Paper presents methods for utilizing the multi-temporal imagery from the Sentinel-2 satellite to detect changes in forest cover, facilitating the updating of forest inventory data. Sentinel-2 images from various dates are combined into a single composite image that is further used as a feature vector set. Digital forest inventory data, which may not always correspond to the actual ground state due to potential inaccuracies or infrequent observations, is employed to identify monospecies regions and assess the classifier’s performance. Misclassified areas within the forest could signal changes in the forest ecosystem. Experimental methods have been developed to filter these areas, enhancing the accuracy in discerning actual changes from classifier errors within complex mixed forest regions. The empirical data presented within this study serves to validate the effectiveness of the proposed solutions.