Read more about Martin Karlson's open defence and download his thesis “Remote Sensing of Woodland Structure and Composition in the Sudano-Sahelian zone: Application of WorldView-2 and Landsat 8” here.

Describe in three words how you feel now when you have conducted the Thesis Nailing ceremony “spikning” at your university?

How did you practically go about to test the capacities of the two new satellite systems on the ground in Burkina Faso?

- Field data on different tree cover attributes, including tree crown size, tree species, tree canopy cover and aboveground biomass, were first collected in a woodland/agroforestry landscape. Satellite data were then obtained for this study site. The study site was selected to be representative of the main landscape type in the Sudano-Sahelian zone. The field data were used to (1) train algorithms that predict, or map, tree cover attributes by using satellite data as input, and (2) to assess the accuracy of the satellite based predictions.

- The field research period in 2012 included lots of hard work to collect the data that was required for subsequent analyses. During five weeks, me and my assistance went out in the field at dawn by motor cycle and then measured trees until the Sun went down.

The main challenge was to cope with heat during the first few days. However, it surprising to see how quickly and well the body adapted to these quite extreme conditions, with the temperature sometimes exceeding 40 ° C. The field data were collected in inventory plots (50 m × 50 m) and included individual tree measurements of position, crown size, height and stem diameter.

Is it actually possible to, through remote sensing, determine which type of trees that grows on the ground?
- Yes, at least to some extent. My research showed that the six main tree species (e.g., Shéa, Mango, Eucalyptus), which account for about 80 percent of the tree in the study site, can be identified rather accurately. The first step in the mapping process was to create a GIS-layer of the tree crown boundaries. Spectral information from satellite images was then extracted from these tree crown polygons and used as input to a classification algorithm called Random Forest.

I used a satellite system called WorldView-2 that provides data with both high spatial and spectral resolution, which is required both to delineate individual tree crowns and to resolve spectral differences between the tree species. Such differences are the result of, for example, the color, shape and structure of leafs, as well as the crown density.

What do you think is the most important contribution of your PhD Thesis for the understanding and application of remote sensing in the region?
- The main contribution from my research is that the capacity of two easily accessible satellite systems for mapping important tree cover attributes has been determined. The research has shown that WorldView-2 and Landsat 8 have high potential for improving the availability of tree cover data in the Sudano-Sahelian zone, where alternative data sources are scarce.

The interview was conducted by the Focali Project Coordinator Maria Ölund.