View the SRWP Facebook Page  View the SRWP Twitter Feed  Signup for the SRWP Email Newsletter

3.3.3 Carbon Stock and Sequestration


Goal: C. Protect and enhance landscape and habitats structure and processes to benefit ecosystem and watershed functions

Objective: 3. Protect and maintain natural variability and rates of primary production and nutrient cycling to support aquatic and terrestrial communities

WAF Attribute: Ecological processes

What is it?

For this indicator we examine two elements of the carbon budget of the Feather River Watershed, carbon sequestration amounts and net primary productivity. Both are of interest in terms of global change issues, in particular because of the potential for offsetting increases in atmospheric carbon dioxide by storage of carbon in terrestrial carbon pools (Dixon et al. 1994). In this analysis we look at carbon standing stock at a single point in time as a measure of watershed condition, and assess trends in carbon storage by examining changes in net primary productivity detected by satellite remote sensing.

Carbon Sequestration

Research on carbon sequestration has focused on measurements of carbon stocks and carbon flux. Measuring carbon flux requires sophisticated instrumentation making fine-scale studies difficult, but measurement of carbon stock is more amenable to landscape-scale studies. The general approach for carbon stock evaluation is to amalgamate remote sensing-based landscape classifications with vegetation plot data that includes above-ground biomass, litter accumulation on the soil floor, and below-ground carbon to estimate total carbon storage across the landscape. Typical units for the metric are in megagrams (Mg) of carbon per hectare for the stock and Mg C per hectare per year for the flux. In this analysis we use the results from a landscape-scale assessment of carbon stocks in California and compare that to a reference condition that assumes all trees are fully mature.

Net Primary Productivity

According to NASA, terrestrial biological productivity (or primary productivity) is the single most fundamental measure of “global change” of practical interest for humankind. Primary productivity is the measure of carbon intake by plants during photosynthesis, and this measure is an important indicator for studying the health for plant communities.

Net Primary Productivity (NPP) is the amount of carbon uptake after subtracting Plant Respiration (RES) from Gross Primary Productivity (GPP). GPP is the total rate at which the ecosystem capture and store carbon as plant biomass, for a given length of time.


Photosynthesis is the process in which the energy from the sun converts carbon dioxide (CO2) from the atmosphere and water (or water vapor) to organic sugar molecules (carbohydrates), which are stored in the plants, and oxygen, which we, and other life on earth, consume. The extra water molecules which are derived in photosynthesis are reused by the plant or transpired into the atmosphere. Below is the chemical formula for photosynthesis:

6CO2 + 12H20(+sunlight) C6H12O6+6O2+6H2O

NPP measures the mass of the new plant growth (chemically-fixed carbon) produced during a given interval. Change in NPP may change with vegetation health, so NPP rates were used to analyze the overall trend of carbon uptake in this region over the past ten years. To analyze trend, we downloaded ten years of monthly satellite data from NASA, which are are reported as grams of carbon uptake per meter square per day (gC/m2/day). With monthly data, we ran a Seasonal-Kendall trend analysis, and with annual data, we ran a Mann-Kendall and Regional-Kendall trend analysis.

Why is it Important?

Humans continue to release CO2 and other greenhouse gases into the atmosphere from the burning of fossil fuels and agricultural practices. Plants cannot convert CO2 into biomass as fast as it is entering the atmosphere, causing a global buildup. These greenhouse gases trap heat from the sun and cause the surface temperature to rise, which has started a chain of events that will have enormous impacts on the globe in the years to come. These changes include glacial melting, sea level rising, and climatic shifting, which in turn can affect the welfare and health of all living things on this planet.

Carbon sequestration is considered an important means to mitigate the impacts of greenhouse gases on climate change (Sedjo & Solomon 1989). Increasing the amount of carbon stored on a watershed may become an important policy goal with economic benefits accruing from the establishment of a carbon offset market (Richards & Stokes 2004).

