Application of the object based image analysis of VHR satellite images in land-use classification

English. Development of geomatic techniques calls for use of the VHR satellite imagery (e.g. IKONOS-2, QuickBird-2) in forest inventory practices. The experiences of Scandinavians, USA and Canada prove, that such practices can be used, although Polish forests differ with regard to tree species and stand structure from coniferous boreal forests in North America and Canada. Polish forest inventory practices are based on old and traditional (time and money consuming) methods of field measurements. The new Polish Instruction of Forest Inventory (2003) allows the use of satellite imagery (VHR; pixel size; 6,0 m) to locate and delineate forest stands and gaps. This paper is a part of research project executed by the Forest Management and Monitoring Unit (Forest Research Institute in Warsaw) in cooperation with the Lab of GIS & Remote Sensing of the Forest Ecology Department (Forestry Faculty; Agricultural University of Krakow). The usefulness of QuickBird technology was demonstrated on the example of the Staszów Forests District (Regional Directorate of the Polish State Forests in Radom) and was supported by the General Directorate of Polish State Forest Holding. In the study, the VHR QuickBird-2 satellite imagery and DGPS measurements were applied and tested for the forest inventory purposes. The selected forest stands were delineated again by interpretation based on QuickBird-2 composition (432 pan-sharpened) and then compared with DGPS measurements and digital forest map (LMN) with regard to location (XY), shape and area. The results show the need for immediaterevision and updating of the newest digital forest map. The technology of LMN is based mainly on the vectorization of scanned old maps without any serious verification. The reference geodata such like VHR QuickBird-2 images may promptly change the quality of the process of creating maps by supporting the delineation of forest stands and gaps. The authors concluded that the so called combined methods, based on visual interpretation and vectorisation of the VHR QuickBird composition, DGPS survey and existing forest digital map (LMN) with related database from the latest forest inventory is the best practice in the Polish Forest Inventory.

Polish. obiektowa analiza obrazu (OBIA); protokół klasyfikacji; analizy przestrzenne GIS; sukcesja leśna; klasy pokrycia terenu

English. Object-Based Image Analysis (OBIA), classification protocols, GIS spatial analysis, forest succession, land-cover

3

227-238

3