Using Terrestrial Ecosystems for Carbon Sequestration
GARY K. JACOBS
Oak Ridge National Laboratory
Using the world’s terrestrial ecosystems to sequester carbon is a familiar idea. I’m going to take a slightly different tack to this question and share some thoughts on research and technology that might help us harness the power of our terrestrial ecosystems. I want to thank and hold harmless many of my colleagues from the U.S. Department of Energy (DOE) Research Consortium for Enhancing Carbon Sequestration in Terrestrial Ecosystems (CSiTE). Some of the results I will present are preliminary and subject to change. I am grateful to my colleagues for allowing me to share their information, any errors are mine and not theirs. I also wish to thank the DOE Office of Science for their support of this research.
What do we really mean when we talk about sequestration in our terrestrial ecosystems? Many interrelated processes go on in these systems. Fundamentally, the first thing we can do in terrestrial ecosystems is to protect, recover, and manage existing resources; that has a certain benefit numerically. Next, there may be opportunities to intervene and enhance existing resources. We are trying to solve a global problem, but some of our options will involve manipulations or interventions that actually make changes at the molecular level—for instance, to protect soil organic carbon. There are three basic intervention strategies: (1) increasing the productivity of ecosystems, thereby creating more biomass and more inputs to the soil-carbon pool; (2) partitioning carbon to the long-lived components of ecosystems (e.g., converting from grassy to woody systems); and (3) altering the longevity of existing pools (e.g., protecting soil by changing agricultural practices). There may also be ways to adjust soil-microbial communities to promote certain behaviors that protect soil organic carbon.
But let’s be honest. Terrestrial ecosystems alone will not be able to balance fossil-fuel emissions. The potential contribution from terrestrial ecosystems is not as minuscule as some have implied, but neither is it the answer. Terrestrial ecosystems can buy some time for our global community, and using ecosystems to sequester carbon would have some other benefits (e.g., erosion protection, improved soil productivity, increased biodiversity, and retention of soil moisture).
The challenge is to shift both the rate of carbon accumulation and the capacity to store it. Right now, we do not even know for certain what the capacity is. And even if we can shift these rates, we don’t know the potential consequences. They could be very positive (e.g., improving soil productivity, reducing erosion, retaining moisture more thoroughly) or less positive (e.g., changing biodiversity, shifting valuable lands from one crop to another). We must consider all the implications.
Right now, the terrestrial biosphere acts as a biological scrubber that removes roughly two gigatons of carbon per year (GtC/y) from the atmosphere. Of course, one must also recognize that global change could alter their current behavior, and ecosystems could become net emitters of carbon. Nevertheless, there are several options for increasing the current sequestration rate. Through reforestation, afforestation, restoration of degraded lands, improved agricultural practices, and the use of biomass feedstocks, the global potential might be as high as 5 GtC/y. These rates could be sustained for perhaps 10 to 50 years, resulting in a total sequestration potential of 50 to 250 GtC—a significant help to meeting the global challenge.
But we are concerned not only about the impact of carbon dioxide (CO2) but also about the total integrated impact of any intervention on the Earth’s climate system. That is the real issue we have to evaluate with any option. For example, growing more forest will have an impact on the reflectivity of the Earth’s surface and the water cycle. If we do too good a job of restoring degraded lands, might we change the amount of wind-blown desert particles that fertilize the ocean? We have to remember that none of these systems can be viewed independently. We can change land area or the below-ground carbon or the above-ground carbon, but we have to keep in mind the dynamics of whole ecosystems and evaluate how we might affect the global system.
Scientific efforts in various disciplines could potentially contribute to important solutions. The first of these is the selection and engineering of species. Consider a tree, and hypothesize that there are genetic controls on the allocation of below-ground and above-ground carbon (i.e., roots versus shoots). Furthermore, there may be genes that are important in the partitioning of the carbon within lignin, hemicellulose, and cellulose, both above and below ground. Stan Wullschleger and Jerry Tuskan of Oak Ridge National Laboratory and their colleagues are trying to discover the genes that might allow us to take advantage of this allocation behavior. Their preliminary results are encouraging.
