Skip to main content
European Commission logo print header

Mire Ecosystem Transition Area Flux

Final Report Summary - METAFLUX (Mire Ecosystem Transition Area Flux)

In this report we summarise the results and conclusions of the METAflux (Mire Ecosystem Transition Area Flux) project undertaken by Dr. John Connolly during the two year Marie Curie Intra-European Fellowship mobility to the Department of Physical Geography and Ecosystem Science, Lund University under the supervision of Prof. Lars Eklundh. The main scientific aims of the project were (1) map the extent and spatial variability of zones undergoing rapid transformation due to permafrost melting at a site in Northern Sweden using LiDAR, hyperspectral data and high resolution imagery from airplane and satellite sensors, (2) integrate this data with site specific micrometeorological data to determine the relationship between transition area hyper/spectral signatures and C-Flux, and (3) to model the spatial and temporal relationships over larger areas of Scandinavia, North America and Eurasia. The aim of the project also encompassed training for the researcher, outreach and the development of a collaborative network working on permafrost in Europe and beyond.

The project website is: http://jcresearch.wix.com/metaflux

Summary
The climate of the Arctic is warming rapidly. Terrestrial Arctic ecosystems are already experiencing irreversible regime changes including change to vegetation and permafrost soils. Arctic soils and particularly frozen peatland soils contain large amount of soil organic carbon. Under current warming stable stocks of soil organic carbon may become vulnerable and, as permafrost soils thaw, release greenhouse gases. An effective framework to monitor Arctic net ecosystem carbon balance is lacking. Remote sensing from satellite and aerial imagery as well as from ground-based sensors may aid the monitoring of these remote and vulnerable ecosystems.

In this project, we collected sensor data using a ground-based hyperspectral ASD Fieldspec spectroradiometer. This instrument allowed us to collect several hundred spectral signatures across a transition gradient (from relatively dry Palsa (permafrost) peatland to very wet fen conditions) on several peatlands in northern Sweden. However, due to near continuous cloud cover during the sampling period only about 10% of the data was usable for analysis. Furthermore, there was a large natural variability among the spectral classes which made it difficult to derive distinct spectral classes for vegetation along the transition gradient. This is an issue that may be solved with more data therefore warrants further work to increase the sample sizes. The unpublished results indicate that the use of hyperspectral data may be useful to detect different plant functional types both on northern palsa peatlands and within them, across transition gradients, however, sampling has to take place under clear sky conditions and large sample sizes are essential. Lund University’s Multispectral Network sites in Abisko offer the opportunity to combine spatial sampling and data to advance the knowledge on vegetation dynamics and data sampling methodologies and could be useful in extending the hyperspectral sampling on these palsa peatlands.

Multispectral data was also acquired for the Palsa peatlands in Abisko to take advantage of the acquisition of recently declassified military aerial imagery from 1943 for the Abisko area. This data, acquired in collaboration with Dr. Andreas Persson, extends the imagery record for Palsa peatlands in this region by approximately 30 years. This gave us the opportunity to examine change in the region over a 70-year period from 1943 to 2013. High resolution satellite data were acquired from the WorldView2 satellite for 2010 and 2013. This dataset now represents one of the longest records of digital imagery for a Palsa peatland site in Europe. A change detection study was implemented. The results show that parts of the mire are relatively stable while other areas are thawing more rapidly. Details and hypothesis of why this is the case are discussed in further detail in a manuscript by Persson et al. (70 years of Permafrost Thaw in sub-Arctic Sweden, an analysis of remotely sensed imagery from 1943 and 2013).

Multispectral data was also acquired for a Boreal peatland area in Canada. Dr. Connolly developed an international collaboration with several researchers from Canada. The aim of this part of the work is to refine estimates of peatland soil organic carbon pool. An object oriented approach was used to map the spatial extent of peatland ponds using high resolution WorldView2 satellite date from 2012. The results show that peatland ponds ranging from 9 m2 to >1000 m2 can be accurately identified. The peatland pond maps have high accuracy levels and will be integrated with carbon flux and pool depth data to overcome current limits in peatland carbon modelling. Further details of this work will be published in a manuscript by Connolly et al. (Mapping small peatland ponds and their impact of carbon pool estimates).

