Abstract
The purpose of this report is to present a suite of methods and products for landscape scale assessment of peatland condition that can contribute to the sustainable management and restoration of peatland habitats. The outputs will also allow for the reporting of condition at the local, regional and national scales. The methods all use satellite measures of peatland surface motion (or ‘bog breathing’) to determine peatland condition and generate maps and summaries of peatland condition and change in peatland condition.
Peatland surface motion is a sensitive indicator of peatland condition and resilience (Howie and Hebda, 2018; Waddington, 2010; Loisel and Gallego-Sala, 2022). It is a mechanical response to changes in water storage that is determined by a range of peatland properties including softness/stiffness, water table depth, plant functional type, land use history and topography. It also provides a measure of the ability of a peatland surface to rise and fall with the water table and hence minimise the risk posed by drought and fire.
This method provides measures of condition based on ground motion that results from the combined effects of ecology, hydrology and mechanics. This view of condition is a different view to that provided solely by ecological or hydrological measures. It is a view that provides new information about the behaviour of the peatland that enhances our understanding of peatland function and is complementary to ecological and hydrological measures of condition. Importantly it does not replace detailed assessment of peatland ecology or hydrology condition should specific measures of these parameters be required.
Previous work in peer reviewed publications (Alshammari et al., 2018; Alshammari et al., 2020; Bradley et al., 2022; Islam et al., 2022; Marshall et al., 2022) and NatureScot Research Reports (Marshall et al., 2021; Bradley et al., 2025) has demonstrated that measures of peatland condition can be derived from surface motion over large areas at up to 20 m spatial resolution from satellite radar data using an interferometric synthetic aperture radar (InSAR) technique. These methods have been developed via funding from University of Nottingham, NERC (NE/P014100/1; NE/T010118/1; NE/T006528/1), Leverhulme Research Leadership Award (RL-2019-002), Peatland ACTION and Forestry and Land Scotland. Much of this research has been supported by RSPB, Plantlife, as well as individual landowners, Highland Rewilding (Bunloit Estate) and Welbeck Estate.
The evolution of these methods is as follows. Using the novel InSAR methods developed by fellow academic Andrew Sowter, David Large developed the initial concept of using InSAR signals for peatland monitoring. These initial concepts were explored at an international workshop held in 2016 funded by the University of Nottingham and held at the Environmental Research Institute in Thurso. Following this workshop the then PhD student Lubna Alshammari demonstrated the first evidence of a systematic relationship between the nature of the InSAR signal and peatland condition (Alshammari et al., 2018; Alshammari et al., 2020). Simultaneously the first in a series of NERC grants was secured jointly between Roxane Andersen, Stuart Marsh and David Large to develop with postdoctoral researchers, Andrew Bradley and Chris Marshall, field validation and signal analysis techniques required to enable condition classification from the InSAR signal (Bradley et al., 2022; Marshall et al., 2022). The condition classification method developed at this stage is referred to as the Key Metrics Method. Towards the end of this period of development the first work was undertaken for NatureScot illustrating the potential of the Key Metrics Method over selected sites in Scotland (Marshall et al., 2021) and then subsequently applied to the Cairnsmore National Nature Reserve (Bradley et al., 2025). Cairnsmore was chosen as its Molinia dominated degraded peatlands were very different to those of the Flow Country where the Key Metrics Method was initially developed. Although the Key Metrics Method was functional it was complex to apply and lacked a strong statistical basis. To address this David Large and Roxane Andersen secured further NERC funding via the Landscape Decisions Programme to work alongside University of Nottingham statisticians Ian Dryden, Chris Fallaize and postdoctoral researchers Emily Mitchell and Andrew Bradley to develop a new Object Oriented Data Analysis (OODA) framework (a statistical machine learning approach) to analyse the data. This method was applied to both Cairnsmore and the Flow Country (Mitchell et al., 2024) where it was shown to be statistically robust and objective. At the same time a companion method for detecting change in the trend of a time series was developed and is detailed in this report. The method is considered valuable as it provides a measure of whether peatland is responding to interventions irrespective of whether there is a substantive change in condition. Both the change detection and OODA condition mapping methods are readily applied over large areas and were finally used to undertake a large area mapping exercise (Large et al., 2025a).
Alongside the development of these condition mapping techniques an entirely different technique was developed, using the same InSAR data, to assess landslide susceptibility (Islam et al., 2022; Large et al., 2025b).
