To support both the climate modelling and carbon cycle science communities, the ESA Climate Change Initiative (CCI) Biomass project is producing maps depicting the global distribution of woody above-ground biomass at 100 m spatial resolution as well as the uncertainty (standard deviation) of the derived estimates. In the first three years of the project, maps have been produced for the years 2010, 2017, and 2018 based on advanced versions of the retrieval algorithms developed in the frame of the predecessor ESA GlobBiomass project (Santoro et al., 2021). The project relies on multi-temporal stacks of spaceborne C- and L-band radar data acquired by the ESA C-band SAR missions ENVISAT ASAR (Wide-Swath mode) and Sentinel-1 (Interferometric Wide-Swath mode) and JAXA’s L-band SAR missions ALOS-1 PALSAR (Fine Beam Dual-Polarization mode) and ALOS-2 PALSAR-2 (Fine Beam Dual-Polarization and ScanSAR modes) for mapping above-ground biomass. In addition, the mapping of above-ground biomass considers spaceborne LiDAR (ICESAT GLAS) data to support the modeling of multi-temporal C- and L-band radar backscatter with information on forest structural differences reflected in varying allometric relationships between forest height, density, and above-ground biomass. An independent validation based on in situ plots distributed across the major forest biomes, albeit not with a systematic sampling design, as well as intercomparisons with airborne LiDAR-derived biomass maps available for various sites in South and North America, Africa, Europe, Southeast Asia, and Australia confirmed that the CCI Biomass products are of better quality than GlobBiomass but still have regional biases and a per-pixel uncertainty of about 30-40%.

The quantification of annual and decadal changes in above-ground biomass is a critical component of CCI Biomass. Based on the above-ground biomass for three different years, CCI Biomass has released change products for the years 2010-2018 and 2017-2018. The change maps are accompanied by quality flags indicating the reliability/probability of the reported changes. However, quantification of pixel-level changes beyond those flagged as probable in the released change products is currently discouraged. Intercomparisons of the three biomass maps and associated uncertainties highlighted the limitations to the estimation of biomass on a global scale, which could be categorized as signal- or processing-dependent. The signal-dependent limitations relate to the varying sensitivity of C- and L-band backscatter to biomass as well as the insufficient characterization of forest structure in the modeling locally. Processing-dependent limitations were a consequence of local or systematic imperfections in the pre-processing of the available radar data to radiometrically terrain-corrected level. Furthermore, radar data acquired by different satellite missions had to be used for the three different epochs and this posed restrictions on the inter-annual harmonization of global maps. This was largely attributed to the different acquisition modes and inconsistent multi-temporal acquisition plans, resulting in a strongly varying number of observations as well as seasonal coverage between years.

Future activities in CCI Biomass will seek to improve the inter-annual consistency of above-ground biomass estimates when reproducing the maps for the years 2010, 2017, and 2018 as well as producing maps for additional years between 2015 and 2022. Continued efforts will be made on optimizing the retrieval algorithms considering new spaceborne LiDAR data acquired by GEDI and ICESAT-2 as well as additional field data. In addition, through an ESA-JAXA cooperation on biomass estimation, JAXA will make available to the CCI Biomass consortium an exclusive dataset of ALOS-1 PALSAR and ALOS-2 PALSAR-2 imagery that will be reprocessed with quality superior to the publicly available data mosaics available so far.

References
Santoro, M., Cartus, O., Carvalhais, N. et al. (2021) The global forest above-ground biomass pool for 2010 estimated from high-resolution satellite observations. Earth System Science Data 13: 3927–3950. doi: 10.5194/essd-13-3927-2021