The DOE Office of Science has just initiated some exciting research with the Joint Genome Institute in this area. The first tree to be sequenced, Populus, is a species with many potential uses in the energy area. Current domesticated varieties are leafy, thin, and competitive with one another. Once the genome sequence is completed, we could begin to address important issues, such as reducing competition within a plantation, optimizing the wood chemistry for either fuel or products, increasing resistance to pests, increasing the yield, enhancing storage of soil organic matter in the below-ground system, and expanding the range of effective plantations to other climate regimes. In 25 years, could we double the yield? Could we reduce the price of harvesting and processing? Could we sequester carbon at the same time? The possibilities are intriguing.
The DOE Office of Science Genomes-to-Life Program offers many other possibilities. As we develop a better understanding of complex soil-microbial communities, we may be able to enhance certain functions that will yield soil-organic materials that are more resistant to further degradation. In addition, Julie Jastrow of Argonne National Laboratory is working on new methods of fractionating various types of organic matter that may yield clues to how we can protect soil-organic matter from microbial degradation.
Research is also being done on promoting behaviors in soils that encourage the formation of humic material that is more resistant to degradation. Jim Amonette of Pacific Northwest National Laboratory is studying relationships among organic monomers, enzymes, and manganese oxides that may promote humification reactions. The challenge is to optimize cycling between oxidizing and reducing conditions while increasing concentrations of monomers and stabilizing the enzymes. It is feasible that we might learn to do this via a combination of crop rotations and new amendments applied during fertilization.
In 1975, DOE’s FermiLab began restoring a tall-grass prairie. Julie Jastrow and others are following the changes in the soil carbon over a significant amount of time in this chronosequence. The good news is that carbon can accumulate and that this accumulation is occurring without fertilization. Sampling of the soil indicates a pretty substantial increase in soil carbon, and preliminary evidence suggests that its mean residence time could be 127 years.
Changes in land-management practices also offer opportunities for substantial accrual of organic carbon. It will be critical to do a full accounting of greenhouse gases for every new method, from conventional to no-till practices. As management strategies change, there will necessarily be consequences, good and bad. It is essential that we develop tools to carry out truly comprehensive evaluations.
We must evaluate all impacts and costs for all sequestration options. We are already doing that systematically for terrestrial ecosystems. Tris West and Gregg Marland of Oak Ridge National Laboratory are developing methods of performing a full accounting of greenhouse gases from shifts in agricultural or forest
management practices. Gregg Marland is also developing relationships so we can compare the values of storing carbon for different lengths of time. It is important to be able to represent the time-value of carbon in products. Cesar Izaurralde and colleagues at Pacific Northwest National Laboratory and several universities are improving models for evaluating the environmental aspects of sequestration options. The factors being addressed include soil carbon, erosion, nutrient availability, runoff, and non-CO2 greenhouse gases. Finally, there is the matter of economics. Bruce McCarl of Texas A&M University is addressing the issue of what is possible within economic constraints. He has shown that the value of carbon will have a large impact on the potential for ecosystem management that contributes to a global solution.
How much difference can innovative research make in realizing a significant role for terrestrial ecosystems? First of all, we already manage a large portion of our Earth’s terrestrial resources. There are estimates that we already “use” 10 to 55 percent of terrestrial production. Agricultural productivity has increased dramatically in the last 50 years, and some studies project that similar increases will continue for the next 50 years.
Earlier, I said that future carbon sequestration rates could be, say, 2 to 5 GtC/y for several decades. Roger Dahlman of the DOE Office of Science has estimates (based on results of the Ameriflux Program) of perhaps 3 gigatons being sequestered globally in forests. Thus, my stretch goal of 5 gigatons may not be unreasonable. I believe that science and technology will enable us to extend what we are already doing for a longer time, increase rates of sequestration, and eventually lead to higher capacities for carbon storage.
What we need most right now is the ability to do things on a practical scale to discover what works and what doesn’t. Implementing these practices over a large land area will, of course, pose some engineering challenges, but only large-scale tests and demonstrations will determine what motivates behavior and what the impediments (technical, social, and economic) are and how they can be overcome. Our challenge is to deploy some near-term options to buy some time and do some learning.
The bottom line is that we know a lot right now. Through innovative research, terrestrial ecosystems could substantially help solve the problem, buy us some time, and provide meaningful ancillary benefits to human and ecological systems. But they are not the sole answer. Fully integrated evaluations will be critical to informed decisions, and we urgently need large-scale test systems operating in parallel with excellent fundamental science.
