The high dimensionality of data generated by Unmanned Aerial Vehicle(UAV)-Lidar makes it difficult to use classical statistical techniques to design accurate predictive models from these data for conducting forest inventories. Machine learning techniques have the potential to solve this problem of modeling forest attributes from remotely sensed data. This work tests four different machine learning approaches – namely Support Vector Regression, Random Forest, Artificial Neural Networks, and Ex-

treme Gradient Boosting – on high-density GatorEye UAV-Lidar point clouds for indirect estimation of individual tree dendrometric metrics (field-derived) such as diameter at breast height, total height, and timber volume.

Accurate forest parameters are essential for forest inventory. Traditionally, parameters such as diameter at breast height (DBH) and total height are measured in the field by level gauges and hypsometers. However, field inventories are usually based on sample plots, which, despite providing valuable and necessary information, are laborious, expensive, and spatially limited. Most of the work developed for remote measurement of DBH has used terrestrial laser scanning (TLS), which has high density point clouds, being an advantage for the accurate forest inventory. However, TLS still has a spatial limitation to application because it needs to be manually carried to reach the area of interest, requires sometimes challenging field access, and often requires a field team. UAV-borne (unmanned aerial vehicle) lidar has great potential to measure DBH as it provides much higher density point

cloud data as compared to aircraft-borne systems. Here, we explore the potential of a UAV-lidar system (GatorEye) to measure individual-tree DBH and total height using an automatic approach in an integrated crop-livestock-forest system with seminal forest plantations of Eucalyptus benthamii. 

We’re excited to announce the release of LiDARMill v2! LiDARMill v2 takes automated post-processing to the next level. In our recent webinar, we covered some of the new improvements and features including:

• Imagery Processing in LiDARMill
• Ground Control Reporting and Adjustments
• Robust Coordinate System Handling
• A Workflow Overview and Demonstration
• Multi-Mission Processing Support
• Advanced Point Cloud Filtering Options
• RGB Thermal & Fusion
• Accuracy Reporting
• Automated LiDAR and Camera Calibration Options
• Near-Real Time (NRT) Reference Station Positioning for Projects Requiring Less Than 24 Hour Turn-around Time
• Trajectory Post-Processing Without Reference Stations

If you have any questions or would like to learn more about LiDARMill v2, please don’t hesitate to get in touch. We’d be happy to help!

The Cerrado, the most biologically diverse savanna in the world, is threatened by anthropogenic activities, and requires development of effective environmental policies spanning local to global scales. Remotely Piloted Aircraft Systems (RPAS) can dramatically reduce the costs and time of surveys and evaluation of these regions. The objective of this article is to demonstrate the potential of visual (RGB) and Light Detection and Ranging (LIDAR) sensors on RPAS for physical characterization of landscapes in the Cerrado biome. Analyses on vegetation structure were performed, with the number of trees automatically counted. The average height of the trees obtained with the RGB sensor was significantly lower than the obtained by LIDAR, demonstrating the limitation of Structure from Motion data in representing the landscape with denser vegetation. Automatic counting of trees with LIDAR data were equal to 1825 on the whole study area, and 245 inside the ecological study area parcels.