Forest ecosystems sequester the most carbon of any terrestrial ecosystem, and most United States surveys of carbon storage to date have emphasized storage in forests, usually working with the USFS Forest Inventory and Assessment plots as a base (Woodbury et al. 2007, Blackard et al. 2008). The forests of the Pacific Northwest, including the Sierra Nevada, may have some of the highest potential to store additional carbon of any forests in the world (Hudiburg et al. 2009).

What is the target or desired condition?

Prior to the industrial revolution, the planet’s carbon cycle was closer to a state of equilibrium. While an increase in solar radiation or an increase in planetary volcanism can drastically change the carbon cycle for a relatively short period of time, it has been shown that human activity has adjusted this cycle by adding more carbon and methane into the atmosphere at higher concentrations than any natural occurrence over the last 650,000 years (Siegenthaler et al., 2005). The carbon cycle is a global phenomena, so to return to a desired condition at equilibrium will be a global, population-wide, effort. To select a desired condition at a regional scale, we look at the carbon holding capacity for each region and compare it with current conditions.

We take the desired condition to be a landscape where all trees are fully mature; that is, they have grown to the point where additional carbon storage on the landscape in aboveground biomass is limited to the rate of trees dying and new ones growing. Such a landscape is at its maximum potential for mitigating climate change through storage of atmospheric carbon dioxide.

We also selected a target for new carbon sequestration, as indicated by NPP, as an increasing trend, or at least not a declining trend. This means that a significant upward trend is a good condition from a climate mitigation point of view, and a declining trend is a poor condition.

What can influence or stress condition?

Any changes in plant cover in the landscape will affect the amount of aboveground carbon storage. Most important are changes in forest cover, given that forests have the greatest amount of biomass of any habitat type. Processes that influence forest cover and hence carbon storage include fire, timber harvest, land development, and disturbances such as pest outbreaks as well as forest regrowth (Brown et al. 2004). In a recent study, scientists found that logging was the greatest impact on reduced carbon storage in forests and “no management” of forests resulted in the greatest sequestration of carbon (Nunery and Keeton, 2010). Fire can also reduce NPP, with reduction depending on fire intensity (Meigs et al., 2009). Remaining and newly-growing plants will tend to grow vigorously, so at the landscape scale, fire temporarily reduces NPP rates.

Regional climate will greatly affect the natural growth of shrubs and trees. Between 2006 and 2009, California experienced three consecutive dry water years. NPP will tend to decline in response to seasonal and drought-related drying. Plants take up CO2 through holes in their leaves called stomata. These will close under very dry conditions in order to reduce water loss by the plant. This means that as conditions dry, rates of carbon sequestration will decline. Because climate change may lead to drier and hotter conditions in many places in California, NPP may decline.

What did we find out/How are we doing?

There were relatively high scores for carbon standing stock, ranging from 86 for the East Branch North Fork Feather to 96 for Deer Creek (Table 1 and Figure 1). There was significant downward trends in annual NPP for the three the western, lower elevation and agricultural rich, subwatersheds (Lower Bear, Lower Feather, Lower Yuba). Despite the high absolute values of the indicator scores, scores should be as close to 100% as possible, because of the need to reach global greenhouse gas mitigation goals.

Table 1. Report Card scores for carbon for subwatersheds

Goal Measurable Objective Subwatershed Score Trend
C. Protect and enhance landscape and habitats structure and processes to benefit ecosystem and watershed functions 3) Protect and maintain natural variability and rates of primary production and nutrient cycling to support aquatic and terrestrial communities NFF 94 n.s.
EBNFF 86 n.s.
MFF 88 n.s.
LF 93
NY 93 n.s.
MY 89 n.s.
SY 93 n.s.
DC 96 n.s.
LY 91
UB 91 n.s.
LB 93

Carbon Standing Stock

The indicator value is a comparison of current standing stock to a potential maximum, which is based on a combination of underlying vegetation types and canopy closure values. Figure 2 is an intermediate layer which shows carbon storage at a 100 meter pixel-resolution and provides additional detail about the patterns in each subwatershed. Low elevation subwatersheds with a predominance of oak woodlands and chaparral have relatively low levels of carbon storage. There is a mid-elevation zone where carbon storage is particularly high, which is related to the wetter, productive conifer and mixed-conifer/hardwood forest types. On the eastern slope watersheds (Middle Fork Feather, East Branch North Fork Feather) there is a large band with relatively low carbon storage.

