Assessing the Relation Between Peatland Vegetation Patterns and Climate Using Remote Sensing

Organised by Laboratory of Geo-information Science and Remote Sensing

Fri 29 June 2018 09:00 to 09:30

Venue Gaia, gebouwnummer 101
Room 2

By Quanxing Wan (China)

Vegetation in boreal peatlands have long evolution periods, so it is hard to monitor considerable development of peatland vegetation patterns over a limited number of years. The objective of this research was to assess the relation between peatland vegetation change and climate development. We fell back on the nationwide climatic variability of Sweden. Temperature and precipitation gradients were referred to as the long-term climate evolution. Sweden was divided into 15 climatic sub-areas according to climatic difference. 39 sample peatlands, distributed nationwide, were de determined by stratified sampling. The dynamics of hummock-hollow microform was used as the indicator of patterning change in peatland vegetation. Apart from hummock and hollow, forest, water and mud-bottom were other used classes for land-cover classification in this study. Owing to the large number of peatlands, 7 signature files from representative peatlands were applied to the rest of peatlands. The robustness of the maximum likelihood classification (MLC) for most peatland locations was acceptable. MLC inputs were  RGB, IR images and digital elevation model (DEM). 1-meter buffer around sample points were generated for extracting signature profile of classes. The overall classification accuracy of representative peatlands achieved 84.35%, validated by random sample points in their own locations. According to both the visual assessment and pattern analysis (with metrics of Major, Minor and Radius of Gyration), patch size of hummocks was showing a decreasing trend from cold region to warm region. Striping patterning of hummocks is clearer in colder areas. The metrics of Anisotropy and Minor had the highest correlations with temperature (r2 = 0.18 and 0.12, respectively). In relation to precipitation variation, average patch area of hummocks is smaller in dryer areas. The metrics of Anisotropy and Percentage of Landscape had the strongest correlations with precipitation (r2 = 0.14 and 0.10, respectively). Correlation between pattern metrics (of Major, Minor, Anisotropy, Percentage of Landscape and Radius of Gyration) has gone through transition to temperature. No obvious transition occurred in correlation to precipitation. It was not evident enough to draw the conclusion that alternative stable state theory exactly applies to peatland pattern change over climate development. The influence of precipitation on peatland patterning needs to be taken into special consideration in follow-up studies.

Keywords: Boreal peatland; hummock; hollow; pattern metrics; Maximum Likelihood Classification