Forest biomass is an essential indicator for monitoring the Earth’s ecosystems and climate. It is a critical input for greenhouse gas accounting, the estimation of forest losses and degradation, renewable energy assessment, the development of climate change mitigation policies such as REDD+, climate regulations and many other applications. Wall-to-wall mapping of biomass is now possible with satellite remote-sensing technology. However, both existing and future RS methods require up-to-date, reliable and comparable in situ data for calibration and validation.
The Forest Observation System – FOS (http://forest-observation-system.net/) is an international cooperation to establish a global in-situ forest biomass database to support earth observation and to encourage investment in relevant field-based observations and science. FOS aims to link the Remote Sensing (RS) community with ecologists who measure forest biomass and estimating biodiversity in the field for a common benefit. The benefit of FOS for the RS community is the partnering of the most established teams and networks that manage permanent forest plots globally; to overcome data sharing issues and introduce a standard biomass data flow from tree level measurement to the plot level aggregation served in the most suitable form for the RS community. Ecologists benefit from the FOS with improved access to global biomass information, data standards, gap identification and potential improved funding opportunities to address the known gaps and deficiencies in the data. FOS closely collaborate with the Center for Tropical Forest Science –CTFS-ForestGEO, the ForestPlots.net (incl. RAINFOR, AfriTRON and T-FORCES), AusCover, Tropical managed Forests Observatory (TmFO) and the IIASA network. FOS is an open initiative with other networks and teams most welcome to join. The online database provides open access for both metadata (e.g. who conducted the measurements, where and which parameters) and actual data for a subset of plots where the authors have granted access. A minimum set of database values include: principal investigator and institution, accurate plot location, number of trees, forest type, wood density, canopy height and above ground biomass of trees. Plot size is 0.25 ha or large. The database is essential for validating and calibrating satellite observations and various models. Comparison of plot biomass data with available global and regional maps (incl. GlobBiomass by Santoro et al., 2018; NASA tropical by Saatchi et al., 2011; WHRC tropical by Baccini et al., 2012; WUR pan-tropical by Avitabile et al., 2015; IB-CAS global map by Hu et al., 2016) shows wide range of uncertainties associated with biomass estimation.

Accurate quantification of above ground biomass (AGB), and it’s pools and fluxes, over large heterogeneous regions is important for national carbon accounting and mitigation of rising CO2 levels. There is an increasing need for consistency in the methodology of nation-wide accounting and the monitoring of changes over time. Within the context of the United Nations framework convention on climate change (UNFCCC), 196 Nations are committed to reporting annual greenhouse gas emissions including reporting on carbon stocks and the above ground biomass (AGB) component. Tree cover is generally a large proportion of AGB, and, a significant proportion of total tree cover is often found outside forests (across Switzerland over 6%). Detailed quantification of AGB for trees, however, is usually limited to forest areas. In many cases, including Switzerland, GHGI biomass reporting for areas outside forests is based on land use type, estimated based on research studies or expert knowledge. The significance of tree biomass on non-forest land use types is becoming more recognised and the ability to model AGB separate from land use type more and more relevant with changing land uses and shifts to more complex land systems. While repeat airborne laser scanning (ALS) data has been used successfully in some case studies for monitoring AGB over time, repeat capture of ALS over large regions is rare and costly and often with considerable variation in data specifications. As a part of the NFI program, Switzerland captures wall-to-wall stereo aerial imagery to a standard specification on an ongoing cycle with repeat coverage every 6 years. Here we consider wall-to-wall canopy height models derived from stereo aerial imagery as an alternative to ALS. We take advantage of repeat acquisitions of nation-wide stereo aerial imagery to model and map AGB at two time steps and use these maps to determine AGB changes over time. This work uses observation ABG data from the Swiss NFI 3 (2004-2006) and NFI 4 (2009-2017) to build models with a logistic regression modelling approach and explanatory variables of canopy height derived from the stereo aerial imagery, elevation and remote sensing based tree type maps. The models are applied wall-to wall across Switzerland inside and outside of forest. The R2 values from cross-validated models ranged from 0.48 to 0.57 (mean 0.54) suggesting a comparable model performance to published ALS-based AGB predictions for Switzerland. We find that the relationship between the canopy height model and AGB within forest area remains stable over time (between two different National Forest Inventories) allowing for AGB change monitoring. Our results showed that the time difference between AGB measurements in the field and image capture is particularly important, with model performance dropping considerably at time differences greater than 2 years. Model performance varies with forest productivity. On the whole, we show that cost-effective repeat stereo aerial imagery can provide reliable estimates of AGB across space and time in a large and topographically heterogeneous area. We conclude that stereo-imagery is a valuable alternative to ALS for AGB monitoring.

