Regional scale identification of peatland environments using European Ground Motion Service data

Peatlands are the most effective terrestrial carbon sinks and play a key role in hydrological regulation and ecosystem services. Their spatial extent, condition and level of degradation are not fully mapped, particularly at a regional scale. This study used Interferometric Synthetic Aperture Radar (InSAR) data extracted from the European Ground Motion Service (EGMS) to characterise the typical vertical surface displacement behaviour of peatland environments. The main goal is to identify peculiar ground displacement signatures to peatlands for refining existing inventories and detecting of previously unclassified peat areas. A two-step unsupervised methodology combining Principal Component Analysis (PCA) and K-means clustering was applied across Great Britain (GB). First, vertical displacement data from EGMS, covering the period from January 2019 to December 2023, were filtered using national land cover datasets in order to remove Measurement Points (MPs) located in urban areas, roads and infrastructure. This filtering allowed the analysis to focus on natural environments where peatlands are expected to occur. PCA was applied to the InSAR time series to reduce its dimensionality and extract the dominant modes of variability in vertical displacement. The resulting principal components were clustered using the K-means algorithm to identify distinct classes of temporal deformation behaviour. Well-documented peatland sites, such as Hatfield Moors, were used as reference areas to interpret and refine the clustering results. Two PCA clusters were identified as representing the typical deformation behaviour of peatlands in GB. This behaviour is characterised by distinct seasonal oscillations related to “breathing” processes in the peat and long-term subsidence trends associated with peat compaction and degradation. The reference behaviour, together with seasonal indicators, was used to screen the full EGMS dataset and identify MPs with similar dynamics. A spatial clustering analysis was then applied to group MPs with high spatial density while excluding isolated or scattered points that are less likely to represent coherent peatland areas. The resulting clusters were then compared with the Copernicus CORINE Land Cover (CLC) 2018 to evaluate their alignment with recognised peatlands and to detect areas potentially affected by peat soils that are not currently classified as peatlands. Cross-correlation analyses were performed between vertical displacement time series and climatic variables, including precipitation, temperature, and a moisture index, in order to validate the identified displacement patterns and investigate their driving mechanisms. These analyses helped to identify the dominant atmospheric controls on peatland seasonality and support the discrimination between different peat types and hydrological conditions. The methodology was validated through three case studies: (i) Hatfield Moors, a well-studied peatbog with ongoing restoration efforts; (ii) New Forest, an extensive peatland complex in southern GB; and (iii) an area not previously classified as peatland but showing comparable behaviour. The results reveal widespread negative vertical displacement across most peatland areas between 2019 and 2023, showing prevailing subsidence and limited rewetting. The study demonstrates the potential of using InSAR data for large-scale peatland monitoring and identification, supporting improved peatland management and restoration strategies also at regional scale.