WP1 Peat carbon stock estimation

Tasks

We will study various peatland types across southern Norway at four study areas: three sites with recent or ongoing development (Okla wind park, Turufjell cabins, Ørskogfjellet E39 road project) and one additional site (Skrimfjella, Viken County). At each site we will estimate carbon stocks in peat soil using a peat volume approach (area * depth * carbon density). We will map peatland area, measure peatland depth by manual and geophysical methods (probing, ground-penetrating radar), and determine carbon density from soil samples. Intensive sampling in the field will allow us to estimate carbon stocks precisely. Where intensive sampling is not feasible, we will collect data from a stratified random sample and extrapolate peat depth and density across the site.

At two of the sites (Ørskogfjellet and Skrimfjella) we will build statistical models that predict peat occurrence (area) and depth at landscape scales. These models will consider topographic, radiometric, and optical predictors, and use field measurements for model training and spatially independent validation. We will examine the predictive accuracy of geostatistical and non-spatial statistical models as well as machine-learning algorithms. Occurrence and depth models will be compared to a national peatland database to assess the potential for improvement in Norway’s peatland inventory.

The WP involves academic partners and the Norwegian Public Roads Administration (SVV).

Research questions

• How much carbon per unit area do various peatland types store, and how uncertain are these estimates?
• How well can remotely sensed (especially radiometric) data predict landscape-scale peat area and depth?
• Which dimension (area, depth, or carbon density) is the largest source of uncertainty in peatland carbon stocks at local and landscape scales?