Figure 1. Carbon stock scores across subwatersheds

Figure 2. Actual carbon standing stocks based on vegetation present

Trend Analysis

To study trends in the carbon budget, NPP was analyzed for each subwatershed. NPP provides a rate of carbon fixation or sequestration into plant material. Ten years of monthly NPP rates were available, allowing for an estimate of change in rate over time.
We used the R statistical program to analyze these data, and used custom-made variations of the Kendall package depending on whether the analysis was for monthly or annual data. Kendall’s rank correlation measures the strength of monotonic association between two vectors, such as year and data value (see Section 4.3 for more information on trends analysis).

Monthly Trend

Monthly-seasonal variation over 10 years (2000 – 2009), was analyzed using a Seasonal Kendall statistical model. Monthly NPP raster data for each subwatershed were aggregated as sum, mean, maximum, minimum, and standard deviation for each subwatershed, and trends across each parameter were calculated for the 10-year analysis period (Table 2).

Table 2. Monthly Net Primary Productivity in each subwatershed: trends in the sum, mean, maximum, minimum, and standard deviation (StdDev) of NPP

Subwatershed Trend
  Sum Mean Max Min StdDev
Deer Creek          
East Branch North Fork Feather          
Lower Bear Negative Negative   Negative  
Lower Feather Negative Negative   Negative  
Lower Yuba          
Middle Fork Feather     Positive   Positive
Middle Yuba         Positive
North Fork Feather     Positive   Positive
North Yuba         Positive
South Yuba          
Upper Bear         Positive

We found a negative trend (decline in NPP) in the Lower Bear and Feather subwatersheds (Figure 3 and 4, respectively). Declines in NPP are associated with changes in vegetation type (e.g., replacement of tree canopy by row-crops), increases in temperature, and/or decreases in available water (from irrigation or precipitation. In these subwatersheds, all of these could be occurring. Maximum NPP rates (which occur in the Spring) increased in the Middle Fork Feather and North Fork Feather regions. This could be because Spring temperatures are higher and water is not limiting growth, resulting in higher Spring photosynthesis rates. A positive trend in the standard deviation tells us that the variability between years is increasing for those regions. Most alarmingly, NPP rates have recently dipped into negative rates in the summer, meaning that there is a net export of carbon dioxide to the atmosphere during these times.

Figure 3. Monthly sums of NPP for the Lower Bear subwatershed, showing a statistically-significant downward trend

Figure 4. Monthly sums of NPP for the Lower Feather subwatershed, showing a statistically-significant downward trend


Annual Trend

NPP values were summed for the year for each subwatershed and trends analyzed using the Mann-Kendall and Regional-Kendall statistical tests. These values were aggregated from the monthly NPP values, where the annual sum is the sum across all months, annual mean is the mean across all months, annual maximum, annual minimum, and the mean standard deviation for all months in the year. There were significant declines in the total NPP (sum), mean NPP, and minimum NPP for the Lower Bear, Feather, and Yuba subwatersheds and declines in the minimum NPP for the Deer Creek and Middle Fork Feather subwatersheds. There were also increasing trends in the standard deviation (Figure 5), a measure of variation within the year among months. These results are consistent with the possibility of NPP declining with land-use or climatic drivers. In addition, the increase in variability could be related to warmer springs allowing greater NPP during the peak natural growing season and hotter/drier summers causing declines and even net losses in carbon (negative NPP).

