Automated Method for Delineating Harvested Stands Based on Harvester Location Data

• Published in: Remote sensing (Basel, Switzerland) Switzerland), 2020-09, Vol.12 (17), p.2754
• Main Authors: Melkas, Timo, Riekki, Kirsi, Sorsa, Juha-Antti
• Format: Article
• Language: English
• Subjects: automated geoprocessing; clustering; forest inventory; forest stand; GNSS positioning; harvester data
• ISSN: 2072-4292; 2072-4292
• DOI: 10.3390/rs12172754

The data produced by cut-to-length harvesters provide new large-scale data source for event-based update of national forest stand inventory by Finnish Forest Centre. This study aimed to automate geoprocessing, which generates delineations of operated areas from harvester location data. Automated algorithms were developed and tested with a dataset of 455 harvested objects, recorded during harvestings. In automated stand delineation, the location points are clustered, the stand points are identified and external strip roads are separated. Then, stand polygons are produced. To validate the results, automatic delineations were compared to 57 observed delineations from field measurements and aerial images. A detailed comparison method was developed to study the correspondence. Stand polygonization parameter was adjusted and areal correspondence with 1% error on average was obtained for stands over 0.75 ha. Good stand shape agreement was observed. Overall, the automated method worked well, and the operative stand delineations were found suitable for updating the forest inventory data. To modify the operative stands towards forest inventory stands, a balancing algorithm is introduced to create a solid, unique stand boundary between overlapping stands. This algorithm is beneficial for upkeep of stand networks. In addition, the Global Navigation Satellite System (GNSS) accuracy of the harvesters was examined and estimated numerically.