Assessment of forest harvesting using digital imagery

Client

Resolu Forest Products

Collaborator

Center for Experimentation and Development in the Boreal Forest (CEDFOB)

Funding

Applied Research and Development (ARD) grant from the Natural Sciences and Engineering Research Council of Canada (NSERC)

Objective

The aim of this project was to determine the post-intervention characteristics required to confirm the operational compliance of a progressive irregular cut (PIC) harvest, using high-precision multispectral and thermal imagery acquired by drone. Advanced multispectral and thermal imaging technologies acquired by UAVs offer an affordable solution to the challenges of IPC inventory. The drone offers great flexibility in acquisition, even in real time over specific territories.

The latest advances in photogrammetry also enable three-dimensional reconstructions of the surfaces observed, giving data of comparable quality to what lidar can obtain.

Interpretation and monitoring of forest condition will be enhanced through the use of new data sources and processing techniques combined with advanced image analysis, such as image segmentation and machine learning-based classification methods.

Methodology

Before and after logging operations, a drone is used to take multispectral and thermal images of the study sites. In order to assess the characteristics of the stands overflown, the drone acquisitions are coordinated with in situ field surveys carried out by foresters.

The images captured are then processed using photogrammetry software to create orthomosaics and a digital surface model.

Finally, the data is analyzed using image processing techniques and machine learning analysis to extract data such as widths of collected strips, collected and residual crowns, and other features of ecological interest.

Equipment

DJI Matrice 210 RTK drone

Zenmuse X7 RGB camera

MicaSense Altum multispectral camera. This is a multispectral camera with 5 bands in the visible and near-infrared spectrum (blue, green, red, red edge and near-infrared) and one band in the thermal infrared.

Software: Pix4D, PyTorch

Field

Resource management