Monitoring Carbon Adsorption in the Terrestrial Ecosphere
Terrestrial ecosystems may offer near-term, cost-effective ways of storing carbon. We are trying to bring good science to the monitoring and verification of carbon storage—to determine if carbon is actually being stored, if so, in what quantities, and if the storage is cost effective. Today I want to talk about monitoring systems and the practical field testing and application of monitoring systems, discuss some examples of large-scale measurements, and put forth future monitoring options we think may change the way business is done.
On a global basis, there is a net emission of carbon from terrestrial ecosystems. Loss of forest cover and changes in land use in recent decades have already put a significant amount of carbon into the atmosphere, and we are trying to monitor what is there now. There is a widespread belief that sequestration is a near-term option, and a number of sequestration projects are ready to go forward.
When creating a monitoring system, the first point to consider is design objectives. What are the criteria? We designed a monitoring system for accuracy and precision at predictable levels. Integrity of results was very important. The methods I will discuss here have been peer reviewed and are available free on our website (www.winrock.org). We will be publishing a revised set of methods, in partnership with the Center for International Forestry Research in Bogor, Indonesia. We do regular updates, so we encourage everybody to make suggestions or recommendations.
A monitoring plan has certain essential parts. First, we stratify projects to minimize the costs of measurement; we use the statistical sampling approach. We are trying to reduce the variability within each stratum so we can get high levels of accuracy and precision with fewer sampling plots. We sample the
variability of the particular class of carbon that is being stored. When we put together standard operating procedures for doing the measurements, we find that setting projects in a long time scale introduces an element of uncertainty.
When setting up a monitoring system, it is essential that the instructions be very specific, because the system may be in place for 50 years and may be run by different people and different management systems. The instructions must address the most minute aspects of the system, for instance, whether to put the tape under the vine or over the vine and how to deal with a forked tree. We put in recommended frequencies for monitoring, quality assurance, quality control plans, and soil blanks in the soil samples to check the calibration of the equipment.
Then the data must be archived. Over the 10 years of this project, the software has been changed half a dozen times. Try to envision deciding on a form to keep your data in so it can be referenced 50 years in the future. That is not a trivial decision.
To stratify sample sites, we try to pool available data. In the United States, it is easy to find digital elevation models and topographic maps, aerial photography, and satellite imagery. When we put in preliminary plots, we can determine the variability within those strata and estimate the number of plots.
There is a trade-off between cost and precision. The more precise you want your numbers to be, the more plots you have to put in. But the more plots, the more the expense; it is always important to keep an eye on costs. Fixed and variable costs are different. Fundamental design cost is unavoidable, no matter how big the project is. But the costs of methodology details can be found in the Intergovernmental Panel on Climate Change year 2000 special report, Land Use, Land-Use Change and Forestry. The lead author on the chapter about project monitoring is one of my coauthors, Sandra Brown (2000).
These techniques do not require cutting-edge science. Monitoring methods are based on the principles of forest inventory, soil sampling, and ecological surveys and have been in place for decades. By talking to scientists involved in those areas, you can arrive at consensus about acceptable methods. There may be some debate, more in the soil community than in the forestry community (e.g., whether a geomorphologist’s approach is more useful than a microbiologist’s approach in soil analysis), but there is general consensus about methods that have been used with confidence over long periods of time by a variety of institutions.
For non-carbon dioxide (CO2) flux gases, however, methods are not as well established. When we talk about the complete package—especially, when we look at methane and nitrous oxide—monitoring methods are not clearly established in the literature, especially for comprehensive accounting that requires manipulating inputs in large land systems. Pilot projects would provide a good opportunity for trying out methods.
We always put in permanent plots. We use mean grass-type diameter for trees, and we use regression equations to convert mean grass-type diameters into
biomass content. When looking at understory litter, we use clip plots with standard rings to dry the samples for carbon analysis.
Dead wood is an important element in most forestry systems; it can be up to 20 percent of the total carbon in the pool. The flux changes a good deal depending on the climate conditions. For a standing biomass, one can use regression equations for dead wood on the ground. We use a line-intersect method and sampling for density to estimate the carbon.
We currently dig soil pits (currently 30 centimeters deep) into a permanent plot. If we go to 50 centimeters, we blend the 30-centimeter and 50-centimeter samples together. We put the samples through a 2-millimeter sieve, air dry them, and then do a carbon analysis. We also take a bulk density from a vertical wall in one of the pits. This process is labor intensive and somewhat expensive. We have been following some interesting work on laser-induced breakdown spectroscopy that we think will reduce the costs of soil sampling. But, for now, we are using the traditional approach.
