Assessing tree modeling algorithm from terrestrial laser scanning in Rumperöd mixed forest

Recent technological advancements have revolutionized tree inventory methods for the forest industry. The development of terrestrial laser scanning (TSL) and automatized tree recognition software to estimate crucial tree parameters such as diameter at breast height (DBH) has made the inventory work faster and more reliable. Forest managers can quickly acquire large amounts of data to safely estimate timber yield and effectively use the resources of the forest. This study assesses the Hough transform algorithms (TreeLS) performance to acquire DBH and tree height in the Rumperöd mixed forest in Scania, Sweden, with continuous cover forestry management. Few studies have been made evaluating the Hough transform script under these challenging conditions with uneven terrain and irregular spatial tree density and dense understory vegetation. The result showed large difficulties for the Hough transform algorithm to correctly identify the trees in the plot due to stones, boulders, and uneven terrain. Only 25% of the trees in the plot were identified correctly. The algorithm also returned artefacts that further impaired the results of the TLS measurements to traditional field measurements. The mean difference in DBH between TLS measurement and field measurement in the plot was 115 mm and for tree height, the mean difference was 7.35 m. This difference in DBH and tree height would have overestimated the timber yield and made this algorithm unsuitable for the forest industry in the conditions present in the Rumperöd mixed forest.