10:45 AM - 12:15 PM
[AHW18-P01] Estimation of Regional Seagrass Biomass and Carbon Content Using UAV Aerial Photography at Multiple Sites
Keywords:coastal seagrass & seaweed beds, UAV, biomass
Seagrass beds play a very important role in the shallow-water ecosystem and have recently come to be defined as "blue carbon" due to their high carbon-fixing capacity. Seagrass beds provide habitat for marine organisms and purify water quality, and their carbon-fixing function in particular has been attracting attention in recent years as a means of reducing carbon dioxide emissions. Despite the wide variety of functions of seaweed beds, the amount of algae present in the sea is decreasing, and it is very important to monitor their habitat status in order to conserve them.
Remote sensing has been attracting attention in recent years as a means of monitoring the habitat conditions of seaweed beds. In particular, remote sensing of seaweed beds using UAVs has been shown to identify seaweed beds with a high degree of accuracy and can accurately identify the distribution of seaweed beds. However, most examples of analysis using UAVs are limited in scope and only identify the distribution of the algal beds.
This study aimed to identify the detailed distribution of extensive algal beds on a scale of several tens of kilometers and to estimate their biomass (SB) and carbon content (SC). The study site was Ikuchi Island, located in the Seto Inland Sea. To understand the detailed distribution of seaweed beds around Ikuchi Island, we divided the island into 32 areas and took aerial photographs at each location using a UAV. After compositing the aerial images, the seaweed beds were identified by applying the maximum likelihood method and the Iso-data method. The results of the accuracy evaluation showed that the maximum likelihood method had good identification accuracy, with an average overall accuracy of 0.92. On the other hand, the average overall accuracy of the Iso-data method was 0.74, which was worse than the maximum likelihood method.
The distribution of seaweed beds by the maximum likelihood method, which had good classification accuracy, was used to distinguish between seaweed bed component species. The difference in habitat location between seaweeds and seagrasses was used to distinguish between seagrasses and seaweeds. Among seaweeds, blue-green algae were distinguished by coloration because the difference in coloration was relatively clear. Since all seaweeds were defined as eelgrass based on literature and survey results, the final classification of seaweed bed component species was eelgrass, blue-green algae, and other seaweeds.
Coverage was calculated from the identification results of the seaweed bed component species, and the relationship equations between coveredness and LAI (Leaf Area Index), leaf area and SB, SC, which were derived from the prior coddling and sampling, were adapted. After converting from coverage to LAI using the regression equation, SB and SC were estimated to be 43 and 14 tons for eelgrass and 37 and 10 kg for blue-green algae, respectively. In this study, eelgrass had a very large habitat area compared to other algae, and the amount of SB and SC was also large. These results indicate the importance of the role of eelgrass in the function of algal beds.
This research was supported by Grant-in-Aid for Scientific Research (B) (No. 21H03650, PI: Mitsuyo Saito) and Grant-in-Aid for International Cooperative Research (A) (No. 20KK0262, PI: Mitsuyo Saito) from JSPS.
Remote sensing has been attracting attention in recent years as a means of monitoring the habitat conditions of seaweed beds. In particular, remote sensing of seaweed beds using UAVs has been shown to identify seaweed beds with a high degree of accuracy and can accurately identify the distribution of seaweed beds. However, most examples of analysis using UAVs are limited in scope and only identify the distribution of the algal beds.
This study aimed to identify the detailed distribution of extensive algal beds on a scale of several tens of kilometers and to estimate their biomass (SB) and carbon content (SC). The study site was Ikuchi Island, located in the Seto Inland Sea. To understand the detailed distribution of seaweed beds around Ikuchi Island, we divided the island into 32 areas and took aerial photographs at each location using a UAV. After compositing the aerial images, the seaweed beds were identified by applying the maximum likelihood method and the Iso-data method. The results of the accuracy evaluation showed that the maximum likelihood method had good identification accuracy, with an average overall accuracy of 0.92. On the other hand, the average overall accuracy of the Iso-data method was 0.74, which was worse than the maximum likelihood method.
The distribution of seaweed beds by the maximum likelihood method, which had good classification accuracy, was used to distinguish between seaweed bed component species. The difference in habitat location between seaweeds and seagrasses was used to distinguish between seagrasses and seaweeds. Among seaweeds, blue-green algae were distinguished by coloration because the difference in coloration was relatively clear. Since all seaweeds were defined as eelgrass based on literature and survey results, the final classification of seaweed bed component species was eelgrass, blue-green algae, and other seaweeds.
Coverage was calculated from the identification results of the seaweed bed component species, and the relationship equations between coveredness and LAI (Leaf Area Index), leaf area and SB, SC, which were derived from the prior coddling and sampling, were adapted. After converting from coverage to LAI using the regression equation, SB and SC were estimated to be 43 and 14 tons for eelgrass and 37 and 10 kg for blue-green algae, respectively. In this study, eelgrass had a very large habitat area compared to other algae, and the amount of SB and SC was also large. These results indicate the importance of the role of eelgrass in the function of algal beds.
This research was supported by Grant-in-Aid for Scientific Research (B) (No. 21H03650, PI: Mitsuyo Saito) and Grant-in-Aid for International Cooperative Research (A) (No. 20KK0262, PI: Mitsuyo Saito) from JSPS.