[AHW32-P20] Classification analysis of coastal seaweed beds using UAV and Near-Infrared camera
Keywords:coastal seaweed beds, UAV, near-infrared camera
Seaweed beds play a role in the coastal sea keeping water quality better and maintain ecosystem itself in shallow waters. However, the coastal environment has been affected by economic development. In Seto Inland Sea separating Honshu from Shikoku and Kyushu of Japan, where water quality has been improved by environmental conservation efforts, and some area appeared to be locally oligotrophic.
Satellite data has been used for remote sensing of seaweed beds, so far, however, it is susceptible to weather and clouds. Furthermore, free satellite data could be used with some problem such as low spatial resolution. Other typical monitoring methods include airborne platforms, grand based surveys and snorkel / scuba surveys. These techniques can suffer from temporal and spatial inconsistency, or are very restricted making it hard to assess seagrass meadows in a structured manner.
This research aims to obtain distribution of shallow seaweed beds by analyzing the combined data of visible and near-infrared images acquired using UAV. The UAVs can reduce effects of clouds and atmosphere. In addition, by conducting surveys at the same location during the daytime in the intertidal zone, the conditions for on-site observations can be made the same. Furthermore use of commercially available UAVs and inexpensive near-infrared cameras also reduce cost of research compared to expensive satellite data and snorkel / scuba surveys. The distribution data of shallow seaweed beds is to be analyzed with ground survey samples and the factors affecting on changes in seagrass bed ecosystems will be revealed.
The study area is Gohonmatsu Beach, Ikuchijima, Onomichi City, Hiroshima Prefecture.
UAV cameras shot about 100 at coasts vertically downward. Next, an ortho mosaic image was created (3.6 cm pixel-1) in Structure from Motion (SfM). The same procedure was used for images from Near-Infrared cameras. Although NDVI were used as first analysis index, NDVI showed negative values for seaweed on the intertidal beach and submerged seagrass. Newly proposed NSGI showed positive value for those ones. Furthermore, we propose SGI1 and SGI2 as a method of seaweed bed classification using the brightness intensity of each band. SGI1 and SGI2 determines the seaweed beds under sea surface and on sandy beach, respectively. By combining these, most seaweed beds in the coastal area could be determined.
Duffy, J.P., Pratt, L., Anderson, K., Land, P.E., Shutler, J.D., 2018. Spatial assessment of intertidal seagrass meadows using optical imaging systems and a lightweight drone. Estuarine, Coastal and Shelf Science. 200, 169-180. https://www.sciencedirect.com/science/article/pii/S0272771417302202
*This work is supported by JSPS Grant-in-Aid for Scientific Research (B) (18H03411, PI: Mitsuyo Saito).
Satellite data has been used for remote sensing of seaweed beds, so far, however, it is susceptible to weather and clouds. Furthermore, free satellite data could be used with some problem such as low spatial resolution. Other typical monitoring methods include airborne platforms, grand based surveys and snorkel / scuba surveys. These techniques can suffer from temporal and spatial inconsistency, or are very restricted making it hard to assess seagrass meadows in a structured manner.
This research aims to obtain distribution of shallow seaweed beds by analyzing the combined data of visible and near-infrared images acquired using UAV. The UAVs can reduce effects of clouds and atmosphere. In addition, by conducting surveys at the same location during the daytime in the intertidal zone, the conditions for on-site observations can be made the same. Furthermore use of commercially available UAVs and inexpensive near-infrared cameras also reduce cost of research compared to expensive satellite data and snorkel / scuba surveys. The distribution data of shallow seaweed beds is to be analyzed with ground survey samples and the factors affecting on changes in seagrass bed ecosystems will be revealed.
The study area is Gohonmatsu Beach, Ikuchijima, Onomichi City, Hiroshima Prefecture.
UAV cameras shot about 100 at coasts vertically downward. Next, an ortho mosaic image was created (3.6 cm pixel-1) in Structure from Motion (SfM). The same procedure was used for images from Near-Infrared cameras. Although NDVI were used as first analysis index, NDVI showed negative values for seaweed on the intertidal beach and submerged seagrass. Newly proposed NSGI showed positive value for those ones. Furthermore, we propose SGI1 and SGI2 as a method of seaweed bed classification using the brightness intensity of each band. SGI1 and SGI2 determines the seaweed beds under sea surface and on sandy beach, respectively. By combining these, most seaweed beds in the coastal area could be determined.
Duffy, J.P., Pratt, L., Anderson, K., Land, P.E., Shutler, J.D., 2018. Spatial assessment of intertidal seagrass meadows using optical imaging systems and a lightweight drone. Estuarine, Coastal and Shelf Science. 200, 169-180. https://www.sciencedirect.com/science/article/pii/S0272771417302202
*This work is supported by JSPS Grant-in-Aid for Scientific Research (B) (18H03411, PI: Mitsuyo Saito).