Accurate estimation of aboveground forest biomass stocks is required to assess the impacts of land use changes such as deforestation and subsequent regrowth on concentrations of atmospheric CO2. The Global Ecosystem Dynamics Investigation (GEDI) is a lidar mission launched by NASA to the International Space Station in 2018 and has now completed 3 years of canopy structure observations required for biomass estimation. GEDI was specifically designed to retrieve vegetation structure within a novel, theoretical sampling design that explicitly quantifies biomass and its uncertainty across a variety of spatial scales. Here we report on GEDI’s approach to biomass estimation and its resulting estimates of pan-tropical and temperate biomass that span areas from 25 m footprints to mean values and uncertainties at sub-national and country levels. We begin with an overview of the GEDI mission and provide details of its technical and mission implementation, including its lidar instrument. We then present a summary of GEDI data products, including canopy metrics, with respect to their quality and quantity. We next briefly describe GEDI’s statistical framework and illustrate the process by which GEDI was designed to support its use. We provide the most current biomass results from the mission and provide comparisons with national forest inventory data from the United States and for other countries. Finally, we address the assumptions and limitations of GEDI’s approach and consider areas where improvements may be warranted. The results reported here represent a watershed product of the first space mission longitudinally coordinated, from engineering to estimation, to generate biomass products in a transparent way with errors that are well-characterized using established probability theory. The GEDI investigation highlights the great value of an approach that explicitly address uncertainty as integral part of mission design and suggests that future space missions should carefully consider adopting a similar strategy, as appropriate.

Forests play a critical role in the global carbon cycle, storing approximately 500 Pg of aboveground biomass (Santoro et al. 2021), and forest loss contributes approximately 14% to atmospheric warming. Forest management is an important avenue for climate mitigation, both through avoiding emissions related to deforestation and degradation, and bolstering carbon sinks from afforestation and regrowth. The carbon estimates associated with forest losses and gains are highly uncertain, largely because of a lack of reliable forest carbon stock (aboveground biomass) maps. Indeed, past estimates of forest aboveground biomass stocks and fluxes vary greatly both within and between forest systems (Spawn et al. 2020). Accurately mapping forest biomass at a global-scale is a priority for a suite of new and upcoming satellite missions from NASA, ESA and JAXA, including GEDI, ICESat-2 ALOS-2, ALOS-4, NISAR and BIOMASS. The first set of new (circa 2020) biomass products have recently become available (https://earthdata.nasa.gov/maap-biomass/), many using inputs from this suite of new satellite instruments. While these products should represent increased accuracies in comparison to pre-2020 products, their relative accuracies have yet to be assessed across the range of biomes they represent. Indeed, discrepancies between products may reduce their uptake and confuse users, and harmonizing these products for policy applications (e.g. the UNFCCC’s Global Stocktake) is highly desirable. Transparent product validation and inter-comparison is critical to facilitate the improvement and uptake of these new biomass products, and other products that come online in future.

A recent international collaborative effort between biomass map producers and users organized under the Committee for Earth Observation Satellites (CEOS) Agriculture Forestry and Other Land User (AFOLU) seeks to fulfill this need. This activity, hosted on the NASA-ESA Multi-Mission Algorithm and Analysis Platform (MAAP), is an Open Science activity aimed at increasing transparency and collaboration for biomass mapping and validation. Here we present early intercomparison and validation results for 2020 biomass products. We include an intercomparison of biomass maps at a biome-by-biome scale (e.g. Moist Tropical Forests, Mangroves, Boreal Forests) to better understand discrepancies between new products. Additionally, independent reference data from airborne lidar biomass maps are used for pixel-level validation of products with samples across a subset of biomes in tropical, temperate and boreal systems. A second approach to validation using the novel plot2map tool provides insights into product performance at the policy-relevant, national and jurisdictional-scale for which harmonization and targeted estimation are planned. These analyses further inform data users on which products may be most suitable for their area of interest, and will be subsequently used to a) improve EO products and b) guide product harmonization at policy-relevant scales. This presentation includes the latest updates from the CEOS biomass harmonization team and links to a second abstract by Melo et al. focused on the importance of engagement with countries in the harmonization process.

The role of forest biomass in the European bioeconomy is of increasing importance. The assessment of the current availability and the modelling of the potential supply of forest biomass require harmonized statistics and maps indicating how much biomass is available for wood supply and its increment.

