Reference Emissions Level [MGD Sections]

REDD+ Reporting [MGD Sections]

3.2   Integration frameworks for estimating emission and removals Previous topic Parent topic Child topic Next topic

Developing systems for reporting greenhouse gas emissions and removals requires combination of data from different sources, with data gaps filled through assumptions and expert judgement where necessary (Box 10). Tools to facilitate this are known as integration frameworks. Integration frameworks that are designed to simulate the impacts of human activities on future carbon stock changes can also support development of scenarios relevant to policy analyses. Integration frameworks can simplify reporting by automatically assigning land uses and emissions to the required classes based on rules set by the user, consistent with national definitions.
Ideally, an integration framework should be scalable and apply to forest stands, projects, regions or countries. It should also be able to start with simple, best available data, and be improved progressively; at each stage meeting IPCC good practice requirements of neither under- nor over-estimation so far as can be judged, and reducing uncertainties as far as is practicable.
Integration frameworks require knowledge of(1):
  1. initial land cover condition of the landscape (i.e. forest, non-forest or other land cover classes)
  2. drivers of change (activity data on human and natural disturbances), and estimates of subsequent land use (where a land use change has occurred)
  3. initial condition of the forest and rates of forest growth
  4. rate of carbon loss from decomposition and transfer between pools (for dead organic matter, soils, wood products as required)
These data, in particular the data on land cover, land cover change and the agents of change are increasingly obtained from remote sensing. These can greatly assist in describing the history of land-cover changes that drive emissions and removals (Chapter 5, Section 5.1). The further back in time these data go on a consistent basis (Box 23), the more reliable and useful they are as the inputs to the integration tools.
The analysis of the impacts of future REDD+ (or forest management scenarios more generally) can be undertaken with integration frameworks (e.g. CBM-CFS3 Opens in new window, FullCAM Opens in new window) that can use scenarios of future activity data to extend historical time series activity data. For example, if the past rate of deforestation activity is estimated from remote sensing observations, this rate can be extended into the future as a baseline (e.g. the average rate of deforestation over the past N years) and be compared against one or more scenarios showing the impacts of reducing deforestation rates by X or Y% per year (e.g. Kurz et al., 2016). Provided the socio-economic drivers can be identified and quantified and the relationship between them understood, it is easier to extend time series of activity data in integration frameworks that use spatially-referenced activity data. Extending the observed time series of activity data with projections about alternate future management regimes in Canada’s National Forest Carbon Monitoring, Accounting and Reporting System have allowed for the evaluation of various climate change mitigation strategies (Smyth et al., 2014).

Box 10: Data, assumptions, models, tools and emissions estimation

All emissions estimation relies on measurement data, assumptions, models and other tools. Understanding each of these components is helpful when developing MRV systems.
Data: Data can be divided into measurement data (such as forest inventory measurements) and derived data (such as biomass estimates derived from the base measurements such as diameter at breast height). Derived data require the application of models such as volume and taper equations to estimate tree volumes or allometric models to estimate biomass. Measurement data have errors associated with the measurement and derived data have errors associated with the model in addition to measurement errors.
Assumptions: To convert input data to numbers that can be used in emissions estimation requires assumptions. For example, emissions factors assume that growth occurs at the same rate between two points in time, while growth curves assume that the forest is following a non-linear growth pattern. These assumptions affect the accuracy of the results at any point in time and cannot be improved merely by increasing the statistical accuracy of an individual point in time.
Models: All systems rely on models of various complexity and all models rely upon data and assumptions. Generally, moving from simple assumptions and models, such as linear growth curves in Tier 1 and 2 emissions factor methods, to more realistic s-curve forest growth and yield assumptions in Tier 3 methods leads to greater accuracy in emissions estimates through time (Figure 10). The increase in complexity between Tier 2 and 3 can be small in the case of empirical growth curves (which are commonly applied in forestry operations worldwide), or large if implementing more complex, physiological models. Models can be brought together through the use of integration tools.
Integration tools: Integration tools combine multiple streams of information, most commonly spatial data, such as from remote sensing, and forest inventory data from ground and plot measurements with models. Models can help obtain estimates for pools that are difficult to measure (e.g. soils), and to extrapolate measurements obtained from plots across space and time. Tools may range from simple excel sheets (e.g. EXACT Opens in new window) through to stand-alone executables (e.g. ALU Opens in new window) and detailed analytical systems (e.g. CBM-CFS3 Opens in new window, FullCAM Opens in new window, and the new system under development: FLINT Opens in new window). Some of these tools may have models and assumptions built into them, but most are flexible and allow for different data and assumptions to be used and modified.
The figure below presents a comparison of an emissions/removals factor model and a typical growth curve. Both predict similar biomass at age 100, but the pattern is different, leading to potential bias in estimates of carbon accumulation rates in biomass. Both models are simplifications of the real biomass accumulation rate which also varies over time as a function of climatic and other environmental conditions. This can have a significant impact over short periods.
Box2-figure.jpg

 (1)
The points listed are most relevant for integration frameworks based on gain-loss approaches.