.12. The uncertainty
analysis was performed using the IPCC-recommended Tier 2 uncertainty estimation methodology, Monte Carlo
Simulation technique. The results of the Tier 2 quantitative uncertainty analysis are summarized in Table 7-8. The
2004 flux estimate for forest C stocks is estimated to be between -794.7 and -476.3 Tg CO2 Eq. at a 95 percent
confidence level (i.e., 19 out of every 20 Monte Carlo stochastic simulations fall within this interval). This indicates
a range of 24.7 percent below to 25.2 percent above the 2004 flux estimate of -637.2 Tg CO2 Eq. The 95 percent
confidence intervals for the two principal components of total flux are -546 to -294 Tg CO2 Eq. for forest
ecosystems and -297 to -136 Tg CO2 Eq. for harvested wood.
8
Table 7-8: Tier 2 Quantitative Uncertainty Estimates for CO2 Net Flux from Forest Land Remaining Forest Land:
Changes in Forest Carbon Stocks (Tg CO2 Eq. and Percent)
2004 Flux
Source
Gas
Uncertainty Range Relative to Flux Estimatea
Estimate
(Tg CO2 Eq.)
(%)
(Tg CO2 Eq.)
Lower
Upper
Lower
Upper
Bound
Bound
Bound
Bound
Forests Remaining
Forests: Changes in Forest
Carbon Stocks
(637.2)
(794.7)
(476.3)
-25%
+25%
CO2
Note: Parentheses indicate negative values or net sequestration.
a
Range of flux estimates predicted by Monte Carlo Stochastic Simulation for a 95 percent confidence interval.
QA/QC and Verification
As discussed above, the FIA program has conducted consistent forest surveys based on extensive statistically-based
sampling of most of the forest land in the conterminous United States since 1952. The main purpose of the Forest
Inventory and Analysis program has been to estimate areas, volume of growing stock, and timber products output
and utilization factors. The FIA program includes numerous quality assurance and quality control procedures,
including calibration among field crews, duplicate surveys of some plots, and systematic checking of recorded data.
Because of the statistically-based sampling, the large number of survey plots, and the quality of the data, the survey
databases developed by the FIA program form a strong foundation for C stock estimates. Field sampling protocols,
summary data, and detailed inventory databases are archived and are publicly available on the Internet (FIA
Database Retrieval System).
Many key calculations for estimating current forest C stocks based on FIA data are based on coefficients from the
FORCARB2 model (see additional discussion in the Methodology section above and in Annex 3.12). The model
has been used for many years to produce national assessments of forest C stocks and stock changes. General quality
control procedures were used in performing calculations to estimate C stocks based on survey data. For example,
the derived C datasets, which include inventory variables such as areas and volumes, were compared with standard
inventory summaries such as Resources Planning Act (RPA) Forest Resource Tables or selected population
estimates generated from the FIA Database (FIADB), which are available at an FIA Internet site (FIA Database
Retrieval System). Agreement between the C datasets and the original inventories is important to verify accuracy of
the data used. Finally, C stock estimates were compared with previous inventory report estimates to assure that any
differences could be explained by either new data or revised calculation methods (see the “Recalculations”
discussion below).
Recalculations Discussion
The overall scheme for developing annualized estimates of C stocks based on the individual state surveys is similar
to that presented in the previous Inventory (EPA 2005). The change from the previous year’s methods involves the
use of survey data. This year, the emphasis was on using all available state surveys in the FIADB, with RPA data
used as necessary to estimate pre-1990 stocks. In the previous inventory, the FIADB was used to supplement the
RPA datasets. Additionally, the FIADB has been updated over the last year.
The modifications and updates to the forest inventory data are detailed in Table A-180 in Annex 3.12 (the forest
carbon methodology annex) and can be compared with forest inventories identified in a similar table in the previous
U.S. Greenhouse Gas Inventory (EPA 2005). These changes are reflected in estimates of forest carbon stocks.
Biomass stocks prior to 1996 were revised upward slightly, and biomass stocks after 1997 were revised downward.
Stocks of dead wood were revised downward throughout, with greater changes in more recent years. The net effect
is an average decrease in estimated forest carbon stocks of less than 1 percent for the period 1990 through 2003.
These comparisons can be independently calculated by referring to Table A-183 in this Inventory and the analogous
table in the previous Inventory (EPA 2005). Overall, these changes resulted in an average annual decrease of 206
Tg CO2 Eq. (24 percent) in the net change in forest carbon stocks for the period 1990 through 2003.
9
Planned Improvements
The ongoing annualized surveys by the FIA Program will improve precision of forest C estimates as new state
surveys become available (Gillespie 1999). In addition, the more intensive sampling of down dead wood, litter, and
soil organic C on some of the permanent plots will substantially improve resolution of C pools at the plot level.
As more information becomes available about historical land use, the ongoing effects of changes in land use and
forest management will be better accounted for in estimates of soil C (Birdsey and Lewis 2003). Currently, soil C
estimates are based on the assumption that soil C density depends only on broad forest type group, not on land use
history. However, many forests in the Eastern United States are re-growing on abandoned agricultural land. During
such regrowth, soil and forest floor C stocks often increase substantially over many years or even decades,
especially on highly eroded agricultural land. In addition, with deforestation, soil C stocks often decrease over many
years. A new methodology is being developed to account for these changes in soil C over time. This methodology
includes estimates of area changes among land uses (especially forest and agriculture), estimates of the rate of soil C
stock gain with afforestation, and estimates of the rate of soil C stock loss with deforestation over time. This topic is
important because soil C stocks are large, and soil C flux estimates contribute substantially to total forest C flux, as
shown in Table 7-6 and Figure 7-2.