Table 3. Annual Net Primary Productivity in each subwatershed: trends in the sum, mean, maximum, minimum, and standard deviation (StdDev) of NPP

Subwatershed Annual Trend
  Sum Mean Max Min StdDev
Deer Creek       Negative  
East Branch North Fork Feather          
Lower Bear Negative Negative   Negative  
Lower Feather Negative Negative   Negative  
Lower Yuba Negative Negative   Negative  
Middle Fork Feather       Negative Positive
Middle Yuba         Positive
North Fork Feather         Positive
North Yuba         Positive
South Yuba          
Upper Bear       Negative Positive

Figure 5. Change in annual mean standard deviations for all subwatersheds

Temporal and spatial resolution

Carbon standing stock was measured at 100 m resolution with vegetation data that was roughly a decade old. The NPP data were at a 0.1 degree resolution, which is equivalent to roughly 10 kilometers squared, using data that are recent and updated. The standing stock is unlikely to change rapidly over areas the size of the subwatersheds, but for planning watersheds or similar units it may. NPP changes rapidly (daily to monthly) and high time-resolution is required for accurate estimations.

How sure are we about our findings? (Things to keep in mind)

Carbon Stock

Carbon stock estimation is difficult for a number of reasons, and the results above should be treated carefully. First, because estimation methodology depends upon combining synoptic land cover data from remote sensing platforms with plot-level measurements of carbon in living and dead plant material, it is important that the remote sensing-derived map has accurate information about vegetation height and cover. This is challenging because remotely sensed imagery usually only gives spectral information about the top level of the canopy and not the canopy depth, the latter corresponding more closely to volume of aboveground biomass. Also, plot-level data tends to be focused on forest stands [e.g. the Forest Inventory and Assessment plots (Woodbury et al. 2007)], with shrublands and grasslands being sampled more poorly. Carbon estimation is even difficult at the plot level, since the usual technique for estimating carbon stored in a tree is to measure diameter and height and then refer to a set of allometric equations (e.g. Jenkins et al. 2004) relating tree biomass to those parameters, and these equations may have been developed from measurements of trees located in a very different landscape than one’s study plot.

For this particular analysis of carbon stocks, a couple things to note are the following. First, using the estimator equations in Brown et al. (2004) involves reducing the land cover data to types that are not very specific to California vegetation. It would be best if this assessment was made using equations based on California vegetation types, if these were available. Second, the reference condition assumes that carbon storage will be maximized if all vegetation types are at dense cover. This introduces error because some localities will not support dense forest (e.g. sparsely forested upper elevation rocky areas).

The standard deviation measure, which was calculated from the values of all pixels within each subwatershed, is relatively high, with values ranging from 9.9 to 17.9 (Table 4). This reflects the fact that only four discrete canopy cover classes were used to calculate the carbon values in each pixel, leading to discrete and well spread apart bins in the output values.

Table 4 — Standard Deviation of the mean in carbon stock estimation per subwatershed

 Subwatershed Confidence: Standard Deviation of Carbon Stock Estimate
Deer Creek ± 9.9
East Branch North Fork Feather ± 17.9
Lower Bear ± 11.4
Lower Feather ± 11.9
Lower Yuba ± 12.2
Middle Fork Feather ± 15.5
Middle Yuba ± 11.8
North Fork Feather ± 15.0
North Yuba ± 12.7
South Yuba ± 12.6
Upper Bear ± 13.2


With regard to NPP, these data were not readily available at the highest resolution provided by NASA. While the GIS processing of the raster data should provide an accurate estimate for calculated parameters, the smallest subwatersheds, for example Deer Creek, contain only a few of the low-resolution data cells that NASA currently provides through their website.

Technical Information

Data Sources

The primary GIS data source for the carbon stock calculations was the CalFire Multi-Source Land Cover layer (Fire and Resource Protection Program 2003), which provides 100 meter resolution habitat data for all of California. This dataset was compiled in 2002, by amalgamating the best available local sources for land cover information in California present at that time. Most of these local data sources were made available in the period from 1993 to 1998. Equations for calculating carbon stock were from Brown et al. (2004), using equations orignialy published in Smith et al. (2003).

