Biomass stored in young forests has enormous potential for the reduction of fossil fuel consumption. However, to ensure long-term sustainability, the measurement accuracy of tree height is crucial for forest biomass and carbon stock monitoring, particularly in young forests. Precise height measurement using traditional field measurements is challenging and time consuming. Remote sensing (RS) methods can, however, replace traditional field-based forest inventory. In our study, we compare individual tree height estimation from Light Detection and Ranging (LiDAR) and Digital Aerial Photogrammetry (DAP) with field measurements. It should be noted, however, that there was a one-year temporal difference between the field measurement and LiDAR/DAP scanning. A total of 130 trees (32 Scots Pine, 29 Norway Spruce, 67 Silver Birch, and 2 Eurasian Aspen) were selected for height measurement in a young private forest in south-east Finland. Statistical correlation based on paired t-tests and analysis of variance (ANOVA, one way) was used to compare the tree height measured with the different methods. Comparative results between the remote sensing methods and field measurements showed that LiDAR measurements had a stronger correlation with the field measurements and higher accuracy for pine (R² = 0.86, bias = 0.70, RMSE = 1.44) and birch (R² = 0.81, bias = 0.86, RMSE = 1.56) than DAP, which had correlation values of (R² = 0.71, bias = 0.82, RMSE = 2.13) for pine and (R² = 0.69, bias = 1.19, RMSE = 2.08) for birch. The correlation of the two remote sensing methods with the field measurements was very similar for spruce: LiDAR (R² = 0.83, bias = 0.30, RMSE = 1.17) and DAP (R² = 0.83, bias = 0.44, RMSE = 1.26). Moreover, the correlation was highly significant, with minimum error and mean difference (R² = 0.79–0.98, MD = 0.12–0.33, RMSD = 0.45–1.67) between LiDAR and DAP for all species. However, the paired t-test suggested that there is a significant difference (p < 0.05) in height observation between the field measurements and remote sensing for pine and birch. The test showed that LiDAR and DAP output are not significantly different for pine and spruce. Presumably, the time difference in field campaign between the methods was the reason for these significant results. Additionally, the ANOVA test indicated that the overall means of estimated height from LiDAR and DAP were not significantly different from field measurements in all species. We concluded that utilization of LiDAR and DAP for estimating individual tree height in young forests is possible with acceptable error and comparable accuracy to field measurement. Hence, forest inventory in young forests can be carried out using LiDAR or DAP for height estimation at the individual tree level as an alternative to traditional field measurement approaches.

The precise identification and classification of tree species in young forests during their early development stages are vital for forest management and silvicultural efforts that support their growth and renewal. However, achieving accurate geolocation and species classification through field-based surveys is often a labor-intensive and complicated task. Remote sensing technologies combined with machine learning techniques present an encouraging solution, offering a more efficient alternative to conventional field-based methods. This study aimed to detect and classify young forest tree species using remote sensing imagery and machine learning techniques. The study mainly involved two different objectives: first, tree species detection using the latest version of You Only Look Once (YOLOv12), and second, semantic segmentation (classification) using random forest, Categorical Boosting (CatBoost), and a Convolutional Neural Network (CNN). To the best of our knowledge, this marks the first exploration utilizing YOLOv12 for tree species identification, along with the study that integrates digital aerial photogrammetry with Planet imagery to achieve semantic segmentation in young forests. The study used two remote sensing datasets: RGB imagery from unmanned aerial vehicle (UAV) ortho photography and RGB-NIR from PlanetScope. For YOLOv12-based tree species detection, only RGB from ortho photography was used, while semantic segmentation was performed with three sets of data: (1) Ortho RGB (3 bands), (2) Ortho RGB + canopy height model (CHM) + Planet RGB-NIR (8 bands), and (3) ortho RGB + CHM + Planet RGB-NIR + 12 vegetation indices (20 bands). With three models applied to these datasets, nine machine learning models were trained and tested using 57 images (1024 × 1024 pixels) and their corresponding mask tiles. The YOLOv12 model achieved 79% overall accuracy, with Scots pine performing best (precision: 97%, recall: 92%, mAP50: 97%, mAP75: 80%) and Norway spruce showing slightly lower accuracy (precision: 94%, recall: 82%, mAP50: 90%, mAP75: 71%). For semantic segmentation, the CatBoost model with 20 bands outperformed other models, achieving 85% accuracy, 80% Kappa, and 81% MCC, with CHM, EVI, NIRPlanet, GreenPlanet, NDGI, GNDVI, and NDVI being the most influential variables. These results indicate that a simple boosting model like CatBoost can outperform more complex CNNs for semantic segmentation in young forests.

