Assessing methane emissions from paddy fields through environmental and UAV remote sensing variables.

Area of Science:

• Agricultural Science
• Environmental Science
• Remote Sensing

Background:

• Rice cultivation is a major source of methane (CH4), a potent greenhouse gas contributing to climate change.
• Accurate quantification of CH4 emissions from rice paddies is essential for climate change mitigation strategies.
• Traditional methods for measuring CH4 emissions are labor-intensive and costly.

Purpose of the Study:

• To develop and validate a novel, cost-effective method for quantifying methane (CH4) emissions from rice fields.
• To integrate machine learning (ML) algorithms with remote sensing data for improved CH4 emission prediction.
• To challenge the limitations of traditional closed chamber methods in rice CH4 emission assessment.

Main Methods:

• Utilized drones equipped with Micasense Altum camera for data collection.
• Integrated ground sensors to capture environmental variables.
• Employed and evaluated over 20 regression models, including the random forest regressor, for CH4 emission prediction.
• Leveraged remote sensing-derived vegetation indices and environmental data as predictors.

Main Results:

• Achieved high predictive accuracy with R-squared values of 0.98 for training and 0.95 for testing data.
• Identified phosphorus, GRVI median, and cumulative soil and water temperature as key predictive variables.
• The random forest regressor demonstrated superior predictive capabilities for CH4 emissions.

Conclusions:

• The developed ML and remote sensing approach offers an innovative, efficient, and cost-effective alternative for quantifying rice CH4 emissions.
• This technology-driven method provides valuable insights for rice growth parameter and vegetation index evaluation.
• The findings represent a significant advancement in monitoring greenhouse gas emissions from agricultural systems.