The methods have now reached a point of development where they can be automated for large area deployment, integration with other datasets and large-scale testing against a spectrum of known conditions. This report summarises the status of these methods and introduces the new additional method of change detection.
Peatland surface motion is a sensitive indicator of peatland condition and resilience (Howie and Hebda, 2018; Waddington, 2010; Loisel and Gallego-Sala, 2022). It is a mechanical response to changes in water storage that is determined by a range of peatland properties including softness/stiffness, water table depth, plant functional type, land use history and topography. It also provides a measure of the ability of a peatland surface to rise and fall with the water table and hence minimise the risk posed by drought and fire.
This method provides measures of condition based on ground motion that results from the combined effects of ecology, hydrology and mechanics. This view of condition is a different view to that provided solely by ecological or hydrological measures. It is a view that provides new information about the behaviour of the peatland that enhances our understanding of peatland function and is complementary to ecological and hydrological measures of condition. Importantly it does not replace detailed assessment of peatland ecology or hydrology condition should specific measures of these parameters be required.
Previous work in peer reviewed publications (Alshammari et al., 2018; Alshammari et al., 2020; Bradley et al., 2022; Islam et al., 2022; Marshall et al., 2022) and NatureScot Research Reports (Marshall et al., 2021; Bradley et al., 2025) has demonstrated that measures of peatland condition can be derived from surface motion over large areas at up to 20 m spatial resolution from satellite radar data using an interferometric synthetic aperture radar (InSAR) technique. These methods have been developed via funding from University of Nottingham, NERC (NE/P014100/1; NE/T010118/1; NE/T006528/1), Leverhulme Research Leadership Award (RL-2019-002), Peatland ACTION and Forestry and Land Scotland. Much of this research has been supported by RSPB, Plantlife, as well as individual landowners, Highland Rewilding (Bunloit Estate) and Welbeck Estate.
The evolution of these methods is as follows. Using the novel InSAR methods developed by fellow academic Andrew Sowter, David Large developed the initial concept of using InSAR signals for peatland monitoring. These initial concepts were explored at an international workshop held in 2016 funded by the University of Nottingham and held at the Environmental Research Institute in Thurso. Following this workshop the then PhD student Lubna Alshammari demonstrated the first evidence of a systematic relationship between the nature of the InSAR signal and peatland condition (Alshammari et al., 2018; Alshammari et al., 2020). Simultaneously the first in a series of NERC grants was secured jointly between Roxane Andersen, Stuart Marsh and David Large to develop with postdoctoral researchers, Andrew Bradley and Chris Marshall, field validation and signal analysis techniques required to enable condition classification from the InSAR signal (Bradley et al., 2022; Marshall et al., 2022). The condition classification method developed at this stage is referred to as the Key Metrics Method. Towards the end of this period of development the first work was undertaken for NatureScot illustrating the potential of the Key Metrics Method over selected sites in Scotland (Marshall et al., 2021) and then subsequently applied to the Cairnsmore National Nature Reserve (Bradley et al., 2025). Cairnsmore was chosen as its Molinia dominated degraded peatlands were very different to those of the Flow Country where the Key Metrics Method was initially developed. Although the Key Metrics Method was functional it was complex to apply and lacked a strong statistical basis. To address this David Large and Roxane Andersen secured further NERC funding via the Landscape Decisions Programme to work alongside University of Nottingham statisticians Ian Dryden, Chris Fallaize and postdoctoral researchers Emily Mitchell and Andrew Bradley to develop a new Object Oriented Data Analysis (OODA) framework (a statistical machine learning approach) to analyse the data. This method was applied to both Cairnsmore and the Flow Country (Mitchell et al., 2024) where it was shown to be statistically robust and objective. At the same time a companion method for detecting change in the trend of a time series was developed and is detailed in this report. The method is considered valuable as it provides a measure of whether peatland is responding to interventions irrespective of whether there is a substantive change in condition. Both the change detection and OODA condition mapping methods are readily applied over large areas and were finally used to undertake a large area mapping exercise (Large et al., 2025a).
Alongside the development of these condition mapping techniques an entirely different technique was developed, using the same InSAR data, to assess landslide susceptibility (Islam et al., 2022; Large et al., 2025b).
The methods have now reached a point of development where they can be automated for large area deployment, integration with other datasets and large-scale testing against a spectrum of known conditions. This report summarises the status of these methods and introduces the new additional method of change detection.
| Original language | English |
|---|---|
| Publication status | Published - 13 Jun 2025 |