Skip to main content
European Commission logo
English English
CORDIS - EU research results
CORDIS
CORDIS Web 30th anniversary CORDIS Web 30th anniversary

A glimpse into the Arctic future: equipping a unique natural experiment for next-generation ecosystem research

Periodic Reporting for period 2 - FutureArctic (A glimpse into the Arctic future: equipping a unique natural experiment for next-generationecosystem research)

Reporting period: 2021-06-01 to 2023-12-31

How much carbon will escape from the (sub)Arctic in future climate?
How do the multitude of ecosystem processes, driven by plant growth, microbial activities and soil characteristics, interact to determine (sub)Arctic soil carbon storage capacity?
These are the central research questions of FutureArctic.

They are addressed at a unique location. The ForHot (www.forhot.is) site in Iceland offers a geothermally controlled soil temperature warming gradient, where Subarctic ecosystem processes are affected by temperature increases as expected through climate change. Given the strong urgency of tackling the climate challenge and the particularly important role herein of (sub)Arctic ecosystems, a rapid assessment of the ecosystem and ambient processes at the ForHot site will provide potentially crucial insight in future carbon cycling.

If this knowledge is to be applied in the climate change challenge, significant advances should be made on short time scales, including an unprecedented search for unknown interactions within the ecosystem. FutureArctic will achieve this challenge by adopting the fast advances made in the field of machine learning and artificial intelligence (AI), unmanned aerial vehicles (UAV) and (remote) sensor technology into environmental research at the ecosystem scale, into a new concept of an ‘ecosystem-of-things’.

FutureArctic embeds this research challenge in an inter-sectoral training initiative, aiming to form ‘ecosystem-of-things’ scientists and engineers, in order to:

- Pave the way for generalized permanently connected data acquisition systems for key environmental variables. Technological advances in scientific instrumentation in regard to both sensor capabilities and connectedness are necessary to achieve permanent analysis and data integration. The interaction with industry partners to achieve these advances is a key to success (WP2).

- Initiate a new machine-learning approach to analyse large environmental data-streams. ForHot, with its unique and large subarctic warming gradients, becomes a pioneer project for the ‘ecosystem-of-things’. Machine learning analysis procedures will be implemented (WP3) to complement hypothesis-based environmental research (WP1). Will big data and machine learning algorithms inspire new ways of thinking into environmental science and ecology, complementing traditional hypothesis-based-research that does not make full use of valuable information contained in data?

- Create an ethical / philosophical framework for the implementation of machine-learning/artificial intelligence approaches into environmental science. What are the potential implications for traditional science, what are the caveats and pitfalls?
All objectives set out in the original workplan are achieved. The second and third networkwide training events took place in 2021 and 2022.
All ESR’s in WP1 have completed their fieldwork. Data have been put together and analyzed in discussion with other ESRs. Some data are published already and some manuscripts are in preparation, about to get published. All 4 ESRs from WP2 have developed their prototypes and installed and tested them in the field. In WP3 the ESRs 14 have been working on implementing Machine learning algorithms using available datasets in Europe, for example, enhancing predictive capabilities for European forest and grassland ecosystems, or modelling the impact of (increasing) soil temperature vegetation phenology using state-of-the-art machine learning models. the “industry standard” methodology of using linear models was extended to also model non-linear interactions, using artificial neural networks.
The results were presented on the EGU conference in Vienna in 2022. The trainings organised by the consortium were open to other participants. And a stakeholder meeting was organised at the end of the project. There were several press releases about the project and a lot of communication on social media.
FutureArctic aims for a strong breakthrough in the assessment of climate change on ecosystems, where currently even plant, soil and atmosphere science is mostly uncoupled at the ecosystem scale. We propose an innovative approach where latest achievements in machine learning-based statistics, sensor technology and efficient digital data management are combined, to assess the ecosystem feedbacks at spatial and temporal scales.
Each ESR trained in FutureArctic will be ideally placed to make a significant contribution to 21st century earth sciences. FutureArctic will create the foundation in Europe for a new “Ecosystem Science 2.0”, where optimal balance is achieved between field research, permanent ecosystem sensing, novel sensing technique development and machine learning assisted data analysis. Main breakthroughs:

• Insights in interactions between soil microbial community, vegetation and climate change drivers
• An ecosystem-health based assessment of subarctic ecosystem functioning in a warming world
• Prediction of future alterations in carbon budget of subarctic ecosystems
• Development of novel ready-to-market techniques in ecosystem science
• The introduction of machine learning in ecosystem model development
• The creation of a science-society-ethics framework for machine-assisted environmental science and modelling


FutureArctic will achieve an enduring impact not only through its lasting effect on earth system model development and improved datasets of detailed (sub)Arctic ecosystem functioning, but also through the strong involvement of inter-sectoral partners. This is achieved through the development of ready-to-market solutions for equipping new research sites for an ecosystem-of-things setup. Industry beneficiaries VSI, SVARMI and DMR all worked on the development of sensors, while IMEC developed ready-to-use algorithms for detecting ecosystem process interactions. This paves the way for self-learning model algorithms that automatically adapt ecosystem models to the newest insights from most recent datasets. The deep inter-sectoral cooperation in FutureArctic opened new horizons for the environmental modelling community. ESRs in FutureArctic have developed unique skill sets and job profiles that are currently rare or even unavailable to environmental science, yet are key to a coordinated scientific and societal responses to some of the most pressing ecological challenges, including climate change.
Project Workflow