Search

Modern forestry relies on an extensive network of well-maintained forest roads. However, the trafficability of forest roads can be limited by poor construction or by insufficient load-bearing capacity caused by, for example, weather conditions. In all forestry operations, the forest roads should be able to support heavy vehicles during the transport of roundwood and harvesting machinery, so logistics must be carefully planned to ensure good road trafficability. To facilitate this planning, it is necessary to know the various characteristics of the forest road in terms of its width, structure and load-bearing capacity. Obtaining such information using remote sensing data would be beneficial. In this study, we used airborne laser scanning (ALS) data to assess the width of the roadway on forest roads. We developed several algorithms that were used to assess roadway width based on the cross-sectional data (derived with ALS) of the roads in 8-m long segments. The ALS echoes within the segments were classified into 20 cm wide strips which the algorithms then analyzed in a stepwise manner. Both height and intensity information from the ALS data were utilized. The algorithm results were then used as predictors in linear models to predict the roadway width. The main focus was on the road level, i.e., aggregated values from 25 different roads were used in model fitting. In terms of root mean square error (RMSE) values, accuracies of approximately 20–30 cm (or 7–10 %) were obtained for the predicted roadway widths with a separate validation data. This level of accuracy can be considered adequate given the characteristics of the ALS data used in the modelling. The results indicate that the presented approach has potential for practical applications to predict the width attributes of forest roads.