Abstract
The purpose of this paper is to evaluate the forest health condition in the area of Hydropower plant Gabčikovo using the digital multispectral aerial images from 2011 in the section Dobrohošt - Sap. We proposed a new innovative procedure of creating the forest mask in the image processing. It concerns a process of defining tree crowns, where we combined subject and raster oriented approach of the image classification. We evaluated forest health condition on a selected forest mask using the method of twophased sampling with regression. The indicators of health condition were tree defoliation and presence of heavily damaged trees and snags in the stands. Defoliation was used to compare the current forest health condition in the concerned area with the data from 2008. Presence of the snags as the result of severe damage is interpreted in terms of stand types related to site conditions. Field survey proved presence of invasive and non-native tree species and plants. Based on the result, we have formulated recommendations for further forest monitoring in the Danube River inundation.