Living England, led by Natural England, is a multi-year project delivering a satellite-derived national habitat map in support of the Environmental Land Management (ELM) system and the Natural Capital and Ecosystem Assessment (NCEA). The Living England satellite derived habitat map uses an object-based image classification technique known as Object-based Image Analysis (OBIA) to predict the most likely broad habitat class for parcels of land known as objects. The national habitat map is published every two years and therefore has potential to assess change in extent of terrestrial habitats.
Accurately detecting change between classified habitat maps is not as straightforward as overlaying iterations. Real change needs to be differentiated from those arising due to errors or variations in methodology. This requires a clear understanding of the methodology, sources of error, and the development of accuracy assessments.
The purpose of this review is to highlight key issues in existing object-based change detection (OBCD) and identify potential solutions and recommendations for performing OBCD to the Living England team. The review concludes by providing a set of recommendations including to maintain the spatial framework from one of the input maps to analyse imagery from both time points to avoid production of sliver polygons and to further investigate approaches to detect changes along habitat boundaries when maintaining the spatial framework. In addition, a reduction to the number of transition classes being detected is suggested to reduce the amount of reference data needed to assess the accuracy of the change detection product.