The estimates of C stored in harvested wood products are currently being revised using more detailed wood products
production and use data, and more detailed parameters on disposition and decay of products.
References
Alerich, C.L., L. Klevgard, C. Liff, P.D. Miles, and B. Knight. (2005) The Forest Inventory and Analysis Database:
Database Description and Users Guide Version 2.0. Available online at <
http://ncrs2.fs.fed.us/4801/fiadb/index.htm>. Accessed September 2005.
Amichev, B. Y. and J. M. Galbraith (2004) A Revised Methodology for Estimation of Forest Soil Carbon from
Spatial Soils and Forest Inventory Data Sets. Environmental Management 33, Supplement 1: S74-S86.
Birdsey, R.A., and L.S. Heath (1995) “Carbon Changes in U.S. Forests.” In: Productivity of America’s Forests and
Climate Change. Gen. Tech. Rep. RM-271. Rocky Mountain Forest and Range Experiment Station, Forest Service,
U.S. Department of Agriculture, Fort Collins, CO, pp. 56-70.
Birdsey, R. (1996) Carbon Storage for Major Forest Types and Regions in the Conterminous United States. Pages 126 and 261-379 (appendices 262 and 263) in R. N. Sampson and D. Hair, editors. Forest and Global Change Volume
2: Forest Management Opportunities for Mitigating Carbon Emissions. American Forests, Washington DC.
Birdsey, R., and L. S. Heath (2001) Forest Inventory Data, Models, and Assumptions for Monitoring Carbon Flux.
Pages 125-135 in Soil Carbon Sequestration and the Greenhouse Effect. Soil Science Society of America, Madison,
WI.
Birdsey, R. A., and G. M. Lewis. (2003) Current and Historical Trends in Use, Management, and Disturbance of U.
S. Forestlands. Pages 15-34 in J. M. Kimble, L. S. Heath, R. A. Birdsey, and R. Lal, editors. The Potential of U.S.
Forest Soils to Sequester Carbon and Mitigate the Greenhouse Effect. CRC Press, New York.
EPA. (2005) Inventory of U. S. Greenhouse Gas Emissions and Sinks: 1990 - 2003. EPA, (Environmental Protection
Agency), Office of Atmospheric Programs, Washington, DC, U. S. A.
FIA Database Retrieval System. Date created is unknown. U.S. Department of Agriculture, Forest Service.
Available online at <http://ncrs2.fs.fed.us/4801/fiadb/index.htm>. Accessed November 2005.
Frayer, W.E., and G.M. Furnival (1999) “Forest Survey Sampling Designs: A History.” Journal of Forestry 97(12):
4-10.
Gillespie, A.J.R. (1999) “Rationale for a National Annual Forest Inventory Program.” Journal of Forestry 97(12):
16-20.
Haynes, R.W. (2003) An Analysis of the Timber Situation in the United States: 1952-2050. Gen. Tech. Rep. PNWGTR-560., U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station, Portland, OR.
Heath, L.S., R.A. Birdsey, C. Row, and A.J. Plantinga (1996) “Carbon Pools and Fluxes in U.S. Forest Products.”
10
In: Apps, M.J. and Price, D.T. (eds.) Forest Ecosystems, Forest Management and the Global Carbon Cycle.
Springer-Verlag, Berlin, pp. 271-278.
Heath, L. S., and J.E. Smith (2000a) “Soil Carbon Accounting and Assumptions for Forestry and Forest-related
Land Use Change.” In: The Impact of Climate Change on America’s Forests. Joyce, L.A., and Birdsey, R.A. Gen.
Tech. Rep. RMRS-59. Rocky Mountain Research Station, Forest Service, U.S. Department of Agriculture, Fort
Collins, CO, pp. 89-101.
Heath, L.S., and J. E. Smith (2000b) “An Assessment of Uncertainty in Forest Carbon Budget Projections.”
Environmental Science & Policy 3:73-82.
Heath, L.S., J.E., Smith, and R.A. Birdsey (2003). Carbon Trends in U. S. Forestlands: A Context for the Role of
Soils in Forest Carbon Sequestration. Pages 35-45 in J. M. Kimble, L. S. Heath, R. A. Birdsey, and R. Lal, editors.
The Potential of U. S. Forest Soils to Sequester Carbon and Mitigate the Greenhouse Effect. Lewis Publishers (CRC
Press), Boca Raton, FL.
Howard, J. L. (2001) U.S. Timber Production, Trade, Consumption, and Price Statistics 1965-1999. Research Paper
RP-595, USDA Forest Service, Forest Products Laboratory, Madison, WI.
Ince, P. J. (1994) Recycling and Long-Range Timber Outlook. Gen. Tech. Rep. RM-242. Rocky Mountain Forest
and Range Experiment Station, Forest Service, U.S. Department of Agriculture. Fort Collins, CO, p. 66.
IPCC (2003) Good Practice Guidance for Land Use, Land-Use Change, and Forestry. J. Penman and others, editors.
IPCC National Greenhouse Gas Inventories Programme. Copy at <http://www.ipccnggip.iges.or.jp/public/gpglulucf/gpglulucf.htm>, August 13, 2004.
IPCC/UNEP/OECD/IEA (1997) Revised 1996 IPCC Guidelines for National Greenhouse Gas Inventories, Paris:
Intergovernmental Panel on Climate Change, United Nations Environment Programme, Organization for Economic
Co-Operation and Development, International Energy Agency. Paris, France.
