Martin Karlson

Satellites for Earth Observation has collected data continuously since the early 70s, and progress regarding the quality and availability has since been massive. The main advantages of satellite systems is that; 

1. large areas can be observed frequently
2. the cost of data is relatively low
3. long, global time series are available. 

Satellite systems helps us therefore to understand different environmental processes that are impossible to observe at ground level. My research is about how these data can be used to improve the knowledge of our environment, with a special focus on the use of Africa.

Currently I am part of a project aimed at improving the knowledge about the role of trees in the West African agricultural landscape, where crops are usually grown in fields with a relatively dense stands of trees consisting of useful tree species. These so-called agroforestry systems or parklands are exposed to intense pressures from climate change, population growth and intensified land use that threaten their long-term sustainability, and thus the population's livelihood. The project conducts research into the trees' impact on

1.  soil conditions,
2. the availability of groundwater
3. the sequestration and storage of carbon dioxide. 

Improved knowledge of these processes is essential for the design and implementation of sustainable land use policies.

In my research, I use satellite imagery and GIS to create maps of tree cover structure and composition, which are important tools for monitoring and modeling of landscape processes. The advantage of these maps compared with field-based measurements is that much larger areas can be inventoried at a significantly lower cost. High resolution images of the same type as shown in Google Earth is used to map the individual trees in relatively small areas. Satellite images with lower detail, e.g., from Landsat, is used to map the canopy cover and biomass at the landscape level. A large part of my work involves developing and evaluating methods that automate the analysis of satellite images. Object-based image analysis, machine learning algorithms and traditional statistics are examples of methods that are important to my research.

I am also involved in environmental science education at various levels. This fall, I and my colleagues for the first time organize a course that is specialized to the use of GIS and remote sensing in environmental science.