Nations need to be able to account for their carbon stock in order to know if they are reducing greenhouse gas emissions or not. Different techniques give different results, depending on the level of precision, and there are certain things that can be done to make it clear, say Meine van Noordwijk, Sonya Dewi, Betha Lusiana, Degi Harja, Fachmudin Agus, Subekti Rahayu, Kurniatun Hairiah, Maswar, Valentina Robiglio, Glen Hyman, Douglas White, Peter Minang, Lou Verchot and Vu Tan Phuong
The international mechanism to reduce deforestation and forest degradation (REDD+) seeks to establish ‘performance-based’ financial instruments to make forests more valuable standing than destroyed.
To achieve this, trusted, reliable and transparent national carbon accounting systems are essential. But the accuracy of the estimates of carbon stock and emissions depends strongly on scale: methods that are sufficient for reliable national accounting may not be accurate at local site level.
The proposed REDD implementation mechanisms thus influence the required levels of precision at specific scales and the benefits that stakeholders can obtain from investment in better data.
Within a general scheme of the type of tree, forest, soil and land management practices that are needed to estimate emissions, we reviewed a number of datasets to assess sources of bias and random error, linked to the level of replication that is needed to achieve specified precision. We also summarized data on the costs of data collection at a number of scales, with different levels of precision. In combination, the costs and benefits of investment in data quality can be weighed and a balance achieved between achievement and ‘transaction costs’ (to which the costs of designing a monitoring system contribute).
To be cost effective, national monitoring systems can build on existing forest inventories and soil data but they need to be analyzed for bias and variability to assess adequacy for carbon-stock appraisals.
Examples for Indonesia show the gap between these data and intensive ecological studies: reconciliation of the data sources requires reanalysis of the site selection for ecological studies and of pre-1990 logging across the country.
Based on our research, we have devised 10 recommendations for national monitoring systems that combine biophysical and institutional dimensions of system design.
1) Start with what you have: forest department data, agricultural statistics, land-cover studies, spatial planning zones, existing use rights, soil maps and soil-fertility databases can all contribute important information.
2) Expect gaps and mismatches between data sets, especially where institutional and biophysical concepts use the same terms (for example, ‘forest’).
3) A national monitoring system is dependent on three characteristics:
a. Salience (does it address key policy issues and respond to policy implementation within a relevant time scale?)
b. Credibility (are the methods up to date and consistent with international standards; are confidence intervals of key parameters known; is error propagation towards final estimate traceable with realistic degrees of confounding of component errors?)
c. Legitimacy (is the work done by agencies and individuals that are, through their combination and cooperation, seen to represent the specific and valid concerns of:
i. local, sub-national and national governments (aligned with reporting obligations);
ii. local people and indigenous group representatives;
iii. the local, national and international private sector bodies with interests in the ‘footprint’ of land use associated with commodity value chains; and
iv. environmental NGOs at local, national and international level).
4) Involve local stakeholders in data collection and international expertise in consistency and validity checks.
5) Invest in methodology, harmonization of legends (operational classification scales), ‘one map’ consistency of spatial data across government agencies, and clarity of operational definitions from a local perspective before investing in new data collection.
6) The Intergovernmental Panel on Climate Change’s Tier 2.5 (agriculture, forestry and other land use) accounting system is a feasible and realistic goal for any country wanting to participate in REDD+ debates; it involves stock-change accounting with sub-national land-use classes, carbon-stock (activity) data that are adjusted to eco-climatic zones, major soil types (including peat and volcanic soils as special classes), and the typical management practices across the life cycle of land-use systems. It also requires area data of land-cover change with matching legend (more than 20 map units may be needed); for Indonesia most of this was achievable with an external investment of about EUR 1 million plus data and staff capacity of national agencies.
7) The protocols for Rapid Carbon Stock Appraisal (Racsa) allow local data collection and reporting at a cost of EUR 10,000–20,000 for areas of 20 x 20 km2 to district scale, if carried out by competent national universities and NGOs, in cooperation with local governments.
8) High-precision, location-specific data (for example, following VCS protocols) are only useful if nested within the national system and its hierarchical legend units, and when possible sampling bias (selective focus on high carbon stock or high emission areas) can be assessed.
9) Monitoring and reporting: The quality of interdepartmental coordination between custodians of various data sets that contribute to the national accounting system determines the quality of the national accounts; it requires considerable effort and adjustment of institutional incentive systems; and Verification: Basic data on soils and tree cover need to be open to public scrutiny in sufficiently high resolution to allow public scrutiny and corrections; an appropriate system for obtaining feedback, verifying local discrepancies and adjustment of databases is needed, and might require appropriate budget.
10) The primary accounting precision target for REDD+ and NAMA is the national scale, consistent with National Communications on Greenhouse Gas Emissions; this implies that bias issues (systematic error) are prominent and require attention in temporal consistency, while random error is less problematic for national reports; local-scale confidence levels at the finest spatial scale that is publicly accessible, however, influence the fraction of local stakeholders that will see their area as misrepresented.
Edited by Robert Finlayson
Read the report
Van Noordwijk M, Dewi S, Lusiana B, Harja D, Agus F, Rahayu S, Hairiah K, Maswar M, Robiglio V, Hyman G, White D, Minang PA, Verchot L, Phuong VT. 2012. Recommendations on the design of national monitoring systems relating the costs of monitoring to the expected benefits of higher quality of data. Project Report. Bogor, Indonesia: World Agroforestry Centre (ICRAF) Southeast Asia Regional Program.
This work relates to the CGIAR Research Program on Forests, Trees and AgroforestryDownload PDF copy