Jenkins, J,C., D.C. Chojnacky, L.S. Heath, and R.A. Birdsey. (2003) “National-scale biomass estimators for United
States tree species.” Forest Science 49(1):12-35.
Johnson, D. W., and P. S. Curtis (2001) Effects of Forest Management on Soil C and N Storage: Meta Analysis.
Forest Ecology and Management 140:227-238.
Mills, J.R., and J.C. Kincaid (1992) The Aggregate Timberland Assessment System-ATLAS: A Comprehensive
Timber Projection Model. Gen. Tech. Rep. PNW-281. Pacific Northwest Research Station, Forest Service, U.S.
Department of Agriculture, Portland, OR, pp. 160.
Phillips, D.L., S.L. Brown, P.E. Schroeder, and R.A. Birdsey (2000) Toward Error Analysis of Large-Scale Forest
Carbon Budgets. Global Ecology and Biogeography 9:305-313.
Row, C., and R. B. Phelp (1996) Wood Carbon Flows and Storage After Timber Harvest. Pages 27-58 in R. N.
Sampson and D. Hair, editors. Forest and Global Change Volume 2: Forest Management Opportunities for
Mitigating Carbon Emissions. American Forests, Washington, DC.
Skog, K.E., and G.A. Nicholson (1998) “Carbon Cycling Through Wood Products: The Role of Wood and Paper
Products in Carbon Sequestration.” Forest Products Journal 48:75-83.
Skog, K.E., K. Pingoud, and J.E. Smith. (2004) “A method countries can use to estimate changes in carbon stored in
harvested wood products and the uncertainty of such estimates.” Environmental Management 33(Supplement
1):S65-S73.
Smith, J.E., and L.S. Heath (2000) “Considerations for Interpreting Probabilistic Estimates of Uncertainty of Forest
Carbon.” In: The Impact of Climate Change on America’s Forests. Gen. Tech. Rep. RMRS-59. Rocky Mountain
Research Station, Forest Service, U.S. Department of Agriculture, Fort Collins, CO, pp. 102-111.
Smith, J.E., and L.S. Heath (2002) A model of forest floor carbon mass for United States forest types. Res. Paper
NE-722. USDA Forest Service, Northeastern Research Station, Newtowne Square, PA.
Smith, J. E., L. S. Heath, and J. C. Jenkins (2003) Forest Volume-to-Biomass Models and Estimates of Mass for Live
and Standing Dead Trees of U.S. Forests. General Technical Report NE-298, USDA Forest Service, Northeastern
Research Station, Newtown Square, PA.
11
Smith, J. E., L. S. Heath, and P. B. Woodbury (2004 a) How to estimate forest carbon for large areas from inventory
data. Journal of Forestry 102:25-31.
Smith, W.B., J.S. Vissage, D.R. Darr, and R.M. Sheffield (2001) Forest Resources of the United States, 1997. Gen.
Tech. Rep. NC-219. North Central Research Station, Forest Service, U.S. Department of Agriculture, St. Paul, MN,
pp. 191.
Smith, W. B., P. D. Miles, J. S. Vissage, and S. A. Pugh (2004 b) Forest Resources of the United States, 2002.
General Technical Report NC-241, USDA Forest Service, North Central Research Station, St. Paul, MN.
Ulrich, A.H. (1989) U.S. Timber Production, Trade, Consumption, and Price Statistics, 1950-1987. USDA
Miscellaneous Publication No. 1471, U.S. Department of Agriculture, Forest Service, Washington, DC, pp. 77.
USDA Forest Service (1964) The Demand and Price Situation for Forest Products, 1964. Misc. Pub. 983,
Washington, DC.
USDA/SCS. 1991. State Soil Geographic (STATSG0) Data Base Data use information. Miscellaneous Publication
Number 1492, USDA, Natural Resources Conservation Service, National Soil Survey Center, Fort Worth, TX.
12
ANNEX3 Methodological Descriptions for Additional
Source or Sink Categories
3.1.
Methodology for Estimating Net Carbon Stock Changes in Forest Lands
Remaining Forest Lands
This annex expands on the methodology used to calculate net changes in carbon (C) stocks in forest ecosystems and
in harvested wood products. Some of the details of C conversion factors and procedures for calculating net CO2 flux
for forests are provided below; more detailed descriptions of selected topics may be found in the cited references.
Carbon Stocks and Net Changes in Forest Ecosystem Carbon Stocks
C stocks were estimated at the inventory plot level for each C pool within each state in the conterminous
U.S. based on availability of inventory data. Forest survey data in the United States were obtained from USDA
Forest Service, Forest Inventory and Analysis (FIA) Resources Planning Act Assessment (RPA) databases or the
individual state surveys in the FIADB, version 2.1. More complete information about these data is available at an
FIA Internet site (FIA Database Retrieval System). All FIADB surveys used for C stock estimates were obtained
from this site on or before September 30, 2005.
The first step in developing C estimates was to identify separate inventory surveys for each state and
associate each with an average year for field collection of data. Most inventory databases provide the year, month,
and day in which the data were collected. If only the year is specified, the date for collection of data is assigned the
midpoint in the year. If data for an individual survey were collected over a number of years, an average value is
calculated. A few surveys had missing or incorrect values for year of field data; in some cases it was possible to
obtain this information from the regional FIA units, otherwise the year was inferred from other data. Some overlap
exists between the RPA and FIADB inventories because the RPA summaries were compiled from the FIADB. Such
overlaps are identified and adjusted to avoid duplication. Older surveys for some states, particularly in the West,
have National Forest System lands surveyed at different times than other forest land in the state. For this reason, C
stocks for National Forests were separately estimated from other forests to account for differences in average year.
