Vegetationsinventering med rumsligt högupplösande satellitdata : en studie av QuickBirddata för kartläggning av gräsmark och konnektivitet i landskapet

Detta är en Magister-uppsats från Lunds universitet/Institutionen för naturgeografi och ekosystemvetenskap

Sammanfattning: Semi-natural grasslands are one of the habitats with the highest biodiversity in northern Europe. The high biodiversity is a result of the land being short in nutrient and being continuously disturbed, an effect of mowing and clearing of bushes and trees. Because the development of agriculture into more effective land use, many of these sites and their associated species are endangered. Besides degradation of seminatural grasslands, fragmentation is also believed to constitute a threat to many of the species concerned. An extensive project of inventory and supervision of semi-natural grassland is taking place in Sweden. Remote sensing is a practical tool when describing and analysing the landscape and can improve the effectiveness of inventory work. The access to spatial high-resolution satellite data makes it also possible to capture finer features that are common in cultural landscapes. In this study spatial high-resolution satellite data, from QuickBird, is used to perform a land use classification, based on spectral signature (Maximum Likelihood Classification), of a cultural landscape on the island of Öland. The possibility to capture grassland and possible dispersal corridors, such as road verges and avenues, was then evaluated from the result of the classification. The detection of less luxuriant grassland succeeded well. Remaining grassland was on the other hand mistaken for arable fields to a high extent. The recognition of avenues was connected to width and spectral separability. Wide avenues were more likely to be detected than thin ones. Avenue detection was higher where the surroundings had a similar spectral separability as the avenues. The possibility to detect road verges could be correlated with neither width nor spectral separability of the surroundings. MLclassification is not suitable for detecting linear elements. To improve the results classification should be based on texture. To be able to preserve and restore semi-natural grasslands it is important to consider the impact of fragmentation. Dispersal possibilities for organisms and genes are considered important to secure the survival of many species. Dispersal corridors are suggested to have a positive effect for many species, including species associated with traditionally managed grasslands. In this study some distance measurements of the isolation of semi-natural grasslands is discussed. When incorporating the road verges (suggested to function as possible dispersal corridors) in the distance calculations, the connectivity of the semi-natural grasslands seems to increase.

  HÄR KAN DU HÄMTA UPPSATSEN I FULLTEXT. (följ länken till nästa sida)