We decide which pools will be measured based on a number of criteria. Not all of the pools have to be measured, but every pool expected to be negative must be measured. Beyond those, we decide which pools to measure based on the cost of measurement, the expected value of the carbon, the amount of carbon we expect to find in the pool, how fast and in what direction we expect the change to be, and how accurate we want to be.
When measuring changes in soil carbon over time, we are trying to detect much smaller changes against a larger background than when we measure trees. Therefore, we need larger sample plots, so the costs are higher for, perhaps, a smaller amount of carbon. Those factors can influence your decisions depending on the project design.
Measuring below-ground biomass is a complicated exercise and quite expensive. We are now measuring more than a million hectares on a global scale. In those projects, our cost estimates are based on regressions. Three categories of measurements were agreed on in the Kyoto negotiations: additionality, baselines, and permanent leakage.
Most regulatory systems and trading systems have guidelines that govern what must be measured. Besides the physical properties and presence of carbon, other concerns may be taken into account, such as the three major classes of error: sampling error, measurement error, and regression error. By far the largest errors are sampling errors, which reflect flaws in the preliminary design and the stratification and estimated variability. Measurement errors are mostly transposed numbers on forms, which are very difficult to eliminate. Regression errors, surprisingly, do not seem to make much difference in measuring carbon in classes of trees. Standard regression equations from 34 different species of eastern hardwoods east of the Mississippi in the United States reveals a 0.99 curve fit. Many people predicted that measurements for tropical forests would be very
different, but a regression for 224 species of tropical hardwoods in a mixed tropical forest was 0.97. Predicting root structure based on these factors for below-ground biomass, we end up with a regression correlation of only 0.83.
As a nonprofit institution interested in environmental sustainability, we emphasize another aspect of measurement—the co-benefits. In this respect, there are clear differences among different types of carbon-storage projects, which may have biodiversity benefits or habitat benefits or watershed-restoration or wetland benefits. We try to measure those types of benefits in the same transparent ways. The same also holds true for socioeconomic benefits, such as changes in income, availability of jobs, and sustainable production systems.
For example, the Noel Kempf project, designed by the Nature Conservancy, is probably the largest carbon project implemented to date. It is a $10-million program that covers about 1.5 million acres (640,000 hectares) in Bolivia. The Nature Conservancy project bought together all of the logging concessions harvesting trees from the forest to maintain the area as a park; the Conservancy also tried to reduce the influx of slash-and-burn farmers on the park perimeters. We broke up the 1.5 million acres into half a dozen different strata and put in 625 permanent sample plots in the park. This is probably the first large-scale carbon assessment of a mixed natural tropical forest. Our conclusion was that the park was sequestering 225 tons of carbon per hectare. Across the site, we surmised that inundated soils would have a lot more carbon in them, and we were very interested in providing data to support that guess. This was also an issue in a number of other projects.
The Nature Conservancy ran another project in Brazil’s Parana state, in the Atlantic forest. The project sponsors bought up buffalo ranches and reforested them to defragment a natural Atlantic forest. Land holdings were added to restore the integrity of the block of forest. When we went in to monitor the carbon content in different areas, we had to give different expected carbon values to different parts of the land holdings. We are studying average carbon density, as well as minimums and maximums. The diverse results we are getting give us a sense of the accuracy and precision that can be achieved across the statistical sampling regime.
In the future, we hope to bring down the cost of measurement. The cost for the work I just described on the Atlantic forest project has been less than 25 cents per ton of carbon. We actually could have achieved 20 to 22 cents for that measurement. To bring down costs in other ways, we are looking into remote-sensing techniques and working on combining those with existing ground-based measurements. The goal is to bring costs into the 10- to 15-cent range.
We have also been working to develop new technological approaches to monitoring. We considered using satellites, but found that, at this point, they are more expensive and unreliable than sending people in on the ground. We tried some aerial techniques but found we had a hard time differentiating crown dimensions. We have now developed a camera system for aerial monitoring that
uses digital cameras and digital video cameras, logged with a global positioning system (GPS) system hooked to a satellite, an internal onboard navigational system to correct for pitch and yaw, and a laser range finder that shoots 2,000 pulses a second so we can get canopy height and ground height in over-flights. We are also using dual cameras to get a stereo image that enables us to differentiate crown heights.
