SOIL-EO: Mitigating transition risk in supply chains with below-ground biomass estimation using Machine Learning and Earth Observation data

点击翻译按钮获取中文摘要

Financial institutions are coming under ever greater regulatory pressure to ensure climate risk is central to investment decisions, governance and risk management, with recent initiatives including the Task Force on CIimate-related Financial Disclosure (TCFD), the EU's Sustainable Finance Disclosure Regulation (SFDR), as well as amendments to UCITS/AIFMD/MiFID-II and IFD/IFR. Almost 500 financial institutions across 45 countries have joined the Glasgow Financial Alliance for Net Zero (GFANZ), aligning $130trillion assets under management with net zero.The TCFD defines climate risk as '_the combination of physical risk of climate change and the transition risks of moving to net zero which may pose financial and reputational damage'_. Under the TCFD, asset managers are expected to report on governance, strategy, risk and metrics. Whilst TCFD disclosures are currently voluntary, they will become mandatory for large UK asset managers in 2022\. Other countries are expected to follow suit.Over 99% of asset managers' emissions are Scope-3 (from their investment portfolio). These are currently impossible to fully verify so managers rely on incomplete corporate reporting and inaccurate third-party providers (e.g. MSCI, Sustainalytics). These high and inaccurately quantified emissions expose them to 'transition risk' from future carbon price regulations that may create unknown liabilities.If asset managers could accurately measure portfolio company emissions, a balance sheet liability could be calculated using carbon shadow pricing, mitigating transition risk.Particular inaccuracy relates to measuring a company's impact on soil carbon from land use in supply chains (e.g. food, agriculture). Land use represents 23% of global CO2 emissions, but land also can be a powerful carbon sink (removing CO2 from the atmosphere), thus reducing company-level net emissions.Soil stores c.2500Gt, or 80% of terrestrial carbon\[8\]), and human land use processes can significantly increase or reduce sequestration. Despite this, no accurate method of frequent monitoring exists without unscalable and cost prohibitive manual sampling, meaning it is effectively excluded from current reporting.Sylvera solve this challenge by leveraging world-first multi-sensor ground truth technique to accurately infer the carbon content of soil using machine learning techniques applied to Earth Observation data. This will unlock the ability of the finance industry to evaluate the net zero status and liabilities of their investment portfolio by independently quantifying the emissions associated with land use throughout supply chains.By setting the standard for land-use emissions, SOIL-EO data will become the industry benchmark, triggering a race to the top in terms of data quality.