The inventories used for each state as well as average year identified for each are provided in Table A-180.
For each inventory summary in each state, each C pool was estimated using coefficients from the
FORCARB2 model (Birdsey and Heath 1995, Birdsey and Heath 2001, Heath et al. 2003, Smith et al. 2004a).
Coefficients of the model are applied to the survey data at the scale of FIA inventory plots; the results are estimates
of C density (Mg per hectare) for a number of separate C pools. C stocks and fluxes for Forests Remaining Forests
are reported in pools following IPCC Good Practice Guidance for Land Use, Land Use Change, and Forestry
(2003). FORCARB2 estimates C density for live trees, standing dead trees, understory vegetation, down dead
wood, forest floor, and soil organic matter. All non-soil pools except forest floor can be separated into aboveground
and belowground components. FORCARB2’s live tree and understory C pools are pooled as biomass in this
Inventory. Similarly, standing dead trees and down dead wood are pooled as dead wood in this Inventory.
Definitions of forest floor and soil organic matter in FORCARB2 correspond to litter and forest soils, respectively in
IPCC Good Practice Guidance for Land Use, Land Use Change, and Forestry (2003).
The tree C pools in FORCARB2 include aboveground and belowground (coarse root) C mass of live trees.
Separate estimates are made for whole-tree and aboveground-only biomass. Thus, the belowground portion is
determined as the difference between the two estimates. Tree C estimates are based on Jenkins et al. (2003) and are
functions of tree species and diameter as well as forest type and region. Some survey data do not provide
measurements of individual trees; tree C in these plots are estimated from plot-level growing stock volume of live
trees and equations based on Smith et al. (2003). C mass of wood is 50 percent of dry weight
(IPCC/UNEP/OECD/IEA 1997). The minimum-sized tree included in these FIA data is one-inch diameter (2.54
cm) at diameter breast height (1.3 meter); this represents the minimum size included in the tree C pools.
A second, but minor, component of biomass is understory vegetation. Understory vegetation is defined in
FORCARB2 as all biomass of undergrowth plants in a forest, including woody shrubs and trees less than one-inch
diameter, measured at breast height. In this Inventory, it is assumed that 10 percent of understory C mass is
belowground. This general root-to-shoot ratio (0.11) is near the lower range of temperate forest values provided in
13
IPCC LULUCF Good Practice Guidance (2003) and was selected based on two general assumptions: ratios are
likely to be lower for light-limited understory vegetation as compared with larger trees, and a greater proportion of
all root mass will be less than 2 mm diameter. C density estimates are based on Birdsey (1996) and were applied at
the inventory plot level (Smith et al. 2004a).
Dead wood includes the FORCARB2 pools of down dead wood and standing dead trees. Down dead wood
is defined as pieces of dead wood greater than 7.5 cm diameter, at transect intersection, that are not attached to live
or standing dead trees. Down dead wood includes stumps and roots of harvested trees. Ratio estimates of down dead
wood to live tree biomass were developed by FORCARB2 simulations and applied at the plot level (Smith et al.
2004a). The standing dead tree C pools in FORCARB2 include aboveground and belowground (coarse root) mass.
Estimates are based on Smith et al. (2003) and are functions of plot level growing stock volume of live trees, C
density of live trees, forest type, and region
Estimates of litter and soil organic carbon (SOC) are not based on C density of trees. Litter C is the pool of
organic C (litter, duff, humus, and fine woody debris) above the mineral soil and includes woody fragments with
diameters of up to 7.5 cm. Estimates are based on equations of Smith and Heath (2002) and applied at the plot level.
Estimates of SOC are based on the national STATSGO spatial database (USDA 1991), and the general approach
described by Amichev and Galbraith (2004). In their procedure, SOC was calculated for the conterminous U.S.
using the STATSGO database, and data gaps were filled by representative values from similar soils. Links to region
and forest type groups were developed with the assistance of the USDA Forest Service FIA Geospatial Service
Center by overlaying FIA forest inventory plots on the soil C map.
A historical focus of the FIA program was to provide information on timber resources of the US. For this
reason, some forest lands, which were less productive or reserved (i.e., land where harvesting was prohibited by
law), were less intensively surveyed. This generally meant that forest type and area were identified but data were
not collected on individual tree measurements. However, all annualized surveys initiated since 1998 have followed
a new national plot design for all forest land (Alerich et al. 2005, FIA Database Retrieval System). The practical
effect that this evolution in inventories has had on estimating forest C stocks from 1990 through the present is that
some older surveys of lands do not have the stand level values for merchantable volume of wood or stand age, which
are necessary inputs to FORCARB2. The data gaps in the surveys taken before 1998 were filled by assigning
regional average C densities calculated from the more complete, later inventories. This overall effect of this is to
generate estimates for C stock with no net change in C density on those lands with gaps in past surveys.
Average C density values for forest ecosystem C pools according to region and forest types within regions
are provided in Table A-181. Note that C densities reflect the most recent survey for each state as available in the
FIADB, not potential maximum C storage. Thus, C densities are affected by the distribution of stand sizes within a
forest type, which can range from regenerating to mature stands. A large proportion of young stands in a particular
forest type is likely to reduce the regional average for C density.
The overall approach for determining forest C stocks and stock change was to estimate forest C stocks
based on data from two forest surveys conducted several years apart (Table A-180). C stocks were calculated
separately for each state based on inventories available since 1990 and for the most recent inventory prior to 1990.
For each pool in each state in each year, C stocks were estimated by linear interpolation between survey years.
