Grid Decarbonization: Examination of Data-Driven Methods for Emission Factor Estimation

Grid decarbonization is a key strategy for mitigating global warming, as the energy sector contributes around 40% of global greenhouse gas emissions. Estimating accurate emission factors is essential for carbon-aware power system operations. However, existing relevant efforts rely on annual or hourly averages, which fail to align with real-time operational needs. Direct high-resolution measurements are often limited by sensor availability and privacy concerns. This study explores emerging data-driven methods: tensor completion, machine learning, and deep learning techniques, to estimate emission factors at a finer temporal granularity. Validated with real-world data, the proposed approach enhances the temporal resolution of emission factors, supporting more responsive and informed carbon-aware decision making.