Following the urgent need for improved mapping of global biomass stocks (Le Toan et al., 2011) to better quantify global carbon stocks and fluxes from land use change, ecosystem services and forest resources, NASA and ESA are currently developing three spaceborne missions (GEDI, NISAR and BIOMASS) to be launched between 2018 and 2021 with the aim of estimating aboveground biomass and carbon stocks in vegetation.
In 2016 NASA, ESA and DLR therefore carried out a joint airborne campaign in Gabon to advance methods for measurements of forest structure and biomass from the upcoming NISAR, BIOMASS and GEDI missions.
The NASA portion of the campaign collected coincident SAR and LiDAR data with the UAVSAR and LVIS instruments over primary rainforests, mangroves, savanna and plantation areas. The aim of the campaign was to improve the development and evaluation of forest structure and biomass retrieval algorithms in one of the regions of the world with the greatest uncertainty. During the NASA AfriSAR campaign multiple datasets were collected, covering ten existing field measurement sites.
The objectives of the ESA/NASA AfriSAR deployments were to:
1) Measure forest canopy height, canopy profiles and biomass density under a variety of forest conditions such as primary and degraded forest and a variety of forest types, including tropical rainforest, mangroves, forested freshwater wetlands and savannas
2) Acquire detailed measurements of airborne SAR data and Lidar data for cross calibration of NASA and ONERA/DLR instruments and for calibration and validation support of the BIOMASS, NISAR, and GEDI missions, and
3) Conduct technology demonstrations of joint SAR and Lidar applications for improved measurement of canopy structure and biomass.
The AfriSAR campaign also provided the opportunity to advance applications of current airborne and future spaceborne missions, notably in the field of tropical forest ecology, conservation and biodiversity. Indeed, tropical forests act as a major carbon sink, are the largest reservoir of terrestrial biodiversity and make up 80 % of the global biomass. The sites imaged as part of the NASA AfriSAR campaign were selected based on preceding ESA acquisitions, the availability of field measurements of forest structure, accessibility and existing research sites by.
As part of the campaign, the NASA teams also produced higher level data products including canopy height, Tomographic SAR profiles, Lidar -based canopy cover fraction, canopy cover profiles, and gridded aboveground biomass with associated uncertainties. The aim of generating these products is to facilitate further analysis and use of waveform lidar and multiple baseline polarimetric SAR datasets by the greater scientific community. Here we provide an overview of all the NASA acquisitions as part of the AfriSAR campaign, the higher level data products that were developed and scientific discoveries made.

Reliable wall-to-wall estimates of aboveground biomass are crucial for quantifying global forest carbon stocks and understanding their variation in space and time. Such maps require remote sensing data due to the limited and uneven distribution of forest field inventory measurements. A number of remote sensing-derived large area aboveground biomass maps have been produced in recent years. However, assessing their accuracy, and quantifying uncertainties associated with biomass estimates, remains a challenge. Validation with a large and representative dataset on above-ground biomass measurements not used during the process of estimating biomass provides a means to do so.

Within the scope of the ESA DUE GlobBiomass and Biomass CCI project, we developed a framework to validate biomass estimates from remote sensing data with inventory data collected at forest plots. We have built a reference aboveground biomass database that includes more than 150.000 entries from forest inventory and research plots across the globe provided by multiple data providers and covering all relevant eco-zones and forest types.

In the presentation we summarize the findings of the work, including:
1. An assessment of the quality of forest inventory data and the related minimum quality criteria necessary for validation purposes
2. Exploration of different options to facilitate a direct comparison between map- and inventory-based biomass estimates taking into account issues related to imprecise geolocations of inventory plots, different acquisition years, variation in plot size and different spatial support of the datasets
3. An explanation of the validation results from the Globbiomass global dataset of forest aboveground biomass, also in comparison with the accuracy reported by existing large-scale biomass maps
4. Discuss various options for accuracy reporting in the context of different biomass map user requirements

The results are synthesized and used to create an outlook of options to further proceed in biomass validation, considering:
1. The validation protocol and procedures foreseen for the validation of biomass estimates in different epochs as foreseen in the context of ESA’s Climate Change Initiative (CCI) Biomass project
2. Contributions and implications for the new CEOS Land Product Validation group protocol for calibration of biomass retrieval algorithms and validation of biomass estimates using remote sensing data
3. The use of remotely-sensed estimates of biomass and associated uncertainties to improve national-level forest carbon estimation and reporting (i.e. related to REDD+)
4. Additional options for validation based on measurements obtained with terrestrial and airborne (with unmanned aircraft systems) laser scanning