Similarly, fluxes were estimated for each pool in each state by dividing the difference between two successive
stocks by the number of intervening years between surveys. Thus, the number of separate stock estimates for each
state was one less than the number of available inventories. Stocks and fluxes since the most recent survey were
based on extrapolating estimates from the last two surveys. C stock and flux estimates for each pool were summed
over all states to form estimates for the conterminous United States. Summed fluxes and stocks are in Table A- 182
and Table A - 183, respectively.
Carbon in Harvested Wood Products
Estimates of C stock changes in wood products and wood discarded in landfills were based on the methods
described by Skog and Nicholson (1998) which were based in turn on earlier efforts using similar approaches (Heath
et al. 1996, Row and Phelps, 1996). C stocks in wood products in use and wood products stored in landfills were
estimated from 1910 onward based on several sets of historical data from the USDA Forest Service. These data
include estimates of wood product demand, trade, and consumption (USDA 1964, Ulrich 1989, Howard 2001).
Annual historical estimates and model projections of the production of wood products were used to divide consumed
14
roundwood into wood product, wood mill residue, and pulp mill residue. To estimate the amount of time products
remain in use before disposal, wood and paper products were divided into 21 categories, each with an estimated
product half-life (Skog and Nicholson 1998). After disposal, the amount of waste that is burned was estimated. For
products entering dumps or landfills, the proportion of C emitted as CO2 or CH4 was estimated using the estimated
maximum proportion of wood and paper converted to CO2 or CH4 in landfills for 5 product types. By following the
fate of C from the wood harvested in each year from 1910 onward, the change in C stocks in wood products and
landfills, and the amount of C emitted to the atmosphere with and without energy recovery were estimated for each
year through 2003. To account for imports and exports, the production approach was used, meaning that C in
exported wood was counted as if it remained in the United States, and C in imported wood was not counted. From
1990 through 2002, the amount of C in exported wood averaged 6 Tg C per year, with little variation from year to
year. For comparison, imports (which were not included in the harvested wood net flux estimates) increased from
7.2 Tg C per year in 1990 to 13 Tg C per year in 2002. Skog and Nicholson (1998) go into further detail in their
description of this methodology. Summaries of net fluxes and stocks for harvested wood in products and landfills
are in Table A-182 and Table A-183.
Uncertainty Analysis
The uncertainty analysis for total net flux of forest C (see Table 7-8 in LULUCF chapter) was consistent with the
IPCC-recommended Tier 2 methodology (IPCC 2003). The estimates were simulated with Monte Carlo sampling of
probability densities representing plot-level C for the forest ecosystem estimates following general methods
described in Heath and Smith (2000b) and Smith and Heath (2000). Estimates of uncertainty for C in harvested
wood were based on Skog et al. (2004). Monte Carlo sampling of all probability densities involved random
sampling of equal-probable intervals. The 95 percent confidence interval about the simulated flux (Table A-180) is
based of the bounds of the central 95 percent of the simulated probability density for flux.
Uncertainty about C density (Mg/ha) was defined for each of six FORCARB2 C pools for each inventory plot. Live
and standing dead trees were assigned normal or truncated normal probability densities, which were defined
according to Jenkins et al. (2003) and the species and number of trees measured on each plot. Down dead wood and
forest floor were assigned skewed distributions, which assume that a small proportion of plots will have relatively
high carbon densities. Understory and soil organic C were assigned uniform distributions to reflect the fact that the
model currently has little information to assign plot-specific values. Monte Carlo sampling of live tree, down dead
wood, and understory probabilities were highly correlated to reflect the same process in FORCARB2. Uncertainty
about plot area was assigned a normal distribution and defined according to accuracy standards defined for the
surveys (Alerich et al. 2005). The uncertainty analysis of Skog et al. (2004) was developed for a slightly different
estimate of C in harvested wood as compared with the method followed here (Skog and Nicholson 1998).
Therefore, the probability densities for annual flux for wood products and landfilled wood were defined as uniform
densities bounded by the summaries in Table 3 of Skog et al. (2004). Two effects of estimating uncertainty at the
plot level and aggregating to state totals for determining net stock change (flux) are: 1) relative uncertainty tends to
decrease, and 2)skewed probability densities for individual plots tend to approach normality as independent samples
among plots are summed.
Table A-180. Source of Forest Inventory and Average Year of Field Survey Used to Estimate Statewide Carbon
Stocks.