In the European context, the biomass assessment is traditionally performed by the National Forest Inventories (NFIs) using extensive field sampling. Yet, satellite and airborne Earth Observation (EO) data are increasingly being used to spatially integrate and intensify the monitoring frequency of ground-based data by mapping forest properties over large areas. However, as European NFIs employ country-specific forest and forest biomass definitions and estimation methods, and since related estimates refer to different periods and spatial scales, it is essential to harmonize the ground-based biomass statistics and EO-based maps to perform a meaningful pan-European biomass assessment. To support this goal, we present a comprehensive study that includes the following components towards a full harmonization and integration of maps and plot-based statistics.

First, the biomass statistics from NFI assessments were harmonized in terms of biomass definition and statistical estimator for expansion from plot level to regional estimates, through collaboration of 26 European NFIs. The biomass regional estimates were then further harmonized to a common reference year using the Carbon Budget Model, a forest growth model developed by the Canadian Forest Service and adapted to the specific European conditions.

Second, the resulting biomass statistics was used as a reference to assess the uncertainties of EO-based biomass maps. The map with the highest accuracy was selected and was then modified with a bias-removal correction, removing the observed systematic difference of this map with the harmonized statistics. The resulting 1-ha forest biomass map for Europe is in line with the reference statistics in terms of forest area and biomass stock.

Third, the national statistics of 22 European NFIs were also harmonized for the Forest area Available for Wood Supply (FAWS) and related biomass stocks, using the same reference definition and common criteria to assess wood availability and related restrictions. These harmonized statistics were used to map the FAWS in Europe using environmental and economic spatially-explicit restrictions. The mapped restrictions considered the following parameters and related datasets: (i) forest accessibility, excluding areas above a certain altitude and slope (derived from the Copernicus EU-DEM) and distance to roads (derived from the Open Street Map database); (ii) the legal restrictions to the forest use, excluding the protected areas with no or minimal management (IUCN reserves and national parks as mapped in the World Database on Protected Areas) and protected tree species (according to their probability of presence derived from the JRC European Atlas of Forest Tree Species); and (iii) the areas where the forest productivity (estimated using the novel kNDVI vegetation index computed from MODIS NDVI 250 m data) is deemed to be too low for sustainable timber extraction. The thresholds of each restriction were adjusted according to the country circumstances to account for the differences in the forestry sectors. The FAWS map was then applied to the harmonized forest biomass map to identify the biomass available for wood supply in Europe.

Lastly, the assessment of the available biomass stock is complemented with accurate and comparable estimates of the biomass increment, which are essential to quantify the sustainable supply of biomass from the forest sector. For this purpose, a dedicated study on forest increment is currently being performed by 11 European NFIs to provide harmonized estimates of the forest gross and net annual increment. The preliminary results of this study are presented and compared with existing satellite-based products related to forest productivity to investigate the coherence between the EO maps and plot-based statistics related to biomass growth.

Above-ground biomass (AGB) and its change (∆AGB) are essential variables for dynamic global climate models and national reporting of carbon profiles (Herold et al., 2019). While there is an increasing availability of space-based estimates of AGB in multiple periods, existing AGB maps cannot be simply subtracted to obtain ∆AGB values, as owing to uncertainties, the map differences are not depicting true changes. The issue is illustrated by the disagreements among different map-based ∆AGB estimates (Figure 1). Here we provide an assessment and inter-comparison of global forest AGB change products (2018-2010) from recent publicly released sources: the European Space Agency Climate Change Initiative (ESA-CCI) 100-m AGB maps version 3 (Santoro and Cartus 2021); the World Resource Institute 2000-2020 Carbon Flux Model (WRI-Flux) (Harris et al., 2021) that we modified to produce 2018-2010 AGB fluxes at 30-m pixel size; and the 10-km “JPL” global time series AGB (Xu et al., 2021). For each map-based product, we also produced bias-adjusted counterparts following our uncertainty assessment framework that includes bias prediction as a function of spatial covariates (Araza et al., under review). The assessments were done at 10-km aggregation level for all products and at finer spatial resolution (500 m and 1 km) without the JPL product. Several independent reference data with uncertainty estimates were used to evaluate the map-based ∆AGB consisting of re-measured National Forest Inventory (NFI) and periodic high-resolution AGB maps from airborne LiDAR (local level) and satellite images (regional level). The reference datasets were from ten countries within the four major ecological zones. The ∆AGB estimates were further compared against the Forest Resource Assessment (FRA) country data.

