Living Wales: Automatic and routine environmental monitoring using multi-source earth observation data

Increasing awareness of the adverse impacts of human-induced environmental change have prompted the need for more sustainable development and proactive planetary restoration. An essential component is to equip stakeholders with timely and reliable data that provide informed understanding of landscape change across varying spatial and temporal scales. The Earth Observation Data for Ecosystem Monitoring (EODESM), which is based on the Food and Agriculture Organisation's (FAO) Land Cover Classification System (LCCS), is an open source system allowing routine and automatic generation of land cover and change maps from Earth Observation (EO) data. It is currently being developed and implemented at national scales through the Living Wales project (https://Wales.livingearth.online) using multi-source freely available EO data, including those provided by the Sentinel-1 and Sentinel-2 sensors. Airborne LiDAR, Open Street Map, Copernicus High Resolution Layers, and National Forest Inventory data have also been integrated. These EO data are transformed into Environmental Descriptors (EDs) which are then combined in EODESM to generate land cover maps. From those maps, changes are detected in the landscape using the evidence-based change module. The system allowed generation of nationally consistent land cover maps for Wales (UK) at 10 m spatial resolution. Using the evidence-based change module, 2017-2019 multi-year forest clearcutting as well as daily changes in water extent associated with flooding were identified and described. As the system is independent of temporal and spatial scale, EODESM has the capacity to classify diverse landscape changes across multiple time frames (e.g., localised episodic events or decadal processes) and provides robust, consistent and interpretable classifications. Furthermore, additional EDs can be ingested, which provides a logical and simple approach to tailoring user requirements. EODESM shows considerable promise for directing short to long-term restoration and enhancing natural resource management in support of greater ecosystem resilience.