The combination of earthbound expertise with satellite imagery and artificial intelligence could prove to be a game-changer in the preservation and restoration of Ireland’s peatlands. This in turn could make a significant contribution to Ireland’s carbon reduction targets.
AI2Peat – All-Ireland, AI-enhanced Peatland Monitoring Platform – uses satellite imagery, data from the National Parks and Wildlife Service (NPWS), and machine learning models to monitor peatland condition at scale. The platform generates ecological maps that reflect the condition of Ireland’s raised bogs and can serve as a decision support mechanism for peatland restoration efforts.
The project, led by Dr Corrado Grappiolo and Dr Eoghan Holohan of University College Dublin with societal impact champion Dr Shane Regan of the NPWS, has won the Future Digital Challenge prize award under the National Challenge Fund which was established under the Government’s National Recovery and Resilience Plan (NRRP), funded by the EU’s Recovery and Resilience Facility. The fund is co-ordinated and administered by Research Ireland.
The €65 million National Challenge Fund supports academic researchers to work with government, enterprise, public sector organisations and societal stakeholders to address national priorities for Ireland.
RM Block
Participation in the challenge involves the teams progressing through four phases to compete for an overall prize award of between €1 and €2 million. The AI2Peat project has successfully come through the phases and has been awarded a total of €1.3 million in funding, which will enable the team to further develop and validate the solution over the coming five years.
Peatlands represent a significant carbon sequestration and storage resource. Found on all continents, peatlands are wetland ecosystems covering about 3.8 per cent of the global surface land area with a carbon stock of up to 650 billion tonnes.
By comparison, forests cover more than 20 per cent of global surface land area and store 373 billion tonnes of carbon. In Ireland, peatlands cover about 21 per cent of surface area – but 85 per cent of these have been degraded. Degradation of peatlands releases greenhouse gases, pollutes watercourses, and damages biodiversity.
‘AI2Peat has collaboration at its core, and our achievements would have never been possible without the expertise and guidance of the National Parks and Wildlife Service and the 200+ stakeholders we have been interacting with’
— Dr Corrado Grappiolo
“Going back to basics, peatlands are made up of old plants which don’t break down because of the wet conditions,” Regan explains. “Anoxic conditions are great at preserving things. Peat is 90 per cent water; even when it’s dried for turf it’s still 60 per cent water. But when you drain bogs, they emit the stored carbon back into the atmosphere.”
Ireland is under increasing legal pressure to meet international commitments to restore and conserve those peatlands. The Nature Restoration Law entered into force last year, for example, setting targets to restore 30 per cent of damaged peatlands by 2030, 40 per cent by 2040 and 50 per cent by 2050.
Peatland restoration requires regular monitoring, at least yearly. However, the scale of that undertaking, encompassing about 11,000 raised bogs around the country, would involve huge cost as well as vast numbers of ecologists who simply are not available.
That’s where the AI2Peat project comes in. Regan points out that only a small proportion of Ireland’s bogs have been accurately mapped. But those maps form the basis of the AI2Peat project.
In simple terms, the project takes earth observation satellite images of vegetation on Ireland’s peatlands and compares them with existing maps. For example, the presence of sphagnum is an indicator of a wet or relatively healthy bog, while heather can indicate dry or degraded conditions.
The application of machine learning and AI to the comparisons with existing maps is then used to interpret what those indicators and other data from the satellite images of unmapped raised bogs are telling us about their condition. That information can be used by ecologists to prioritise areas to survey, locate areas where in-situ sensors such as water loggers should be installed, and estimate the potential for restoration.
Continuous improvement is at the core of the project. “We want to improve the predictive performance of the models,” says Grappiolo. “We want to validate and test the performance with the mapped bogs and then apply that to the other peatlands.”
Adds Regan: “We can map things from space, but we need boots on the ground for validation and verification.”
The maps produced by AI2Peat are published on the project’s custom-built knowledge and data-sharing PeatSense platform (peatsense.org). The aim is for this to become a decision support mechanism for peatland restoration actions, as the published data and knowledge can be interpreted by ecologists to further monitor and verify results.
“We are deeply honoured for having been selected to enter the prize phase of the National Challenge Fund – Future Digital Challenge,” says Grappiolo. “AI2Peat has collaboration at its core, and our achievements would have never been possible without the expertise and guidance of the National Parks and Wildlife Service and the 200+ stakeholders we have been interacting with. A huge portion of our success is due to the massive support provided by Research Ireland. Thanks to their mentoring and training activities, we mastered design thinking and theory of change skills; that allowed us to truly work towards a game-changing solution to monitoring peatland condition at national scale that is the most impactful and useful as possible.”