Average Year Assigned
Statea
Source of Inventory Datab
to Inventoryc
Alabama
1987 RPA
1982
FIADB, cycle 1
1990
FIADB, cycle 7
1999
FIADB, cycle 4
2002
Arizona, NFS
1987 RPA
1985
FIADB, cycle 2
1996
FIADB, cycle 3
2003
Arizona, all other
FIADB, cycle 1
1986
FIADB, cycle 2
1992
FIADB, cycle 3
2003
15
Arkansas
California, NFS
California, all other
Colorado, NFS
Colorado, all other
Connecticut
Delaware
Florida
Georgia
Idaho, NFS
Idaho, all other
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
1987 RPA
FIADB, cycle 1
FIADB, cycle 3
1987 RPA
1997 RPA
FIADB, cycle 5
1987 RPA
FIADB, cycle 3
FIADB, cycle 5
2002 RPA
FIADB, cycle 2
2002 RPA
FIADB, cycle 2
FIADB, cycle 3
FIADB, cycle 4
FIADB, cycle 5
FIADB, cycle 3
FIADB, cycle 4
FIADB, cycle 1
FIADB, cycle 2
FIADB, cycle 1
FIADB, cycle 7
FIADB, cycle 4
1987 RPA
FIADB, cycle 1
FIADB, cycle 2
1987 RPA
FIADB, cycle 1
FIADB, cycle 2
FIADB, cycle 3
FIADB, cycle 4
FIADB, cycle 5
FIADB, cycle 3
FIADB, cycle 4
FIADB, cycle 5
FIADB, cycle 3
FIADB, cycle 4
FIADB, cycle 3
FIADB, cycle 4
FIADB, cycle 5
FIADB, cycle 1
FIADB, cycle 4
1987 RPA
FIADB, cycle 1
FIADB, cycle 3
1987 RPA
FIADB, cycle 4
FIADB, cycle 5
FIADB, cycle 4
FIADB, cycle 5
FIADB, cycle 3
FIADB, cycle 4
FIADB, cycle 5
FIADB, cycle 4
FIADB, cycle 5
FIADB, cycle 6
1978
1996
2002
1981
1993
2003
1983
1993
2003
1986
2004
1979
2004
1985
1998
2004
1986
1999
1987
1995
1989
1997
2001
1982
1998
2005
1981
1990
2005
1985
1998
2003
1986
1998
2001
1990
2002
1981
1994
2003
1987
2002
1984
1991
2003
1983
1995
2002
1986
2000
1985
1998
2004
1980
1993
2002
16
Minnesota
Mississippi
Missouri
Montana, NFS
Montana, all other
Nebraska
Nevada, NFS
Nevada, all other
New Hampshire
New Jersey
New Mexico, NFS
New Mexico, all other
New York
North Carolina
North Dakota
Ohio
Oklahoma
Oregon, eastern NFS
Oregon, eastern all other
Oregon, western NFS
Oregon, western all other
FIADB, cycle 4
FIADB, cycle 5
FIADB, cycle 12
1987 RPA
FIADB, cycle 1
FIADB, cycle 4
FIADB, cycle 5
1987 RPA
FIADB, cycle 1
FIADB, cycle 2
FIADB, cycle 1
FIADB, cycle 2
FIADB, cycle 2
FIADB, cycle 3
FIADB, cycle 4
1987 RPA
FIADB, cycle 1
FIADB, cycle 2
FIADB, cycle 1
FIADB, cycle 2
FIADB, cycle 4
FIADB, cycle 5
FIADB, cycle 6
FIADB, cycle 3
FIADB, cycle 4
1987 RPA
FIADB, cycle 2
FIADB, cycle 1
FIADB, cycle 2
1987 RPA
2002 RPA
FIADB, cycle 5
FIADB, cycle 1
FIADB, cycle 2
FIADB, cycle 3
FIADB, cycle 2
FIADB, cycle 3
FIADB, cycle 4
1987 RPA
FIADB, cycle 4
FIADB, cycle 5
1987 RPA
FIADB, cycle 1
1987 RPA
2002 RPA
FIADB, cycle 5
1987 RPA
FIADB, cycle 3
FIADB, cycle 4
FIADB, cycle 5
1987 RPA
2002 RPA
FIADB, cycle 5
1997 RPA
2002 RPA
FIADB, cycle 5
1977
1989
2001
1977
1994
1988
2002
1987
1996
2004
1989
2004
1983
1995
2003
1984
1997
2005
1981
2005
1983
1997
2004
1987
1999
1985
1997
1987
1991
1987
1993
2003
1984
1990
2001
1979
1995
2003
1987
1991
2003
1986
1992
1987
1995
2003
1976
1991
1999
2003
1986
1996
2003
1989
1997
2003
17
Pennsylvania
Rhode Island
South Carolina
South Dakota, NFS
South Dakota, all other
Tennessee
Texas
Utah
Vermont
Virginia
Washington, eastern NFS
Washington, eastern all other
Washington, western NFS
Washington, western all other
West Virginia
Wisconsin
Wyoming, NFS
Wyoming, all other
FIADB, cycle 4
FIADB, cycle 5
FIADB, cycle 3
FIADB, cycle 4
FIADB, cycle 5
FIADB, cycle 1
FIADB, cycle 2
FIADB, cycle 3
1997 RPA
FIADB, cycle 4
FIADB, cycle 5
1987 RPA
FIADB, cycle 4
FIADB, cycle 5
FIADB, cycle 5
FIADB, cycle 6
FIADB, cycle 4
1987 RPA
FIADB, cycle 1
FIADB, cycle 3
1987 RPA
FIADB, cycle 1
FIADB, cycle 2
FIADB, cycle 4
FIADB, cycle 5
FIADB, cycle 6
FIADB, cycle 1
FIADB, cycle 2
FIADB, cycle 3
1987 RPA
2002 RPA
FIADB, cycle 5
1987 RPA
FIADB, cycle 3
FIADB, cycle 5
1987 RPA
2002 RPA
FIADB, cycle 5
1987 RPA
FIADB, cycle 3
FIADB, cycle 5
FIADB, cycle 4
FIADB, cycle 5
FIADB, cycle 4
FIADB, cycle 5
FIADB, cycle 6
1997 RPA
2002 RPA
FIADB, cycle 2
2002 RPA
FIADB, cycle 2
1990
2002
1985
1999
2004
1986
1993
2001
1986
1999
2003
1987
1995
2003
1989
1998
2002
1986
1992
2003
1977
1993
2003
1983
1997
2004
1985
1991
2000
1987
1995
2004
1981
1992
2004
1987
1995
2004
1979
1990
2004
1988
2001
1982
1995
2002
1982
1992
2000
1984
2001
a
Inventories for 11 western states were separated into National Forest System (NFS) and all other forestlands (all other). Oregon
and Washington were also divided into eastern and western forests (east or west of the crest of the Cascade Mountains).
b
FIADB is version 2.1 as available on Internet September, 2005.
c
Based on forestland survey plots and rounded to the nearest integer year.
