14:45 〜 15:00
[BCG04-05] Estimation of Methane Emissions from Eutrophic Lake of Japan Using Satellite Observation and Machine Learning Method
キーワード：富栄養湖、水生植物、水温
Introduction
Methane is the second largest greenhouse gas contributing to global warming after carbon dioxide. And it has about 28times greenhouse effect that of carbon dioxide. Freshwater areas, including lakes, are the largest natural source of methane so estimating methane emissions from lakes is crucial. However, currently, there is a significant discrepancy in methane emission estimates from lakes depending on the research methods, making it challenging to obtain accurate values. Combining remote sensing data with on-site data from flux observation points and utilizing machine learning to construct estimation models is expected to enable more precise methane emission estimates.
Objective
In this study, focusing on Lake Suwako in Nagano Prefecture, we used satellite observations to gather information on the extent of aquatic plant proliferation, water temperature, and methane flux data from flux observation points. We employed the Random Forest machine learning technique to construct a model for estimating methane flux.
Material and method
The on-site methane flux data were obtained using the eddy covariance method. High spatial and temporal resolution Planet Scope satellite data were utilized for estimating the extent of macrophytes proliferation, while Landsat 8 was employed for estimating water temperature. This study has four study methods. First is Obtaining and pre-processing methane flux data, second is satellite data calculation, third is methane emissions prediction by Random Forest and fourth is application of models to entire area of study area.
Result and discussion
In winter, methane flux was high at lakeshore area because of water temperature is high at these regions. In summer, it is high at region of macrophytes growth. Spatial distribution of methane flux showed similar pattern in winter and summer but a little bit different because of macrophytes growth and summer has more than twelve times methane flux compared with winter of it.
Conclusion
From our result, macrophytes were shown as important index for estimation of methane emission. Future task of this study is to improve of prediction accuracy in no macrophytes season model and to estimate methane emissions with Landsat pixel-sized spatial resolution.
Methane is the second largest greenhouse gas contributing to global warming after carbon dioxide. And it has about 28times greenhouse effect that of carbon dioxide. Freshwater areas, including lakes, are the largest natural source of methane so estimating methane emissions from lakes is crucial. However, currently, there is a significant discrepancy in methane emission estimates from lakes depending on the research methods, making it challenging to obtain accurate values. Combining remote sensing data with on-site data from flux observation points and utilizing machine learning to construct estimation models is expected to enable more precise methane emission estimates.
Objective
In this study, focusing on Lake Suwako in Nagano Prefecture, we used satellite observations to gather information on the extent of aquatic plant proliferation, water temperature, and methane flux data from flux observation points. We employed the Random Forest machine learning technique to construct a model for estimating methane flux.
Material and method
The on-site methane flux data were obtained using the eddy covariance method. High spatial and temporal resolution Planet Scope satellite data were utilized for estimating the extent of macrophytes proliferation, while Landsat 8 was employed for estimating water temperature. This study has four study methods. First is Obtaining and pre-processing methane flux data, second is satellite data calculation, third is methane emissions prediction by Random Forest and fourth is application of models to entire area of study area.
Result and discussion
In winter, methane flux was high at lakeshore area because of water temperature is high at these regions. In summer, it is high at region of macrophytes growth. Spatial distribution of methane flux showed similar pattern in winter and summer but a little bit different because of macrophytes growth and summer has more than twelve times methane flux compared with winter of it.
Conclusion
From our result, macrophytes were shown as important index for estimation of methane emission. Future task of this study is to improve of prediction accuracy in no macrophytes season model and to estimate methane emissions with Landsat pixel-sized spatial resolution.