In February 2000, the Moderate Resolution Imaging Spectroradiometer (MODIS), aboard NASA’s Terra and Aqua satellites, began producing regular global estimates for GPP and NPP at a spatial resolution of one square-km. When analyzing data from satellites, the scale, or resolution, which the data is collected can greatly influence the analysis. We downloaded these data from NASA Earth Observations, which provides global NPP data at a 0.1 degree scale (equivalent to approximately 8.5 km east/west and 11 km north/south at the study site). While this analysis could be improved with a finer-scaled dataset, with an average of 16.5 pixels for each subwatershed, this provided enough data to make estimates of general trends. The full dataset available was downloaded, with a temporal scale from February 2000 to January 2010 (120 GIS layers). These data were downloaded as georeferenced .tif files at the highest resolution (0.1 degrees) and as floating point pixel values. Each pixel represents the rate of NPP as grams of carbon uptake per meter squared per day (gC/m2/day), averaged over the 0.1 degree box and for that month.

The downloaded MOD 17 data is a product consisting of 8-day Net Photosynthesis (PSN) and NPP. Annual NPP is the time integral of the PSN product over a year.

These NPP data were used to provide an estimate of NPP for this study region. It has been previously found that areas recently affected by fire can cause the MODIS algorithm which is used to estimate NPP (MODIS 4.1 fPar) to overestimate NPP for many terrestrial ecosystems (Cheng, et al. 2006), and therefore, if the specific values were important, another data source should be used to validate MODIS data. Since this study has a coarse spatial resolution with a fairly stable ecosystem, we use these data to analyze the overall trend and assume a consistent variation of NPP estimates.

Data Transformations

We calculated the indicator value for each subwatershed in two steps. First, in a raster calculation we divided the estimated carbon stock layer by the target condition stock layer to produce a fraction giving the percent of maximum carbon storage for each pixel. Second, we averaged the carbon stock values for all pixels within each subwatershed to produce a value for each subwatershed. We calculated a measure of variation for each subwatershed in a similar way by computing the standard deviation of the values of all pixels in each subwatershed.


Carbon Stock

Brown et al. (2004) provided the first comprehensive evaluation of carbon storage and greenhouse gas emissions across agricultural lands, forests, and rangelands in California. We followed their methodology at a watershed scale in this analysis. They used the CalFire, FRAP, Multi-Source Land Cover (MSLC) layer as well as Land Cover Mapping and Monitoring Program (LCMMP) change maps to assess changes in carbon stock in the 1990s, referring to Smith et al. (2003) for measures of carbon content by forest cover type.

In particular, the CalFire MSLC layer provides habitat mapping for the state to 100 meter resolution using the vegetation classification from the California Wildlife Habitat Relationships (CWHR) mapping system (Mayer and Laudenslayer 1988). In addition to the vegetation type, this dataset gives information on vegetation canopy cover and canopy size where source data was available. The methodology in Brown 2004, calls for crosswalking the CWHR vegetation types to 5 forest types given in Smith (2003), namely Douglas fir, hardwoods, redwoods, fir-spruce, and other conifers. Taken together with canopy cover information, the equations in Brown (2004) allow for estimation of the carbon content (Table 5).

Table 5. Summary of equations available to estimate carbon standing stock in forest from Brown et al. (2004). In these equations, x is the canopy cover in percent, and y is the amount of carbon in Mg C/ha.

Habitat type Carbon estimation equation
Douglas fir y = -101 + 96 ln x
Fir-spruce y = -125 + 83 ln x
Hardwoods y = -70 + 52 ln x
Other conifer y = 59 + 2 x
Redwood/sequoia No equation provided, instead use carbon values of ~90 Mg C/ha for canopy densities < 40 %, and carbon values of ~300 Mg C/ha for canopy densities > 40% (the graph provides only 4 points because of scarcity of input data)

For shrublands and grasslands, Brown et al. (2004) use estimates for carbon content derived from other literature. In their report, Brown et al. (2004) do not provide carbon content values for woodlands, so the USDA Forest Service Carbon Online Estimator (NCASI 2010) was used to give carbon estimates for different age classes of blue oak, blue oak woodland being the dominant woodland habitat in the Feather River Watershed.