18
Table A-181. Average carbon density (Mg/ha) by carbon pool and forest area (1000 ha) according to region
and forest type, based on the most recent inventory survey available for each State from the FIADB (see
Table A-180)
Region
AboveBelowSoil
Forest
(States)
Dead
Organic
Area
ground
ground
Wood
Forest Types
Litter
Carbon
Biomass
Biomass
Carbon Density (Mg/ha)
1000 ha
Northeast
(CT,DE,MA,MD,ME,NH,NJ,NY,OH,PA,RI,VT,WV)
White/Red/Jack Pine
91.9
19.0
11.3
13.6
78.1
1,966
Spruce/Fir
51.4
10.9
11.8
30.6
98.0
2,972
Oak/Pine
73.6
14.5
8.9
27.1
66.9
1,403
Oak/Hickory
77.6
14.7
10.1
7.9
53.1
11,802
Elm/Ash/Cottonwood
51.2
9.7
8.1
23.9
111.7
1,266
Maple/Beech/Birch
75.1
14.4
12.4
26.4
69.6
15,239
Aspen/Birch
46.2
9.1
7.7
8.5
87.4
1,659
Minor Types and Nonstocked
42.9
8.6
6.2
13.8
81.8
1,218
Northern Lake States
(MI,MN,WI)
White/Red/Jack Pine
52.7
11.0
7.9
12.2
120.8
1,794
Spruce/Fir
41.1
8.7
8.3
32.5
261.8
3,081
Oak/Hickory
70.2
13.3
10.3
7.8
97.1
2,920
Elm/Ash/Cottonwood
50.1
9.6
8.6
25.5
179.9
1,652
Maple/Beech/Birch
71.4
13.7
10.9
26.4
134.3
5,110
Aspen/Birch
42.1
8.2
8.4
8.3
146.1
5,346
Minor Types and Nonstocked
37.4
7.4
6.2
11.1
127.2
886
Northern Prairie States
(IA,IL,IN,KS,MO,ND,NE,SD)
Ponderosa Pine
42.0
8.9
6.9
14.3
48.5
563
Oak/Pine
51.8
10.1
7.1
25.3
39.7
573
Oak/Hickory
68.4
12.9
9.2
7.5
49.1
8,154
Elm/Ash/Cottonwood
72.1
13.5
11.1
23.6
83.0
1,760
Maple/Beech/Birch
62.5
11.8
8.7
24.8
71.0
1,017
Minor Types and Nonstocked
36.1
7.2
5.8
12.5
58.8
884
South Central
(AL,AR,KY,LA,MS,OK,TN,TX)
Longleaf-slash pine
37.3
7.6
3.9
10.7
55.5
1,321
Loblolly-shortleaf pine
42.2
8.6
4.7
9.6
41.9
12,701
Oak-pine
51.6
10.0
6.4
9.7
41.7
6,928
Oak-hickory
63.5
11.9
7.3
6.4
38.6
18,841
Oak-gum-cypress
71.3
13.6
8.7
6.3
52.8
5,303
Elm-ash-cottonwood
57.7
10.9
7.8
5.8
49.9
2,455
Minor types and nonstocked
52.1
10.0
7.1
8.1
46.4
1,155
Southeast
(FL,GA,NC,SC,VA)
Longleaf/Slash Pine
30.5
6.2
3.3
9.5
110.0
4,185
Loblolly/Shortleaf Pine
44.8
9.2
5.5
9.2
72.9
8,691
Oak/Pine
49.4
9.6
5.5
9.1
61.4
4,928
Oak/Hickory
71.5
13.5
8.2
6.4
45.3
11,006
Oak/Gum/Cypress
71.4
13.8
9.0
6.3
158.0
4,643
Elm/Ash/Cottonwood
70.3
13.3
11.0
6.2
95.7
666
Minor Types and Nonstocked
40.5
7.8
5.7
6.3
87.2
1,129
Pacific Northwest, Westside
(Western OR and WA)
19
Douglas-fir
Fir/Spruce/Mt. Hemlock
Hemlock/Sitka Spruce
Alder/Maple
Minor Types and Nonstocked
Pacific Northwest, Eastside
(Eastern OR and WA)
Pinyon/Juniper
Douglas-fir
Ponderosa Pine
Fir/Spruce/Mt. Hemlock
Lodgepole Pine
Western Larch
Minor Types and Nonstocked
Pacific Southwest
(CA)
Pinyon/Juniper
Douglas-fir
Ponderosa Pine
Fir/Spruce/Mt. Hemlock
Lodgepole Pine
Redwood
California Mixed Conifer
Western Oak
Tanoak/Laurel
Minor Types and Nonstocked
Rocky Mountains, North
(ID,MT)
Douglas-fir
Ponderosa Pine
Fir/Spruce/Mt. Hemlock
Lodgepole Pine
Western Larch
Minor Types and Nonstocked
Rocky Mountains, South
(AZ,CO,NM,NV,UT,WY)
Pinyon/Juniper
Douglas-fir
Ponderosa Pine
Fir/Spruce/Mt. Hemlock
Lodgepole Pine
Aspen/Birch
Western Oak
Minor Types and Nonstocked
143.4
144.1
175.6
82.5
69.4
30.1
30.4
37.0
16.2
13.8
31.3
37.5
45.0
21.0
12.0
31.4
37.9
38.4
7.4
13.7
94.8
62.1
116.3
115.2
83.0
5,594
1,215
1,659
1,359
1,276
13.3
79.4
50.0
95.5
41.2
70.7
29.0
