Conference Agenda

.

.

.

.

.

.

.

Moderator: Spencer Plumb, Senior Manager, Forest Carbon Innovations(Verra)

.

.

In the context of the European Green Deal and the Zero pollution Action Plan for 2050, satellite information is recognized as a crucial tool to measure, monitor and verify carbon storage.

Above-ground tree biomass refers to the weight of that portion of the tree found above the ground surface, when oven-dried until a constant weight is reached, has been traditionally quantifying using land cover stratification combined with ground sampling. The need of large and continuous spatial coverage makes satellite information as a crucial tool to help Governments and enterprises need to reduce the uncertainty of carbon emissions and sequestration, from both to control and to compensate their carbon emissions.

To measure, monitor and verify carbon storage makes necessary an accurate calculation of the amount of biomass in terrestrial ecosystems and carbon sequestration estimation based on earth observation optical sensor and in-situ field measurements have usually used moderate resolution information images (Blackard J.A. et al, 2008, Muukkonen P. et al, 2006).

Detecting the need to enhance the spatial resolution and the accuracy of the biomass estimation GEOSAT has developed a more precise model in an Iberian Peninsula area, taking advance of some variables used in a biomass estimation procedure (Dengsheng Lu., et al, 2014) and of Machine Learning (ML) techniques, putting in value de Iberian Peninsula Coverage. Being in this case, the combination of very high resolution (VHR) of 75 cm Ground Sample Distance (GSD) multispectral GEOSAT-2 images spectral bands, that would be enhanced till 40 cm applying Super Resolution (SR) techniques, both orthorectified and atmospherically corrected, vegetation indexes together with forest maps, National Forest Inventory (NFI) data and allometric model equations (Ruiz-Peinado, R. et al, 2011; Ruiz-Peinado, R. et al, 2012), which allows GEOSAT to obtain a sub-metric biomass Map and so an accurate Carbon storage measure.

The ESA Forest Carbon Monitoring project (https://www.forestcarbonplatform.org/) developed EO based user-centric approaches to support forestry stakeholders to meet the requirements of compliance and voluntary carbon markets. A prototype of a cloud processing platform which offers a selection of biomass and carbon monitoring approaches for large and small areas was developed.

Different approaches for EO-based forest structural variables and biomass monitoring were evaluated. Key findings include the benefits of (1) optical and radar data fusion, (2) multivariable methods (such as kNN) in forest structure variable mapping and (3) deep learning approaches. Cost of the datasets and robustness of the methods were also considered important aspects for operational monitoring platform.

Eight demonstrations were conducted, seven in Europe and one in Peru. The demonstrations used primarily Sentinel 1 and 2 data with varying physical and empirical methods and process-based ecosystem modelling. The output products included (1) European wide forest biomass maps for 2020 and 2021, (2) European wide biomass clearance map 2020-2021, (3) biomass and growth maps of Finland for 2017 and 2019 and (4) numerous local to provincial level maps of forest structure variables (diameter, basal area, height, conifer proportion etc), biomass and clearance in demonstration areas in Austria, Ireland, Romania, Spain and Peru.

The usability of the products was evaluated in cooperation with eight user partners. Overall, the project highlighted the great potential of EO based approaches. Particularly public sector users found the products potentially useful for large area forest biomass and carbon monitoring with some improvements in the accuracy and its reporting. Private sector users would have additionally needed higher accuracy in pixel to stand level to support forest management decisions. New deep learning-based methods are in development to better meet also the private sector demands.

ESA WORLDSOILS project is developing a pre-operational Soil Monitoring System to provide yearly estimations of SOC at regional and continental scale, exploiting space-based EO data, leveraging large soil data archives and modelling techniques to improve the spatial resolution and accuracy of SOC maps. The system characteristics are:

• Annual delivery of two products: prediction map and associated uncertainty metrics.

• Spatial resolution of 100m x 100m at continental scale and 50m x 50m at regional scale over test site regions of Central Macedonia, Wallonia (Belgium), Czech Republic and Piracicaba (Brazil).

• A modular implementation, capable of future add-ons, of required new soil indices.

• Use of large time-series of a minimum of three years.

• Validation over the abovementioned regions.