In a raster GIS, the portion of the MSLC layer that covered the Feather River Watershed was selected and analysis was restricted to the boundaries of the watershed using a raster mask. Using the CWHR habitat types in the MSLC layer and the crosswalk described above, vegetation pixels within the watershed were classified to one of eight vegetation types: either the five forest types listed above, shrublands, grasslands, or oak woodlands. Agricultural lands and developed lands were also masked out. The MSLC layer provides canopy cover information using the four canopy cover classes described in CWHR, namely sparse (10-24% cover), open (25-39%), moderate (40-59%) or dense cover (60% or greater). In pixels where the MSLC layer did not identify a canopy cover value. It was assumed this value was moderate cover. Using the mean values of the canopy cover class intervals, the carbon estimation relationships described above for the eight vegetation types were used to create a lookup table from which each pixel was assigned a carbon content value. All carbon stock GIS calculations were performed in the GIS GRASS (Neteler & Mitasova 2008).

A target condition layer was calculated using the same method, except that instead of taking the canopy cover value to be the actual value from the MSLC layer, dense cover was assigned. Because the carbon estimation relationships all reach their maximum value in the dense cover condition, this forces the output layer to have the maximum stock possible while keeping vegetation types the same for each pixel. The raster data for carbon standing stock and the subwatershed boundaries were intersected to generate a mean value per subwatershed.


NPP spatial analysis was done with ArcMap 9.3 and a series of Python scripts using the ArcGIS scripting engine. The Feather River Watershed was detailed by a vector polygon, and the zonal statistics aggregated the raster (pixel-based) dataset and summarized the results. A third party product, Hawth’s Tools version 3.27, was used to perform raster analysis, specifically the Zonal Statistics, on the set of NPP raster layers. Zonal Statistics produces a data table which includes the summation, minimum, maximum, mean, and standard deviation of the raster NPP values for each subwatershed. These data were then transformed from a column format (where each column represents a montly results) to a “long format,” where month is its own column and has subsequent columns for the corresponding data value.


Beyer, H. L. 2004. Hawth’s Analysis Tools for ArcGIS.

Blackard, J. et al., 2008. Mapping U.S. forest biomass using nationwide forest inventory data and moderate resolution information. Remote Sensing of Environment, 112(4), 1658-1677.

Brown, Sandra, Timothy Pearson, Aaron Dushku, John Kadyzewski, and Ye Qi. 2004. Baseline Greenhouse Gas Emissions and Removals for Forest, Range, and Agricultural Lands in California. Winrock International, for the California Energy Commission, PIER Energy-Related Environmental Research, March.

Cheng, Y., J. A Gamon, D. A Fuentes, Z. Mao, D. A Sims, H. Qiu, H. Claudio, A. Huete, and A. F Rahman. 2006. A multi-scale analysis of dynamic optical signals in a Southern California chaparral ecosystem: A comparison of field, AVIRIS and MODIS data. Remote Sensing of Environment 103, no. 3: 369–378.

Dixon, R. K., S. Brown, R. A. Houghton, A. M. Solomon, M. C. Trexler, and J. Wisniewski. 1994. Carbon pools and flux of global forest ecosystems. Science 263:185-190.

ESRI, 2009. ArcGIS 9.3. Environmental Systems Research Institute, Redlands, CA.

Fire and Resource Assessment Program, California Department of Forestry and Fire Protection. 2003. FRAP Data - Multi-source Land Cover Data (v02_2). Retrieved April 6, 2010.

Hudiburg, T., B. Law, D. P Turner, J. Campbell, D. Donato, and M. Duane. “Carbon dynamics of Oregon and Northern California forests and potential land-based carbon storage.” Ecological Applications 19, no. 1 (2009): 163–180.

