Beyond the panacea narrative: Evaluating the capabilities and limitations of remote sensing for biodiversity net gain in England

Résumé fourni par la source

Abstract In England, biodiversity net gain (BNG) is a relatively new mandatory approach applied to many development sites, intended to leave biodiversity in a measurably better state following completion. BNG assessments typically rely on field‐based ecological surveys that are labour and time‐intensive, requiring seasonal access and expert interpretation. There are concerns around the scalability and adequate resourcing of these field surveys, as BNG delivery is now subject to legally binding agreements. The rapid growth of nature markets, including off‐site habitat creation, together with the impending introduction of BNG requirements for Nationally Significant Infrastructure Projects, will further increase the need for scalable and reliable monitoring systems. Remote sensing (RS) has been posited to complement and, in some cases, streamline the survey effort required for BNG delivery by providing habitat data over large and inaccessible areas. Furthermore, machine learning approaches are increasingly incorporated into commercial products for BNG applications alongside RS data. RS can support survey planning by producing pre‐survey base maps discriminating broad habitat types (e.g. grassland, cropland, woodland). These products are valuable for early scoping; however, there is currently limited published evidence that RS can reliably classify all habitats to the required level of detail needed for BNG assessments. RS has the potential to support long‐term monitoring by providing consistent, seasonally appropriate time‐series data. However, effective use depends on standardisation and transparency in methodological approaches to ensure reliable detection of habitat change over the 30‐year monitoring period required for BNG. Using a scoring framework, we evaluated each BNG condition criterion against current RS capabilities, assigning scores from A (impossible with current data) to D (mature‐operational products). This analysis revealed RS‐derived data cannot currently replace in situ BNG condition assessments, as the available spectral and structural information is insufficient to reliably identify many of the fine‐scale indicators required to assess habitat condition. Practical implication . RS should be integrated as a complementary tool alongside field‐based ecological practices. This integration must be supported by consistent data standards, transparent workflows, and close collaboration between ecologists and RS specialists. Approaches that operate as ‘black‐boxes’ do not meet these objectives.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.