Raised Peatland Ecohydrology Evaluation through Sentinel-1 InSAR data and Machine Learning – RaiPEAT_InSAR

The “Raised Peatland Ecohydrology Evaluation through Sentinel-1 InSAR data and Machine Learning” (RaiPEAT_InSAR) project will investigate linkages between ecohydrological parameters of raised peatlands and InSAR estimates to classify of ecohydrological dynamics of peatlands. The main objectives are as follows:

1. Ground validation of InSAR-derived surface motions of temperate raised peatlands by using co-located soil moisture, water table and in-situ displacement measurements.
2. Derivation of empirical relationships between soil moisture, groundwater levels, true surface displacement and InSAR-derived displacement for temperate raised peatlands.
3. Comparison of peatland surface motion data from the new European Ground Motion Service to both the in-situ ground measurements and local area InSAR data processing.
4. Upscaling of InSAR-derived water table fluctuation at temperate raised peatlands from site-scale to regional scales.
5. Classification of eco-hydrological dynamics of temperate raised peatlands from InSAR data via machine learning.

Raised peatlands are highly sensitive ecosystems subject to landmark European Union (EU) environmental legislation. They are also a major long-term carbon reservoir, both in Northern Europe and in many other parts of the globe. Protection and restoration of peatland is central to European policy goals for biodiversity, environment, and climate. Peatlands form by accumulation of plant and animal remains under anoxic and highly acidic groundwater conditions in areas with a high and stable groundwater table. The maintenance of the groundwater table is critical for peatlands in terms of their ecological functions, their geotechnical stability, their impact on downstream water quality and their greenhouse gas emissions. Long- and short-term water table levels are thus a key environmental variable for peatland conservation and restoration, as well as related climate change mitigation efforts.

A challenge is to monitor the progress and effectiveness of peatland conservation and restoration at large scales (regional, national, or global). Remote sensing data from ESA’s Sentinel missions potentially hold the key. Recent work on temperate peatlands in Ireland and Britain indicates that Interferometry of Synthetic Aperture Radar (InSAR) data derived from Sentinel-1 imagery may enable the mapping of ground surface motions at peatlands in space and time. Furthermore, these motions may be tied to changes in groundwater level. A problem is that there has been no systematic ground validation of the satellite-derived surface motions to date. Consequently, it is unclear how exactly InSAR data for peatlands relate to the underlying ecohydrological variables. Such validation is critical for underpinning an upscaling of peatland monitoring from local in-situ data to a regional or national scale via, for instance, the Copernicus program’s European Ground Motion Service (EGMS). Additionally, a robust methodology for peatland ecohydrological classification based on InSAR needs to be developed and tested.

The RaiPEAT_InSAR aims to address these scientific gaps by conducting a detailed analysis of Sentinel-1 derived InSAR time-series datasets over several well-studied and well-instrumented ‘flagship’ raised peatlands in Ireland and Britain. To ground validate the InSAR surface motion estimates, the project will take advantage of novel camera-based instruments (PeatCams) that have been specifically designed for measuring peatland surface motions at sub-mm precision. By co-locating the PeatCams for the first time with both piezometers and soil moisture sensors, the project will offer unprecedented insight into the relationship between InSAR-derived apparent ground motion, true ground motion, soil moisture and water table levels at temperate raised peatlands. The new data will be used to derive empirical relations between InSAR-estimated ground motion and water table fluctuation that can be used to upscale the in-situ data for regional or even global peatland monitoring. The data will also help to check the accuracy of the EGMS over these peatlands to evaluate its suitability for upscaling of peatland monitoring. Finally, it is aimed to develop and test a new machine learning approach for the classification of peatland water table dynamics from Sentinel-1 InSAR data to facilitate the upscaling of peatland monitoring and provide a basis for future implementation across Europe.