The baseline definition of the system begun in 2020, gathering end-user requirements and carrying out feasibility analyses of intermediate SOC indexes (bare/vegetated soils, forest/grassland) from satellite reflectance composites of three to five year-series of Sentinel-2 data. Intermediate SOC indexes were selected to cope with land cover diversity and aligned with bare soils and permanently vegetated areas. Thereafter, the design of the cloud-based monitoring system was implemented, verified and tested. WORLDSOILS is currently in the production and validation phase, yielding early outputs, some of which are presented below:

• The verification of disturbing factors (e.g., haze, soil moisture, crop residues) of the SOC prediction accuracy within the satellite multitemporal composites for selecting the adequate pixels reflecting the calibration samples to run the classification algorithms. The use of regional thresholds for bare soil compositing showed an improved performance.

• The spectral harmonisation of soil spectral libraries (SSLs) across different continents along with the models. A conversion factor between laboratory SSLs to field SSLs has been found.

• Testing of hyperspectral missions (e.g., CHIME resampled data) has improved SOC prediction, explaining disturbances more accurately, which can be implemented in the prediction system as soon as new mission capabilities come online enabling more robust SOC estimations from spectral data.

• The threefold validation plan, i.e.: EO input data, internal and external validation and demonstration of final product to the National Reporting Centers on soil (NRCs).

All nations of the Paris Agreement must report on their annual anthropogenic GHG emissions and removals in their national inventory reports (NIR) to the United Nations Framework Convention on Climate Change (UNFCCC). Existing and future satellite missions provide the opportunity to fulfil the need for independent GHG Monitoring, Reporting and Verification (MRV) capacities.

GeoVille, one of the main Earth Observation (EO) service providers of the Copernicus Land Monitoring Service (CLMS), has developed an expertise in the field of climate change mitigation by implementing new products and services such as land use land use change and forestry (LULUCF), Above-Ground-Biomass (AGB), forest disturbance mapping, and CO2 emissions from biomass changes and wildfires. GeoVille led or is leading several national and international projects such as SMARTCO2, LULUCF Türkiye, GHG-KIT, BREATHE, CLC+ LULUCF Instances and contributing to AI4TREES and GDA Climate Resilience.

The project SMARTCO2 aimed at evaluating the requirements, gaps, and challenges in GHG reporting for public and private sectors. This study also revealed a high potential for EO applications related to the establishment of a controlling and verification service for the forest carbon offset market in Europe. Within the framework of a EUROPAID project, GeoVille achieved the first EO-based LULUCF maps for the 1990-2015 period in Türkiye.

The Austrian Research Promotion Agency and The Austrian Federal Ministry of Climate Action has awarded the flagship “GHG-KIT: Keep it traceable” project to a consortium under the lead of GeoVille with research and industry partners GeoSphere, University of Vienna, Cloudflight, Sistema, EODC and Technical University of Vienna. Our advisory board is an international expert team on satellite data, e.g., ESA Sentinel 5P-Tropomi and CO2M missions, inverse modelling (NILU), LULUCF (FERS), and exploitation strategy via digital marketplace (EOX). GHG-KIT is also supported by GHG-SAT. The overarching aim of GHG-KIT is to develop a Proof-of-Concept for an EO-based integrated GHG overall reporting system, including two prototypes: (i) LULUCF Inventory Reporter (bottom-up approach), and (ii) Verification Element (top-down approach), in support to the Environment Agency Austria (UBA). The resulting blueprint for system and service implementation will be a major milestone towards a new national GHG monitoring expertise with high potential for European export.

The EUREKA/FFG project BREATHE – “EO-based enhancement and verification of LULUCF Inventories for Forest & Biomass”, in partnership with Globetech, Türkiye, UBA, Sistema and University of Vienna aims at developing a Proof-of-Concept and prototyping the 1st operational EO MRV system in Europe, with a focus on biomass changes in Austria and wildfires in Türkiye. The benefits and applicability of BREATHE will apply not only to the public GHG reporting sector, but also to the forest carbon trading, corporate sustainability reporting and to the voluntary carbon market.

Finally, GeoVille is currently executing the project CLC+ Instances for the European Environment Agency (EEA) with a dedicated task to set up LULUCF Instances. The latter shall support the EEA to establish a MRV system mainly based on European CLMS products and CLC+ Core, the data container that allows ingestion, referencing and applied extraction of different spatial LC/LU datasets.

.