Satellite image and computer vision to assess biodiversity value-at-risk

Challenge to overcome

The impact of economic activity on the environment, also called “Nature Risks”, is becoming a serious concern that can have far-reaching and lasting consequences. These consequences include threats to global health, food and water security, aggravation of poverty and migration, and even geopolitical destabilisation.

However, sustainable finance faces a core obstacle: the lack of reliable, science-based data on the environmental impacts of corporate activities.

This undermines the credibility of sustainability metrics and restricts the financial sector’s capacity to drive significant, systemic change. ZHAW is working to address this issue by bridging the gap between measurements of environmental impacts and the resulting financial costs – a key assumption being that financial cost is the most effective actionable quantity in the financial practices of corporate management.

AI application / use case

ZHAW, in collaboration with the University of Zurich, is leading the BRIDGE Discovery project, Spatial Sustainable Finance, with the purpose to develop a scientifically robust approach for assessing the contribution and exposure to environmental degradation of companies, in particular biodiversity loss and water risk. The improvement and increased availability of Earth observation data and complementary geodata offers significant potential to assess biodiversity and water systems changes on an objective and scientific basis.

Geoprocessing workflows were developed and adapted to analyze and model the spatial footprints of companies’ production sites, relating Earth observation data to geolocations of company assets and developing methods to assess and to monitor company-caused risks, like biodiversity loss and water degradation, across sectors of activity.

One of the core innovations of this initiative is to translate the measured environmental impact locally into a monetary value, for use in company valuation, financial portfolio construction, and credit risk analysis (bottom-up approach). This is translated in the concept of “BioVar», or “Biodiversity Valueat-Risk” as a critical bridge between environmental data and financial risk management:

• Step 1: Biodiversity loss and water impact are measured at site-level by using direct footprint data obtained from satellite imagery. For each production site (e.g., mine, farm, factory, and the like), a normalized biodiversity loss metric is calculated, and then scaled by a metric relevant for the business activity of the company at that site. This provides a measure for the absolute change or loss in biodiversity.
• Step 2: The absolute biodiversity loss is then translated into financial loss by a cost function that has been empirically calibrated using real-world financial data, such as documented mine rehabilitation expenses or environmental liabilities reported in company filings.

This cost function considers many attributes, including, e.g., the type of industrial activity, the legal and regulatory framework of the country in which the site operates, and the size or market capitalisation of the company. For
example, a mine in a highly regulated jurisdiction would be assigned a different financial impact of biodiversity degradation than a similar mine operating in a less regulated setting. This ensures that the monetary valuation is grounded in real liabilities rather than theoretical assumptions.

• Step 3: Once the site-level financial losses are known for all sites of a company, they can be aggregated and integrated into the classical risk framework (e.g., as part of operational risk), balance sheet provisions (future liabilities), or discounted cash flows (for equity valuation).

As a first industry sector of application, ZHAW is collaborating with mining engineers from various European countries within the COST Action REMINDNET (a pan-European research network focused on sustainable mine closure and legacy management) to apply this scheme to mining locations.

Link to demonstrator website: www.biovar.ch to test the environmental indicators (see screenshot below)

Use case key beneficiaries

☐ Relationship Managers

☒ Portfolio Managers

☒ Research teams, macroeconomists

☒ Control functions

☐ Support functions (HR, CFO, …)

Benefits for the financial services sector

The initiative provides site-specific estimates of financial loss driven by environmental harm (Nature Risks) at companies’ production sites. Due to the bottom-up approach selected, these can easily be aggregated across a company’s portfolio of operations, with BioVaR laying the foundation for integration into the portfolio manager’s risk management framework. For portfolio managers, these risk assessments provide relevant insights for initial or ongoing investment due diligence.

Supporting technology

The project leverages large data and computer vision applied to satellite imagery, including:

• Meta’s open-source model SegmentAnything, which was retrained to work on satellite imagery (“transfer learning”);
• MineNetCD, a comprehensive benchmark designed for global mining change detection using remote sensing imagery;
• Preexisting datasets of mining areas such as Maus et al. and the GESTIM database, Gouvernement du Québec (2025);
• Current and historical satellite imagery from Copernicus Sentinel-2, Landsat 5-8 and others.