Browsing by Subject "kuntoinventointi"

In Finland, forest road network has played a significant role in the society throughout history by serving landowners, stakeholders of timber trade, forest management operators, agricultural- and other entrepreneurs. Different forest recreational users such as berry pickers, mushroom pickers and hunters benefit also from good quality forest roads. Wide forest road network help also in preventing forest fires, building fires and it provides help for human and animal rescue missions. In Finland, large number of private forest roads have reached end of their working life and require therefore wide renovations in near future so that the high quality can be maintained. The large-scale determination of forest roads quality is vital so that situation of lower level road network can be followed, and decisions can be maid whether forest roads can be utilized in timber harvesting operations for example. The growing trend in size and weight of timber transport vehicles will cause more careful route planning to the harvest site when forest roads are in bad shape. Good quality forest roads will reduce fuel consumption in timber transport, vehicle damages and road damages. The main objective in this study was to determine the potential of open access geographic information data and especially open access low-density airborne laser scanning data to evaluate the quality of forest roads. Area-based laser scanning inventory method was used with reference data from field plots. Field data was collected from area of research in November 2018 and it consisted from predefined sample plots that were evaluated with the means of traditional forest road quality factors. The aim was to find these quality factors from ALS data and from other open access data and predict forest road quality class using non-parametric k-nearest neighbor method. The results show that metrics calculated from ALS data were quite important in evaluating forest road quality classes. Metrics that illustrate point height distribution, height averages and metrics extracted from digital elevation model which illustrate slope were the most significant in this study. The results show also that the correlation of individual metrics and forest road quality class from reference data was not very high. However, the quality class of forest roads could be predicted correctly at least 69,8 % accuracy when k-nearest neighbor method was used, and all metrics were used. The method used in this study can be utilized to predict forest road quality class relatively accurately, but the accuracy could still be improved. One way to improve this method would be to use high density ALS data and more accurate reference data. It could also be interesting to use this method in another area of research and inspect how the results would differ from this study.