2.6
16.6
10.4
20.2
8.7
14.8
5.7
2.4
18.6
10.1
27.0
9.7
18.9
13.1
21.1
36.5
22.8
37.9
21.0
36.1
22.3
46.9
94.8
50.7
62.1
52.0
45.1
79.7
832
2,004
2,925
1,573
1,034
288
1,486
25.6
156.8
51.9
163.6
94.8
200.4
116.7
67.1
125.7
37.2
5.0
32.4
10.8
34.6
20.0
41.8
24.5
12.8
24.6
7.2
1.9
32.8
10.1
45.0
19.8
42.0
28.8
7.4
18.5
9.0
21.1
34.8
35.1
38.3
39.2
60.8
37.6
29.0
27.1
23.9
26.3
40.1
41.3
51.9
35.2
53.8
49.8
27.6
27.6
38.0
789
442
376
777
396
274
3,825
3,677
790
1,935
73.8
43.5
68.1
55.2
63.2
27.4
15.6
9.1
14.4
11.8
13.3
5.5
13.8
7.9
21.2
10.4
14.9
9.6
37.2
23.1
37.3
23.3
35.9
24.7
38.8
34.3
44.1
37.2
34.2
42.5
5,917
1,772
4,574
2,622
411
4,010
22.1
72.6
48.5
81.3
53.8
56.2
19.8
16.6
4.5
15.4
10.2
17.3
11.4
10.8
3.8
3.1
0.8
16.4
8.2
23.0
13.0
11.6
2.2
4.1
21.1
38.0
23.6
38.8
24.1
28.5
27.1
23.7
19.7
30.9
24.1
31.5
27.0
58.8
38.0
25.3
19,809
1,719
3,453
4,180
2,157
2,589
2,874
5,164
20
Table A-182. Net Annual Changes in Carbon Stocks (Tg C yr-1) in Forest and Harvested Wood Pools, 1990-2004
Carbon Pool
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
Forest
(154)
(157)
(145)
(112)
(109)
(78)
(74)
(82)
(113)
(115)
Live, aboveground
(92)
(92)
(87)
(71)
(70)
(63)
(62)
(67)
(78)
(84)
Live, belowground
(18)
(18)
(17)
(13)
(13)
(12)
(11)
(13)
(15)
(16)
Dead Wood
(12)
(13)
(11)
(9)
(10)
(8)
(8)
(8)
(11)
(10)
Litter
(23)
(20)
(17)
(11)
(7)
1
2
3
(3)
(7)
Soil Organic Carbon
(9)
(14)
(12)
(8)
(9)
4
5
3
(4)
1
Harvested Wood
(57)
(54)
(55)
(56)
(57)
(55)
(57)
(58)
(56)
(59)
Wood Products
(13)
(11)
(13)
(15)
(17)
(15)
(15)
(16)
(14)
(17)
Landfilled Wood
(44)
(43)
(43)
(41)
(41)
(41)
(41)
(42)
(42)
(42)
Total Net Flux
(211)
(211)
(200)
(167)
(166)
(133)
(131)
(140)
(169)
(174)
Table A-183. Carbon Stocks (Tg C) in Forest and Harvested Wood Pools, 1990-2005
Carbon Pool
1990
1991
1992
1993
1994
1995
1996
Forest
39,508 39,661 39,818 39,963 40,074 40,183 40,261
Live, aboveground
14,334 14,426 14,518 14,605 14,676 14,746 14,809
Live, belowground
2,853 2,871 2,888 2,905 2,918 2,931 2,943
Dead Wood
2,409 2,421 2,434 2,445 2,454 2,464 2,472
Litter
4,492 4,515 4,535 4,553 4,563 4,570 4,570
Soil Organic Carbon
15,420 15,429 15,443 15,455 15,463 15,472 15,467
Harvested Wood
1,915 1,973 2,027 2,082 2,137 2,195 2,250
Wood Products
1,134 1,147 1,158 1,171 1,186 1,202 1,217
Landfilled Wood
781
825
868
911
952
992 1,033
Total Carbon Stock
41,423 41,634 41,845 42,044 42,212 42,378 42,511
1997
40,335
14,871
2,954
2,479
4,568
15,463
2,307
1,232
1,074
42,642
1998
40,417
14,938
2,967
2,488
4,565
15,460
2,365
1,248
1,117
42,782
1999
40,529
15,016
2,982
2,499
4,569
15,464
2,421
1,262
1,159
42,951
2000
(115)
(85)
(16)
(9)
(7)
3
(57)
(16)
(41)
(172)
2000
40,645
15,100
2,998
2,509
4,575
15,463
2,480
1,279
1,200
43,125
2001
(115)
(85)
(16)
(9)
(7)
3
(58)
(16)
(42)
(173)
2001
40,760
15,184
3,014
2,518
4,583
15,460
2,537
1,295
1,242
43,297
2002
(115)
(85)
(16)
(9)
(7)
3
(58)
(16)
(42)
(173)
2002
40,874
15,269
3,031
2,527
4,590
15,458
2,595
1,311
1,284
43,470
2003
(115)
(85)
(16)
(9)
(7)
3
(59)
(16)
(42)
(173)
2003
40,989
15,354
3,047
2,536
4,597
15,455
2,654
1,327
1,327
43,643
2004
(115)
(85)
(16)
(9)
(7)
3
(59)
(17)
(43)
(174)
2004
41,103
15,438
3,064
2,545
4,604
15,452
2,713
1,344
1,369
43,816
21