Oil heating systems are abundant in rural Finland and they need to be replaced by renewable energy as Finland aims to be carbon neutral by 2035. Bioenergy, one of the renewable energies, is a common source of energy in Finland as the country is rich in forest resources. In Finland, combined heat and power plants utilize such resource to produce district heat and electricity but Finnish rural areas do not have access to the district heating network. However, there are potential scenarios where community heating could be possible using portable chip-fired heating systems (heat containers). Ultimately, the cost of heating is an important factor for the consumers and the cost of investment is likely to put off any interest from the communities. In this research, we explored the cost and profitability of heat container investments in rural Finland and examined the challenges for the energy transition away from oil heating systems, as well as the opportunities decentralized biomass-fired heating systems might bring. The results of this research indicate that the price of heat produced in heat containers is comparatively higher than district heating, which is commonly used in cities in Finland, but is cost-competitive compared to oil heating depending on the price of oil. For example, the current price of LFO (~1 EUR/l) generates costlier heat than the 300 kW heat container provides. Firing wood pellets in the heat container is not economically viable due to expensive raw material but smaller-sized heat container (110 kW) firing wood chips could provide cost-competitive heat if uptime is raised to >2700 h/year. There are socio-economic impacts and value-added effects on the rural region due to utilization of local resource instead of imported LFO but there remain challenges and barriers such as high initial investment, low investment support and lack of policies focused on decentralised energy enterprises.

The precise identification and classification of tree species in young forests during their early development stages are vital for forest management and silvicultural efforts that support their growth and renewal. Yet, achieving accurate geolocation and species classification is often a labor-intensive and complex task through field-based surveys. Remote sensing technologies combined with machine learning techniques present an encouraging solution, offering a more efficient alternative to conventional field-based methods. This study aimed to detect and classify young forest tree species by employing remote sensing imagery and machine learning techniques. The study involved mainly two different objectives: first, the tree species detection using the latest version of You only Look Once (YOLOv12) and second, semantic segmentation (classification) using random forest, Categorical Boosting (CatBoost) and Convolutional Neural Network (CNN). To the best of our knowledge, this marks the first exploration utilizing YOLOv12 for tree species identification, along with the groundbreaking study that integrates digital aerial photogrammetry with Planet imagery to achieve semantic segmentation in young forest. The study utilized two remote sensing datasets: RGB imagery from UAV ortho photography and RGB-NIR from PlanetScope. For YOLOv12-based tree species detection, only RGB from ortho photography was used, while semantic segmentation was performed with three sets of data: (1) Ortho RGB, (2) Ortho RGB + canopy height model (CHM) + Planet RGB-NIR (8 Bands), and (3) ortho RGB + CHM + Planet RGB-NIR + 12 vegetation indices (20 Bands). With three models applied to these datasets, a total of nine machine learning models were trained and tested using total of 57 images (1024×1024 pixels) and their corresponding mask tiles. The YOLOv12 model achieved 79% overall accuracy, with Scots pine performing best (precision: 97%, recall: 92%, mAP50: 97%, mAP75: 80%) and Norway spruce showing slightly lower accuracy (precision: 94%, recall: 82%, mAP50: 90%, mAP75: 71%). For semantic segmentation, the CatBoost model with 20 Bands outperformed other models, achieving 85% accuracy, 80% Kappa, and 81% MCC, with CHM, EVI, NIRPlanet, GreenPlanet, NDGI, GNDVI, and NDVI being the most influential variables. These results indicate that a simple boosting model like CatBoost can outperform more complex CNNs for semantic segmentation in young forests.