Jenkins, J. C., D. C. Chojnacky, L. S. Heath, and R. A. Birdsey. 2004. Comprehensive database of diameter-based biomass regressions for North American tree species. General Technical Report NE-319, USDA Forest Service.

Karl-Heinz Erb, Fridolin Krausmann, Veronika Gaube, Simone Gingrich, Alberte Bondeau, Marina Fischer-Kowalski, Helmut Haberl, Analyzing the global human appropriation of net primary production -- processes, trajectories, implications. An introduction, Ecological Economics, Volume 69, Issue 2, Special Section: Analyzing the global human appropriation of net primary production - processes, trajectories, implications, 15 December 2009, Pages 250-259, ISSN 0921-8009, DOI: 10.1016/j.ecolecon.2009.07.001.

Mayer, K.E. and W.F. Laudenslayer Jr. 1988. A Guide to Wildlife Habitats of California. State of California, Resources Agency, Department of Fish and Game, Sacramento, CA. 166 pp.

Meigs, G.W., D.C. Donato, J.L. Campbell, J.G. Martin, and B.E. Law. 2009. Forest fire impacts on carbon uptake, storage, and emission: The role of burn severity in the Eastern cascades, Oregon. Ecosystems (2009) 12: 1246–1267

NASA Earth Observations, Net Primary Productivity / Terra MODIS, 1 month  (Accessed February 15th, 2010)

NCASI & The USDA Forest Service Research Work Unit 404, 2010. COLE Carbon Estimator. Retrieved April 6, 2010.

Neteler, M. & H. Mitasova. 2008. Open Source GIS: A GRASS GIS Approach. Third Edition. The International Series in Engineering and Computer Science: Volume 773. Springer, New York. 406 pp.

Nunery, J.S. and W.S. Keeton. 2010. Forest carbon storage in the northeastern United States: Net effects of harvest frequency, post-harvest retention, and wood products. Forest Ecology and Management, doi:10.1016/j.foreco.2009.12.029

Richards, K.R. & Stokes, C., 2004. A Review of Forest Carbon Sequestration Cost Studies: A Dozen Years of Research. Climatic Change, 63(1), 1-48.

Running, S., R. R. Nemani, F. A. Heinsch, M. Zhao, M. Reeves, and H. Hashimoto (2004), A continuous satellite-derived measure of global terrestrial primary production , BioScience, 54(6): 547-560.

Sedjo, R. A., and A. M. Solomon. 1989. Greenhouse warming: abatement and adaptation. Pages 110-119 in P. Crosson, J. Darmstadter, W. Easterling, and N. Rosenberg, editors. RFF Proceedings.

Siegenthaler, U., T. F Stocker, E. Monnin, D. Luthi, J. Schwander, B. Stauffer, D. Raynaud, et al. 2005. Stable carbon cycle-climate relationship during the late Pleistocene. Science 310, no. 5752: 1313.

Smith, J.E., Heath, L.S. & Jenkins, J.C., 2003. Forest volume-to-biomass models and estimates of mass for live and standing dead trees of U.S. forests, USDA Forest Service Northeastern Research Station. GTR NE-298

Williams, J. W, E. W Seabloom, D. Slayback, D. M Stoms, and J. H Viers. 2005. Anthropogenic impacts upon plant species richness and net primary productivity in California. Ecology Letters 8, no. 2: 127–137.

Woodbury, P.B., Smith, J.E. & Heath, L.S., 2007. Carbon sequestration in the U.S. forest sector from 1990 to 2010. Forest Ecology and Management, 241(1-3), 14-27.

van Kooten, G. 2009. Biological carbon sequestration and carbon trading re-visited. Climatic Change 95:449-463. doi: 10.1007/s10584-009-9572-8.

Zhao, M., S. W. Running, and R. R. Nemani (2006), Sensitivity of Moderate Resolution Imaging Spectroradiometer (MODIS) terrestrial primary production to the accuracy of meteorological reanalyses, J. Geophys. Res., 111, G01002, doi:10.1029/2004JG000004.