Laser Technology and AI in Support of Forest Management
Forest management planning demands many types of information related to the forest resources like quantity, composition and condition also taking into account terrain characteristics, road layout and so on. This study aim to discover new data mining techniques providing more information on forests to support forest management and planning.
There are many ways to obtain detailed spatial information of forests. In last decades, remote sensing data have been widely used to describe forest features [1]. These techniques can’t replaced field surveys completely, but they can provide some sort of information in very short time period about large areas. In this study, the main data sources are from airborne laser scanning (ALS) and National forest inventory (NFI). ALS provides three-dimensional point clouds capturing forest structure at large areas level. ALS data can accurately represent canopy with some capability to capture sub-canopy forest structure. Otherwise, NFI provides accurate ground data. With these limitations in mind, we can combine data from airborne laser scanning with detailed field data from sample plots used in National forest inventory in so called Area-based approach [2].
Based on statistical relationship between metrics describing point clouds and dendrometric features from NFI plots, we can create models for predicting the features across large areas. These predicted values can be transformed into dendrometric maps — crucial source of information for forest management planning. Conventional forest stand maps are very time-consuming to produce. In Czechia, we also cannot say, how precise they are. In fact, they are often not very much, neither numerical or graphical part. This method should fix both parts at large-scale.
The challenge for artificial intelligence in this research will be to determine appropriate metrics describing point clouds to create the most accurate models for forest feature prediction. It will also detect and classify tree species and optimize silvicultural interventions for forest management and planning, including thinning and felling proposals [3]. Sustainable forest management importance gains as forests become more endangered by global warming.
The scope of outputs will be at the level of the natural forest area supporting informed decision-making and sustainable forest management practices.
[1] P. Surový a K. Kuželka, Aplikace dálkového průzkumu Země v lesnictví, Vydání první. V Praze: Česká zemědělská univerzita, 2019.
[2] E. Næsset, „Area-Based Inventory in Norway — From Innovation to an Operational Reality”, in Forestry Applications of Airborne Laser Scanning, roč. 27, M. Maltamo, E. Næsset, a J. Vauhkonen, Ed., in Managing Forest Ecosystems, vol. 27. , Dordrecht: Springer Netherlands, 2014, s. 215–240. doi: 10.1007/978–94–017–8663–8_11.
[3] P. Surový a K. Kuželka, „Acquisition of Forest Attributes for Decision Support at the Forest Enterprise Level Using Remote-Sensing Techniques — A Review”, Forests, roč. 10, č. 3, s. 273, bře. 2019, doi: 10.3390/f10030273.
