The CUP4Soil Project

Land Monitoring Land Monitoring The CUP4Soil Project 1 1st CLMS General Assembly Presented by Uta Heiden (DLR) 03.06.2024 Land Monitoring •High-resolution soil property service development for National and European soil carbon reporting •DLR and ISRIC partnered for the 2-years project (2023 –2024) •Funded by FPCUP - Framework Partnership Agreement on Copernicus User Uptake •Objectives: ▪Prepare a potential Copernicus downstream service to support national and European agencies for reporting on soil health/quality. ▪Generate European-wide example data products characterising soil health/quality ▪Develop a user community that tests and validates data products ▪Close cooperation with the ESA WorldSoils project activities ▪Cooperation with other EU soil related projects and initiatives (e.g. EJP SOIL, Soil Mission, …) 2 CUP4SOIL - Project Introduction Land Monitoring •European-wide EO data products and soil maps (20 m pixel size): ▪Soil property maps (e.g. soil organic carbon, soil texture) and ▪Information about soil and vegetation dynamics including quality indicators •Documents (User requirement document, key soil product description, scientific and grey publications, …) •Finished User Survey (83 institutions, companies, NGO’s) •Meetings + Workshops: ▪07.12.2024 (online) –Virtual meeting for discussing and consolidating User Requirements ▪07.03.2024 –First soil information products are presented, user requirements will be updated (session during the ESA Symposium on EO for Soil Protection and Restoration) ▪Q4/2024 –Final project workshop to assess key user feedback, recommendations and future directions 3 CUP4SOIL –First achievements CUP4SOIL Webpage Land Monitoring 5 CUP4SOIL –Online User Survey 23 questions, sent out to people across Europe on soils and EO (Oct 2023 –Feb 2024) •54 / 83 users are missing soil related information at the Copernicus service •Spatial patterns are useful but quantitative correct values with confidence intervals are more important •Majority needs information for monitoring yearly, but if not possible less often is still useful. •Purposes are MRV, monitoring, CAP •Spatial resolution winner is 10-20 m pixel size, but coarser pixel sizes are still useful Simple properties Derived properties Land Monitoring 6 CUP4SOIL –EO and soil data products •Temporal compositing using Sentinel-2 •Digital soil mapping LUCAS Covariates: SCMaP outputs Copernicus (DEM, LC, Geology (JRC), Sentinel-1 Model: quantile random forest Outputs: Primary soil properties Uncertainty index Land Monitoring 7 CUP4SOIL –Soil composite information Soil reflectance composite (Sentinel-2, 2018 –2022, Mean, RGB –B4/B3/B2) Soil frequency (Sentinel-2, 2018 –2022) Land Monitoring 8 CUP4SOIL –Soil composite information (Champagne –France) Soil reflectance composite (Sentinel-2, 2018 –2022, Mean, RGB –B4/B3/B2) Google Satellite Soil reflectance composite Land Monitoring 9 CUP4SOIL –Soil composite information (Champagne –France) Soil organic carbon content (top soil) Land Monitoring 10 CUP4SOIL –Cross validation •DEM, land cover, etc. (Copernicus layers) •Geology/parent material (JRC) •Simple radar products from Sentinel1 1 •DEM, land cover, etc. (Copernicus layers) •Geology/parent material (JRC) •Simple radar products from Sentinel1 •SCMaP: Mean reflectance composite (MEAN) 2 •DEM, land cover, etc. (Copernicus layers) •Geology/parent material (JRC) •Simple radar products from Sentinel1 •SCMaP: Mosaic of MEAN and soil reflectance composite (SRC) 3 SOC 1 2 3 1 2 3 MEC* RMSE** pH (water) 1 2 3 1 2 3 MEC RMSE Bulk density (oven dry) 1 2 3 1 2 3 MEC RMSE *MEC - model efficiency coefficient, equivalent to R² ** RMSE - Root Mean Squared Error