DITCH EDGE EXTRACTION AS A RESOURCE FOR ENVIRONMENTAL CONSERVATION AND MANAGEMENT

Land consolidation is one of the most important operations carried out on agricultural land to ensure effective land management, whether for agricultural production, land conservation or the preservation of historic land structures. It is a time- and cost-intensive process, typically spanning several years; therefore, any advancements that can reduce its duration are valuable. This study focuses on the identification of drainage and roadside ditch edges, which are measured at an initial stage of the consolidation process to update land use records. The aim of this study was to assess the accuracy with which ditch banks in agricultural areas can be delineated using publicly available, high-resolution LiDAR data from airborne laser scanning (ALS), with a point cloud density of 4 points per square meter. The study area was classified and a ditch class was extracted with a classification accuracy of Kappa index = 0.77. Then the edges were compared with ground-truth data obtained via GNSS RTK measurements. Survey points were recorded every 0.5 meters along the ditch edges and the distances to the nearest corresponding points were analyzed. The mean deviation across the entire study area was 0.92 meters, with a standard deviation of ±1.5 meters and a median of 0.42 meters. A significant impact of outlier data was observed. The obtained results indicate that this method is not sufficiently accurate for application in the land consolidation process, particularly in cases where ditch edges coincide with property boundaries. However, the accuracy achieved is adequate for updating land use records. To assess the upper limit of classification accuracy, two ditches were selected for in-depth analysis: a drainage ditch (mean deviation: 0.42 meters, median: 0.30 meters) and a roadside ditch (mean deviation: 0.46 meters, median: 0.38 meters). The improved accuracy observed in these cases highlights the potential for proposed method improvement.