Abstract:This study investigates the correlation between dairy farm characteristics and methane concentrations as derived from satellite observations in Eastern Canada. Utilizing data from 11 dairy farms collected between January 2020 and December 2022, we integrated Sentinel-5P satellite methane data with critical farm-level attributes, including herd genetics, feeding practices, and management strategies. Initial analyses revealed significant correlations with methane concentrations, leading to the application of Variance Inflation Factor (VIF) and Principal Component Analysis (PCA) to address multicollinearity and enhance model stability. Subsequently, machine learning models - specifically Random Forest and Neural Networks - were employed to evaluate feature importance and predict methane emissions. Our findings indicate a strong negative correlation between the Estimated Breeding Value (EBV) for protein percentage and methane concentrations, suggesting that genetic selection for higher milk protein content could be an effective strategy for emissions reduction. The integration of atmospheric transport models with satellite data further refined our emission estimates, significantly enhancing accuracy and spatial resolution. This research underscores the potential of advanced satellite monitoring, machine learning techniques, and atmospheric modeling in improving methane emission assessments within the dairy sector. It emphasizes the critical role of farm-specific characteristics in developing effective mitigation strategies. Future investigations should focus on expanding the dataset and incorporating inversion modeling for more precise emission quantification. Balancing ecological impacts with economic viability will be essential for fostering sustainable dairy farming practices.

What problem does this paper attempt to address?

### What problems does this paper attempt to solve? This paper aims to explore the correlation between dairy farm characteristics and methane concentration, especially to evaluate the methane emissions of dairy farms in eastern Canada through satellite data and machine - learning techniques. Specifically, the research focuses on the following aspects: 1. **The relationship between dairy farm characteristics and methane emissions**: - By analyzing the data of 11 dairy farms from January 2020 to December 2022 and combining the methane concentration observed by Sentinel - 5P satellite, the research explores the impact of dairy farm characteristics (such as herd genetics, feeding methods and management strategies) on methane emissions. 2. **Multicollinearity and model stability**: - Variance inflation factor (VIF) and principal component analysis (PCA) are used to deal with the multicollinearity problem and improve the stability and interpretability of the model. The formula is as follows: \[ \text{VIF}=\frac{1}{1 - R^{2}} \] where \(R^{2}\) is the coefficient of determination of the regression model. 3. **Application of machine - learning models**: - Two machine - learning models, random forest and neural networks, are adopted to evaluate the importance of features and predict methane emissions. The random forest model is used for feature importance ranking, while the neural network model captures complex non - linear relationships. 4. **The impact of genetic selection on methane emissions**: - The research finds that there is a significant negative correlation between the protein percentage in the estimated breeding value (EBV) and the methane concentration. The formula is expressed as: \[ \text{Correlation Coefficient}(\text{EBV for Protein Percentage},\text{Methane Concentration}) < 0 \] This indicates that selecting genetic characteristics with higher milk protein content can effectively reduce methane emissions. 5. **Integration of atmospheric transport models and satellite data**: - Combine the atmospheric transport model with satellite data to estimate methane emissions more accurately and improve spatial resolution and accuracy. 6. **Guidance for sustainable agricultural practices**: - The research emphasizes the importance of improving the assessment of methane emissions in the dairy industry through advanced satellite monitoring, machine - learning techniques and atmospheric modeling, and points out the crucial role of farm - specific characteristics in formulating effective emission reduction strategies. In general, this paper is committed to revealing the complex relationship between dairy farm characteristics and methane emissions through the integration of satellite data, machine - learning techniques and atmospheric models, thereby providing a scientific basis for formulating effective emission reduction strategies.

Mapping Methane -- The Impact of Dairy Farm Practices on Emissions Through Satellite Data and Machine Learning

Satellite Data and Machine Learning for Benchmarking Methane Concentrations in the Canadian Dairy Industry

Estimating Air Methane and Total Hydrocarbon Concentrations in Alberta, Canada Using Machine Learning

Artificial Intelligence driven Benchmarking Tool for Emission Reduction in Canadian Dairy Farms

From Microbes to Methane: AI-Based Predictive Modeling of Feed Additive Efficacy in Dairy Cows

Approaches for Predicting Dairy Cattle Methane Emissions: From Traditional Methods to Machine Learning

Unveiling microbial biomarkers of ruminant methane emission through machine learning

Screening and Functional Prediction of Rumen Microbiota Associated with Methane Emissions in Dairy Cows

Factors Affecting Enteric Emission Methane and Predictive Models for Dairy Cows

A 130‐year Global Inventory of Methane Emissions from Livestock: Trends, Patterns, and Drivers

Evaluating the use of an Unmanned Aerial Vehicle (UAV)-based active AirCore system to quantify methane emissions from dairy cows

Prediction of enteric methane emissions from cattle

Prediction of enteric methane production, yield, and intensity in dairy cattle using an intercontinental database

Spatiotemporal variations and influencing factors of methane emissions from livestock in China: A spatial econometric analysis

Combining the Eddy Covariance Method and Dry Matter Intake Measurements for Enteric Methane Emission Estimation from Grazing Dairy Cows

Modeling N2O emissions with remotely sensed variables using machine learning

Modelling methane emissions from pig manure using statistical and machine learning methods

An integral assessment of the impact of diet and manure management on whole-farm greenhouse gas and nitrogen emissions in dairy cattle production systems using process-based models

Accounting for minimum data required to train a machine learning model to accurately monitor Australian dairy pastures using remote sensing

Verifying Methane Inventories and Trends With Atmospheric Methane Data

Spatiotemporal characteristics and driven forces of nitrogen flow of dairy farms in China: Based on a comprehensive model combined with production practices and embedding localized parameters