The preliminary results revealed that the assessment of map-based AGB losses and gains depends on the aggregation scale and the choice of reference data. At 1-km scale, map-based AGB losses and gains compare well with the LiDAR data and they are slightly underestimated when compared with NFIs and regional maps. The underestimation of AGB losses is evident to all reference data at 10-km comparisons especially when using NFI and at the country-level regardless of the FRA reporting capacity. The underestimated AGB gains are lessened at such coarser levels except when using NFI. This slight improvement of map-based AGB gains at coarser levels is also the added value of the bias adjustment. Moreover, correcting for map biases also helps reduce the map disagreements particularly in dry Africa woodlands and boreal regions in Figure 1. Outcomes of the map assessments will be the basis to use any individual product or a harmonized product for national carbon accounting in pilot countries.


*******INSERT FIGURE 1********

Figure 1. Overlap of ∆AGB (loss as < -10 Mg/ha; gain as > 10 Mg/ha; no change as < 10 to > -10 Mg/ha) among the three global AGB change products (ESA-CCI, WRI-Flux, JPL) epoch 2018-2010 at 0.1° spatial resolution and without bias adjustment. The map classes portray whether the 3 products agree/disagree or only 2 of them agree/disagree. An example where all products disagree is when a certain pixel depicts: loss (ESA-CCI), gain (JPL), and no change (WRI-Flux).



 
References:


Araza, A. et al. (2021). A comprehensive framework for assessing the accuracy and uncertainty of global 375 above-ground biomass maps. (Manuscript under review)

Harris, N. L., Gibbs, D. A., Baccini, A., Birdsey, R. A., De Bruin, S., Farina, M., ... & Tyukavina, A. (2021). Global maps of twenty-first century forest carbon fluxes. Nature Climate Change, 11(3), 234-240.

Herold, M., Carter, S., Avitabile, V., Espejo, A. B., Jonckheere, I., Lucas, R., ... & De Sy, V. (2019). The role and need for space-based forest biomass-related measurements in environmental management and policy. Surveys in Geophysics, 40(4), 757-778.

Santoro, M.; Cartus, O. (2021): ESA Biomass Climate Change Initiative (Biomass_cci): Global datasets of forest above-ground biomass for the years 2010, 2017 and 2018, v3. NERC EDS Centre for Environmental Data Analysis.

Xu, L., Saatchi, S. S., Yang, Y., Yu, Y., Pongratz, J., Bloom, A. A., ... & Schimel, D. (2021). Changes in global terrestrial live biomass over the 21st century. Science Advances, 7(27), eabe9829.

The forests and savannahs of Africa are amongst the most pristine and biodiverse ecosystems on Earth and collectively contain large carbon stocks in the form of biomass. Despite its global importance, the African continent is one of the weakest links in our understanding of the global carbon cycle due to its sparse observation network. The CarboAfrica project estimated that the biogenic carbon balance of sub-Saharan Africa is currently a net sink of between 0.16 and 1.00 Pg C yr-1 [1]. However, other studies indicated a very small sink or a near to neutral balance [2,3], but that this may be declining or already transitioning into a source [3-5]. In contrast, process-based model estimates present a far larger and unrealistic sink of 3.23 Pg C yr-1 (ranging from 1.3 - 3.9 Pg C yr-1) [1].

