Characterization of the variability and controls of the carbon dioxide exchange in northern peatlands

Peatlands play an important role within the global carbon cycle as they account for ∼20-30% of the global soil carbon, yet it is uncertain whether peatlands will continue to accumulate carbon with projected changes in climate. To improve the understanding of the carbon cycle in peatlands, it is important to know the spatial and temporal variability of the carbon dioxide (CO2) exchange and other carbon components of peatlands' carbon cycle. Studying the carbon cycle in peatlands at regional and global scales is challenging because of their remote locations and their extensive areas. Studies have used ecosystem models in combination with ground measured data and/or satellite data to obtain regional to global information about ecosystem function and production, but peatlands have not been generally included in these studies. Within this context, my research was guided by two main questions: 1) how does CO2 exchange and phenology vary in and across peatlands? 2) and how can satellite data be used to obtain estimates of CO 2 exchange at larger spatial extents than the currently available field measurements? To answer these questions I used a combination of light use efficiency (LUE) models, satellite data (vegetation and phenological indices; gross primary production, GPP), and field-measured data (climate, CO 2 exchange) from four functionally different northern peatlands in Canada and Europe. Analysis of the CO2 exchange data showed higher annual accumulated GPP and net ecosystem production (NEP) at the warmer peatlands. The two warmer sites had longer growing seasons and carbon uptake periods than the two colder sites. The carbon uptake period, for example was ∼6 months at the warmer sites and ∼3 months at the colder sites. Within each of these groups (i.e. long vs. short carbon uptake period), the sites with the higher leaf area index and with circum-neutral pH levels had higher annual GPP and NEP. Variations in interannual accumulated GPP and NEP were mainly explained by variations in monthly or seasonal precipitation. Satellite-derived vegetation indices captured the seasonal variations in GPP and NEP well, and satellite-derived phenological indices showed modest relationships with annual GPP and NEP. The LUE-based GPP from the moderate resolution imaging spectroradiometer (MODIS) performed best in capturing the seasonal variation in GPP and NEP at all sites. A more in depth analysis of LUE models identified two major challenges: 1) uncertainties in satellite input data (e.g. vegetation indices) and their relationship with ground biophysical information; and 2) fine-tuning of scalars that reduce the maximum LUE parameter (i.e. epsilon) values over time: scalars should better reflect changes in soil temperature, seasonal water availability and snow presence. These findings are important for improvement of LUE models that ultimately should lead to better characterization of regional and global peatland production. The applicability of satellite data for obtaining information on peatland function and production was showed in this study; yet, the results from the final analysis showed that it is important to examine if the available satellite data (with its predefined spatial and temporal resolutions) can provide the necessary information about the vegetation structure and function in peatlands.