Objectives of the Product

ClearBNG is a scalable, AI-driven Earth observation (EO) insights solution designed to address the complex challenges of Biodiversity Net Gain (BNG) compliance and long-term monitoring. Delivered by Aspia Space, ClearBNG integrates satellite imagery with ground-truth inputs with advanced analytics to provide a trusted, automated, and regulatory-aligned biodiversity assessment tool. At its core, the solution leverages EarthPT, Aspia’s multimodal spatio-temporal foundation model for geospatial data. This foundation ensures that biodiversity insights are scientifically validated, scalable, and seamlessly integrated into existing planning and environmental workflows.

ClearBNG delivers trusted, high-resolution habitat classification and Biodiversity Net Change metrics, derived from detailed, map-based outputs tracking habitat classification, condition and change over time. It is designed to be aligned with Defra’s Biodiversity Metric 4.0, Natural England frameworks, and UKHab classifications for audit-ready reporting. It supports local planning authorities (LPAs) in assessing biodiversity uplift feasibility, developer compliance and enforcement.

ClearBNG established a replicable model for:

• Automating biodiversity net gain monitoring across LPAs, reducing costs and administrative burdens.
• Enhancing spatial planning and policy enforcement through high-resolution EO insights.
• Increasing confidence in EO-derived insights, allowing for a measured transition away from expensive, manual survey reliance.

By ensuring seamless adoption by LPAs, developers and environmental agencies, ClearBNG sets the benchmark for scalable, EO-powered biodiversity monitoring.

---

Customers and their Needs

Initially, the target customers will be local and national government organisations (e.g. county councils, LPAs, agencies), as well as enterprise-level institutes with ESG and/or stewardship responsibilities (e.g. large food producers, agri-tech companies, real estate developers) in the UK, scaling to other geographies in due course.

---

Targeted customer/users countries

UK, scaling to Europe and beyond.

---

Product description

ClearBNG is a data product that informs users of past and current habitat makeup and condition of a given area of interest. ClearBNG is derived from satellite data and exploits two state-of-the-art and proprietary GenAI technologies: ClearSky, an algorithm that predicts cloud-free multi-spectral optical imagery from Sentinel-1, and EarthPT, a geospatial/temporal foundation model for EO data. EarthPT is fine-tuned to dynamically predict habitat type aligned with the UKHab hierarchical classification system. It significantly reduces the burden of biodiversity monitoring and reporting by providing a consistent, validated, remotely-sensed approach.

ClearBNG outputs are designed with interoperability in mind in order to maximise uptake and utility. From GIS-ready data products such as high-frequency habitat zonal polygons, to summative reporting and interactive dashboards. ClearBNG quantifies, for any area of interest:

• Habitat demographics over a multi-year baseline and on-going monitoring.
• Statistical quantification of habitat changes and direction relative to baseline, identifying significant loss or gain in biodiversity.
• Indication of relative habitat ‘condition’ where appropriate.

ClearBNG enables organisations to meet compliance with BNG legislation, monitor local nature recovery strategies, and deliver actionable biodiversity insights to guide planning and policy. It offers high-quality analytics as a service and meets customer needs for scalable, validated tools for environmental monitoring.

---

Added Value

ClearBNG differs from existing offerings in its use of innovative, proprietary GenAI:

ClearSky: Aspia’s proprietary technology eliminates the challenges of cloud cover that affect 70% of optical EO imagery. By integrating Sentinel-1 Synthetic Aperture Radar (SAR) data with AI, ClearSky produces consistent, cloud-free multi-spectral optical images, ensuring uninterrupted data availability. In commercial use for over four years, ClearSky has supported more than 15 million individual image downloads in the UK. ClearSky is the bedrock technology that allows ClearBNG to deliver regular, reliable and consistent remote sensing monitoring of the ground.

EarthPT: As the world’s first GPT-based Large Observation Model (LOM), EarthPT leverages cutting-edge AI to translate complex, multimodal, and temporal geospatial data into actionable insights. Scaling to more than two billion parameters, it is already delivering answers to complex challenges. Recently, it achieved >98% performance (AUC-ROC) in dynamic tree classification compared to the National Forest Inventory, even distinguishing individual large trees. This classification capability is central to ClearBNG’s habitat classification requirements at the heart of biodiversity tracking.

For Biodiversity Net Gain (BNG) monitoring, Aspia combines AI, cloud-free EO imagery, and multi-source data to deliver trusted, scalable, and cost-effective biodiversity assessments. This innovation supports compliance with the UK’s mandatory 10% BNG legislation.

A key additional outcome for this project is its application to support the monitoring and stewardship of forestry and woodlands. Predicting tree types, density and change on a national scale, significantly reducing effort in manual inventory surveys.

---

Current Status

Project achievement to date:

• Engagement with key stakeholders to identify pain points, assemble subject matter expert ground truth/validation data across 9 key biodiversity sites across Cornwall.
• Workshop with David Langton and Prof. Duccio Rocchini to establish a credible technical approach and identification of challenges/risks, as well as appropriate ecological informatic methodology.
• Assembly of UK-wide training data samples, including UKHab-aligned Priority Habitats Inventory, National Forest Inventory and Trees Outside Woodland datasets.
• Preliminary training of fine-tuned EarthPT habitat classification model and first analysis of results.