In this study we analysed continent wide aboveground woody biomass (AGB) dynamics using a time series of AGB maps for 2007 to 2017. We developed these maps at a spatial resolution of 100m using Global Ecosystem Dynamics Investigation (GEDI) LiDAR footprints [6], Airborne Laser Scanner (ALS)-based AGB maps, temporally cross-calibrated Synthetic Aperture Radar (SAR) ALOS PALSAR / ALOS-2 PALSAR-2 mosaics [7], and Landsat Percent Tree Cover [8]. Our approach consisted of a Random Forests Regression algorithm within a spatial k-fold cross-validation framework, followed by empirical modelling to generate AGB predictions and uncertainty outputs as in Rodríguez-Veiga et al. [9]. We validated our AGB maps with a large dataset of reference field data distributed across the continent (circa 11,000 field plots). Our results show that the AGB stocks in Africa were approximately 120.5 Pg during the study period. When estimating AGB gains and losses in Africa, we observe that the AGB inter-annual stock changes are nearly zero at the beginning of the period, but a continuous increase in the annual rate of deforestation, especially in the Congo Basin, is driving a negative trend in inter-annual AGB stock changes in recent years.



References:
1. Bombelli, A.; Henry, M.; Castaldi, S.; Adu-Bredu, S.; Arneth, A.; Grandcourt, A.d.; Grieco, E.; Kutsch, W.L.; Lehsten, V.; Rasile, A. An outlook on the Sub-Saharan Africa carbon balance. Biogeosciences 2009, 6, 2193-2205.
2. Ciais, P.; Piao, S.L.; Cadule, P.; Friedlingstein, P.; Chédin, A. Variability and recent trends in the African terrestrial carbon balance. Biogeosciences 2009, 6, 1935-1948, doi:10.5194/bg-6-1935-2009.
3. Williams, C.A.; Hanan, N.P.; Neff, J.C.; Scholes, R.J.; Berry, J.A.; Denning, A.S.; Baker, D.F. Africa and the global carbon cycle. Carbon balance and management 2007, 2, 1-13.
4. Hubau, W.; Lewis, S.L.; Phillips, O.L.; Affum-Baffoe, K.; Beeckman, H.; Cuní-Sanchez, A.; Daniels, A.K.; Ewango, C.E.; Fauset, S.; Mukinzi, J.M. Asynchronous carbon sink saturation in African and Amazonian tropical forests. Nature 2020, 579, 80-87.
5. Baccini, A.; Walker, W.; Carvalho, L.; Farina, M.; Sulla-Menashe, D.; Houghton, R.A. Tropical forests are a net carbon source based on aboveground measurements of gain and loss. Science 2017, 10.1126/science.aam5962, doi:10.1126/science.aam5962.
6. Dubayah, R.; Blair, J.B.; Goetz, S.; Fatoyinbo, L.; Hansen, M.; Healey, S.; Hofton, M.; Hurtt, G.; Kellner, J.; Luthcke, S. The Global Ecosystem Dynamics Investigation: High-resolution laser ranging of the Earth’s forests and topography. Science of remote sensing 2020, 1, 100002.
7. Shimada, M.; Ohtaki, T. Generating Large-Scale High-Quality SAR Mosaic Datasets: Application to PALSAR Data for Global Monitoring. Selected Topics in Applied Earth Observations and Remote Sensing, IEEE Journal of 2010, 3, 637-656.
8. Hansen, M.C.; Potapov, P.V.; Moore, R.; Hancher, M.; Turubanova, S.A.; Tyukavina, A.; Thau, D.; Stehman, S.V.; Goetz, S.J.; Loveland, T.R., et al. High-Resolution Global Maps of 21st-Century Forest Cover Change. Science 2013, 342, 850-853, doi:10.1126/science.1244693.
9. Rodríguez-Veiga, P.; Carreiras, J.; Smallman, T.L.; Exbrayat, J.-F.; Ndambiri, J.; Mutwiri, F.; Nyasaka, D.; Quegan, S.; Williams, M.; Balzter, H. Carbon Stocks and Fluxes in Kenyan Forests and Wooded Grasslands Derived from Earth Observation and Model-Data Fusion. Remote Sensing 2020, 12, 2380.