9:15 AM - 9:30 AM
[MIS12-02] 3D-measuring of alpine vegetation using action cameras
Keywords:Alpine Vegetation, Photogrametry, Monitoring Sites 1000
Quantitative observation of alpine vegetation changes is an urgent issue as earlier snowmelt due to climate change and severe deer grazing continue to progress. However, surveys in alpine regions are costly, and the declining and aging workforce of researchers poses additional challenges. This study introduces a simplified alpine vegetation survey method using action camera video recording, currently being tested in the Ministry of the Environment's "Monitoring Sites 1000 Alpine Zone Survey" project. We also discuss the comparison between this method and traditional quadrat surveys conducted by human observers.
In the proposed method, an action camera such as the GoPro Hero12 Black (GoPro Inc.) is used to record videos of a 1 m × 1 m quadrat from a height of approximately 1.5 m, moving back and forth about five times. Markers that can be easily recognized through image analysis are placed at the corners of the quadrat to enable accurate area measurement and alignment of images taken in different years. The recorded videos are processed using the Structure from Motion (SfM) software "OpenDroneMap" to generate orthophotos with a resolution of 1 mm and a Digital Surface Model (DSM) representing vegetation height.
In the 2024 trial surveys, data were collected at Mt. Hakusan, the Japanese Southern Alps, and Mt. Fuji. At four sites in Hakusan and five sites in the Southern Alps, we compared the number of plant species identified from the SfM-derived orthophotos with those recorded through field surveys. The results showed that 50–70% of the species identified in the field could also be distinguished in the orthophotos. However, understory vegetation was not visible in the orthophotos, and some species were difficult to identify solely from the images.
This survey method, combining video recording and SfM processing, offers a relatively low-effort approach to monitoring alpine vegetation while also enabling the accumulation of quantitative digital data such as orthophotos and vegetation height models. Future work should focus on developing automated vegetation classification techniques from the orthophotos.
In the proposed method, an action camera such as the GoPro Hero12 Black (GoPro Inc.) is used to record videos of a 1 m × 1 m quadrat from a height of approximately 1.5 m, moving back and forth about five times. Markers that can be easily recognized through image analysis are placed at the corners of the quadrat to enable accurate area measurement and alignment of images taken in different years. The recorded videos are processed using the Structure from Motion (SfM) software "OpenDroneMap" to generate orthophotos with a resolution of 1 mm and a Digital Surface Model (DSM) representing vegetation height.
In the 2024 trial surveys, data were collected at Mt. Hakusan, the Japanese Southern Alps, and Mt. Fuji. At four sites in Hakusan and five sites in the Southern Alps, we compared the number of plant species identified from the SfM-derived orthophotos with those recorded through field surveys. The results showed that 50–70% of the species identified in the field could also be distinguished in the orthophotos. However, understory vegetation was not visible in the orthophotos, and some species were difficult to identify solely from the images.
This survey method, combining video recording and SfM processing, offers a relatively low-effort approach to monitoring alpine vegetation while also enabling the accumulation of quantitative digital data such as orthophotos and vegetation height models. Future work should focus on developing automated vegetation classification techniques from the orthophotos.