Looking for new solutions in delineating forest stands: segmentation of multi-source data

Abstract

Up to date the main source of forest related information in Lithuania remains the standwise forest inventory. Delineation of stands all the time was based on the use of aerial photography. Changing economic conditions and new approaches in using satellite and GIS data requires looking for new solutions in conventional forest inventory techniques. This presentation would dea] with testing segmentation method for generating possibly homogeneous in terms of forestry variables units, corresponding to forest stands. The data sei of multi-source information, to be used for segmentation, consists of Spot and Landsai images, digital aerial photographs, old forest maps, boundaries of cutting areas, soil maps, digital elevation model, general GIS dala. A systematic network of field plots (Apr. 8 plots per ha) was developed for the test area. Minimization of within stand variance of main foresi parameters and correspondence with the expert-done stand discretization are used as the main criterions for assessment the segmentation results. The very first results indicate the possibility to achieve segmentation accuracy acceptable for forest inventory aims