LivingEarth

Existing, globally available, quantitative biophysical products will be reviewed, and where gaps are identified, approaches will be developed to retrieving quantitative, biophysical, environmental descriptors from EO data acquired from various sensor types (i.e., optical, radar, thermal, lidar) routinely observing at spatial resolutions of at least 10-30 m. The methods will be transparent (i.e., compliant with FAIR principles), replicable (e.g., across sensors and platforms), and adaptable (across computational resources). The global capacity and transferability of the approach will be demonstrated by implementing the system in various national and regional big data systems (focusing primarily on Open Data Cubes) available around the world. Specific focus will be on Switzerland, Australia, Wales, and Papua New Guinea, with examples also undertaken in South America and Africa. For this project, LC descriptions on at least an annual basis and at 10-30 m spatial resolution will be generated for study sites in these countries/regions (with these including information on various climate, topography, and land management practices). Uncertainty analysis of both the newly developed quantitative LC description system and its inputs and outputs, as well as comparative assessments, will be undertaken to ensure that the new products and approach are sufficiently reliable to support national reporting obligations and environmental and territorial planning, as well as many Multilateral Environmental Agreements (MEA) and Global Environmental Goals (GEG), such as UNFCCC and associated greenhouse gas inventories, the Ramsar Convention on Wetlands, and the UN Sustainable Development Goals, to which many land organizations and stakeholders are committed. In terms of scientific research, the Dynamic Land project will address specifically the following questions: i) How to develop a full quantitative biophysical approach for land cover description from EO that does not involve discretising land into a priori classes? ii) How can this generic quantitative land cover system, being scalable across space and time dimensions, be developed and optimised to address multiple agendas/issues? iii) Can a full quantitative biophysical approach unlock the accuracy limitations reached by ML-based methods for national to global scale mapping? The resulting quantitative LC description system will provide new global capacity for scientists to investigate the impacts (components and dynamics) of human activities and natural events or processes on landscapes, with this informing (as examples) carbon cycle and climate science, nature and human well-being, and ecosystem and biodiversity dynamics and responses to change.