This is how the system works. When flying over a site, the cameras take two positions equidistant from a point in the frame; we then put together an epipolar model that allows us to come up with a three-dimensional terrain map of a complex forest. Between the height of the trees and the crown diameters, we can come up with good estimates of biomass. This gives us a fairly high degree of confidence and certainty that the trees we are measuring are there. Over a period of time, we can go back and look at a specific tree (all trees are given georeference points). The data are all there. At any point, we can zoom in on a particular tree and flip it into a three-dimensional model, doing a real-time measurement on the crown and the crown height. With this technique, circling the crowns and taking their heights, we can do a sampling routine based on flying over a strip and identifying trees and crown diameters, rather than having to send in a ground crew.
Currently, we are working on automating the process of drawing crown diameters, which, in two dimensions, had been very difficult. In three dimensions, however, it appears to be feasible.
Early results of fly-overs with a large number of samples and higher levels of precision have been surprisingly good. When we tried to do aerial photography, the R-squared were about 0.6. But with three-dimensional terrain modeling, we have been getting R-squared higher than 0.9 on a regular basis. So far, we have only tried it in about a dozen places, so we are not sure what the limitations are. In one example, the ground plots gave us 89 tons of carbon per hectare with a 95-percent competence interval, whereas our aerial measurements gave us 87.7 tons of carbon per hectare with a 95-percent competence interval. We think we will be able to reduce error with new regression equations that go directly from crown diameters to biomass instead of going through existing algorithms.
The aerial monitoring equipment, which was designed to be inexpensive, can be flown on any small aircraft; we put it into a suitcase and take it on the Cessna with us. Weather constraints have not been a problem for collecting data and imagery, which was one of the problems we had with satellite data.
Aerial monitoring with this camera system can also be used for other kinds of projects, such as soil projects. For example, we could monitor compliance of a group of farmers who had agreed to use no-till management. With this kind of imagery at low altitude, with the benefits of a three-dimensional model, we can see the furrows in a field to determine if farmers had been plowing. If there are minimum setback requirements, we could measure, say, the distance between a field and a river. We could look at the impact of water and flooding regimes because we would have a georeferenced digital elevation model for the area. We
could also determine access points—how to get into a particular site to do measurements.
Most of all, over time, this kind of information gives people a high level of confidence that, in fact, what you say is there, is there. Whether in the United States or in a remote place in Bolivia, pictures are sometimes more convincing than statistical reports full of tables of data. We are excited that this aerial monitoring system cannot only take measurements but can also increase the confidence in near-term markets because of their potential to buy time in the short term while other technologies are being developed.
Thanks to developments on the Internet in the last two years, this information may also be layered into a data-rich field. It is now possible to add all kinds of data layers. We can put down a topographic map, overlay the hydrology on top of it, put in the road system, gas pipelines, transmission lines, put down land-sat photos. When you go to a particular project site and lay in that image, the new data you collect can be layered onto all of the existing data from the Internet. This means this tool can be used for a much broader range of environmental assessments and change detection.
At any point in a flight, you can look down at a specific location, ask what is going on, and take a measurement. By moving my cursor over a tree, for instance, I can tell how tall the tree is. If I want to know the distance between a road and a particular point, I can do the same kinds of measurements in real time off a dataset, knowing all of the other details that are embedded in this image for tracking purposes. For forest certification stewardship, just in terms of harvesting, we can tell how many trees were taken out in a selective harvest; we can also measure the size of those trees and how many board-feet of timber was probably in those trees. This method of data collection could be used not just for single points but also for large areas.
Let me close by saying that we are very concerned about the integrity of measurement systems. We need transparent standards based on good science. We believe new imaging tools will reduce overall monitoring costs and enable us to measure a broader range of environmental attributes for projects and, at the same time, make the projects more credible. All in all, we think that it will be possible to make accurate measurements for forestry projects anywhere in the world at low cost.
Brown, S., O. Masera, and J. Sathaye. 2000. Project-Based Activities. Chapter 5 in Land Use, Land-Use Change and Forestry, R.T. Watson et al., eds. Special Report by the Intergovernmental Panel on Climate Change. Available online at http://www.grida.no/climate/ipcc/land_use/index.htm