Updated, harmonized and accurate maps of forest aboveground biomass in Europe are becoming increasingly important to support the assessment of regional carbon balances and of woody biomass resources. Almost every European country has a National Forest Inventory (NFI) system from which it is possible to obtain reliable statistics on forest biomass and related carbon stocks. However, the data of NFIs are updated at non-regular time intervals according to survey planning in the countries (e.g. annually, every five years or ten years), usually do not provide the fine-scale spatial distribution of biomass, and are based on country-specific inventory designs and definitions that make it difficult their integration for a European assessments. Thus, in the recent years, maps of aboveground forest biomass have been produced using satellite data calibrated with ground observations. However, the accuracy assessment of the existing maps is limited by the lack of reference ground data consistent over the study region, which should be comparable with the maps and representative of the map cells. Clearly, the uncertainty in the reliability of maps products is a severe limit to their operational use.

Here, we use harmonized forest biomass data from 26 European countries, calculated and provided by the NFI organizations under the European National Forest Inventory Network (ENFIN), and based on their commonly elaborated definition for aboveground biomass. Six biomass maps are assessed, including the recent global biomass maps produced by the ESA GLOBBIOMASS project and the upcoming Baccini et al. (in prep.) dataset. The assessment is performed at European, national and sub-national scales using harmonized statistics derived from almost half-million ground measurement plots of NFIs, and at pixel level using a subset of 22,166 plots covering most European forest types. The NFI field plots are temporally aligned with the maps using growth rates and further screened using an innovative approach based on tree cover variability to remove the plots not representative of the map cells.

We quantify separately the effect of harmonizing national statistics with regard to the forest definition, the biomass definitions and the temporal domain, by considering their impact on the maps accuracy. The temporal harmonization is performed by updating the statistics to the reference year of the biomass maps using the Carbon Budget Model (CBM-CFS3, developed by the Canadian Forest Service and adapted to the specific European conditions), while the effect of different forest definitions is quantified by selecting only the spatial units where statistics and maps present comparable forest cover.

Our first results show that harmonized and representative reference data are essential to assess properly the accuracy of biomass maps. The harmonization of the forest, biomass and temporal definitions and domains increased the comparability of the datasets, resulting in improved maps accuracy compared to the values based on non-harmonized reference data. In terms of maps performance, our harmonized reference dataset indicates that the biomass maps tend to overestimate at low biomass and underestimate at medium – high biomass density, resulting in an overall negative bias. Such effects are due to different reasons, depending on the data and methodology used by each dataset. Ad example, maps based on average biomass values per forest type do not capture the full biomass variability existing within the forests, while maps based on optical or radar satellite data present limited sensitivity to variations in canopy height and tree diameter, and signal saturations effects. In this study, we also plan to include other relevant biomass maps covering Europe that may become available in the coming months, further assess and describe the factors affecting the maps accuracy, and provide indications for their improvements.

Forest biomass has been recognized as a Global Climate Observing System (GCOS) Essential Climate Variable (ECV), a critical input to the United Nations’ Reducing Emissions from Deforestation and Degradation-plus (REDD+) program, and an important input to Earth system models. Spatially explicit maps of forest biomass are therefore important inputs for decreasing uncertainties in the global carbon cycle, particularly from areas where insufficient ground or airborne lidar data are available. Accurate biomass products are of great importance for forest management and climate mitigation. However, due to a previous dearth of satellite data specifically designed for producing accurate estimates of forest structure few global-scale forest biomass products are currently available, and the assessment of their accuracy is challenged by a lack of appropriate reference data. To overcome this critical carbon accounting gap, several upcoming Earth Observing (EO) missions will collect satellite data sensitive to forest structure and aboveground biomass, e.g. NASA’s GEDI, NASA/ISRO’s NISAR, and ESA’s BIOMASS. We anticipate a multitude of new global forest biomass products in the coming decade both from these missions and through data fusion, but foresee challenges in inter-comparison and validation across biomass products. These challenges may hinder the effective adoption of biomass products for various policy, management and science applications.

To overcome confusion in user uptake of forthcoming biomass products, and also to inform the science community of strengths and limitations of products, the Committee for Earth Observing Satellites (CEOS)’s Land Product Validation (LPV) has developed a good practices guidebook, or protocol, for biomass product validation. This protocol has been developed with a large group of international scientists from the Earth Observation of biomass community, and includes recommendations for 1) collecting new in situ field data, 2) collecting new airborne or UAV lidar data, 3) generating airborne lidar biomass reference maps, 4) selecting the appropriate scales and scopes of validation for a given reference dataset, 5) applying statistical validation and accuracy reporting, and finally 6) propagating errors from in situ data through airborne maps to spaceborne validation. This presentation will give an overview of the biomass protocol, particularly highlighting high level recommendations for product validation, and knowledge and data gaps required to implement the protocol.