Methane (CH₄) is a potent greenhouse gas, responsible for nearly 30% of the global warming effect since the pre-industrial era. Due to its significant role in climate change and its shorter atmospheric lifetime, accurately monitoring methane levels is critical for effective climate action. Precise methane data helps researchers understand sources, sinks, and overall atmospheric behavior, enabling more targeted and effective mitigation strategies.

    In this study, we aim to improve global methane concentration (XCH₄) estimates by applying machine learning techniques. We use data from multiple satellite missions, including GOSAT, GOSAT-2, and Sentinel-5P, which provide important methane measurements but often contain biases due to differences in sensors, retrieval algorithms, and atmospheric conditions. To address these issues, we employ advanced machine learning approaches to integrate satellite data with ground-based observations from networks like TCCON. This allows us to correct biases and achieve more accurate methane concentration estimates on a global scale. The goal of our project is to deliver improved XCH₄ data that can provide a more accurate understanding of global methane distributions, ultimately contributing to better-informed climate policies and strategies for mitigating methane emissions.​