3. PRODUCTS 15
3.3. Terrestrial 59
3.3.9. ECV Above-ground Biomass 87
Biomass is defined as total mass of living plant material per unit area, but this ECV adopts a more limited definition, dealing only with above-ground biomass (AGB) in forests and woodlands. This is a pragmatic restriction, since, although a significant proportion of biomass may be stored below ground, this is difficult to measure, even by in situ methods, and it is often estimated from AGB using allometric equations. In
addition, little is recorded on non-forest biomass, except through agricultural yield statistics. AGB can be measured at stand level in temperate and boreal forests with an accuracy of 10 to 20 per cent using in situ
methods, but tropical forests present a greater challenge. Estimates of AGB can vary greatly depending on the allometric equations used to convert the in situ measurements (for example, diameter at breast
height for trees above a certain size in a given area) to AGB.
AGB is an important fraction of the carbon stored in the terrestrial domain, and its dynamics play two major roles in the climate system:
• Photosynthesis withdraws CO2 from the atmosphere and stores it as biomass, which then provides a source of soil carbon through plant detritus and mortality, with associated respiration;
• The amount of CO2, other trace gases and aerosols emitted by fire depends on the quantity of biomass consumed.
Differences in the assumed value of the average biomass gave rise to differences of 1PgCyr-1 in the range of estimates of emissions from tropical deforestation. Biomass information is also important in the UNFCCC COP’s invitation to Parties, relevant organizations and stakeholders to support ongoing efforts, capacity building, demonstration activities and mobilization of resources relating to reducing global greenhouse gas emissions from deforestation and forest degradation in developing countries, and to
enhance forest maintenance, sustainable forestry management and carbon storage by forest lands (REDD+; Bali Action Plan, UNFCCC Decision 1/CP.13).
Many developed countries have national forest inventories that span decades and contain data from a large number of sampling locations; but many forest biomes, in particular those in developing countries, have little or no inventory data. Nonetheless, inventories form the main basis for the periodic (typically five-year) Global Forest Resource Assessments produced by the FAO, the most recent of which was issued in 2010. The annual national reporting on Land Use, Land Use Change and Forestry required by the UNFCCC also mainly relies on inventories.
The labour-intensive nature of establishing an inventory and its requirements for infrastructure have motivated the search for ways to infer AGB from remote sensing data. Both passive optical sensors (such as Landsat or MODIS) and active sensors (lidar and radar) have been used, but each these provides fundamentally different information.
Passive optical sensors respond to upper canopy biochemistry and structure and to the topography of the canopy caused by shadows and gaps. The information they provide on biomass is therefore indirect and typically requires the data to be combined with other environmental and forest data, in some form of inference structure. This approach is used successfully, for example, in Sweden, where there is a lot of supporting data, but it is not readily extended to the global scale.
Lidar signals tend to penetrate the canopy and give information affected by both components of the canopy and the ground. Lidars can measure forest-crown sizes, gaps and tree height and, if operated in waveform mode, can also estimate the vertical distribution of material in the canopy from which biomass can be inferred. The ICEsat mission (2003-2009) provided a global dataset of forest heights. This has the potential to be combined with other data to infer biomass.
Radar signals can also penetrate the canopy and tend to scatter from elements of the canopy comparable to their wavelength, so longer wavelengths provide information on the larger elements of the canopy. Hence the range and upper limit of biomass they can measure increases with wavelength, as demonstrated by airborne synthetic aperture radars (SARs) using wavelengths from a few cm to several m (and also by scattering models). From space, however, the longest usable wavelength is ~70cm (P- band) because of international regulations and ionospheric effects.
Various radar techniques have been used to measure biomass from space. L-band backscatter measured by the JAXA Daichi (ALOS) satellite is sensitive to biomass up to a saturation level around 60-80t/ha and so has proved useful in mapping biomass in areas such as African miombo forests. Good correlation is observed between biomass and information derived from long temporal sequences of Envisat C-band (6cm) data in large area studies in boreal and temperate forests. In boreal forests, strong correlations have been found between biomass and coherence for pairs of winter images acquired by the ALOS radar (coherence is a measure of pixel correlation between an image pair). Neither of these last two methods displays saturation as biomass increases.