Some further project results:
• Dr. Connolly was appointed as a Marie Curie IEF researcher at the Department of Physical Geography and Ecosystem Science at Lund University. This exchange delivered the development and reinforcement of collaborations between the researcher and colleagues in several EU countries in the field of remote sensing and soil organic carbon monitoring in sub-Arctic environments. It also enabled the transfer of knowledge and information between the project partners. Knowledge transfer to University level occurred through his involvement in a PhD. field course where the researcher delivered lectures on remote sensing, as well as demonstrations and interactive use of the SOMI hyperspectral platform (see below) and supervised a group of students in their used and analysis of hyperspectral data. He was also involved in the mentoring and supervision of two PhD students and two MSc. Students. The researcher took part in several workshops and presented METAflux results at several national and international conferences and symposiums. Knowledge transfer to the researcher occurred with his attendance on several remote sensing and GIS courses. He was also mentored in the use of Matlab and trained in the use of DGPS and an ASD FieldSpec Hyperspectral sensor and associated software. He was involved in Spring and Summer field campaigns at the Abisko Scientific Research station located in the sub-Arctic. Outreach activity included the development of a project website and social media, lectures and seminars to PhD students from across Scandinavia and to secondary school students in Lund. Several international collaborations were also developed during this mobility including with Aarhus University, McGill University, Université de Montreal and further developments with Lund University.
• The researcher developed a Single Operator Multi-Instrument platform (SOMI) for mounting the field equipment thus enabling the system to be operated by one person in the field. The SOMI enabled a hyperspectral spectroradiometer, a DGPS, a digital camera and a Spectralon to be mounted on one platform (a tripod). The Spectralon was mounted on an adjustable gimble to allow rapid calibration of spectroradiometer in the field. The SOMI system enables relatively rapid data acquisitions on very undulating and wet surfaces.
• Ground-based hyperspectral data for different plant communities were acquired at three transitional palsa mires in Abisko: Stordalen; Storflaket and at the Heliport site. The acquisition of these data in high latitude regions is difficult to plan due to cloudy weather conditions. Ideally, an operator would be based at a field site for a season to capture clear sky days. The clear sky data that was captured were processed and statistically analyzed. The results were ambiguous due to large natural variability among spectral classes. This resulted in non-distinct spectral classes along the transition gradient.

The collaborations developed through the METAflux project facilitated the development and training of the researcher as well as the development of two new projects: 1. An examination of the impact of the spatial extent of ponds on Boreal peatlands and 2. A Google Earth Engine Research Award funded project examining Net Ecosystem Carbon Balance across the Arctic. Research conducted into the Mire Ecosystem Transition Areas in Abisko, Northern Sweden, during this mobility, will also continue with the examination of the temporal change in the 70-year imagery record as well as the development of further hyperspectral imaging capability in Abisko.

In achieving most of the aims and objectives of the METAflux project, we used geoinformatics to develop new knowledge on the impact of these transition zones on mire ecosystem C-balance. This new data may aid with the prediction of the spatial extent of thawing palsa peatlands and help to overcome limits in peatland carbon modelling at regional levels.

Potential impact of project:
As the Arctic warms and is opened up for development, it is essential that the natural ecosystems that exist there are monitored for change and in the context of this project, change in permafrost peatland extent and soil organic carbon stocks, fluxes and carbon (C) cycling in the Arctic. The spatial and temporal refinement of datasets enables more accurate calculations of carbon in natural ecosystems. In the METAflux project, both hyperspectral and multispectral datasets were created and utilised. High spatial resolution datasets were used to examine the impact of climate change on a permafrost palsa peatland in Northern Sweden and to study the spatial extent of peatland pools on a Boreal peatland in Canada. Both regions contain large amounts of soil organic carbon in the form of peatlands. This peatland carbon, particularly in permafrost areas, is vulnerable to changes in the climate and to anthropogenic influences. The results of our 70-year data set (1943-2013) from Northern Sweden may be useful in deriving a new understanding of how Palsa mires degrade and what causes that degradation. This could have an impact on the modelling of C-balance in these high latitude regions. The high resolution mapping of small area Boreal peatland ponds may lead to the refinement of carbon stock estimates in northern latitudes. Hyperspectral data may be useful for monitoring changes in vegetation and permafrost in the northern high latitudes. Palsa mires are of particular interest as many are transitioning from dry to much wetter conditions, as permafrost thaws, with a corresponding changes in the carbon flux. The development of a sampling methodology that enables a single operator to acquire field based data under difficult conditions in remote areas (SOMI) is useful for monitoring vegetation change on the ground and also for creating spectral libraries for hyperspectral space missions such as EnMAP (http://www.enmap.org/). The work completed in this project has the potential to impact on our knowledge of the effect of climate warming on permafrost, peatlands and soil organic carbon in these regions.