5:15 PM - 7:15 PM
[AAS05-P10] Spatial Analysis and Modeling of Methane Emissions from Rice Paddies Using Machine Learning Based on Satellite Data and Ground Observation
★Invited Papers
Keywords:methane, methane emission, machine learning, GBM, rice paddy
As global warming accelerates, the importance of reducing greenhouse gas emissions is becoming increasingly significant. Methane (CH4) has approximately 21 times the greenhouse effect of carbon dioxide (CO2), making it a major contributor to climate change. In South Korea, the agricultural sector accounts for about 43% of total methane emissions, with rice paddies being the primary source. Accurately quantifying methane emissions from submerged soils is essential for developing climate change mitigation strategies.
This study employs a Gradient Boosting Machine (GBM) model to estimate and visualize methane emissions using ERA5 meteorological data and ground-based observations from FluxNet for rice paddy regions in South Korea, the United States, and the Philippines. Additionally, MODIS satellite imagery was utilized to incorporate NDVI (Normalized Difference Vegetation Index) and NDWI (Normalized Difference Water Index) data, reflecting the growth status and moisture conditions of rice paddies. The study focuses on rice paddies in South Korea, utilizing key meteorological variables such as temperature, humidity, soil moisture, and evapotranspiration, along with satellite-based vegetation and water indices, to predict methane concentrations. These variables are critical determinants of methane production in rice paddies, closely related to the formation of anaerobic conditions in organic matter decomposition, microbial activity, and redox state changes.
Daily methane concentration predictions were performed and spatially visualized as gridded maps to quantitatively analyze methane emissions from rice paddies. Additionally, methane concentration maps reflecting seasonal variations were generated to assess spatiotemporal fluctuations. The findings of this study contribute to improving the accuracy of methane emission estimations in South Korean rice paddies and provide valuable data for formulating greenhouse gas reduction strategies and policy decisions in the agricultural sector.
This research was supported by Particulate Matter Management Specialized Graduate Program through the Korea Environmental Industry & Technology Institute(KEITI) funded by the Ministry of Environment(MOE).
This study employs a Gradient Boosting Machine (GBM) model to estimate and visualize methane emissions using ERA5 meteorological data and ground-based observations from FluxNet for rice paddy regions in South Korea, the United States, and the Philippines. Additionally, MODIS satellite imagery was utilized to incorporate NDVI (Normalized Difference Vegetation Index) and NDWI (Normalized Difference Water Index) data, reflecting the growth status and moisture conditions of rice paddies. The study focuses on rice paddies in South Korea, utilizing key meteorological variables such as temperature, humidity, soil moisture, and evapotranspiration, along with satellite-based vegetation and water indices, to predict methane concentrations. These variables are critical determinants of methane production in rice paddies, closely related to the formation of anaerobic conditions in organic matter decomposition, microbial activity, and redox state changes.
Daily methane concentration predictions were performed and spatially visualized as gridded maps to quantitatively analyze methane emissions from rice paddies. Additionally, methane concentration maps reflecting seasonal variations were generated to assess spatiotemporal fluctuations. The findings of this study contribute to improving the accuracy of methane emission estimations in South Korean rice paddies and provide valuable data for formulating greenhouse gas reduction strategies and policy decisions in the agricultural sector.
This research was supported by Particulate Matter Management Specialized Graduate Program through the Korea Environmental Industry & Technology Institute(KEITI) funded by the Ministry of Environment(MOE).