Improved mapping of global AGB from space is likely to require low-frequency radar, preferably in association with a lidar. The first of these technologies is in the advanced planning stage for a space mission, namely the ESA P-band BIOMASS mission (which includes a polarimetric interferometry capability, allowing forest height to be measured); but the lidar capability has recently been dropped from the NASA DesDynI mission, and its retention carrying an L-band radar is under review. The documentation for both missions includes thorough analysis of the case for measuring biomass as an ECV and the properties of the sensors needed to measure it from space. The proposed NASA Icesat-2 mission will carry a lidar; although optimized for land and sea ice applications, it will provide estimates of forest canopy height.
The following is required for this ECV:
Product T.9 Regional and global above-ground forest biomass Benefits
• Biomass information can help to initialize and test the carbon cycle models that are embedded in the latest generation of climate models;
• Biomass provides an estimate of carbon stocks in terrestrial ecosystems and hence carbon emissions due to fire;
• Biomass change is a direct measurement of carbon sequestration or loss. Target Requirements Variable/ Parameter Horizontal Resolution Vertical Resolution Temporal
Resolution Accuracy Stability
Gridded above- ground biomass (dry weight of woody matter (t/ha)) 500m-1km averages based on 100- 200m observations N/A Annual biomass maps, with coverage of all major forest areas on globe <20% error for biomass values > 50t/ha, and 10t/ha for biomass values
≤ 50t/ha
10%
Rationale: The requirements are driven by the need to quantify the distribution of above-ground biomass (carbon stocks), to initialize and test the carbon cycle models that are embedded in climate models (see ESA BIOMASS report (2008)), and to provide a basis for estimating carbon in the context of national reporting. Changes in forest biomass over time need to be mapped to estimate carbon fluxes, either through emissions (loss of forest) or uptake (forest growth). Horizontal resolution is linked to the typical size of forest disturbance (~1ha), but coarser scale maps averaging this high-resolution information are adequate for some applications. The accuracy and stability requirement is linked to the current ~20 per cent uncertainty of net ocean carbon uptake.
Currently achievable performance
Current capabilities vary by region, as indicated above. Several of the methods and datasets have not yet been peer-reviewed, and there is considerable variation between products. Threshold requirements are perhaps best specified through the landuse-change emission flux. Recent estimates indicate that the error in this flux is 80 per cent of its mean value, and this error is directly proportional to the error in biomass. This suggests a threshold requirement of ~40 per cent error relative to the mean biomass.
Requirements for satellite instruments and satellite datasets
FCDRs of long wavelength radar and lidar are needed, supplemented by high-temporal C-band and moderate to high resolution optical data; the radar instruments require polarimetry and an interferometric capability; the BIOMASS mission, if selected by ESA, will meet these specifications; Lidar is highly desirable and will be carried on the proposed NASA Icesat-2 mission but not optimized for forest canopy measurements.
Calibration, validation and data archiving needs
• For validation, ground-based biomass and height measurements are needed at a range of boreal, temperate, and tropical sites;
• Systematic global measurements of radar and lidar need to be archived; models for this are provided by the ALOS datasets for all the Earth’s forest biomes acquired under the JAXA Kyoto Protocol and Carbon Initiative and ICEsat datasets of forest height (e.g. as held at Colorado State University).
Adequacy/inadequacy of current holdings
• Regional maps have been generated, but most of them have not yet been subject to peer-review;
• There are large differences between different products, particularly over tropical regions. Immediate action, partnerships and international coordination
• Fostering the systematic acquisition of observations from multiple sensors needed for biomass mapping;
• Increase in coordination between active research and implementation groups (e.g. the GOFC-GOLD Biomass Working Group) in order to (a) ensure proposed missions important for biomass measurement receive strong scientific backing, and (b) develop the combination of multiple data sources to generate biomass products;
• Increase in the amount, access to, and quality of biomass in situ data for validating biomass
monitoring from space and development of a strategy for independent assessment of existing and proposed biomass products.
Link to GCOS Implementation Plan
[IP-10 Action T32] Develop demonstration datasets of above ground biomass across all biomes.
Other applications
• Dataset valuable for forest management but only at coarse resolution;
• Consistent input for the FAO Forest Resource Assessment Updates.