Skip to main content
Login | Suomeksi | På svenska | In English

Browsing by Subject "objektiperustainen luokittelu"

Sort by: Order: Results:

  • Rauhala, Jarmo (2020)
    Understanding the factors that affect the climate and the resulting chemical and physical processes will help to develop better climate models. This requires long-term measurements of carbon exchange between the earth and the atmosphere, as well as information on the spatial distribution of different bog nutrient classes and their microtopographic forms, in order to better understand the effects of climate change on different temporal and spatial scales. My master's thesis focuses on the separation of hummocks, interfaces and wet surfaces in the Simoskanaapa bog, which belongs to the Ostrobothnia-Kainuu bog zone, by means of remote sensing, image processing and guided classification. The material used was a high-resolution Optical satellite image of WorldView-2 and an altitude model interpolated to a pixel size of 2 meters, created from the laser scanning data of the National Land Survey of Finland. Object-based classification and support vector machines were used in the guided classification. Object-based classification is suitable for data containing noise, such as remote sensing data taken from bogs. The classification was successful in the ombrotrophic raised bog area of the Simoskanaava bog: the classification accuracy of the six microtopographic forms was calculated to be 84.1% (kappa 0.672). At the aapa-mire, the accuracy of the overall classification was slightly lower for the five classes (76.3%, kappa 0.650), due to the mixing of the interface wet surface levels and intermediate-wet surface levels. Object-based classification is well suited for the classification of certain bog microsites. In my study, it was possible to distinguish well the ridges and wet surfaces of the aapa mire, ridges and wet surfaces of the ombrotrophic raised bog area, and the intermediate sphagnum sp. surfaces. Further research can use more accurate laser scanning data as well as high-resolution satellite imagery to classify the bog into bog types for which emission factors are calculated using the bog's carbon balances