Sviluppo di strumenti predittivi per l'ottimizzazione delle attività di raccolta sostenibile del legname a livello di popolamento forestale

The effective management and preservation of ecosystem services and goods, such as carbon sequestration, infrastructure protection, recreational spaces, and the provision of wood and non-wood forest products, require specific silvicultural treatments. These treatments can only be implemented through the proper execution of forest operations. Therefore, even forests that are not economically viable for timber production and harvesting must be managed to maintain their ability to provide these services. A significant challenge in many European forestry regions is the low competitiveness of the forestry sector. Often, the revenues from timber harvesting are insufficient to cover the associated costs. This issue is particularly pronounced in mountainous areas, where harvesting operations are generally more difficult due to steep terrain and limited accessibility. To address this, timber harvesting in mountainous regions frequently utilizes cable yarding systems, which are inherently less efficient than ground-based systems. This inefficiency is largely due to the considerable setup time required before the cable systems can become operational. In response to this context, the following thesis work aimed to develop new decision support tools that employ mathematical programming techniques to optimize forest planning. The goal was to improve the efficiency of timber harvesting activities, especially those utilizing cable yarding systems. In the first phase of the research, two optimization models for tree selection in mountain forest stands were created, aimed at maximizing the economic benefits of harvesting. This was achieved by first establishing a productivity model for the harvesting system, utilizing StanForD 2010 data, which was then integrated with a detailed forest inventory derived from airborne laser scanner data by means of individual tree detection process. The two models were compared against actual harvesting parameters to determine which formulation offered a tangible improvement in harvesting system efficiency. Following this, the optimization model that was effective in supporting forest operations planning was further refined. This refinement improved its operational aspects to better represent the typical procedures used in harvesting activities with cable yarding systems. Additionally, a second silvicultural objective was incorporated alongside the goal of maximizing economic benefit to create optimal tree selection trade-off solutions. The aim was to create a tool that allows decision-makers to choose the solution that best fits the specific conditions of their forest stand. Finally, the investigation extended to enhancing the efficiency of bucking operations, as this significantly influences the final value of each processed tree. To achieve this, a detailed forest inventory of an industrial forest plantation was created through the collection of accurate LiDAR data using a terrestrial laser scanner. The resulting point cloud was then processed using an advanced individual tree detection and segmentation algorithm. The eventual information was utilized to develop an optimization model aimed at maximizing the quality and volume of timber assortments at a predictive level. In conclusion, it was observed that mathematical programming techniques are effective for optimizing the planning of forest operations and can facilitate the development of new decision support tools for practitioners to evaluate potential intervention strategies. However, it is important to note that these tools rely on very accurate data, which can require considerable effort to collect.